putty/doc/config.but

\C{config} Configuring PuTTY

This chapter describes all the \i{configuration options} in PuTTY.

PuTTY is configured using the control panel that comes up before you
start a session. Some options can also be changed in the middle of a
session, by selecting \q{Change Settings} from the window menu.

\H{config-session} The Session panel

The Session configuration panel contains the basic options you need
to specify in order to open a session at all, and also allows you to
save your settings to be reloaded later.

\S{config-hostname} The \i{host name} section

The top box on the Session panel, labelled \q{Specify the destination
you want to connect to}, contains the details that need to be filled
in before PuTTY can open a session at all.

\b The \q{Host Name} box is where you type the name, or the \i{IP
address}, of the server you want to connect to.

\b The \q{Connection type} controls let you choose what type of
connection you want to make: an \i{SSH} network connection, a
connection to a local \i{serial line}, or various other kinds of
network connection.

\lcont{

\b See \k{which-one} for a summary of the
differences between the network remote login protocols SSH, Telnet,
Rlogin, and SUPDUP.

\b See \k{using-serial} for information about using a serial line.

\b See \k{using-rawprot} for an explanation of \q{raw}
connections.

\b See \k{using-telnet} for a little information about Telnet.

\b See \k{using-rlogin} for information about using Rlogin.

\b See \k{using-supdup} for information about using SUPDUP.

\b The \q{Bare ssh-connection} option in the \q{Connection type}
control is intended for specialist uses not involving network
connections. See \k{config-psusan} for some information about it.

}

\b The \q{Port} box lets you specify which \i{port number} on the
server to connect to. If you select Telnet, Rlogin, SUPDUP, or SSH,
this box will be filled in automatically to the usual value, and you
will only need to change it if you have an unusual server. If you
select Raw mode, you will almost certainly need to fill in the
\q{Port} box yourself.

If you select \q{Serial} from the \q{Connection type} radio buttons,
the \q{Host Name} and \q{Port} boxes are replaced by \q{Serial line}
and \q{Speed}; see \k{config-serial} for more details of these.

\S{config-saving} \ii{Loading and storing saved sessions}

The next part of the Session configuration panel allows you to save
your preferred PuTTY options so they will appear automatically the
next time you start PuTTY. It also allows you to create \e{saved
sessions}, which contain a full set of configuration options plus a
host name and protocol. A saved session contains all the information
PuTTY needs to start exactly the session you want.

\b To save your default settings: first set up the settings the way
you want them saved. Then come back to the Session panel. Select the
\q{\i{Default Settings}} entry in the saved sessions list, with a single
click. Then press the \q{Save} button.

If there is a specific host you want to store the details of how to
connect to, you should create a saved session, which will be
separate from the Default Settings.

\b To save a session: first go through the rest of the configuration
box setting up all the options you want. Then come back to the
Session panel. Enter a name for the saved session in the \q{Saved
Sessions} input box. (The server name is often a good choice for a
saved session name.) Then press the \q{Save} button. Your saved
session name should now appear in the list box.

\lcont{
You can also save settings in mid-session, from the \q{Change Settings}
dialog. Settings changed since the start of the session will be saved
with their current values; as well as settings changed through the
dialog, this includes changes in window size, window title changes
sent by the server, and so on.
}

\b To reload a saved session: single-click to select the session
name in the list box, and then press the \q{Load} button. Your saved
settings should all appear in the configuration panel.

\b To modify a saved session: first load it as described above. Then
make the changes you want. Come back to the Session panel, and press
the \q{Save} button. The new settings will be saved over the top of
the old ones.

\lcont{
To save the new settings under a different name, you can enter the new
name in the \q{Saved Sessions} box, or single-click to select a
session name in the list box to overwrite that session. To save
\q{Default Settings}, you must single-click the name before saving.
}

\b To start a saved session immediately: double-click on the session
name in the list box.

\b To delete a saved session: single-click to select the session
name in the list box, and then press the \q{Delete} button.

Each saved session is independent of the Default Settings
configuration. If you change your preferences and update Default
Settings, you must also update every saved session separately.

Saved sessions are stored in the \i{Registry}, at the location

\c HKEY_CURRENT_USER\Software\SimonTatham\PuTTY\Sessions

If you need to store them in a file, you could try the method
described in \k{config-file}.

\S{config-closeonexit} \q{\ii{Close window} on exit}

Finally in the Session panel, there is an option labelled \q{Close
window on exit}. This controls whether the PuTTY \i{terminal window}
disappears as soon as the session inside it terminates. If you are
likely to want to copy and paste text out of the session after it
has terminated, or restart the session, you should arrange for this
option to be off.

\q{Close window on exit} has three settings. \q{Always} means always
close the window on exit; \q{Never} means never close on exit
(always leave the window open, but \I{inactive window}inactive). The
third setting, and the default one, is \q{Only on clean exit}. In this
mode, a session which terminates normally will cause its window to
close, but one which is aborted unexpectedly by network trouble or a
confusing message from the server will leave the window up.

\H{config-logging} The Logging panel

The Logging configuration panel allows you to save \i{log file}s of your
PuTTY sessions, for debugging, analysis or future reference.

The main option is a radio-button set that specifies whether PuTTY
will log anything at all. The options are:

\b \q{None}. This is the default option; in this mode PuTTY will not
create a log file at all.

\b \q{Printable output}. In this mode, a log file will be
created and written to, but only printable text will be saved into
it. The various terminal control codes that are typically sent down
an interactive session alongside the printable text will be omitted.
This might be a useful mode if you want to read a log file in a text
editor and hope to be able to make sense of it.

\b \q{All session output}. In this mode, \e{everything} sent by
the server into your terminal session is logged. If you view the log
file in a text editor, therefore, you may well find it full of
strange control characters. This is a particularly useful mode if
you are experiencing problems with PuTTY's terminal handling: you
can record everything that went to the terminal, so that someone
else can replay the session later in slow motion and watch to see
what went wrong.

\b \I{SSH packet log}\q{SSH packets}. In this mode (which is only used
by SSH connections), the SSH message packets sent over the encrypted
connection are written to the log file (as well as \i{Event Log}
entries). You might need this to debug a network-level problem, or
more likely to send to the PuTTY authors as part of a bug report.
\e{BE WARNED} that if you log in using a password, the password can
appear in the log file; see \k{config-logssh} for options that may
help to remove sensitive material from the log file before you send it
to anyone else.

\b \q{SSH packets and raw data}. In this mode, as well as the
decrypted packets (as in the previous mode), the \e{raw} (encrypted,
compressed, etc) packets are \e{also} logged. This could be useful to
diagnose corruption in transit. (The same caveats as the previous mode
apply, of course.)

Note that the non-SSH logging options (\q{Printable output} and
\q{All session output}) only work with PuTTY proper; in programs
without terminal emulation (such as Plink), they will have no effect,
even if enabled via saved settings.

\S{config-logfilename} \q{Log file name}

In this edit box you enter the name of the file you want to log the
session to. The \q{Browse} button will let you look around your file
system to find the right place to put the file; or if you already
know exactly where you want it to go, you can just type a pathname
into the edit box.

There are a few special features in this box. If you use the \c{&}
character in the file name box, PuTTY will insert details of the
current session in the name of the file it actually opens. The
precise replacements it will do are:

\b \c{&Y} will be replaced by the current year, as four digits.

\b \c{&M} will be replaced by the current month, as two digits.

\b \c{&D} will be replaced by the current day of the month, as two
digits.

\b \c{&T} will be replaced by the current time, as six digits
(HHMMSS) with no punctuation.

\b \c{&H} will be replaced by the host name you are connecting to
(or the serial line, for a serial connection).

\b \c{&P} will be replaced by the port number you are connecting to on
the target host.

(These are all case-insensitive.)

For example, if you enter the file name
\c{c:\\puttylogs\\log-&h-&y&m&d-&t.dat}, you will end up with files looking
like

\c log-server1.example.com-20010528-110859.dat
\c log-unixbox.somewhere.org-20010611-221001.dat

\S{config-logfileexists} \q{What to do if the log file already exists}

This control allows you to specify what PuTTY should do if it tries
to start writing to a log file and it finds the file already exists.
You might want to automatically destroy the existing log file and
start a new one with the same name. Alternatively, you might want to
open the existing log file and add data to the \e{end} of it.
Finally (the default option), you might not want to have any
automatic behaviour, but to ask the user every time the problem
comes up.

\S{config-logflush} \I{log file, flushing}\q{Flush log file frequently}

This option allows you to control how frequently logged data is
flushed to disc. By default, PuTTY will flush data as soon as it is
displayed, so that if you view the log file while a session is still
open, it will be up to date; and if the client system crashes, there's
a greater chance that the data will be preserved.

However, this can incur a performance penalty. If PuTTY is running
slowly with logging enabled, you could try unchecking this option. Be
warned that the log file may not always be up to date as a result
(although it will of course be flushed when it is closed, for instance
at the end of a session).

\S{config-logheader} \I{log file, header}\q{Include header}

This option allows you to choose whether to include a header line
with the date and time when the log file is opened. It may be useful to
disable this if the log file is being used as realtime input to other
programs that don't expect the header line.

\S{config-logssh} Options specific to \i{SSH packet log}ging

These options only apply if SSH packet data is being logged.

The following options allow particularly sensitive portions of
unencrypted packets to be automatically left out of the log file.
They are only intended to deter casual nosiness; an attacker could
glean a lot of useful information from even these obfuscated logs
(e.g., length of password).

\S2{config-logssh-omitpw} \q{Omit known password fields}

When checked, decrypted password fields are removed from the log of
transmitted packets. (This includes any user responses to
challenge-response authentication methods such as
\q{keyboard-interactive}.) This does not include X11 authentication
data if using X11 forwarding.

Note that this will only omit data that PuTTY \e{knows} to be a
password. However, if you start another login session within your
PuTTY session, for instance, any password used will appear in the
clear in the packet log. The next option may be of use to protect
against this.

This option is enabled by default.

\S2{config-logssh-omitdata} \q{Omit session data}

When checked, all decrypted \q{session data} is omitted; this is
defined as data in terminal sessions and in forwarded channels (TCP,
X11, and authentication agent). This will usually substantially reduce
the size of the resulting log file.

This option is disabled by default.

\H{config-terminal} The Terminal panel

The Terminal configuration panel allows you to control the behaviour
of PuTTY's \i{terminal emulation}.

\S{config-autowrap} \q{Auto wrap mode initially on}

\ii{Auto wrap mode} controls what happens when text printed in a PuTTY
window reaches the right-hand edge of the window.

With auto wrap mode on, if a long line of text reaches the
right-hand edge, it will wrap over on to the next line so you can
still see all the text. With auto wrap mode off, the cursor will
stay at the right-hand edge of the screen, and all the characters in
the line will be printed on top of each other.

If you are running a full-screen application and you occasionally
find the screen scrolling up when it looks as if it shouldn't, you
could try turning this option off.

Auto wrap mode can be turned on and off by \i{control sequence}s sent by
the server. This configuration option controls the \e{default}
state, which will be restored when you reset the terminal (see
\k{reset-terminal}). However, if you modify this option in
mid-session using \q{Change Settings}, it will take effect
immediately.

\S{config-decom} \q{DEC Origin Mode initially on}

\i{DEC Origin Mode} is a minor option which controls how PuTTY
interprets cursor-position \i{control sequence}s sent by the server.

The server can send a control sequence that restricts the \i{scrolling
region} of the display. For example, in an editor, the server might
reserve a line at the top of the screen and a line at the bottom,
and might send a control sequence that causes scrolling operations
to affect only the remaining lines.

With DEC Origin Mode on, \i{cursor coordinates} are counted from the top
of the scrolling region. With it turned off, cursor coordinates are
counted from the top of the whole screen regardless of the scrolling
region.

It is unlikely you would need to change this option, but if you find
a full-screen application is displaying pieces of text in what looks
like the wrong part of the screen, you could try turning DEC Origin
Mode on to see whether that helps.

DEC Origin Mode can be turned on and off by control sequences sent
by the server. This configuration option controls the \e{default}
state, which will be restored when you reset the terminal (see
\k{reset-terminal}). However, if you modify this option in
mid-session using \q{Change Settings}, it will take effect
immediately.

\S{config-crlf} \q{Implicit CR in every LF}

Most servers send two control characters, \i{CR} and \i{LF}, to start a
\i{new line} of the screen. The CR character makes the cursor return to the
left-hand side of the screen. The LF character makes the cursor move
one line down (and might make the screen scroll).

Some servers only send LF, and expect the terminal to move the
cursor over to the left automatically. If you come across a server
that does this, you will see a \I{stair-stepping}stepped effect on the
screen, like this:

\c First line of text
\c                   Second line
\c                              Third line

If this happens to you, try enabling the \q{Implicit CR in every LF}
option, and things might go back to normal:

\c First line of text
\c Second line
\c Third line

\S{config-lfcr} \q{Implicit LF in every CR}

Most servers send two control characters, \i{CR} and \i{LF}, to start a
\i{new line} of the screen. The CR character makes the cursor return to the
left-hand side of the screen. The LF character makes the cursor move
one line down (and might make the screen scroll).

Some servers only send CR, and so the newly
written line is overwritten by the following line. This option causes
a line feed so that all lines are displayed.

\S{config-erase} \q{Use \i{background colour} to erase screen}

Not all terminals agree on what colour to turn the screen when the
server sends a \q{\i{clear screen}} sequence. Some terminals believe the
screen should always be cleared to the \e{default} background
colour. Others believe the screen should be cleared to whatever the
server has selected as a background colour.

There exist applications that expect both kinds of behaviour.
Therefore, PuTTY can be configured to do either.

With this option disabled, screen clearing is always done in the
default background colour. With this option enabled, it is done in
the \e{current} background colour.

Background-colour erase can be turned on and off by \i{control
sequences} sent by the server. This configuration option controls the
\e{default} state, which will be restored when you reset the
terminal (see \k{reset-terminal}). However, if you modify this
option in mid-session using \q{Change Settings}, it will take effect
immediately.

\S{config-blink} \q{Enable \i{blinking text}}

The server can ask PuTTY to display text that blinks on and off.
This is very distracting, so PuTTY allows you to turn blinking text
off completely.

When blinking text is disabled and the server attempts to make some
text blink, PuTTY will instead display the text with a \I{background
colour, bright}bolded background colour.

Blinking text can be turned on and off by \i{control sequence}s sent by
the server. This configuration option controls the \e{default}
state, which will be restored when you reset the terminal (see
\k{reset-terminal}). However, if you modify this option in
mid-session using \q{Change Settings}, it will take effect
immediately.

\S{config-answerback} \q{\ii{Answerback} to ^E}

This option controls what PuTTY will send back to the server if the
server sends it the ^E \i{enquiry character}. Normally it just sends
the string \q{PuTTY}.

If you accidentally write the contents of a binary file to your
terminal, you will probably find that it contains more than one ^E
character, and as a result your next command line will probably read
\q{PuTTYPuTTYPuTTY...} as if you had typed the answerback string
multiple times at the keyboard. If you set the answerback string to
be empty, this problem should go away, but doing so might cause
other problems.

Note that this is \e{not} the feature of PuTTY which the server will
typically use to determine your terminal type. That feature is the
\q{\ii{Terminal-type} string} in the Connection panel; see
\k{config-termtype} for details.

You can include control characters in the answerback string using
\c{^C} notation. (Use \c{^~} to get a literal \c{^}.)

\S{config-localecho} \q{\ii{Local echo}}

With local echo disabled, characters you type into the PuTTY window
are not echoed in the window \e{by PuTTY}. They are simply sent to
the server. (The \e{server} might choose to \I{remote echo}echo them
back to you; this can't be controlled from the PuTTY control panel.)

Some types of session need local echo, and many do not. In its
default mode, PuTTY will automatically attempt to deduce whether or
not local echo is appropriate for the session you are working in. If
you find it has made the wrong decision, you can use this
configuration option to override its choice: you can force local
echo to be turned on, or force it to be turned off, instead of
relying on the automatic detection.

\S{config-localedit} \q{\ii{Local line editing}}

Normally, every character you type into the PuTTY window is sent
immediately to the server the moment you type it.

If you enable local line editing, this changes. PuTTY will let you
edit a whole line at a time locally, and the line will only be sent
to the server when you press Return. If you make a mistake, you can
use the Backspace key to correct it before you press Return, and the
server will never see the mistake.

Since it is hard to edit a line locally without being able to see
it, local line editing is mostly used in conjunction with \i{local echo}
(\k{config-localecho}). This makes it ideal for use in raw mode
\#{FIXME} or when connecting to \i{MUD}s or \i{talker}s. (Although some more
advanced MUDs do occasionally turn local line editing on and turn
local echo off, in order to accept a password from the user.)

Some types of session need local line editing, and many do not. In
its default mode, PuTTY will automatically attempt to deduce whether
or not local line editing is appropriate for the session you are
working in. If you find it has made the wrong decision, you can use
this configuration option to override its choice: you can force
local line editing to be turned on, or force it to be turned off,
instead of relying on the automatic detection.

\S{config-printing} \ii{Remote-controlled printing}

A lot of VT100-compatible terminals support printing under control
of the remote server (sometimes called \q{passthrough printing}).
PuTTY supports this feature as well, but it is turned off by default.

To enable remote-controlled printing, choose a printer from the
\q{Printer to send ANSI printer output to} drop-down list box. This
should allow you to select from all the printers you have installed
drivers for on your computer. Alternatively, you can type the
network name of a networked printer (for example,
\c{\\\\printserver\\printer1}) even if you haven't already
installed a driver for it on your own machine.

When the remote server attempts to print some data, PuTTY will send
that data to the printer \e{raw} - without translating it,
attempting to format it, or doing anything else to it. It is up to
you to ensure your remote server knows what type of printer it is
talking to.

Since PuTTY sends data to the printer raw, it cannot offer options
such as portrait versus landscape, print quality, or paper tray
selection. All these things would be done by your PC printer driver
(which PuTTY bypasses); if you need them done, you will have to find
a way to configure your remote server to do them.

To disable remote printing again, choose \q{None (printing
disabled)} from the printer selection list. This is the default
state.

\H{config-keyboard} The Keyboard panel

The Keyboard configuration panel allows you to control the behaviour
of the \i{keyboard} in PuTTY.  The correct state for many of these
settings depends on what the server to which PuTTY is connecting
expects.  With a \i{Unix} server, this is likely to depend on the
\i\c{termcap} or \i\c{terminfo} entry it uses, which in turn is likely to
be controlled by the \q{\ii{Terminal-type} string} setting in the Connection
panel; see \k{config-termtype} for details.  If none of the settings here
seems to help, you may find \k{faq-keyboard} to be useful.

\S{config-backspace} Changing the action of the \ii{Backspace key}

Some terminals believe that the Backspace key should send the same
thing to the server as \i{Control-H} (ASCII code 8). Other terminals
believe that the Backspace key should send ASCII code 127 (usually
known as \i{Control-?}) so that it can be distinguished from Control-H.
This option allows you to choose which code PuTTY generates when you
press Backspace.

If you are connecting over SSH, PuTTY by default tells the server
the value of this option (see \k{config-ttymodes}), so you may find
that the Backspace key does the right thing either way. Similarly,
if you are connecting to a \i{Unix} system, you will probably find that
the Unix \i\c{stty} command lets you configure which the server
expects to see, so again you might not need to change which one PuTTY
generates. On other systems, the server's expectation might be fixed
and you might have no choice but to configure PuTTY.

If you do have the choice, we recommend configuring PuTTY to
generate Control-? and configuring the server to expect it, because
that allows applications such as \c{emacs} to use Control-H for
help.

(Typing \i{Shift-Backspace} will cause PuTTY to send whichever code
isn't configured here as the default.)

\S{config-homeend} Changing the action of the \i{Home and End keys}

The Unix terminal emulator \i\c{rxvt} disagrees with the rest of the
world about what character sequences should be sent to the server by
the Home and End keys.

\i\c{xterm}, and other terminals, send \c{ESC [1~} for the Home key,
and \c{ESC [4~} for the End key. \c{rxvt} sends \c{ESC [H} for the
Home key and \c{ESC [Ow} for the End key.

If you find an application on which the Home and End keys aren't
working, you could try switching this option to see if it helps.

\S{config-funkeys} Changing the action of the \i{function keys} and
\i{keypad}

This option affects the function keys (F1 to F12) and the top row of
the numeric keypad.

\b In the default mode, labelled \c{ESC [n~}, the function keys
generate sequences like \c{ESC [11~}, \c{ESC [12~} and so on. This
matches the general behaviour of Digital's terminals.

\b In Linux mode, F6 to F12 behave just like the default mode, but
F1 to F5 generate \c{ESC [[A} through to \c{ESC [[E}. This mimics the
\i{Linux virtual console}.

\b In \I{xterm}Xterm R6 mode, F5 to F12 behave like the default mode, but F1
to F4 generate \c{ESC OP} through to \c{ESC OS}, which are the
sequences produced by the top row of the \e{keypad} on Digital's
terminals.

\b In \i{VT400} mode, all the function keys behave like the default
mode, but the actual top row of the numeric keypad generates \c{ESC
OP} through to \c{ESC OS}.

\b In \i{VT100+} mode, the function keys generate \c{ESC OP} through to
\c{ESC O[}

\b In \i{SCO} mode, the function keys F1 to F12 generate \c{ESC [M}
through to \c{ESC [X}.  Together with shift, they generate \c{ESC [Y}
through to \c{ESC [j}.  With control they generate \c{ESC [k} through
to \c{ESC [v}, and with shift and control together they generate
\c{ESC [w} through to \c{ESC [\{}.

\b In \I{xterm}Xterm 216 mode, the unshifted function keys behave the
same as Xterm R6 mode. But pressing a function key together with Shift
or Alt or Ctrl generates a different sequence containing an extra
numeric parameter of the form (1 for Shift) + (2 for Alt) + (4 for
Ctrl) + 1. For F1-F4, the basic sequences like \c{ESC OP} become
\cw{ESC [1;}\e{bitmap}\cw{P} and similar; for F5 and above,
\cw{ESC[}\e{index}\cw{~} becomes
\cw{ESC[}\e{index}\cw{;}\e{bitmap}\cw{~}.

If you don't know what any of this means, you probably don't need to
fiddle with it.

\S{config-sharrow} Changing the action of the \i{shifted arrow keys}

This option affects the arrow keys, if you press one with any of the
modifier keys Shift, Ctrl or Alt held down.

\b In the default mode, labelled \c{Ctrl toggles app mode}, the Ctrl
key toggles between the default arrow-key sequences like \c{ESC [A} and
\c{ESC [B}, and the sequences Digital's terminals generate in
\q{application cursor keys} mode, i.e. \c{ESC O A} and so on. Shift
and Alt have no effect.

\b In the \q{xterm-style bitmap} mode, Shift, Ctrl and Alt all
generate different sequences, with a number indicating which set of
modifiers is active.

If you don't know what any of this means, you probably don't need to
fiddle with it.

\S{config-appcursor} Controlling \i{Application Cursor Keys} mode

Application Cursor Keys mode is a way for the server to change the
control sequences sent by the arrow keys. In normal mode, the arrow
keys send \c{ESC [A} through to \c{ESC [D}. In application mode,
they send \c{ESC OA} through to \c{ESC OD}.

Application Cursor Keys mode can be turned on and off by the server,
depending on the application. PuTTY allows you to configure the
initial state.

You can also disable application cursor keys mode completely, using
the \q{Features} configuration panel; see
\k{config-features-application}.

\S{config-appkeypad} Controlling \i{Application Keypad} mode

Application Keypad mode is a way for the server to change the
behaviour of the numeric keypad.

In normal mode, the keypad behaves like a normal Windows keypad:
with \i{NumLock} on, the number keys generate numbers, and with NumLock
off they act like the arrow keys and Home, End etc.

In application mode, all the keypad keys send special control
sequences, \e{including} Num Lock. Num Lock stops behaving like Num
Lock and becomes another function key.

Depending on which version of Windows you run, you may find the Num
Lock light still flashes on and off every time you press Num Lock,
even when application mode is active and Num Lock is acting like a
function key. This is unavoidable.

Application keypad mode can be turned on and off by the server,
depending on the application. PuTTY allows you to configure the
initial state.

You can also disable application keypad mode completely, using the
\q{Features} configuration panel; see
\k{config-features-application}.

\S{config-nethack} Using \i{NetHack keypad mode}

PuTTY has a special mode for playing NetHack. You can enable it by
selecting \q{NetHack} in the \q{Initial state of numeric keypad}
control.

In this mode, the numeric keypad keys 1-9 generate the NetHack
movement commands (\cw{hjklyubn}). The 5 key generates the \c{.}
command (do nothing).

In addition, pressing Shift or Ctrl with the keypad keys generate
the Shift- or Ctrl-keys you would expect (e.g. keypad-7 generates
\cq{y}, so Shift-keypad-7 generates \cq{Y} and Ctrl-keypad-7
generates Ctrl-Y); these commands tell NetHack to keep moving you in
the same direction until you encounter something interesting.

For some reason, this feature only works properly when \i{Num Lock} is
on. We don't know why.

\S{config-compose} Enabling a DEC-like \ii{Compose key}

DEC terminals have a Compose key, which provides an easy-to-remember
way of typing \i{accented characters}. You press Compose and then type
two more characters. The two characters are \q{combined} to produce
an accented character. The choices of character are designed to be
easy to remember; for example, composing \q{e} and \q{`} produces
the \q{\u00e8{e-grave}} character.

If your keyboard has a Windows \i{Application key}, it acts as a Compose
key in PuTTY. Alternatively, if you enable the \q{\i{AltGr} acts as
Compose key} option, the AltGr key will become a Compose key.

\S{config-ctrlalt} \q{Control-Alt is different from \i{AltGr}}

Some old keyboards do not have an AltGr key, which can make it
difficult to type some characters. PuTTY can be configured to treat
the key combination Ctrl + Left Alt the same way as the AltGr key.

By default, this checkbox is checked, and the key combination Ctrl +
Left Alt does something completely different. PuTTY's usual handling
of the left Alt key is to prefix the Escape (Control-\cw{[})
character to whatever character sequence the rest of the keypress
would generate. For example, Alt-A generates Escape followed by
\c{a}. So Alt-Ctrl-A would generate Escape, followed by Control-A.

If you uncheck this box, Ctrl-Alt will become a synonym for AltGr,
so you can use it to type extra graphic characters if your keyboard
has any.

(However, Ctrl-Alt will never act as a Compose key, regardless of the
setting of \q{AltGr acts as Compose key} described in
\k{config-compose}.)

\H{config-bell} The Bell panel

The Bell panel controls the \i{terminal bell} feature: the server's
ability to cause PuTTY to beep at you.

In the default configuration, when the server sends the character
with ASCII code 7 (Control-G), PuTTY will play the \i{Windows Default
Beep} sound. This is not always what you want the terminal bell
feature to do; the Bell panel allows you to configure alternative
actions.

\S{config-bellstyle} \q{Set the style of bell}

This control allows you to select various different actions to occur
on a terminal bell:

\b Selecting \q{None} \I{terminal bell, disabling}disables the bell
completely. In this mode, the server can send as many Control-G
characters as it likes and nothing at all will happen.

\b \q{Make default system alert sound} is the default setting. It
causes the Windows \q{Default Beep} sound to be played. To change
what this sound is, or to test it if nothing seems to be happening,
use the Sound configurer in the Windows Control Panel.

\b \q{\ii{Visual bell}} is a silent alternative to a beeping computer. In
this mode, when the server sends a Control-G, the whole PuTTY window
will flash white for a fraction of a second.

\b \q{Beep using the \i{PC speaker}} is self-explanatory.

\b \q{Play a custom \i{sound file}} allows you to specify a particular
sound file to be used by PuTTY alone, or even by a particular
individual PuTTY session. This allows you to distinguish your PuTTY
beeps from any other beeps on the system. If you select this option,
you will also need to enter the name of your sound file in the edit
control \q{Custom sound file to play as a bell}.

\S{config-belltaskbar} \q{\ii{Taskbar}/\I{window caption}caption
indication on bell}

This feature controls what happens to the PuTTY window's entry in
the Windows Taskbar if a bell occurs while the window does not have
the input focus.

In the default state (\q{Disabled}) nothing unusual happens.

If you select \q{Steady}, then when a bell occurs and the window is
not in focus, the window's Taskbar entry and its title bar will
change colour to let you know that PuTTY session is asking for your
attention. The change of colour will persist until you select the
window, so you can leave several PuTTY windows minimised in your
terminal, go away from your keyboard, and be sure not to have missed
any important beeps when you get back.

\q{Flashing} is even more eye-catching: the Taskbar entry will
continuously flash on and off until you select the window.

\S{config-bellovl} \q{Control the \i{bell overload} behaviour}

A common user error in a terminal session is to accidentally run the
Unix command \c{cat} (or equivalent) on an inappropriate file type,
such as an executable, image file, or ZIP file. This produces a huge
stream of non-text characters sent to the terminal, which typically
includes a lot of bell characters. As a result of this the terminal
often doesn't stop beeping for ten minutes, and everybody else in
the office gets annoyed.

To try to avoid this behaviour, or any other cause of excessive
beeping, PuTTY includes a bell overload management feature. In the
default configuration, receiving more than five bell characters in a
two-second period will cause the overload feature to activate. Once
the overload feature is active, further bells will \I{terminal bell,
disabling} have no effect at all, so the rest of your binary file
will be sent to the screen in silence. After a period of five seconds
during which no further bells are received, the overload feature will
turn itself off again and bells will be re-enabled.

If you want this feature completely disabled, you can turn it off
using the checkbox \q{Bell is temporarily disabled when over-used}.

Alternatively, if you like the bell overload feature but don't agree
with the settings, you can configure the details: how many bells
constitute an overload, how short a time period they have to arrive
in to do so, and how much silent time is required before the
overload feature will deactivate itself.

Bell overload mode is always deactivated by any keypress in the
terminal. This means it can respond to large unexpected streams of
data, but does not interfere with ordinary command-line activities
that generate beeps (such as filename completion).

\H{config-features} The Features panel

PuTTY's \i{terminal emulation} is very highly featured, and can do a lot
of things under remote server control. Some of these features can
cause problems due to buggy or strangely configured server
applications.

The Features configuration panel allows you to disable some of
PuTTY's more advanced terminal features, in case they cause trouble.

\S{config-features-application} Disabling application keypad and cursor keys

\I{Application Keypad}Application keypad mode (see
\k{config-appkeypad}) and \I{Application Cursor Keys}application
cursor keys mode (see \k{config-appcursor}) alter the behaviour of
the keypad and cursor keys. Some applications enable these modes but
then do not deal correctly with the modified keys. You can force
these modes to be permanently disabled no matter what the server
tries to do.

\S{config-features-mouse} Disabling \cw{xterm}-style \i{mouse reporting}

PuTTY allows the server to send \i{control codes} that let it take over
the mouse and use it for purposes other than \i{copy and paste}.
Applications which use this feature include the text-mode web
browser \i\c{links}, the Usenet newsreader \i\c{trn} version 4, and the
file manager \i\c{mc} (Midnight Commander).

If you find this feature inconvenient, you can disable it using the
\q{Disable xterm-style mouse reporting} control. With this box
ticked, the mouse will \e{always} do copy and paste in the normal
way.

Note that even if the application takes over the mouse, you can
still manage PuTTY's copy and paste by holding down the Shift key
while you select and paste, unless you have deliberately turned this
feature off (see \k{config-mouseshift}).

\S{config-features-resize} Disabling remote \i{terminal resizing}

PuTTY has the ability to change the terminal's size and position in
response to commands from the server. If you find PuTTY is doing
this unexpectedly or inconveniently, you can tell PuTTY not to
respond to those server commands.

\S{config-features-altscreen} Disabling switching to the \i{alternate screen}

Many terminals, including PuTTY, support an \q{alternate screen}.
This is the same size as the ordinary terminal screen, but separate.
Typically a screen-based program such as a text editor might switch
the terminal to the alternate screen before starting up. Then at the
end of the run, it switches back to the primary screen, and you see
the screen contents just as they were before starting the editor.

Some people prefer this not to happen. If you want your editor to
run in the same screen as the rest of your terminal activity, you
can disable the alternate screen feature completely.

\S{config-features-retitle} Disabling remote \i{window title} changing

PuTTY has the ability to change the window title in response to
commands from the server. If you find PuTTY is doing this
unexpectedly or inconveniently, you can tell PuTTY not to respond to
those server commands.

\S{config-features-qtitle} Response to remote \i{window title} querying

PuTTY can optionally provide the xterm service of allowing server
applications to find out the local window title. This feature is
disabled by default, but you can turn it on if you really want it.

NOTE that this feature is a \e{potential \i{security hazard}}. If a
malicious application can write data to your terminal (for example,
if you merely \c{cat} a file owned by someone else on the server
machine), it can change your window title (unless you have disabled
this as mentioned in \k{config-features-retitle}) and then use this
service to have the new window title sent back to the server as if
typed at the keyboard. This allows an attacker to fake keypresses
and potentially cause your server-side applications to do things you
didn't want. Therefore this feature is disabled by default, and we
recommend you do not set it to \q{Window title} unless you \e{really}
know what you are doing.

There are three settings for this option:

\dt \q{None}

\dd PuTTY makes no response whatsoever to the relevant escape
sequence. This may upset server-side software that is expecting some
sort of response.

\dt \q{Empty string}

\dd PuTTY makes a well-formed response, but leaves it blank. Thus,
server-side software that expects a response is kept happy, but an
attacker cannot influence the response string. This is probably the
setting you want if you have no better ideas.

\dt \q{Window title}

\dd PuTTY responds with the actual window title. This is dangerous for
the reasons described above.

\S{config-features-clearscroll} Disabling remote \i{scrollback clearing}

PuTTY has the ability to clear the terminal's scrollback buffer in
response to a command from the server. If you find PuTTY is doing this
unexpectedly or inconveniently, you can tell PuTTY not to respond to
that server command.

\S{config-features-dbackspace} Disabling \i{destructive backspace}

Normally, when PuTTY receives character 127 (^?) from the server, it
will perform a \q{destructive backspace}: move the cursor one space
left and delete the character under it. This can apparently cause
problems in some applications, so PuTTY provides the ability to
configure character 127 to perform a normal backspace (without
deleting a character) instead.

\S{config-features-charset} Disabling remote \i{character set}
configuration

PuTTY has the ability to change its character set configuration in
response to commands from the server. Some programs send these
commands unexpectedly or inconveniently. In particular, \i{BitchX} (an
IRC client) seems to have a habit of reconfiguring the character set
to something other than the user intended.

If you find that accented characters are not showing up the way you
expect them to, particularly if you're running BitchX, you could try
disabling the remote character set configuration commands.

\S{config-features-shaping} Disabling \i{Arabic text shaping}

PuTTY supports shaping of Arabic text, which means that if your
server sends text written in the basic \i{Unicode} Arabic alphabet then
it will convert it to the correct display forms before printing it
on the screen.

If you are using full-screen software which was not expecting this
to happen (especially if you are not an Arabic speaker and you
unexpectedly find yourself dealing with Arabic text files in
applications which are not Arabic-aware), you might find that the
\i{display becomes corrupted}. By ticking this box, you can disable
Arabic text shaping so that PuTTY displays precisely the characters
it is told to display.

You may also find you need to disable bidirectional text display;
see \k{config-features-bidi}.

\S{config-features-bidi} Disabling \i{bidirectional text} display

PuTTY supports bidirectional text display, which means that if your
server sends text written in a language which is usually displayed
from right to left (such as \i{Arabic} or \i{Hebrew}) then PuTTY will
automatically flip it round so that it is displayed in the right
direction on the screen.

If you are using full-screen software which was not expecting this
to happen (especially if you are not an Arabic speaker and you
unexpectedly find yourself dealing with Arabic text files in
applications which are not Arabic-aware), you might find that the
\i{display becomes corrupted}. By ticking this box, you can disable
bidirectional text display, so that PuTTY displays text from left to
right in all situations.

You may also find you need to disable Arabic text shaping;
see \k{config-features-shaping}.

\S{config-features-bracketed paste} Disabling \i{bracketed paste} mode

By default, when you paste text into the terminal window, it's sent to
the server as terminal input, exactly as if you'd typed the same text
into the terminal window using the keyboard (except that it's all sent
at once, much faster than you could type it).

However, a terminal application can change that, by asking the
terminal to enable \q{bracketed paste mode}. In this mode, pasted data
is marked in the input stream, by sending a special control sequence
before the paste, and another one at the end.

A terminal application can use this information to treat pasted data
differently from keyboard input. For example, a terminal-based text
editor can treat the input as literal data, even if some of its
characters would normally trigger special editor functions. A shell
can treat pasted input as less trusted, in case another application
somehow sneaked a malicious shell command into your clipboard: modern
versions of \cw{bash} will highlight pasted data on the command line,
and not run it until you've confirmed it by pressing Return, even if
the pasted data contained a newline character.

In edge cases, it's possible that bracketed paste mode introduces
bigger problems than the ones it solves. So you can use this checkbox
to turn it off completely. If you do that, then PuTTY will always send
your paste data exactly as if it had been typed at the keyboard,
whether or not the server asked for bracketed paste mode.

\H{config-window} The Window panel

The Window configuration panel allows you to control aspects of the
\i{PuTTY window}.

\S{config-winsize} Setting the \I{window size}size of the PuTTY window

The \q{\ii{Columns}} and \q{\ii{Rows}} boxes let you set the PuTTY
window to a precise size. Of course you can also \I{window resizing}drag
the window to a new size while a session is running.

\S{config-winsizelock} What to do when the window is resized

These options allow you to control what happens when the user tries
to \I{window resizing}resize the PuTTY window using its window furniture.

There are four options here:

\b \q{Change the number of rows and columns}: the font size will not
change. (This is the default.)

\b \q{Change the size of the font}: the number of rows and columns in
the terminal will stay the same, and the \i{font size} will change.

\b \q{Change font size when maximised}: when the window is resized,
the number of rows and columns will change, \e{except} when the window
is \i{maximise}d (or restored), when the font size will change. (In
this mode, holding down the Alt key while resizing will also cause the
font size to change.)

\b \q{Forbid resizing completely}: the terminal will refuse to be
resized at all.

\S{config-scrollback} Controlling \i{scrollback}

These options let you configure the way PuTTY keeps text after it
scrolls off the top of the screen (see \k{using-scrollback}).

The \q{Lines of scrollback} box lets you configure how many lines of
text PuTTY keeps. The \q{Display scrollbar} options allow you to
hide the \i{scrollbar} (although you can still view the scrollback using
the keyboard as described in \k{using-scrollback}). You can separately
configure whether the scrollbar is shown in \i{full-screen} mode and in
normal modes.

If you are viewing part of the scrollback when the server sends more
text to PuTTY, the screen will revert to showing the current
terminal contents. You can disable this behaviour by turning off
\q{Reset scrollback on display activity}. You can also make the
screen revert when you press a key, by turning on \q{Reset
scrollback on keypress}.

\S{config-erasetoscrollback} \q{Push erased text into scrollback}

When this option is enabled, the contents of the terminal screen
will be pushed into the scrollback when a server-side application
clears the screen, so that your scrollback will contain a better
record of what was on your screen in the past.

If the application switches to the \i{alternate screen} (see
\k{config-features-altscreen} for more about this), then the
contents of the primary screen will be visible in the scrollback
until the application switches back again.

This option is enabled by default.

\H{config-appearance} The Appearance panel

The Appearance configuration panel allows you to control aspects of
the appearance of \I{PuTTY window}PuTTY's window.

\S{config-cursor} Controlling the appearance of the \i{cursor}

The \q{Cursor appearance} option lets you configure the cursor to be
a block, an underline, or a vertical line. A block cursor becomes an
empty box when the window loses focus; an underline or a vertical
line becomes dotted.

The \q{\ii{Cursor blinks}} option makes the cursor blink on and off. This
works in any of the cursor modes.

\S{config-font} Controlling the \i{font} used in the terminal window

This option allows you to choose what font, in what \I{font size}size,
the PuTTY terminal window uses to display the text in the session.

By default, you will be offered a choice from all the fixed-width
fonts installed on the system, since VT100-style terminal handling
expects a fixed-width font. If you tick the box marked \q{Allow
selection of variable-pitch fonts}, however, PuTTY will offer
variable-width fonts as well: if you select one of these, the font
will be coerced into fixed-size character cells, which will probably
not look very good (but can work OK with some fonts).

\S{config-mouseptr} \q{Hide \i{mouse pointer} when typing in window}

If you enable this option, the mouse pointer will disappear if the
PuTTY window is selected and you press a key. This way, it will not
obscure any of the text in the window while you work in your
session. As soon as you move the mouse, the pointer will reappear.

This option is disabled by default, so the mouse pointer remains
visible at all times.

\S{config-winborder} Controlling the \i{window border}

PuTTY allows you to configure the appearance of the window border to
some extent.

The checkbox marked \q{Sunken-edge border} changes the appearance of
the window border to something more like a DOS box: the inside edge
of the border is highlighted as if it sank down to meet the surface
inside the window. This makes the border a little bit thicker as
well. It's hard to describe well. Try it and see if you like it.

You can also configure a completely blank gap between the text in
the window and the border, using the \q{Gap between text and window
edge} control. By default this is set at one pixel. You can reduce
it to zero, or increase it further.

\H{config-behaviour} The Behaviour panel

The Behaviour configuration panel allows you to control aspects of
the behaviour of \I{PuTTY window}PuTTY's window.

\S{config-title} Controlling the \i{window title}

The \q{Window title} edit box allows you to set the title of the
PuTTY window. By default the window title will contain the \i{host name}
followed by \q{PuTTY}, for example \c{server1.example.com - PuTTY}.
If you want a different window title, this is where to set it.

PuTTY allows the server to send \c{xterm} \i{control sequence}s which
modify the title of the window in mid-session (unless this is disabled -
see \k{config-features-retitle}); the title string set here
is therefore only the \e{initial} window title.

As well as the \e{window} title, there is also an \c{xterm}
sequence to modify the \I{icon title}title of the window's \e{icon}.
This makes sense in a windowing system where the window becomes an
icon when minimised, such as Windows 3.1 or most X Window System
setups; but in the Windows 95-like user interface it isn't as
applicable.

By default, PuTTY only uses the server-supplied \e{window} title, and
ignores the icon title entirely. If for some reason you want to see
both titles, check the box marked \q{Separate window and icon titles}.
If you do this, PuTTY's window title and Taskbar \I{window caption}caption will
change into the server-supplied icon title if you \i{minimise} the PuTTY
window, and change back to the server-supplied window title if you
restore it. (If the server has not bothered to supply a window or
icon title, none of this will happen.)

\S{config-warnonclose} \q{Warn before \i{closing window}}

If you press the \i{Close button} in a PuTTY window that contains a
running session, PuTTY will put up a warning window asking if you
really meant to close the window. A window whose session has already
terminated can always be closed without a warning.

If you want to be able to close a window quickly, you can disable
the \q{Warn before closing window} option.

\S{config-altf4} \q{Window closes on \i{ALT-F4}}

By default, pressing ALT-F4 causes the \I{closing window}window to
close (or a warning box to appear; see \k{config-warnonclose}). If you
disable the \q{Window closes on ALT-F4} option, then pressing ALT-F4
will simply send a key sequence to the server.

\S{config-altspace} \q{\ii{System menu} appears on \i{ALT-Space}}

If this option is enabled, then pressing ALT-Space will bring up the
PuTTY window's menu, like clicking on the top left corner. If it is
disabled, then pressing ALT-Space will just send \c{ESC SPACE} to
the server.

Some \i{accessibility} programs for Windows may need this option
enabling to be able to control PuTTY's window successfully. For
instance, \i{Dragon NaturallySpeaking} requires it both to open the
system menu via voice, and to close, minimise, maximise and restore
the window.

\S{config-altonly} \q{\ii{System menu} appears on \i{Alt} alone}

If this option is enabled, then pressing and releasing ALT will
bring up the PuTTY window's menu, like clicking on the top left
corner. If it is disabled, then pressing and releasing ALT will have
no effect.

\S{config-alwaysontop} \q{Ensure window is \i{always on top}}

If this option is enabled, the PuTTY window will stay on top of all
other windows.

\S{config-fullscreen} \q{\ii{Full screen} on Alt-Enter}

If this option is enabled, then pressing Alt-Enter will cause the
PuTTY window to become full-screen. Pressing Alt-Enter again will
restore the previous window size.

The full-screen feature is also available from the \ii{System menu}, even
when it is configured not to be available on the Alt-Enter key. See
\k{using-fullscreen}.

\H{config-translation} The Translation panel

The Translation configuration panel allows you to control the
translation between the \i{character set} understood by the server and
the character set understood by PuTTY.

\S{config-charset} Controlling character set translation

During an interactive session, PuTTY receives a stream of 8-bit
bytes from the server, and in order to display them on the screen it
needs to know what character set to interpret them in. Similarly,
PuTTY needs to know how to translate your keystrokes into the encoding
the server expects. Unfortunately, there is no satisfactory
mechanism for PuTTY and the server to communicate this information,
so it must usually be manually configured.

There are a lot of character sets to choose from. The \q{Remote
character set} option lets you select one.

By default PuTTY will use the \i{UTF-8} encoding of \i{Unicode}, which
can represent pretty much any character; data coming from the server
is interpreted as UTF-8, and keystrokes are sent UTF-8 encoded. This
is what most modern distributions of Linux will expect by default.
However, if this is wrong for your server, you can select a different
character set using this control.

A few other notable character sets are:

\b The \i{ISO-8859} series are all standard character sets that include
various accented characters appropriate for different sets of
languages.

\b The \i{Win125x} series are defined by Microsoft, for similar
purposes. In particular Win1252 is almost equivalent to ISO-8859-1,
but contains a few extra characters such as matched quotes and the
Euro symbol.

\b If you want the old IBM PC character set with block graphics and
line-drawing characters, you can select \q{\i{CP437}}.

If you need support for a numeric \i{code page} which is not listed in
the drop-down list, such as code page 866, then you can try entering
its name manually (\c{\i{CP866}} for example) in the list box. If the
underlying version of Windows has the appropriate translation table
installed, PuTTY will use it.

\S{config-cjk-ambig-wide} \q{Treat \i{CJK} ambiguous characters as wide}

There are \I{East Asian Ambiguous characters}some Unicode characters
whose \I{character width}width is not well-defined. In most contexts, such
characters should be treated as single-width for the purposes of \I{wrapping,
terminal}wrapping and so on; however, in some CJK contexts, they are better
treated as double-width for historical reasons, and some server-side
applications may expect them to be displayed as such. Setting this option
will cause PuTTY to take the double-width interpretation.

If you use legacy CJK applications, and you find your lines are
wrapping in the wrong places, or you are having other display
problems, you might want to play with this setting.

This option only has any effect in \i{UTF-8} mode (see \k{config-charset}).

\S{config-cyr} \q{\i{Caps Lock} acts as \i{Cyrillic} switch}

This feature allows you to switch between a US/UK keyboard layout
and a Cyrillic keyboard layout by using the Caps Lock key, if you
need to type (for example) \i{Russian} and English side by side in the
same document.

Currently this feature is not expected to work properly if your
native keyboard layout is not US or UK.

\S{config-linedraw} Controlling display of \i{line-drawing characters}

VT100-series terminals allow the server to send \i{control sequence}s that
shift temporarily into a separate character set for drawing simple
lines and boxes. However, there are a variety of ways in which PuTTY
can attempt to find appropriate characters, and the right one to use
depends on the locally configured \i{font}. In general you should probably
try lots of options until you find one that your particular font
supports.

\b \q{Use Unicode line drawing code points} tries to use the box
characters that are present in \i{Unicode}. For good Unicode-supporting
fonts this is probably the most reliable and functional option.

\b \q{Poor man's line drawing} assumes that the font \e{cannot}
generate the line and box characters at all, so it will use the
\c{+}, \c{-} and \c{|} characters to draw approximations to boxes.
You should use this option if none of the other options works.

\b \q{Font has XWindows encoding} is for use with fonts that have a
special encoding, where the lowest 32 character positions (below the
ASCII printable range) contain the line-drawing characters. This is
unlikely to be the case with any standard Windows font; it will
probably only apply to custom-built fonts or fonts that have been
automatically converted from the X Window System.

\b \q{Use font in both ANSI and OEM modes} tries to use the same
font in two different character sets, to obtain a wider range of
characters. This doesn't always work; some fonts claim to be a
different size depending on which character set you try to use.

\b \q{Use font in OEM mode only} is more reliable than that, but can
miss out other characters from the main character set.

\S{config-linedrawpaste} Controlling \i{copy and paste} of line drawing
characters

By default, when you copy and paste a piece of the PuTTY screen that
contains VT100 line and box drawing characters, PuTTY will paste
them in the form they appear on the screen: either \i{Unicode} line
drawing code points, or the \q{poor man's} line-drawing characters
\c{+}, \c{-} and \c{|}. The checkbox \q{Copy and paste VT100 line
drawing chars as lqqqk} disables this feature, so line-drawing
characters will be pasted as the \i{ASCII} characters that were printed
to produce them. This will typically mean they come out mostly as
\c{q} and \c{x}, with a scattering of \c{jklmntuvw} at the corners.
This might be useful if you were trying to recreate the same box
layout in another program, for example.

Note that this option only applies to line-drawing characters which
\e{were} printed by using the VT100 mechanism. Line-drawing
characters that were received as Unicode code points will paste as
Unicode always.

\S{config-utf8linedraw} Combining VT100 line-drawing with UTF-8

If PuTTY is configured to treat data from the server as encoded in
UTF-8, then by default it disables the older VT100-style system of
control sequences that cause the lower-case letters to be temporarily
replaced by line drawing characters.

The rationale is that in UTF-8 mode you don't need those control
sequences anyway, because all the line-drawing characters they access
are available as Unicode characters already, so there's no need for
applications to put the terminal into a special state to get at them.

Also, it removes a risk of the terminal \e{accidentally} getting into
that state: if you accidentally write uncontrolled binary data to a
non-UTF-8 terminal, it can be surprisingly common to find that your
next shell prompt appears as a sequence of line-drawing characters and
then you have to remember or look up how to get out of that mode. So
by default, UTF-8 mode simply doesn't \e{have} a confusing mode like
that to get into, accidentally or on purpose.

However, not all applications will see it that way. Even UTF-8
terminal users will still sometimes have to run software that tries to
print line-drawing characters in the old-fashioned way. So the
configuration option \q{Enable VT100 line drawing even in UTF-8 mode}
puts PuTTY into a hybrid mode in which it understands the VT100-style
control sequences that change the meaning of the ASCII lower case
letters, \e{and} understands UTF-8.

\H{config-selection} The Selection panel

The Selection panel allows you to control the way \i{copy and paste}
work in the PuTTY window.

\S{config-mouse} Changing the actions of the mouse buttons

PuTTY's copy and paste mechanism is by default modelled on the Unix
\i\c{xterm} application. The X Window System uses a three-button mouse,
and the convention in that system is that the \i{left button}
\I{selecting text}selects, the \i{right button} extends an existing
selection, and the \i{middle button} pastes.

Windows often only has two mouse buttons, so when run on Windows,
PuTTY is configurable. In PuTTY's default configuration
(\q{Compromise}), the \e{right} button pastes, and the \e{middle}
button (if you have one) \I{adjusting a selection}extends a
selection.

If you have a \i{three-button mouse} and you are already used to the
\c{xterm} arrangement, you can select it using the \q{Action of
mouse buttons} control.

Alternatively, with the \q{Windows} option selected, the middle
button extends, and the right button brings up a \i{context menu} (on
which one of the options is \q{Paste}). (This context menu is always
available by holding down Ctrl and right-clicking, regardless of the
setting of this option.)

(When PuTTY itself is running on Unix, it follows the X Window System
convention.)

\S{config-mouseshift} \q{Shift overrides application's use of mouse}

PuTTY allows the server to send \i{control codes} that let it
\I{mouse reporting}take over the mouse and use it for purposes other
than \i{copy and paste}.
Applications which use this feature include the text-mode web
browser \c{links}, the Usenet newsreader \c{trn} version 4, and the
file manager \c{mc} (Midnight Commander).

When running one of these applications, pressing the mouse buttons
no longer performs copy and paste. If you do need to copy and paste,
you can still do so if you hold down Shift while you do your mouse
clicks.

However, it is possible in theory for applications to even detect
and make use of Shift + mouse clicks. We don't know of any
applications that do this, but in case someone ever writes one,
unchecking the \q{Shift overrides application's use of mouse}
checkbox will cause Shift + mouse clicks to go to the server as well
(so that mouse-driven copy and paste will be completely disabled).

If you want to prevent the application from taking over the mouse at
all, you can do this using the Features control panel; see
\k{config-features-mouse}.

\S{config-rectselect} Default selection mode

As described in \k{using-selection}, PuTTY has two modes of
selecting text to be copied to the clipboard. In the default mode
(\q{Normal}), dragging the mouse from point A to point B selects to
the end of the line containing A, all the lines in between, and from
the very beginning of the line containing B. In the other mode
(\q{Rectangular block}), dragging the mouse between two points
defines a rectangle, and everything within that rectangle is copied.

Normally, you have to hold down Alt while dragging the mouse to
select a rectangular block. Using the \q{Default selection mode}
control, you can set \i{rectangular selection} as the default, and then
you have to hold down Alt to get the \e{normal} behaviour.

\S{config-clipboards} Assigning copy and paste actions to clipboards

Here you can configure which clipboard(s) are written or read by
PuTTY's various copy and paste actions.

Most platforms, including Windows, have a single system clipboard.
On these platforms, PuTTY provides a second clipboard-like facility by
permitting you to paste the text you last selected in \e{this window},
whether or not it is currently also in the system clipboard. This is
not enabled by default.

The X Window System (which underlies most Unix graphical interfaces)
provides multiple clipboards (or \q{\i{selections}}), and many
applications support more than one of them by a different user
interface mechanism. When PuTTY itself is running on Unix, it has
more configurability relating to these selections.

The two most commonly used selections are called \cq{\i{PRIMARY}} and
\cq{\I{CLIPBOARD selection}CLIPBOARD}; in applications supporting both,
the usual behaviour is that \cw{PRIMARY} is used by mouse-only actions
(selecting text automatically copies it to \cw{PRIMARY}, and
\i{middle-clicking} pastes from \cw{PRIMARY}), whereas \cw{CLIPBOARD}
is used by explicit Copy and Paste menu items or keypresses such as
\i{Ctrl-C} and \i{Ctrl-V}.

\S2{config-selection-autocopy} \q{Auto-copy selected text}

The checkbox \q{Auto-copy selected text to system clipboard} controls
whether or not selecting text in the PuTTY terminal window
automatically has the side effect of copying it to the system
clipboard, without requiring a separate user interface action.

On X, the wording of this option is changed slightly so that
\cq{CLIPBOARD} is mentioned in place of the \q{system clipboard}. Text
selected in the terminal window will \e{always} be automatically
placed in the \cw{PRIMARY} selection, as is conventional, but if you
tick this box, it will \e{also} be placed in \cq{CLIPBOARD} at the
same time.

\S2{config-selection-clipactions} Choosing a clipboard for UI actions

PuTTY has three user-interface actions which can be configured to
paste into the terminal (not counting menu items). You can click
whichever mouse button (if any) is configured to paste (see
\k{config-mouse}); you can press \i{Shift-Ins}; or you can press
\i{Ctrl-Shift-V}, although that action is not enabled by default.

You can configure which of the available clipboards each of these
actions pastes from (including turning the paste action off
completely). On platforms with a single system clipboard (such as
Windows), the available options are to paste from that clipboard or
to paste from PuTTY's internal memory of the \i{last selected text}
within that window. On X, the standard options are \cw{CLIPBOARD} or
\cw{PRIMARY}.

(\cw{PRIMARY} is conceptually similar in that it \e{also} refers to
the last selected text \dash just across all applications instead of
just this window.)

The two keyboard options each come with a corresponding key to copy
\e{to} the same clipboard. Whatever you configure Shift-Ins to paste
from, \i{Ctrl-Ins} will copy to the same location; similarly,
\i{Ctrl-Shift-C} will copy to whatever Ctrl-Shift-V pastes from.

On X, you can also enter a selection name of your choice. For example,
there is a rarely-used standard selection called \cq{\i{SECONDARY}}, which
Emacs (for example) can work with if you hold down the Meta key while
dragging to select or clicking to paste; if you configure a PuTTY
keyboard action to access this clipboard, then you can interoperate
with other applications' use of it. Another thing you could do would
be to invent a clipboard name yourself, to create a special clipboard
shared \e{only} between instances of PuTTY, or between just instances
configured in that particular way.

\S{config-paste-ctrl-char} \q{Permit control characters in pasted text}

It is possible for the clipboard to contain not just text (with
newlines and tabs) but also control characters such as ESC which could
have surprising effects if pasted into a terminal session, depending
on what program is running on the server side. Copying text from a
mischievous web page could put such characters onto the clipboard.

By default, PuTTY filters out the more unusual control characters,
only letting through the more obvious text-formatting characters
(newlines, tab, backspace, and DEL).

Setting this option stops this filtering; on paste, any character on
the clipboard is sent to the session uncensored. This might be useful
if you are deliberately using control character pasting as a simple
form of scripting, for instance.

\H{config-selection-copy} The Copy panel

The Copy configuration panel controls behaviour specifically related to
copying from the terminal window to the clipboard.

\S{config-charclasses} Character classes

PuTTY will \I{word-by-word selection}select a word at a time in the
terminal window if you \i{double-click} to begin the drag. This section
allows you to control precisely what is considered to be a word.

Each character is given a \e{class}, which is a small number
(typically 0, 1 or 2). PuTTY considers a single word to be any
number of adjacent characters in the same class. So by modifying the
assignment of characters to classes, you can modify the word-by-word
selection behaviour.

In the default configuration, the \i{character classes} are:

\b Class 0 contains \i{white space} and control characters.

\b Class 1 contains most \i{punctuation}.

\b Class 2 contains letters, numbers and a few pieces of punctuation
(the double quote, minus sign, period, forward slash and
underscore).

So, for example, if you assign the \c{@} symbol into character class
2, you will be able to select an e-mail address with just a double
click.

In order to adjust these assignments, you start by selecting a group
of characters in the list box. Then enter a class number in the edit
box below, and press the \q{Set} button.

This mechanism currently only covers ASCII characters, because it
isn't feasible to expand the list to cover the whole of Unicode.

Character class definitions can be modified by \i{control sequence}s
sent by the server. This configuration option controls the
\e{default} state, which will be restored when you reset the
terminal (see \k{reset-terminal}). However, if you modify this
option in mid-session using \q{Change Settings}, it will take effect
immediately.

\S{config-rtfcopy} Copying in \i{Rich Text Format}

If you enable \q{Copy to clipboard in RTF as well as plain text},
PuTTY will write formatting information to the clipboard as well as
the actual text you copy. The effect of this is
that if you paste into (say) a word processor, the text will appear
in the word processor in the same \i{font}, \i{colour}, and style
(e.g. bold, underline) PuTTY was using to display it.

This option can easily be inconvenient, so by default it is
disabled.

\H{config-colours} The Colours panel

The Colours panel allows you to control PuTTY's use of \i{colour}.

\S{config-ansicolour} \q{Allow terminal to specify \i{ANSI colours}}

This option is enabled by default. If it is disabled, PuTTY will
ignore any \i{control sequence}s sent by the server to request coloured
text.

If you have a particularly garish application, you might want to
turn this option off and make PuTTY only use the default foreground
and background colours.

\S{config-xtermcolour} \q{Allow terminal to use xterm \i{256-colour mode}}

This option is enabled by default. If it is disabled, PuTTY will
ignore any control sequences sent by the server which use the
extended 256-colour mode supported by recent versions of \cw{xterm}.

If you have an application which is supposed to use 256-colour mode
and it isn't working, you may find you need to tell your server that
your terminal supports 256 colours. On Unix, you do this by ensuring
that the setting of \i\cw{TERM} describes a 256-colour-capable
terminal. You can check this using a command such as \c{infocmp}:

\c $ infocmp | grep colors
\c         colors#256, cols#80, it#8, lines#24, pairs#256,
\e         bbbbbbbbbb

If you do not see \cq{colors#256} in the output, you may need to
change your terminal setting. On modern Linux machines, you could
try \cq{xterm-256color}.

\S{config-truecolour} \q{Allow terminal to use 24-bit colour}

This option is enabled by default. If it is disabled, PuTTY will
ignore any control sequences sent by the server which use the control
sequences supported by modern terminals to specify arbitrary 24-bit
RGB colour value.

\S{config-boldcolour} \q{Indicate bolded text by changing...}

When the server sends a \i{control sequence} indicating that some text
should be displayed in \i{bold}, PuTTY can handle this in several
ways. It can either change the \i{font} for a bold version, or use the
same font in a brighter colour, or it can do both (brighten the colour
\e{and} embolden the font). This control lets you choose which.

By default bold is indicated by colour, so non-bold text is displayed
in light grey and bold text is displayed in bright white (and
similarly in other colours). If you change the setting to \q{The font}
box, bold and non-bold text will be displayed in the same colour, and
instead the font will change to indicate the difference. If you select
\q{Both}, the font and the colour will both change.

Some applications rely on \q{\i{bold black}} being distinguishable
from a black background; if you choose \q{The font}, their text may
become invisible.

\S{config-logpalette} \q{Attempt to use \i{logical palettes}}

Logical palettes are a mechanism by which a Windows application
running on an \i{8-bit colour} display can select precisely the colours
it wants instead of going with the Windows standard defaults.

If you are not getting the colours you ask for on an 8-bit display,
you can try enabling this option. However, be warned that it's never
worked very well.

\S{config-syscolour} \q{Use \i{system colours}}

Enabling this option will cause PuTTY to ignore the configured colours
for \I{default background}\I{default foreground}\q{Default
Background/Foreground} and \I{cursor colour}\q{Cursor Colour/Text} (see
\k{config-colourcfg}), instead going with the system-wide defaults.

Note that non-bold and \i{bold text} will be the same colour if this
option is enabled. You might want to change to indicating bold text
by font changes (see \k{config-boldcolour}).

\S{config-colourcfg} Adjusting the colours in the \i{terminal window}

The main colour control allows you to specify exactly what colours
things should be displayed in. To modify one of the PuTTY colours,
use the list box to select which colour you want to modify. The \i{RGB
values} for that colour will appear on the right-hand side of the
list box. Now, if you press the \q{Modify} button, you will be
presented with a colour selector, in which you can choose a new
colour to go in place of the old one. (You may also edit the RGB
values directly in the edit boxes, if you wish; each value is an
integer from 0 to 255.)

PuTTY allows you to set the \i{cursor colour}, the \i{default foreground}
and \I{default background}background, and the precise shades of all the
\I{ANSI colours}ANSI configurable colours (black, red, green, yellow, blue,
magenta, cyan, and white). You can also modify the precise shades used for
the \i{bold} versions of these colours; these are used to display bold text
if you have chosen to indicate that by colour (see \k{config-boldcolour}),
and can also be used if the server asks specifically to use them. (Note
that \q{Default Bold Background} is \e{not} the background colour used for
bold text; it is only used if the server specifically asks for a bold
background.)

\H{config-connection} The Connection panel

The Connection panel allows you to configure options that apply to
more than one type of \i{connection}.

\S{config-keepalive} Using \i{keepalives} to prevent disconnection

If you find your sessions are closing unexpectedly (most often with
\q{Connection reset by peer}) after they have been idle for a while,
you might want to try using this option.

Some network \i{routers} and \i{firewalls} need to keep track of all
connections through them. Usually, these firewalls will assume a
connection is dead if no data is transferred in either direction
after a certain time interval. This can cause PuTTY sessions to be
unexpectedly closed by the firewall if no traffic is seen in the
session for some time.

The keepalive option (\q{Seconds between keepalives}) allows you to
configure PuTTY to send data through the session at regular
intervals, in a way that does not disrupt the actual terminal
session. If you find your firewall is cutting \i{idle connections} off,
you can try entering a non-zero value in this field. The value is
measured in seconds; so, for example, if your firewall cuts
connections off after ten minutes then you might want to enter 300
seconds (5 minutes) in the box.

Note that keepalives are not always helpful. They help if you have a
firewall which drops your connection after an idle period; but if
the network between you and the server suffers from \i{breaks in
connectivity} then keepalives can actually make things worse. If a
session is idle, and connectivity is temporarily lost between the
endpoints, but the connectivity is restored before either side tries
to send anything, then there will be no problem - neither endpoint
will notice that anything was wrong. However, if one side does send
something during the break, it will repeatedly try to re-send, and
eventually give up and abandon the connection. Then when
connectivity is restored, the other side will find that the first
side doesn't believe there is an open connection any more.
Keepalives can make this sort of problem worse, because they
increase the probability that PuTTY will attempt to send data during
a break in connectivity. (Other types of periodic network activity
can cause this behaviour; in particular, SSH-2 re-keys can have
this effect. See \k{config-ssh-kex-rekey}.)

Therefore, you might find that keepalives help
connection loss, or you might find they make it worse, depending on
what \e{kind} of network problems you have between you and the
server.

Keepalives are only supported in Telnet and SSH; the Rlogin, SUPDUP, and
Raw protocols offer no way of implementing them. (For an alternative, see
\k{config-tcp-keepalives}.)

Note that if you are using SSH-1 and the server has a bug that makes
it unable to deal with SSH-1 ignore messages (see
\k{config-ssh-bug-ignore1}), enabling keepalives will have no effect.

\S{config-nodelay} \q{Disable \i{Nagle's algorithm}}

Nagle's algorithm is a detail of TCP/IP implementations that tries
to minimise the number of small data packets sent down a network
connection. With Nagle's algorithm enabled, PuTTY's \i{bandwidth} usage
will be slightly more efficient; with it disabled, you may find you
get a faster response to your keystrokes when connecting to some
types of server.

The Nagle algorithm is disabled by default for \i{interactive connections}.

\S{config-tcp-keepalives} \q{Enable \i{TCP keepalives}}

\e{NOTE:} TCP keepalives should not be confused with the
application-level keepalives described in \k{config-keepalive}. If in
doubt, you probably want application-level keepalives; TCP keepalives
are provided for completeness.

The idea of TCP keepalives is similar to application-level keepalives,
and the same caveats apply. The main differences are:

\b TCP keepalives are available on \e{all} network connection types,
including Raw, Rlogin, and SUPDUP.

\b The interval between TCP keepalives is usually much longer,
typically two hours; this is set by the operating system, and cannot
be configured within PuTTY.

\b If the operating system does not receive a response to a keepalive,
it may send out more in quick succession and terminate the connection
if no response is received.

TCP keepalives may be more useful for ensuring that \i{half-open connections}
are terminated than for keeping a connection alive.

TCP keepalives are disabled by default.

\S{config-address-family} \q{\i{Internet protocol version}}

This option allows the user to select between the old and new
Internet protocols and addressing schemes (\i{IPv4} and \i{IPv6}).
The selected protocol will be used for most outgoing network
connections (including connections to \I{proxy}proxies); however,
tunnels have their own configuration, for which see
\k{config-ssh-portfwd-address-family}.

The default setting is \q{Auto}, which means PuTTY will do something
sensible and try to guess which protocol you wanted. (If you specify
a literal \i{Internet address}, it will use whichever protocol that
address implies. If you provide a \i{hostname}, it will see what kinds
of address exist for that hostname; it will use IPv6 if there is an
IPv6 address available, and fall back to IPv4 if not.)

If you need to force PuTTY to use a particular protocol, you can
explicitly set this to \q{IPv4} or \q{IPv6}.

\S{config-loghost} \I{logical host name}\q{Logical name of remote host}

This allows you to tell PuTTY that the host it will really end up
connecting to is different from where it thinks it is making a
network connection.

You might use this, for instance, if you had set up an SSH port
forwarding in one PuTTY session so that connections to some
arbitrary port (say, \cw{localhost} port 10022) were forwarded to a
second machine's SSH port (say, \cw{foovax} port 22), and then
started a second PuTTY connecting to the forwarded port.

In normal usage, the second PuTTY will access the \i{host key cache}
under the host name and port it actually connected to (i.e.
\cw{localhost} port 10022 in this example). Using the logical host
name option, however, you can configure the second PuTTY to cache
the host key under the name of the host \e{you} know that it's
\e{really} going to end up talking to (here \c{foovax}).

This can be useful if you expect to connect to the same actual
server through many different channels (perhaps because your port
forwarding arrangements keep changing): by consistently setting the
logical host name, you can arrange that PuTTY will not keep asking
you to reconfirm its host key. Conversely, if you expect to use the
same local port number for port forwardings to lots of different
servers, you probably didn't want any particular server's host key
cached under that local port number. (For this latter case, you
could instead explicitly configure host keys in the relevant sessions;
see \k{config-ssh-kex-manual-hostkeys}.)

If you just enter a host name for this option, PuTTY will cache the
SSH host key under the default SSH port for that host, irrespective
of the port you really connected to (since the typical scenario is
like the above example: you connect to a silly real port number and
your connection ends up forwarded to the normal port-22 SSH server
of some other machine). To override this, you can append a port
number to the logical host name, separated by a colon. E.g. entering
\cq{foovax:2200} as the logical host name will cause the host key to
be cached as if you had connected to port 2200 of \c{foovax}.

If you provide a host name using this option, it is also displayed
in other locations which contain the remote host name, such as the
default window title and the default SSH password prompt. This
reflects the fact that this is the host you're \e{really} connecting
to, which is more important than the mere means you happen to be
using to contact that host. (This applies even if you're using a
protocol other than SSH.)

\H{config-data} The Data panel

The Data panel allows you to configure various pieces of data which
can be sent to the server to affect your connection at the far end.

Each option on this panel applies to more than one protocol.
Options which apply to only one protocol appear on that protocol's
configuration panels.

\S{config-username} \q{\ii{Auto-login username}}

All three of the SSH, Telnet, and Rlogin protocols allow you to
specify what user name you want to log in as, without having to type
it explicitly every time. (Some Telnet servers don't support this.)

In this box you can type that user name.

\S{config-username-from-env} Use of system username

When the previous box (\k{config-username}) is left blank, by default,
PuTTY will prompt for a username at the time you make a connection.

In some environments, such as the networks of large organisations
implementing \i{single sign-on}, a more sensible default may be to use
the name of the user logged in to the local operating system (if any);
this is particularly likely to be useful with \i{GSSAPI} key exchange
and user authentication (see \k{config-ssh-auth-gssapi} and
\k{config-ssh-gssapi-kex}). This control allows you to change the default
behaviour.

The current system username is displayed in the dialog as a
convenience. It is not saved in the configuration; if a saved session
is later used by a different user, that user's name will be used.

\S{config-termtype} \q{\ii{Terminal-type} string}

Most servers you might connect to with PuTTY are designed to be
connected to from lots of different types of terminal. In order to
send the right \i{control sequence}s to each one, the server will need
to know what type of terminal it is dealing with. Therefore, each of
the SSH, Telnet, and Rlogin protocols allow a text string to be sent
down the connection describing the terminal.  On a \i{Unix} server,
this selects an entry from the \i\c{termcap} or \i\c{terminfo} database
that tells applications what \i{control sequences} to send to the
terminal, and what character sequences to expect the \i{keyboard}
to generate.

PuTTY attempts to emulate the Unix \i\c{xterm} program, and by default
it reflects this by sending \c{xterm} as a terminal-type string. If
you find this is not doing what you want - perhaps the remote
system reports \q{Unknown terminal type} - you could try setting
this to something different, such as \i\c{vt220}.

If you're not sure whether a problem is due to the terminal type
setting or not, you probably need to consult the manual for your
application or your server.

\S{config-termspeed} \q{\ii{Terminal speed}s}

The Telnet, Rlogin, and SSH protocols allow the client to specify
terminal speeds to the server.

This parameter does \e{not} affect the actual speed of the connection,
which is always \q{as fast as possible}; it is just a hint that is
sometimes used by server software to modify its behaviour. For
instance, if a slow speed is indicated, the server may switch to a
less \i{bandwidth}-hungry display mode.

The value is usually meaningless in a network environment, but
PuTTY lets you configure it, in case you find the server is reacting
badly to the default value.

The format is a pair of numbers separated by a comma, for instance,
\c{38400,38400}. The first number represents the output speed
(\e{from} the server) in bits per second, and the second is the input
speed (\e{to} the server). (Only the first is used in the Rlogin
protocol.)

This option has no effect on Raw connections.

\S{config-environ} Setting \i{environment variables} on the server

The Telnet protocol provides a means for the client to pass
environment variables to the server. Many Telnet servers have
stopped supporting this feature due to security flaws, but PuTTY
still supports it for the benefit of any servers which have found
other ways around the security problems than just disabling the
whole mechanism.

Version 2 of the SSH protocol also provides a similar mechanism,
which is easier to implement without security flaws. Newer \i{SSH-2}
servers are more likely to support it than older ones.

This configuration data is not used in the SSH-1, rlogin or raw
protocols.

To add an environment variable to the list transmitted down the
connection, you enter the variable name in the \q{Variable} box,
enter its value in the \q{Value} box, and press the \q{Add} button.
To remove one from the list, select it in the list box and press
\q{Remove}.

\H{config-proxy} The Proxy panel

The \ii{Proxy} panel allows you to configure PuTTY to use various types
of proxy in order to make its network connections. The settings in
this panel affect the primary network connection forming your PuTTY
session, and also any extra connections made as a result of SSH \i{port
forwarding} (see \k{using-port-forwarding}).

Note that unlike some software (such as web browsers), PuTTY does not
attempt to automatically determine whether to use a proxy and (if so)
which one to use for a given destination. If you need to use a proxy,
it must always be explicitly configured.

\S{config-proxy-type} Setting the proxy type

The \q{Proxy type} drop-down allows you to configure what type of
proxy you want PuTTY to use for its network connections. The default
setting is \q{None}; in this mode no proxy is used for any
connection.

\b Selecting \I{HTTP proxy}\q{HTTP CONNECT} allows you to proxy your
connections through a web server supporting the HTTP \cw{CONNECT} command,
as documented in \W{https://www.rfc-editor.org/rfc/rfc2817}{RFC 2817}.

\b Selecting \q{SOCKS 4} or \q{SOCKS 5} allows you to proxy your
connections through a \i{SOCKS server}.

\b Many firewalls implement a less formal type of proxy in which a
user can make a Telnet or TCP connection directly to the firewall machine
and enter a command such as \c{connect myhost.com 22} to connect
through to an external host. Selecting \I{Telnet proxy}\q{Telnet}
allows you to tell PuTTY to use this type of proxy, with the precise           
command specified as described in \k{config-proxy-command}.

\b There are several ways to use a SSH server as a proxy. All of
these cause PuTTY to make a secondary SSH connection to the proxy host
(sometimes called a \q{\i{jump host}} in this context).

\lcont{
The \q{Proxy hostname} field will be interpreted as the name of a
PuTTY saved session if one exists, or a hostname if not. This
allows multi-hop jump paths, if the referenced saved session is
itself configured to use an SSH proxy; and it allows combining SSH
and non-SSH proxying.

\b \q{SSH to proxy and use port forwarding} causes PuTTY to use the
secondary SSH connection to open a port-forwarding channel to the
final destination host (similar to OpenSSH's \cw{-J} option).

\b \q{SSH to proxy and execute a command} causes PuTTY to run an
arbitrary remote command on the proxy SSH server and use that
command's standard input and output streams to run the primary
connection over. The remote command line is specified as described in
\k{config-proxy-command}.

\b \q{SSH to proxy and invoke a subsystem} is similar but causes PuTTY
to start an SSH \q{\i{subsystem}} rather than an ordinary command line.
This might be useful with a specially set up SSH proxy server.
}

\b Selecting \I{Local proxy}\q{Local} allows you to specify an arbitrary
command on the local machine to act as a proxy. When the session is
started, instead of creating a TCP connection, PuTTY runs the command
(specified in \k{config-proxy-command}), and uses its standard input and
output streams.

\lcont{
This could be used, for instance, to talk to some kind of network proxy
that PuTTY does not natively support; or you could tunnel a connection
over something other than TCP/IP entirely.

You can also enable this mode on the command line; see
\k{using-cmdline-proxycmd}.
}

\S{config-proxy-exclude} Excluding parts of the network from proxying

Typically you will only need to use a proxy to connect to non-local
parts of your network; for example, your proxy might be required for
connections outside your company's internal network. In the
\q{Exclude Hosts/IPs} box you can enter ranges of IP addresses, or
ranges of DNS names, for which PuTTY will avoid using the proxy and
make a direct connection instead.

The \q{Exclude Hosts/IPs} box may contain more than one exclusion
range, separated by commas. Each range can be an IP address or a DNS
name, with a \c{*} character allowing wildcards. For example:

\c *.example.com

This excludes any host with a name ending in \c{.example.com} from
proxying.

\c 192.168.88.*

This excludes any host with an IP address starting with 192.168.88
from proxying.

\c 192.168.88.*,*.example.com

This excludes both of the above ranges at once.

Connections to the local host (the host name \i\c{localhost}, and any
\i{loopback IP address}) are never proxied, even if the proxy exclude
list does not explicitly contain them. It is very unlikely that this
behaviour would ever cause problems, but if it does you can change
it by enabling \q{Consider proxying local host connections}.

Note that if you are doing \I{proxy DNS}DNS at the proxy (see
\k{config-proxy-dns}), you should make sure that your proxy
exclusion settings do not depend on knowing the IP address of a
host. If the name is passed on to the proxy without PuTTY looking it
up, it will never know the IP address and cannot check it against
your list.

\S{config-proxy-dns} \I{proxy DNS}\ii{Name resolution} when using a proxy

If you are using a proxy to access a private network, it can make a
difference whether \i{DNS} name resolution is performed by PuTTY itself
(on the client machine) or performed by the proxy.

The \q{Do DNS name lookup at proxy end} configuration option allows
you to control this. If you set it to \q{No}, PuTTY will always do
its own DNS, and will always pass an IP address to the proxy. If you
set it to \q{Yes}, PuTTY will always pass host names straight to the
proxy without trying to look them up first.

If you set this option to \q{Auto} (the default), PuTTY will do
something it considers appropriate for each type of proxy. Most
types of proxy (HTTP, SOCK5, SSH, Telnet, and local) will have host
names passed straight to them; SOCKS4 proxies will not.

Note that if you are doing DNS at the proxy, you should make sure
that your proxy exclusion settings (see \k{config-proxy-exclude}) do
not depend on knowing the IP address of a host. If the name is
passed on to the proxy without PuTTY looking it up, it will never
know the IP address and cannot check it against your list.

The original SOCKS 4 protocol does not support proxy-side DNS. There
is a protocol extension (SOCKS 4A) which does support it, but not
all SOCKS 4 servers provide this extension. If you enable proxy DNS
and your SOCKS 4 server cannot deal with it, this might be why.

If you want to avoid PuTTY making \e{any} DNS query related to your
destination host name (for example, because your local DNS resolver is
very slow to return a negative response in that situation), then as
well as setting this control to \q{Yes}, you may also need to turn off
GSSAPI authentication and GSSAPI key exchange in SSH (see
\k{config-ssh-auth-gssapi} and \k{config-ssh-gssapi-kex}
respectively). This is because GSSAPI setup also involves a DNS query
for the destination host name, and that query is performed by the
separate GSSAPI library, so PuTTY can't override or reconfigure it.

\S{config-proxy-auth} \I{proxy username}Username and \I{proxy password}password

You can enter a username and a password in the \q{Username} and
\q{Password} boxes, which will be used if your proxy requires
\I{proxy authentication}authentication.

\I{security hazard}Note that if you save your session, the proxy
password will be saved in plain text, so anyone who can access your PuTTY
configuration data will be able to discover it.

If PuTTY discovers that it needs a proxy username or password and you
have not specified one here, PuTTY will prompt for it interactively in
the terminal window.

Authentication is not fully supported for all forms of proxy:

\b Username and password authentication is supported for HTTP
proxies and SOCKS 5 proxies.

\lcont{

\b With SOCKS 5, authentication is via \i{CHAP} if the proxy
supports it (this is not supported in \i{PuTTYtel}); otherwise the
password is sent to the proxy in \I{plaintext password}plain text.

\b With HTTP proxying, authentication is via \q{\i{HTTP Digest}} if
possible (again, not supported in PuTTYtel), or \q{\i{HTTP Basic}}. In
the latter case, the password is sent to the proxy in \I{plaintext
password}plain text.

}

\b SOCKS 4 can use the \q{Username} field, but does not support
passwords.

\b SSH proxying can use all the same forms of SSH authentication
supported by PuTTY for its main connection. If the SSH server requests
password authentication, any configured proxy password will be used,
but other authentication methods such as public keys and GSSAPI will
be tried first, just as for a primary SSH connection, and if they
require credentials such as a key passphrase, PuTTY will interactively
prompt for these.

\b You can specify a way to include a username and password in the
Telnet/Local proxy command (see \k{config-proxy-command}). If you do
so, and don't also specify the actual username and/or password in the
configuration, PuTTY will interactively prompt for them.

\S{config-proxy-command} Specifying the Telnet, SSH, or Local proxy command

If you are using the \i{Telnet proxy} type, the usual command required
by the firewall's Telnet server is \c{connect}, followed by a host
name and a port number. If your proxy needs a different command,
you can enter an alternative in the \q{Command to send to proxy} box.

If you are using the \i{Local proxy} type, the local command to run
is specified here.

If you are using the \q{SSH to proxy and execute a command} type, the
command to run on the SSH proxy server is specified here. Similarly, if
you are using \q{SSH to proxy and invoke a subsystem}, the subsystem
name is constructed as specified here.

In this string, you can use \c{\\n} to represent a new-line, \c{\\r}
to represent a carriage return, \c{\\t} to represent a tab
character, and \c{\\x} followed by two hex digits to represent any
other character. \c{\\\\} is used to encode the \c{\\} character
itself.

Also, the special strings \c{%host} and \c{%port} will be replaced
by the host name and port number you want to connect to. For Telnet
and Local proxy types, the strings \c{%user} and \c{%pass} will be
replaced by the proxy username and password (which, if not specified
in the configuration, will be prompted for) \dash this does not happen
with SSH proxy types (because the proxy username/password are used
for SSH authentication). The strings \c{%proxyhost} and \c{%proxyport}
will be replaced by the host details specified on the \e{Proxy} panel,
if any (this is most likely to be useful for proxy types using a
local or remote command). To get a literal \c{%} sign, enter \c{%%}.

If a Telnet proxy server prompts for a username and password
before commands can be sent, you can use a command such as:

\c %user\n%pass\nconnect %host %port\n

This will send your username and password as the first two lines to
the proxy, followed by a command to connect to the desired host and
port. Note that if you do not include the \c{%user} or \c{%pass}
tokens in the Telnet command, then anything specified in \q{Username}
and \q{Password} configuration fields will be ignored.

\S{config-proxy-logging} Controlling \i{proxy logging}

Often the proxy interaction has its own diagnostic output; this is
particularly the case for local proxy commands.

The setting \q{Print proxy diagnostics in the terminal window} lets
you control how much of the proxy's diagnostics are printed to the main
terminal window, along with output from your main session.

By default (\q{No}), proxy diagnostics are only sent to the Event Log;
with \q{Yes} they are also printed to the terminal, where they may get
mixed up with your main session. \q{Only until session starts} is a
compromise; proxy messages will go to the terminal window until the main
session is deemed to have started (in a protocol-dependent way), which
is when they're most likely to be interesting; any further proxy-related
messages during the session will only go to the Event Log.

\H{config-ssh} The SSH panel

The \i{SSH} panel allows you to configure options that only apply to
SSH sessions.

\S{config-command} Executing a specific command on the server

In SSH, you don't have to run a general shell session on the server.
Instead, you can choose to run a single specific command (such as a
mail user agent, for example). If you want to do this, enter the
command in the \q{\ii{Remote command}} box.

Note that most servers will close the session after executing the
command.

\S{config-ssh-noshell} \q{Don't start a \I{remote shell}shell or
\I{remote command}command at all}

If you tick this box, PuTTY will not attempt to run a shell or
command after connecting to the remote server. You might want to use
this option if you are only using the SSH connection for \i{port
forwarding}, and your user account on the server does not have the
ability to run a shell.

This feature is only available in \i{SSH protocol version 2} (since the
version 1 protocol assumes you will always want to run a shell).

This feature can also be enabled using the \c{-N} command-line
option; see \k{using-cmdline-noshell}.

If you use this feature in Plink, you will not be able to terminate
the Plink process by any graceful means; the only way to kill it
will be by pressing Control-C or sending a kill signal from another
program.

\S{config-ssh-comp} \q{Enable \i{compression}}

This enables data compression in the SSH connection: data sent by
the server is compressed before sending, and decompressed at the
client end. Likewise, data sent by PuTTY to the server is compressed
first and the server decompresses it at the other end. This can help
make the most of a low-\i{bandwidth} connection.

\S{config-ssh-prot} \q{\i{SSH protocol version}}

This allows you to select whether to use \i{SSH protocol version 2}
or the older \I{SSH-1}version 1.

You should normally leave this at the default of \q{2}. As well as
having fewer features, the older SSH-1 protocol is no longer
developed, has many known cryptographic weaknesses, and is generally
not considered to be secure. PuTTY's protocol 1 implementation is
provided mainly for compatibility, and is no longer being enhanced.

If a server offers both versions, prefer \q{2}. If you have some
server or piece of equipment that only talks SSH-1, select \q{1}
here, and do not treat the resulting connection as secure.

PuTTY will not automatically fall back to the other version of the
protocol if the server turns out not to match your selection here;
instead, it will put up an error message and abort the connection.
This prevents an active attacker downgrading an intended SSH-2
connection to SSH-1.

\S{config-ssh-sharing} Sharing an SSH connection between PuTTY tools

The controls in this box allow you to configure PuTTY to reuse an
existing SSH connection, where possible.

The SSH-2 protocol permits you to run multiple data channels over the
same SSH connection, so that you can log in just once (and do the
expensive encryption setup just once) and then have more than one
terminal window open.

Each instance of PuTTY can still run at most one terminal session, but
using the controls in this box, you can configure PuTTY to check if
another instance of itself has already connected to the target host,
and if so, share that instance's SSH connection instead of starting a
separate new one.

To enable this feature, just tick the box \q{Share SSH connections if
possible}. Then, whenever you start up a PuTTY session connecting to a
particular host, it will try to reuse an existing SSH connection if
one is available. For example, selecting \q{Duplicate Session} from
the system menu will launch another session on the same host, and if
sharing is enabled then it will reuse the existing SSH connection.

When this mode is in use, the first PuTTY that connected to a given
server becomes the \q{upstream}, which means that it is the one
managing the real SSH connection. All subsequent PuTTYs which reuse
the connection are referred to as \q{downstreams}: they do not connect
to the real server at all, but instead connect to the upstream PuTTY
via local inter-process communication methods.

For this system to be activated, \e{both} the upstream and downstream
instances of PuTTY must have the sharing option enabled.

The upstream PuTTY can therefore not terminate until all its
downstreams have closed. This is similar to the effect you get with
port forwarding or X11 forwarding, in which a PuTTY whose terminal
session has already finished will still remain open so as to keep
serving forwarded connections.

In case you need to configure this system in more detail, there are
two additional checkboxes which allow you to specify whether a
particular PuTTY can act as an upstream or a downstream or both.
(These boxes only take effect if the main \q{Share SSH connections if
possible} box is also ticked.) By default both of these boxes are
ticked, so that multiple PuTTYs started from the same configuration
will designate one of themselves as the upstream and share a single
connection; but if for some reason you need a particular PuTTY
configuration \e{not} to be an upstream (e.g. because you definitely
need it to close promptly) or not to be a downstream (e.g. because it
needs to do its own authentication using a special private key) then
you can untick one or the other of these boxes.

I have referred to \q{PuTTY} throughout the above discussion, but all
the other PuTTY tools which make SSH connections can use this
mechanism too. For example, if PSCP or PSFTP loads a configuration
with sharing enabled, then it can act as a downstream and use an
existing SSH connection set up by an instance of GUI PuTTY. The one
special case is that PSCP and PSFTP will \e{never} act as upstreams.

It is possible to test programmatically for the existence of a live
upstream using Plink. See \k{plink-option-shareexists}.

\H{config-ssh-kex} The Kex panel

The Kex panel (short for \q{\i{key exchange}}) allows you to configure
options related to SSH-2 key exchange.

Key exchange occurs at the start of an SSH connection (and
occasionally thereafter); it establishes a \i{shared secret} that is used
as the basis for all of SSH's security features. It is therefore very
important for the security of the connection that the key exchange is
secure.

Key exchange is a cryptographically intensive process; if either the
client or the server is a relatively slow machine, the slower methods
may take several tens of seconds to complete.

If connection startup is too slow, or the connection hangs
periodically, you may want to try changing these settings.

If you don't understand what any of this means, it's safe to leave
these settings alone.

This entire panel is only relevant to SSH protocol version 2; none of
these settings affect SSH-1 at all.

\S{config-ssh-kex-order} \ii{Key exchange algorithm} selection

PuTTY supports a variety of SSH-2 key exchange methods, and allows you
to choose which one you prefer to use; configuration is similar to
cipher selection (see \k{config-ssh-encryption}).

PuTTY currently supports the following key exchange methods:

\b \q{NTRU Prime / Curve25519 hybrid}: \q{\i{Streamlined NTRU Prime}}
is a lattice-based algorithm intended to resist \i{quantum attacks}.
In this key exchange method, it is run in parallel with a conventional
Curve25519-based method (one of those included in \q{ECDH}), in such
a way that it should be no \e{less} secure than that commonly-used
method, and hopefully also resistant to a new class of attacks.

\b \q{\i{ECDH}}: elliptic curve Diffie-Hellman key exchange,
with a variety of standard curves and hash algorithms.

\b The original form of \i{Diffie-Hellman key exchange}, with a
variety of well-known groups and hashes:

\lcont{
\b \q{Group 18}, a well-known 8192-bit group, used with the SHA-512
hash function.

\b \q{Group 17}, a well-known 6144-bit group, used with the SHA-512
hash function.

\b \q{Group 16}, a well-known 4096-bit group, used with the SHA-512
hash function.

\b \q{Group 15}, a well-known 3072-bit group, used with the SHA-512
hash function.

\b \q{Group 14}: a well-known 2048-bit group, used with the SHA-256
hash function or, if the server doesn't support that, SHA-1.

\b \q{Group 1}: a well-known 1024-bit group, used with the SHA-1
hash function. Neither we nor current SSH standards recommend using
this method any longer, and it's not used by default in new
installations; however, it may be the only method supported by very
old server software.
}

\b \q{Diffie-Hellman \i{group exchange}}: with this method, instead
of using a fixed group, PuTTY requests that the server suggest a group
to use for a subsequent Diffie-Hellman key exchange; the server can
avoid groups known to be weak, and possibly invent new ones over time,
without any changes required to PuTTY's configuration. This key
exchange method uses the SHA-256 hash or, if the server doesn't
support that, SHA-1.

\b \q{\i{RSA-based key exchange}}: this requires much less computational
effort on the part of the client, and somewhat less on the part of
the server, than Diffie-Hellman key exchange.

\b \q{GSSAPI key exchange}: see \k{config-ssh-gssapi-kex}.

If the first algorithm PuTTY finds is below the \q{warn below here}
line, you will see a warning box when you make the connection, similar
to that for cipher selection (see \k{config-ssh-encryption}).

\S2{config-ssh-gssapi-kex} GSSAPI-based key exchange

PuTTY supports a set of key exchange methods that also incorporates
GSSAPI-based authentication. They are enabled with the
\q{Attempt GSSAPI key exchange} checkbox (which also appears on the
\q{GSSAPI} panel).

PuTTY can only perform the GSSAPI-authenticated key exchange methods
when using Kerberos V5, and not other GSSAPI mechanisms. If the user
running PuTTY has current Kerberos V5 credentials, then PuTTY will
select the GSSAPI key exchange methods in preference to any of the
ordinary SSH key exchange methods configured in the preference list.
There's a GSSAPI-based equivalent to most of the ordinary methods
listed in \k{config-ssh-kex-order}; server support determines which
one will be used. (PuTTY's preference order for GSSAPI-authenticated
key exchange methods is fixed, not controlled by the preference list.)

The advantage of doing GSSAPI authentication as part of the SSH key
exchange is apparent when you are using credential delegation (see
\k{config-ssh-auth-gssapi-delegation}). The SSH key exchange can be
repeated later in the session, and this allows your Kerberos V5
credentials (which are typically short-lived) to be automatically
re-delegated to the server when they are refreshed on the client.
(This feature is commonly referred to as \q{\i{cascading credentials}}.)

If your server doesn't support GSSAPI key exchange, it may still
support GSSAPI in the SSH user authentication phase. This will still
let you log in using your Kerberos credentials, but will only allow
you to delegate the credentials that are active at the beginning of
the session; they can't be refreshed automatically later, in a
long-running session. See \k{config-ssh-auth-gssapi} for how to
control GSSAPI user authentication in PuTTY.

Another effect of GSSAPI key exchange is that it replaces the usual
SSH mechanism of permanent host keys described in \k{gs-hostkey}.
So if you use this method, then you won't be asked any interactive
questions about whether to accept the server's host key. Instead, the
Kerberos exchange will verify the identity of the host you connect to,
at the same time as verifying your identity to it.

\S{config-ssh-kex-rekey} \ii{Repeat key exchange}

If the session key negotiated at connection startup is used too much
or for too long, it may become feasible to mount attacks against the
SSH connection. Therefore, the SSH-2 protocol specifies that a new key
exchange should take place every so often; this can be initiated by
either the client or the server.

While this renegotiation is taking place, no data can pass through
the SSH connection, so it may appear to \q{freeze}. (The occurrence of
repeat key exchange is noted in the Event Log; see
\k{using-eventlog}.) Usually the same algorithm is used as at the
start of the connection, with a similar overhead.

These options control how often PuTTY will initiate a repeat key
exchange (\q{rekey}). You can also force a key exchange at any time
from the Special Commands menu (see \k{using-specials}).

\# FIXME: do we have any additions to the SSH-2 specs' advice on
these values? Do we want to enforce any limits?

\b \q{Max minutes before rekey} specifies the amount of time that is
allowed to elapse before a rekey is initiated. If this is set to zero,
PuTTY will not rekey due to elapsed time. The SSH-2 protocol
specification recommends a timeout of at most 60 minutes.

You might have a need to disable time-based rekeys completely for the same
reasons that \i{keepalives} aren't always helpful. If you anticipate
suffering a network dropout of several hours in the middle of an SSH
connection, but were not actually planning to send \e{data} down
that connection during those hours, then an attempted rekey in the
middle of the dropout will probably cause the connection to be
abandoned, whereas if rekeys are disabled then the connection should
in principle survive (in the absence of interfering \i{firewalls}). See
\k{config-keepalive} for more discussion of these issues; for these
purposes, rekeys have much the same properties as keepalives.
(Except that rekeys have cryptographic value in themselves, so you
should bear that in mind when deciding whether to turn them off.)
Note, however, the the SSH \e{server} can still initiate rekeys.

\b \q{Minutes between GSSAPI checks}, if you're using GSSAPI key
exchange, specifies how often the GSSAPI credential cache is checked
to see whether new tickets are available for delegation, or current
ones are near expiration. If forwarding of GSSAPI credentials is
enabled, PuTTY will try to rekey as necessary to keep the delegated
credentials from expiring. Frequent checks are recommended; rekeying
only happens when needed.

\b \q{Max data before rekey} specifies the amount of data (in bytes)
that is permitted to flow in either direction before a rekey is
initiated. If this is set to zero, PuTTY will not rekey due to
transferred data. The SSH-2 protocol specification recommends a limit
of at most 1 gigabyte.

\lcont{

As well as specifying a value in bytes, the following shorthand can be
used:

\b \cq{1k} specifies 1 kilobyte (1024 bytes).

\b \cq{1M} specifies 1 megabyte (1024 kilobytes).

\b \cq{1G} specifies 1 gigabyte (1024 megabytes).

}

Disabling data-based rekeys entirely is a bad idea.  The \i{integrity},
and to a lesser extent, \i{confidentiality} of the SSH-2 protocol depend
in part on rekeys occurring before a 32-bit packet sequence number
wraps around.  Unlike time-based rekeys, data-based rekeys won't occur
when the SSH connection is idle, so they shouldn't cause the same
problems.  The SSH-1 protocol, incidentally, has even weaker integrity
protection than SSH-2 without rekeys.

\H{config-ssh-hostkey} The Host Keys panel

The Host Keys panel allows you to configure options related to
\i{host key management}.

Host keys are used to prove the server's identity, and assure you that
the server is not being spoofed (either by a man-in-the-middle attack
or by completely replacing it on the network). See \k{gs-hostkey} for
a basic introduction to host keys.

Much of this panel is only relevant to SSH protocol version 2; SSH-1
only supports one type of host key.

\S{config-ssh-hostkey-order} \ii{Host key type} selection

PuTTY supports a variety of SSH-2 host key types, and allows you to
choose which one you prefer to use to identify the server.
Configuration is similar to cipher selection (see
\k{config-ssh-encryption}).

PuTTY currently supports the following host key types:

\b \q{\i{Ed25519}}: \I{EdDSA}Edwards-curve DSA using a twisted Edwards
curve with modulus \cw{2^255-19}.

\b \q{\i{Ed448}}: another \I{EdDSA}Edwards-curve DSA type, using a
larger elliptic curve with a 448-bit instead of 255-bit modulus (so it
has a higher security level than Ed25519).

\b \q{ECDSA}: \i{elliptic curve} \i{DSA} using one of the
\i{NIST}-standardised elliptic curves.

\b \q{DSA}: straightforward \i{DSA} using modular exponentiation.

\b \q{RSA}: the ordinary \i{RSA} algorithm.

If PuTTY already has one or more host keys stored for the server,
it will by default prefer to use one of those, even if the server has
a key type that is higher in the preference order. You can add such a
key to PuTTY's cache from within an existing session using the
\q{Special Commands} menu; see \k{using-specials}.

Otherwise, PuTTY will choose a key type based purely on the
preference order you specify in the configuration.

If the first key type PuTTY finds is below the \q{warn below here}
line, you will see a warning box when you make the connection, similar
to that for cipher selection (see \k{config-ssh-encryption}).

\S{config-ssh-prefer-known-hostkeys} Preferring known host keys

By default, PuTTY will adjust the preference order for SSH-2 host key
algorithms so that any host keys it already knows are moved to the top
of the list.

This prevents you from having to check and confirm a new host key for
a server you already had one for (e.g. because the server has
generated an alternative key of a type higher in PuTTY's preference
order, or because you changed the preference order itself).

However, on the other hand, it can leak information to a listener in
the network about \e{whether} you already know a host key for this
server.

For this reason, this policy is configurable. By turning this checkbox
off, you can reset PuTTY to always use the exact order of host key
algorithms configured in the preference list described in
\k{config-ssh-hostkey-order}, so that a listener will find out nothing
about what keys you had stored.

\S{config-ssh-kex-manual-hostkeys} \ii{Manually configuring host keys}

In some situations, if PuTTY's automated host key management is not
doing what you need, you might need to manually configure PuTTY to
accept a specific host key, or one of a specific set of host keys.

One reason why you might want to do this is because the host name
PuTTY is connecting to is using round-robin DNS to return one of
multiple actual servers, and they all have different host keys. In
that situation, you might need to configure PuTTY to accept any of a
list of host keys for the possible servers, while still rejecting any
key not in that list.

Another reason is if PuTTY's automated host key management is
completely unavailable, e.g. because PuTTY (or Plink or PSFTP, etc) is
running in a Windows environment without access to the Registry. In
that situation, you will probably want to use the \cw{-hostkey}
command-line option to configure the expected host key(s); see
\k{using-cmdline-hostkey}.

For situations where PuTTY's automated host key management simply
picks the wrong host name to store a key under, you may want to
consider setting a \q{logical host name} instead; see
\k{config-loghost}.

To configure manual host keys via the GUI, enter some text describing
the host key into the edit box in the \q{Manually configure host keys
for this connection} container, and press the \q{Add} button. The text
will appear in the \q{Host keys or fingerprints to accept} list box.
You can remove keys again with the \q{Remove} button.

The text describing a host key can be in one of the following formats:

\b An \I{SHA256 fingerprint}SHA-256-based host key fingerprint of the
form displayed in PuTTY's Event Log and host key dialog boxes,
i.e. \cq{SHA256:} followed by 43 case-sensitive characters.

\b An \I{MD5 fingerprint}MD5-based host key fingerprint, i.e. sixteen
2-digit hex numbers separated by colons, optionally preceded by the
prefix \cq{MD5:}. (The case of the characters does not matter.)

\b A base64-encoded blob describing an SSH-2 public key in
OpenSSH's one-line public key format. How you acquire a public key in
this format is server-dependent; on an OpenSSH server it can typically
be found in a location like \c{/etc/ssh/ssh_host_rsa_key.pub}.

If this box contains at least one host key or fingerprint when PuTTY
makes an SSH connection, then PuTTY's automated host key management is
completely bypassed: the connection will be permitted if and only if
the host key presented by the server is one of the keys listed in this
box, and the \I{host key cache}host key store in the Registry will be
neither read \e{nor written}, unless you explicitly do so.

If the box is empty (as it usually is), then PuTTY's automated host
key management will work as normal.

\S{config-ssh-kex-cert} Configuring PuTTY to accept host \i{certificates}

In some environments, the SSH host keys for a lot of servers will all
be signed in turn by a central \q{certification authority} (\q{CA} for
short). This simplifies host key configuration for users, because if
they configure their SSH client to accept host keys certified by that
CA, then they don't need to individually confirm each host key the
first time they connect to that server.

In order to do this, press the \q{Configure host CAs} button in the
\q{Host keys} configuration panel. This will launch a secondary
configuration dialog box where you can configure what CAs PuTTY will
accept signatures from.

\s{Note that this configuration is common to all saved sessions}.
Everything in the main PuTTY configuration is specific to one saved
session, and you can prepare a separate session with all the
configuration different. But there's only one copy of the host CA
configuration, and it applies to all sessions PuTTY runs, whether
saved or not.

(Otherwise, it would be useless \dash configuring a CA by hand for
each new host wouldn't be any more convenient than pressing the
\q{confirm} button for each new host's host key.)

To set up a new CA using this config box:

First, load the CA's public key from a file, or paste it directly into
the \q{Public key of certification authority} edit box. If your
organisation signs its host keys in this way, they will publish the
public key of their CA so that SSH users can include it in their
configuration.

Next, in the \q{Valid hosts this key is trusted to certify} box,
configure at least one hostname wildcard to say what servers PuTTY
should trust this CA to speak for. For example, suppose you work for
Example Corporation (\cw{example.com}), and the Example Corporation IT
department has advertised a CA that signs all the Example internal
machines' host keys. Then probably you want to trust that CA to sign
host keys for machines in the domain \cw{example.com}, but not for
anything else. So you might enter \cq{*.example.com} into the \q{Valid
hosts} box.

\s{It's important to limit what the CA key is allowed to sign}. Don't
just enter \cq{*} in that box! If you do that, you're saying that
Example Corporation IT department is authorised to sign a host key for
\e{anything at all} you might decide to connect to \dash even if
you're connecting out of the company network to a machine somewhere
else, such as your own personal server. So that configuration would
enable the Example IT department to act as a \q{man-in-the-middle}
between your PuTTY process and your server, and listen in to your
communications \dash exactly the thing SSH is supposed to avoid.

So, if the CA was provided to you by the sysadmins responsible for
\cw{example.com} (or whatever), make sure PuTTY will \e{only} trust it
for machines in the \cw{example.com} domain.

For the full syntax of the \q{Valid hosts} expression, see
\k{config-ssh-cert-valid-expr}.

Finally, choose an identifying name for this CA; enter that name in
the \q{Name for this CA} edit box at the top of the window, and press
\q{Save} to record the CA in your configuration. The name you chose
will appear in the list of saved CAs to the left of the \q{Save}
button.

The identifying name can be anything you like. It's there so that if
you store multiple certificates you can tell which is which later when
you want to edit or delete them. It also appears in the PuTTY Event
Log when a server presents a certificate signed by that CA.

To reload an existing CA configuration, select it in the list box and
press \q{Load}. Then you can make changes, and save it again.

To remove a CA from your configuration completely, select it in the
list and press \q{Delete}.

\S2{config-ssh-cert-valid-expr} Expressions you can enter in \q{Valid
hosts}

The simplest thing you can enter in the \q{Valid hosts this key is
trusted to certify} edit box is just a hostname wildcard such as
\cq{*.example.com}. This matches any host in any subdomain, so
both \cq{ssh.example.com} and \cq{login.dept.example.com} would
match, but \cq{prod.example.net} would not.

But you can also enter multiple host name wildcards, and port number
ranges, and make complicated Boolean expressions out of them using the
operators \cq{&&} for \q{and}, \cq{||} for \q{or}, \cq{!} for \q{not},
and parentheses.

For example, here are some other things you could enter.

\b \cq{*.foo.example.com || *.bar.example.com}. This means the CA is
trusted to sign the host key for a connection if the host name matches
\q{*.foo.example.com} \e{or} it matches \q{*.bar.example.com}. In
other words, the CA has authority over those two particular subdomains
of \cw{example.com}, but not for anything else, like
\cw{www.example.com}.

\b \cq{*.example.com && ! *.extrasecure.example.com}. This means the
CA is trusted to sign the host key for a connection if the host name
matches \q{*.example.com} \e{but does not} match
\q{*.extrasecure.example.com}. (Imagine if there was one top-secret
set of servers in your company that the main IT department didn't have
security clearance to administer.)

\b \cq{*.example.com && port:22}. This means the CA is trusted to sign
the host key for a connection if the host name matches
\q{*.example.com} \e{and} the port number is 22. SSH servers running
on other ports would not be covered.

\b \cq{(*.foo.example.com || *.bar.example.com) && port:0-1023}. This
matches two subdomains of \cw{example.com}, as before, but \e{also}
restricts the port number to the range 0-1023.

A certificate configuration expression consists of one or more
individual requirements which can each be a hostname wildcard, a
single port number, or a port number range, combined together with
these Boolean operators.

Unlike other languages such as C, there is no implied priority between
\cq{&&} and \cq{||}. If you write \cq{A && B || C} (where \cw{A},
\cw{B} and \cw{C} are some particular requirements), then PuTTY will
report a syntax error, because you haven't said which of the \cq{&&}
and \cq{||} takes priority tightly. You will have to write either
\cq{(A && B) || C}, meaning \q{both of \cw{A} and \cw{B}, or
alternatively just \cw{C}}, or \cq{A && (B || C)} (\q{\cw{A}, and also
at least one of \cw{B} and \cw{C}}), to make it clear.

\S2{config-ssh-cert-rsa-hash} RSA signature types in certificates

RSA keys can be used to generate signatures with a choice of secure
hash function. Typically, any version of OpenSSH new enough to support
certificates at all will also be new enough to avoid using SHA-1, so
the default settings of accepting the more modern SHA-256 and SHA-512
should be suitable for nearly all cases. For completeness, however,
you can configure which types of RSA signature PuTTY will accept in a
certificate from a CA using an RSA key.

\H{config-ssh-encryption} The Cipher panel

PuTTY supports a variety of different \i{encryption algorithm}s, and
allows you to choose which one you prefer to use. You can do this by
dragging the algorithms up and down in the list box (or moving them
using the Up and Down buttons) to specify a preference order. When
you make an SSH connection, PuTTY will search down the list from the
top until it finds an algorithm supported by the server, and then
use that.

PuTTY currently supports the following algorithms:

\b \i{ChaCha20-Poly1305}, a combined cipher and \i{MAC} (SSH-2 only)

\b \i{AES} (Rijndael) - 256, 192, or 128-bit SDCTR or CBC, or
256 or 128-bit GCM (SSH-2 only)

\b \i{Arcfour} (RC4) - 256 or 128-bit stream cipher (SSH-2 only)

\b \i{Blowfish} - 256-bit SDCTR (SSH-2 only) or 128-bit CBC

\b \ii{Triple-DES} - 168-bit SDCTR (SSH-2 only) or CBC

\b \ii{Single-DES} - 56-bit CBC (see below for SSH-2)

If the algorithm PuTTY finds is below the \q{warn below here} line,
you will see a warning box when you make the connection:

\c The first cipher supported by the server
\c is single-DES, which is below the configured
\c warning threshold.
\c Do you want to continue with this connection?

This warns you that the first available encryption is not a very
secure one. Typically you would put the \q{warn below here} line
between the encryptions you consider secure and the ones you
consider substandard. By default, PuTTY supplies a preference order
intended to reflect a reasonable preference in terms of security and
speed.

In SSH-2, the encryption algorithm is negotiated independently for
each direction of the connection, although PuTTY does not support
separate configuration of the preference orders. As a result you may
get two warnings similar to the one above, possibly with different
encryptions.

Single-DES is not recommended in the SSH-2 protocol
standards, but one or two server implementations do support it.
PuTTY can use single-DES to interoperate with
these servers if you enable the \q{Enable legacy use of single-DES in
SSH-2} option; by default this is disabled and PuTTY will stick to
recommended ciphers.

\H{config-ssh-auth} The Auth panel

The Auth panel allows you to configure \i{authentication} options for
SSH sessions.

\S{config-ssh-banner} \q{Display pre-authentication banner}

SSH-2 servers can provide a message for clients to display to the
prospective user before the user logs in; this is sometimes known as a
pre-authentication \q{\i{banner}}. Typically this is used to provide
information about the server and legal notices.

By default, PuTTY displays this message before prompting for a
password or similar credentials (although, unfortunately, not before
prompting for a login name, due to the nature of the protocol design).
By unchecking this option, display of the banner can be suppressed
entirely.

\S{config-ssh-noauth} \q{Bypass authentication entirely}

In SSH-2, it is in principle possible to establish a connection
without using SSH's mechanisms to identify or prove who you are
to the server. An SSH server could prefer to handle authentication
in the data channel, for instance, or simply require no user
authentication whatsoever.

By default, PuTTY assumes the server requires authentication (we've
never heard of one that doesn't), and thus must start this process
with a username. If you find you are getting username prompts that
you cannot answer, you could try enabling this option. However,
most SSH servers will reject this.

This is not the option you want if you have a username and just want
PuTTY to remember it; for that see \k{config-username}.
It's also probably not what if you're trying to set up passwordless
login to a mainstream SSH server; depending on the server, you
probably wanted public-key authentication (\k{pubkey})
or perhaps GSSAPI authentication (\k{config-ssh-auth-gssapi}).
(These are still forms of authentication, even if you don't have to
interact with them.)

This option only affects SSH-2 connections. SSH-1 connections always
require an authentication step.

\S{config-ssh-notrivialauth} \q{Disconnect if authentication succeeds
trivially}

This option causes PuTTY to abandon an SSH session and disconnect from
the server, if the server accepted authentication without ever having
asked for any kind of password or signature or token.

This might be used as a security measure. There are some forms of
attack against an SSH client user which work by terminating the SSH
authentication stage early, and then doing something in the main part
of the SSH session which \e{looks} like part of the authentication,
but isn't really.

For example, instead of demanding a signature from your public key,
for which PuTTY would ask for your key's passphrase, a compromised or
malicious server might allow you to log in with no signature or
password at all, and then print a message that \e{imitates} PuTTY's
request for your passphrase, in the hope that you would type it in.
(In fact, the passphrase for your public key should not be sent to any
server.)

PuTTY's main defence against attacks of this type is the \q{trust
sigil} system: messages in the PuTTY window that are truly originated
by PuTTY itself are shown next to a small copy of the PuTTY icon,
which the server cannot fake when it tries to imitate the same message
using terminal output.

However, if you think you might be at risk of this kind of thing
anyway (if you don't watch closely for the trust sigils, or if you
think you're at extra risk of one of your servers being malicious),
then you could enable this option as an extra defence. Then, if the
server tries any of these attacks involving letting you through the
authentication stage, PuTTY will disconnect from the server before it
can send a follow-up fake prompt or other type of attack.

On the other hand, some servers \e{legitimately} let you through the
SSH authentication phase trivially, either because they are genuinely
public, or because the important authentication step happens during
the terminal session. (An example might be an SSH server that connects
you directly to the terminal login prompt of a legacy mainframe.) So
enabling this option might cause some kinds of session to stop
working. It's up to you.

\S{config-ssh-tryagent} \q{Attempt authentication using Pageant}

If this option is enabled, then PuTTY will look for Pageant (the SSH
private-key storage agent) and attempt to authenticate with any
suitable public keys Pageant currently holds.

This behaviour is almost always desirable, and is therefore enabled
by default. In rare cases you might need to turn it off in order to
force authentication by some non-public-key method such as
passwords.

This option can also be controlled using the \c{-noagent}
command-line option. See \k{using-cmdline-agentauth}.

See \k{pageant} for more information about Pageant in general.

\S{config-ssh-tis} \q{Attempt \I{TIS authentication}TIS or
\i{CryptoCard authentication}}

TIS and CryptoCard authentication are (despite their names) generic
forms of simple \I{challenge/response authentication}challenge/response
authentication available in SSH protocol version 1 only. You might use
them if you were using \i{S/Key} \i{one-time passwords}, for example,
or if you had a physical \i{security token} that generated responses
to authentication challenges.  They can even be used to prompt for
simple passwords.

With this switch enabled, PuTTY will attempt these forms of
authentication if the server is willing to try them. You will be
presented with a challenge string (which may be different every
time) and must supply the correct response in order to log in. If
your server supports this, you should talk to your system
administrator about precisely what form these challenges and
responses take.

\S{config-ssh-ki} \q{Attempt \i{keyboard-interactive authentication}}

The SSH-2 equivalent of TIS authentication is called
\q{keyboard-interactive}. It is a flexible authentication method
using an arbitrary sequence of requests and responses; so it is not
only useful for \I{challenge/response authentication}challenge/response
mechanisms such as \i{S/Key}, but it can also be used for (for example)
asking the user for a \I{password expiry}new password when the old one
has expired.

PuTTY leaves this option enabled by default, but supplies a switch
to turn it off in case you should have trouble with it.

\S{config-ssh-agentfwd} \q{Allow \i{agent forwarding}}

This option allows the SSH server to open forwarded connections back
to your local copy of \i{Pageant}. If you are not running Pageant, this
option will do nothing.

See \k{pageant} for general information on Pageant, and
\k{pageant-forward} for information on agent forwarding. Note that
there is a security risk involved with enabling this option; see
\k{pageant-security} for details.

\S{config-ssh-changeuser} \q{Allow attempted \i{changes of username} in SSH-2}

In the SSH-1 protocol, it is impossible to change username after
failing to authenticate. So if you mis-type your username at the
PuTTY \q{login as:} prompt, you will not be able to change it except
by restarting PuTTY.

The SSH-2 protocol \e{does} allow changes of username, in principle,
but does not make it mandatory for SSH-2 servers to accept them. In
particular, \i{OpenSSH} does not accept a change of username; once you
have sent one username, it will reject attempts to try to
authenticate as another user. (Depending on the version of OpenSSH,
it may quietly return failure for all login attempts, or it may send
an error message.)

For this reason, PuTTY will by default not prompt you for your
username more than once, in case the server complains. If you know
your server can cope with it, you can enable the \q{Allow attempted
changes of username} option to modify PuTTY's behaviour.

\H{config-ssh-auth-creds} The Credentials panel

This subpane of the Auth panel contains configuration options that
specify actual \e{credentials} to present to the server: key files and
certificates.

\S{config-ssh-privkey} \q{\ii{Private key} file for authentication}

This box is where you enter the name of your private key file if you
are using \i{public key authentication}. See \k{pubkey} for information
about public key authentication in SSH.

This key must be in PuTTY's native format (\c{*.\i{PPK}}). If you have a
private key in another format that you want to use with PuTTY, see
\k{puttygen-conversions}.

You can use the authentication agent \i{Pageant} so that you do not
need to explicitly configure a key here; see \k{pageant}.

If a private key file is specified here with Pageant running, PuTTY
will first try asking Pageant to authenticate with that key, and
ignore any other keys Pageant may have. If that fails, PuTTY will ask
for a passphrase as normal. You can also specify a \e{public} key file
in this case (in RFC 4716 or OpenSSH format), as that's sufficient to
identify the key to Pageant, but of course if Pageant isn't present
PuTTY can't fall back to using this file itself.

\S{config-ssh-cert} \q{\ii{Certificate} to use with the private key}

(This is optional. If you don't know you need it, you can leave this
blank.)

In some environments, user authentication keys can be signed in turn
by a \q{certifying authority} (\q{CA} for short), and user accounts on
an SSH server can be configured to automatically trust any key that's
certified by the right signature.

This can be a convenient setup if you have a very large number of
servers. When you change your key pair, you might otherwise have to
edit the \cw{authorized_keys} file on every server individually, to
make them all accept the new key. But if instead you configure all
those servers \e{once} to accept keys signed as yours by a CA, then
when you change your public key, all you have to do is to get the new
key certified by the same CA as before, and then all your servers will
automatically accept it without needing individual reconfiguration.

One way to use a certificate is to incorporate it into your private
key file. \K{puttygen-cert} explains how to do that using PuTTYgen.
But another approach is to tell PuTTY itself where to find the public
certificate file, and then it will automatically present that
certificate when authenticating with the corresponding private key.

To do this, enter the pathname of the certificate file into the
\q{Certificate to use with the private key} file selector.

When this setting is configured, PuTTY will honour it no matter
whether the private key is found in a file, or loaded into Pageant.

\S{config-ssh-authplugin} \q{\ii{Plugin} to provide authentication responses}

An SSH server can use the \q{keyboard-interactive} protocol to present
a series of arbitrary questions and answers. Sometimes this is used
for ordinary passwords, but sometimes the server will use the same
mechanism for something more complicated, such as a one-time password
system.

Some of these systems can be automated. For this purpose, PuTTY allows
you to provide a separate program to act as a \q{plugin} which will
take over the authentication and send answers to the questions on your
behalf.

If you have been provided with a plugin of this type, you can
configure it here, by entering a full command line in the \q{Plugin
command to run} box.

(If you want to \e{write} a plugin of this type, see \k{authplugin}
for the full specification of how the plugin is expected to behave.)

\H{config-ssh-auth-gssapi} The \i{GSSAPI} panel

The \q{GSSAPI} subpanel of the \q{Auth} panel controls the use of
GSSAPI authentication. This is a mechanism which delegates the
authentication exchange to a library elsewhere on the client
machine, which in principle can authenticate in many different ways
but in practice is usually used with the \i{Kerberos} \i{single sign-on}
protocol to implement \i{passwordless login}.

GSSAPI authentication is only available in the SSH-2 protocol.

PuTTY supports two forms of GSSAPI-based authentication. In one of
them, the SSH key exchange happens in the normal way, and GSSAPI is
only involved in authenticating the user. The checkbox labelled
\q{Attempt GSSAPI authentication} controls this form.

In the other method, GSSAPI-based authentication is combined with the
SSH key exchange phase. If this succeeds, then the SSH authentication
step has nothing left to do. See \k{config-ssh-gssapi-kex} for more
information about this method. The checkbox labelled \q{Attempt GSSAPI
key exchange} controls this form. (The same checkbox appears on the
\q{Kex} panel.)

If one or both of these controls is enabled, then GSSAPI
authentication will be attempted in one form or the other, and
(typically) if your client machine has valid Kerberos credentials
loaded, then PuTTY should be able to authenticate automatically to
servers that support Kerberos logins.

If both of those checkboxes are disabled, PuTTY will not try any form
of GSSAPI at all, and the rest of this panel will be unused.

\S{config-ssh-auth-gssapi-delegation} \q{Allow GSSAPI credential
delegation}

\i{GSSAPI credential delegation} is a mechanism for passing on your
Kerberos (or other) identity to the session on the SSH server. If
you enable this option, then not only will PuTTY be able to log in
automatically to a server that accepts your Kerberos credentials,
but also you will be able to connect out from that server to other
Kerberos-supporting services and use the same credentials just as
automatically.

(This option is the Kerberos analogue of SSH agent forwarding; see
\k{pageant-forward} for some information on that.)

Note that, like SSH agent forwarding, there is a security
implication in the use of this option: the administrator of the
server you connect to, or anyone else who has cracked the
administrator account on that server, could fake your identity when
connecting to further Kerberos-supporting services. However,
Kerberos sites are typically run by a central authority, so the
administrator of one server is likely to already have access to the
other services too; so this would typically be less of a risk than
SSH agent forwarding.

If your connection is not using GSSAPI key exchange, it is possible
for the delegation to expire during your session. See
\k{config-ssh-gssapi-kex} for more information.

\S{config-ssh-auth-gssapi-libraries} Preference order for GSSAPI
libraries

GSSAPI is a mechanism which allows more than one authentication
method to be accessed through the same interface. Therefore, more
than one authentication library may exist on your system which can
be accessed using GSSAPI.

PuTTY contains native support for a few well-known such libraries
(including Windows' \i{SSPI}), and will look for all of them on your system
and use whichever it finds. If more than one exists on your system and
you need to use a specific one, you can adjust the order in which it
will search using this preference list control.

One of the options in the preference list is to use a user-specified
GSSAPI library. If the library you want to use is not mentioned by
name in PuTTY's list of options, you can enter its full pathname in
the \q{User-supplied GSSAPI library path} field, and move the
\q{User-supplied GSSAPI library} option in the preference list to
make sure it is selected before anything else.

On Windows, such libraries are files with a \I{DLL}\cw{.dll}
extension, and must have been built in the same way as the PuTTY
executable you're running; if you have a 32-bit DLL, you must run a
32-bit version of PuTTY, and the same with 64-bit (see
\k{faq-32bit-64bit}). On Unix, shared libraries generally have a
\cw{.so} extension.

\H{config-ssh-tty} The TTY panel

The TTY panel lets you configure the remote pseudo-terminal.

\S{config-ssh-pty} \I{pseudo-terminal allocation}\q{Don't allocate
a pseudo-terminal}

When connecting to a \i{Unix} system, most \I{interactive
connections}interactive shell sessions are run in a \e{pseudo-terminal},
which allows the Unix system to pretend it's talking to a real physical
terminal device but allows the SSH server to catch all the data coming
from that fake device and send it back to the client.

Occasionally you might find you have a need to run a session \e{not}
in a pseudo-terminal. In PuTTY, this is generally only useful for
very specialist purposes; although in Plink (see \k{plink}) it is
the usual way of working.

\S{config-ttymodes} Sending \i{terminal modes}

The SSH protocol allows the client to send \q{terminal modes} for
the remote pseudo-terminal. These usually control the server's
expectation of the local terminal's behaviour.

If your server does not have sensible defaults for these modes, you
may find that changing them here helps, although the server is at
liberty to ignore your changes. If you don't understand any of this,
it's safe to leave these settings alone.

(None of these settings will have any effect if no pseudo-terminal
is requested or allocated.)

You can change what happens for a particular mode by selecting it in
the list, choosing one of the options and specifying the exact value
if necessary, and hitting \q{Set}. The effect of the options is as
follows:

\b If the \q{Auto} option is selected, the PuTTY tools will decide
whether to specify that mode to the server, and if so, will send
a sensible value.

\lcont{

PuTTY proper will send modes that it has an opinion on (currently only
the code for the Backspace key, \cw{ERASE}, and whether the character
set is UTF-8, \cw{IUTF8}). Plink on Unix will propagate appropriate
modes from the local terminal, if any.

}

\b If \q{Nothing} is selected, no value for the mode will be
specified to the server under any circumstances.

\b If a value is specified, it will be sent to the server under all
circumstances. The precise syntax of the value box depends on the
mode.

By default, all of the available modes are listed as \q{Auto},
which should do the right thing in most circumstances.

The precise effect of each setting, if any, is up to the server. Their
names come from \i{POSIX} and other Unix systems, and they are most
likely to have a useful effect on such systems. (These are the same
settings that can usually be changed using the \i\c{stty} command once
logged in to such servers.)

Some notable modes are described below; for fuller explanations, see
your server documentation.

\b \I{ERASE special character}\cw{ERASE} is the character that when typed
by the user will delete one space to the left. When set to \q{Auto}
(the default setting), this follows the setting of the local Backspace
key in PuTTY (see \k{config-backspace}).

\lcont{
This and other \i{special character}s are specified using \c{^C} notation
for Ctrl-C, and so on. Use \c{^<27>} or \c{^<0x1B>} to specify a
character numerically, and \c{^~} to get a literal \c{^}. Other
non-control characters are denoted by themselves. Leaving the box
entirely blank indicates that \e{no} character should be assigned to
the specified function, although this may not be supported by all
servers.
}

\b \I{QUIT special character}\cw{QUIT} is a special character that
usually forcefully ends the current process on the server
(\cw{SIGQUIT}). On many servers its default setting is Ctrl-backslash
(\c{^\\}), which is easy to accidentally invoke on many keyboards. If
this is getting in your way, you may want to change it to another
character or turn it off entirely.

\b Boolean modes such as \cw{ECHO} and \cw{ICANON} can be specified in
PuTTY in a variety of ways, such as \cw{true}/\cw{false},
\cw{yes}/\cw{no}, and \cw{0}/\cw{1}. (Explicitly specifying a value of
\cw{no} is different from not sending the mode at all.)

\b The boolean mode \I{IUTF8 terminal mode}\cw{IUTF8} signals to the
server whether the terminal character set is \i{UTF-8} or not, for
purposes such as basic line editing; if this is set incorrectly,
the backspace key may erase the wrong amount of text, for instance.
However, simply setting this is not usually sufficient for the server
to use UTF-8; POSIX servers will generally also require the locale to
be set (by some server-dependent means), although many newer
installations default to UTF-8. Also, since this mode was added to the
SSH protocol much later than the others, \#{circa 2016} many servers
(particularly older servers) do not honour this mode sent over SSH;
indeed, a few poorly-written servers object to its mere presence, so
you may find you need to set it to not be sent at all. When set to
\q{Auto}, this follows the local configured character set (see
\k{config-charset}).

\b Terminal speeds are configured elsewhere; see \k{config-termspeed}.

\H{config-ssh-x11} The X11 panel

The X11 panel allows you to configure \i{forwarding of X11} over an
SSH connection.

If your server lets you run X Window System \i{graphical applications},
X11 forwarding allows you to securely give those applications access to
a local X display on your PC.

To enable X11 forwarding, check the \q{Enable X11 forwarding} box.
If your X display is somewhere unusual, you will need to enter its
location in the \q{X display location} box; if this is left blank,
PuTTY will try to find a sensible default in the environment, or use the
primary local display (\c{:0}) if that fails.

See \k{using-x-forwarding} for more information about X11
forwarding.

\S{config-ssh-x11auth} Remote \i{X11 authentication}

If you are using X11 forwarding, the virtual X server created on the
SSH server machine will be protected by authorisation data. This
data is invented, and checked, by PuTTY.

The usual authorisation method used for this is called
\i\cw{MIT-MAGIC-COOKIE-1}. This is a simple password-style protocol:
the X client sends some cookie data to the server, and the server
checks that it matches the real cookie. The cookie data is sent over
an unencrypted X11 connection; so if you allow a client on a third
machine to access the virtual X server, then the cookie will be sent
in the clear.

PuTTY offers the alternative protocol \i\cw{XDM-AUTHORIZATION-1}. This
is a cryptographically authenticated protocol: the data sent by the
X client is different every time, and it depends on the IP address
and port of the client's end of the connection and is also stamped
with the current time. So an eavesdropper who captures an
\cw{XDM-AUTHORIZATION-1} string cannot immediately re-use it for
their own X connection.

PuTTY's support for \cw{XDM-AUTHORIZATION-1} is a somewhat
experimental feature, and may encounter several problems:

\b Some X clients probably do not even support
\cw{XDM-AUTHORIZATION-1}, so they will not know what to do with the
data PuTTY has provided.

\b This authentication mechanism will only work in SSH-2. In SSH-1,
the SSH server does not tell the client the source address of
a forwarded connection in a machine-readable format, so it's
impossible to verify the \cw{XDM-AUTHORIZATION-1} data.

\b You may find this feature causes problems with some SSH servers,
which will not clean up \cw{XDM-AUTHORIZATION-1} data after a
session, so that if you then connect to the same server using
a client which only does \cw{MIT-MAGIC-COOKIE-1} and are allocated
the same remote display number, you might find that out-of-date
authentication data is still present on your server and your X
connections fail.

PuTTY's default is \cw{MIT-MAGIC-COOKIE-1}. If you change it, you
should be sure you know what you're doing.

\S{config-ssh-xauthority} X authority file for local display

If you are using X11 forwarding, the local X server to which your
forwarded connections are eventually directed may itself require
authorisation.

Some Windows X servers do not require this: they do authorisation by
simpler means, such as accepting any connection from the local
machine but not from anywhere else. However, if your X server does
require authorisation, then PuTTY needs to know what authorisation
is required.

One way in which this data might be made available is for the X
server to store it somewhere in a file which has the same format
as the Unix \c{.Xauthority} file. If this is how your Windows X
server works, then you can tell PuTTY where to find this file by
configuring this option. By default, PuTTY will not attempt to find
any authorisation for your local display.

\H{config-ssh-portfwd} \I{port forwarding}The Tunnels panel

The Tunnels panel allows you to configure tunnelling of arbitrary
connection types through an SSH connection.

Port forwarding allows you to tunnel other types of \i{network
connection} down an SSH session. See \k{using-port-forwarding} for a
general discussion of port forwarding and how it works.

The port forwarding section in the Tunnels panel shows a list of all
the port forwardings that PuTTY will try to set up when it connects
to the server. By default no port forwardings are set up, so this
list is empty.

To add a port forwarding:

\b Set one of the \q{Local} or \q{Remote} radio buttons, depending
on whether you want to \I{local port forwarding}forward a local port
to a remote destination (\q{Local}) or \I{remote port forwarding}forward
a remote port to a local destination (\q{Remote}). Alternatively,
select \q{Dynamic} if you want PuTTY to \I{dynamic port forwarding}provide
a local SOCKS 4/4A/5 proxy on a local port (note that this proxy only
supports TCP connections; the SSH protocol does not support forwarding
\i{UDP}).

\b Enter a source \i{port number} into the \q{Source port} box. For
local forwardings, PuTTY will listen on this port of your PC. For
remote forwardings, your SSH server will listen on this port of the
remote machine. Note that most servers will not allow you to listen
on \I{privileged port}port numbers less than 1024.

\b If you have selected \q{Local} or \q{Remote} (this step is not
needed with \q{Dynamic}), enter a hostname and port number separated
by a colon, in the \q{Destination} box. Connections received on the
source port will be directed to this destination. For example, to
connect to a POP-3 server, you might enter
\c{popserver.example.com:110}. (If you need to enter a literal
\i{IPv6 address}, enclose it in square brackets, for instance
\cq{[::1]:2200}.)

\b Click the \q{Add} button. Your forwarding details should appear
in the list box.

To remove a port forwarding, simply select its details in the list
box, and click the \q{Remove} button.

In the \q{Source port} box, you can also optionally enter an \I{listen
address}IP address to listen on, by specifying (for instance)
\c{127.0.0.5:79}.
See \k{using-port-forwarding} for more information on how this
works and its restrictions.

In place of port numbers, you can enter \i{service names}, if they are
known to the local system. For instance, in the \q{Destination} box,
you could enter \c{popserver.example.com:pop3}.

You can \I{port forwarding, changing mid-session}modify the currently
active set of port forwardings in mid-session using \q{Change
Settings} (see \k{using-changesettings}). If you delete a local or
dynamic port forwarding in mid-session, PuTTY will stop listening for
connections on that port, so it can be re-used by another program. If
you delete a remote port forwarding, note that:

\b The SSH-1 protocol contains no mechanism for asking the server to
stop listening on a remote port.

\b The SSH-2 protocol does contain such a mechanism, but not all SSH
servers support it. (In particular, \i{OpenSSH} does not support it in
any version earlier than 3.9.)

If you ask to delete a remote port forwarding and PuTTY cannot make
the server actually stop listening on the port, it will instead just
start refusing incoming connections on that port. Therefore,
although the port cannot be reused by another program, you can at
least be reasonably sure that server-side programs can no longer
access the service at your end of the port forwarding.

If you delete a forwarding, any existing connections established using
that forwarding remain open. Similarly, changes to global settings
such as \q{Local ports accept connections from other hosts} only take
effect on new forwardings.

If the connection you are forwarding over SSH is itself a second SSH
connection made by another copy of PuTTY, you might find the
\q{logical host name} configuration option useful to warn PuTTY of
which host key it should be expecting. See \k{config-loghost} for
details of this.

\S{config-ssh-portfwd-localhost} Controlling the visibility of
forwarded ports

The source port for a forwarded connection usually does not accept
connections from any machine except the \I{localhost}SSH client or
server machine itself (for local and remote forwardings respectively).
There are controls in the Tunnels panel to change this:

\b The \q{Local ports accept connections from other hosts} option
allows you to set up local-to-remote port forwardings in such a way
that machines other than your client PC can connect to the forwarded
port. (This also applies to dynamic SOCKS forwarding.)

\b The \q{Remote ports do the same} option does the same thing for
remote-to-local port forwardings (so that machines other than the
SSH server machine can connect to the forwarded port.) Note that
this feature is only available in the SSH-2 protocol, and not all
SSH-2 servers support it (\i{OpenSSH} 3.0 does not, for example).

\S{config-ssh-portfwd-address-family} Selecting \i{Internet protocol
version} for forwarded ports

This switch allows you to select a specific Internet protocol (\i{IPv4}
or \i{IPv6}) for the local end of a forwarded port. By default, it is
set on \q{Auto}, which means that:

\b for a local-to-remote port forwarding, PuTTY will listen for
incoming connections in both IPv4 and (if available) IPv6

\b for a remote-to-local port forwarding, PuTTY will choose a
sensible protocol for the outgoing connection.

This overrides the general Internet protocol version preference
on the Connection panel (see \k{config-address-family}).

Note that some operating systems may listen for incoming connections
in IPv4 even if you specifically asked for IPv6, because their IPv4
and IPv6 protocol stacks are linked together. Apparently \i{Linux} does
this, and Windows does not. So if you're running PuTTY on Windows
and you tick \q{IPv6} for a local or dynamic port forwarding, it
will \e{only} be usable by connecting to it using IPv6; whereas if
you do the same on Linux, you can also use it with IPv4. However,
ticking \q{Auto} should always give you a port which you can connect
to using either protocol.

\H{config-ssh-bugs} \I{SSH server bugs}The Bugs and More Bugs panels

Not all SSH servers work properly. Various existing servers have
bugs in them, which can make it impossible for a client to talk to
them unless it knows about the bug and works around it.

Since most servers announce their software version number at the
beginning of the SSH connection, PuTTY will attempt to detect which
bugs it can expect to see in the server and automatically enable
workarounds. However, sometimes it will make mistakes; if the server
has been deliberately configured to conceal its version number, or
if the server is a version which PuTTY's bug database does not know
about, then PuTTY will not know what bugs to expect.

The Bugs and More Bugs panels (there are two because we have so many
bug compatibility modes) allow you to manually configure the bugs
PuTTY expects to see in the server. Each bug can be configured in
three states:

\b \q{Off}: PuTTY will assume the server does not have the bug.

\b \q{On}: PuTTY will assume the server \e{does} have the bug.

\b \q{Auto}: PuTTY will use the server's version number announcement
to try to guess whether or not the server has the bug. (This option is
not available for bugs that \e{cannot} be detected from the server
version, e.g. because they must be acted on before the server version
is known.)

(The PuTTY project has a defined policy about when we're prepared to
add auto-detection for a bug workaround. See \k{feedback-workarounds}.)

\S{config-ssh-bug-ignore2} \q{Chokes on SSH-2 \i{ignore message}s}

An ignore message (SSH_MSG_IGNORE) is a message in the SSH protocol
which can be sent from the client to the server, or from the server
to the client, at any time. Either side is required to ignore the
message whenever it receives it. PuTTY uses ignore messages in SSH-2
to confuse the encrypted data stream and make it harder to
cryptanalyse. It also uses ignore messages for connection
\i{keepalives} (see \k{config-keepalive}).

If it believes the server to have this bug, PuTTY will stop using
ignore messages. If this bug is enabled when talking to a correct
server, the session will succeed, but keepalives will not work and
the session might be less cryptographically secure than it could be.

\S{config-ssh-bug-rekey} \q{Handles SSH-2 key re-exchange badly}

Some SSH servers cannot cope with \i{repeat key exchange} at
all, and will ignore attempts by the client to start one. Since
PuTTY pauses the session while performing a repeat key exchange, the
effect of this would be to cause the session to hang after an hour
(unless you have your rekey timeout set differently; see
\k{config-ssh-kex-rekey} for more about rekeys).
Other, very old, SSH servers handle repeat key exchange even more
badly, and disconnect upon receiving a repeat key exchange request.

If this bug is detected, PuTTY will never initiate a repeat key
exchange. If this bug is enabled when talking to a correct server,
the session should still function, but may be less secure than you
would expect.

This is an SSH-2-specific bug.

\S{config-ssh-bug-winadj} \q{Chokes on PuTTY's SSH-2 \cq{winadj} requests}

PuTTY sometimes sends a special request to SSH servers in the middle
of channel data, with the name \cw{winadj@putty.projects.tartarus.org}
(see \k{sshnames-channel}). The purpose of this request is to measure
the round-trip time to the server, which PuTTY uses to tune its flow
control. The server does not actually have to \e{understand} the
message; it is expected to send back a \cw{SSH_MSG_CHANNEL_FAILURE}
message indicating that it didn't understand it. (All PuTTY needs for
its timing calculations is \e{some} kind of response.)

It has been known for some SSH servers to get confused by this message
in one way or another \dash because it has a long name, or because
they can't cope with unrecognised request names even to the extent of
sending back the correct failure response, or because they handle it
sensibly but fill up the server's log file with pointless spam, or
whatever. PuTTY therefore supports this bug-compatibility flag: if it
believes the server has this bug, it will never send its
\cq{winadj@putty.projects.tartarus.org} request, and will make do
without its timing data.

\S{config-ssh-bug-chanreq} \q{Replies to requests on closed channels}

The SSH protocol as published in RFC 4254 has an ambiguity which
arises if one side of a connection tries to close a channel, while the
other side simultaneously sends a request within the channel and asks
for a reply. RFC 4254 leaves it unclear whether the closing side
should reply to the channel request after having announced its
intention to close the channel.

Discussion on the \cw{ietf-ssh} mailing list in April 2014 formed a
clear consensus that the right answer is no. However, because of the
ambiguity in the specification, some SSH servers have implemented the
other policy; for example,
\W{https://bugzilla.mindrot.org/show_bug.cgi?id=1818}{OpenSSH used to}
until it was fixed.

Because PuTTY sends channel requests with the \q{want reply} flag
throughout channels' lifetime (see \k{config-ssh-bug-winadj}), it's
possible that when connecting to such a server it might receive a
reply to a request after it thinks the channel has entirely closed,
and terminate with an error along the lines of \q{Received
\cw{SSH2_MSG_CHANNEL_FAILURE} for nonexistent channel 256}.

\S{config-ssh-bug-maxpkt2} \q{Ignores SSH-2 \i{maximum packet size}}

When an SSH-2 channel is set up, each end announces the maximum size
of data packet that it is willing to receive for that channel.  Some
servers ignore PuTTY's announcement and send packets larger than PuTTY
is willing to accept, causing it to report \q{Incoming packet was
garbled on decryption}.

If this bug is detected, PuTTY never allows the channel's
\i{flow-control window} to grow large enough to allow the server to
send an over-sized packet.  If this bug is enabled when talking to a
correct server, the session will work correctly, but download
performance will be less than it could be.

\S{config-ssh-bug-dropstart} \q{Discards data sent before its greeting}

Just occasionally, an SSH connection can be established over some
channel that will accidentally discard outgoing data very early in the
connection.

This is not typically seen as a bug in an actual SSH server, but it
can sometimes occur in situations involving a complicated proxy
process. An example is
\W{https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=991958}{Debian
bug #991958}, in which a connection going over the console of a User
Mode Linux kernel can lose outgoing data before the kernel has fully
booted.

You can work around this problem by manually enabling this bug flag,
which will cause PuTTY to wait to send its initial SSH greeting until
after it sees the greeting from the server.

Note that this bug flag can never be automatically detected, since
auto-detection relies on the version string in the server's greeting,
and PuTTY has to decide whether to expect this bug \e{before} it sees
the server's greeting. So this is a manual workaround only.

\S{config-ssh-bug-filter-kexinit} \q{Chokes on PuTTY's full \cw{KEXINIT}}

At the start of an SSH connection, the client and server exchange long
messages of type \cw{SSH_MSG_KEXINIT}, containing lists of all the
cryptographic algorithms they're prepared to use. This is used to
negotiate a set of algorithms that both ends can speak.

Occasionally, a badly written server might have a length limit on the
list it's prepared to receive, and refuse to make a connection simply
because PuTTY is giving it too many choices.

A workaround is to enable this flag, which will make PuTTY wait to
send \cw{KEXINIT} until after it receives the one from the server, and
then filter its own \cw{KEXINIT} to leave out any algorithm the server
doesn't also announce support for. This will generally make PuTTY's
\cw{KEXINIT} at most the size of the server's, and will otherwise make
no difference to the algorithm negotiation.

This flag is a minor violation of the SSH protocol, because both sides
are supposed to send \cw{KEXINIT} proactively. It still works provided
\e{one} side sends its \cw{KEXINIT} without waiting, but if both
client and server waited for the other one to speak first, the
connection would deadlock. We don't know of any servers that do this,
but if there is one, then this flag will make PuTTY unable to speak to
them at all.

\S{config-ssh-bug-rsa-sha2-cert-userauth} \q{Old RSA/SHA2 cert
algorithm naming}

If PuTTY is trying to do SSH-2 user authentication using an RSA key,
and the server is using one of the newer SHA-2 based versions of the
SSH RSA protocol, and the user's key is also a certificate, then
earlier versions of OpenSSH (up to 7.7) disagree with later versions
about the right key algorithm string to send in the
\cw{SSH2_MSG_USERAUTH_REQUEST} packet. Modern versions send a string
that indicates both the SHA-2 nature and the certificate nature of the
key, such as \cq{rsa-sha2-512-cert-v01@openssh.com}. Earlier versions
would reject that, and insist on seeing
\cq{ssh-rsa-cert-v01@openssh.com} followed by a SHA-2 based signature.

PuTTY should auto-detect the presence of this bug in earlier OpenSSH
and adjust to send the right string.

\S{config-ssh-bug-sig} \q{Requires padding on SSH-2 \i{RSA} \i{signatures}}

Versions below 3.3 of \i{OpenSSH} require SSH-2 RSA signatures to be
padded with zero bytes to the same length as the RSA key modulus.
The SSH-2 specification says that an unpadded signature MUST be
accepted, so this is a bug. A typical symptom of this problem is
that PuTTY mysteriously fails RSA authentication once in every few
hundred attempts, and falls back to passwords.

If this bug is detected, PuTTY will pad its signatures in the way
OpenSSH expects. If this bug is enabled when talking to a correct
server, it is likely that no damage will be done, since correct
servers usually still accept padded signatures because they're used
to talking to OpenSSH.

This is an SSH-2-specific bug.

\S{config-ssh-bug-oldgex2} \q{Only supports pre-RFC4419 SSH-2 DH GEX}

The SSH key exchange method that uses Diffie-Hellman group exchange
was redesigned after its original release, to use a slightly more
sophisticated setup message. Almost all SSH implementations switched
over to the new version. (PuTTY was one of the last.) A few old
servers still only support the old one.

If this bug is detected, and the client and server negotiate
Diffie-Hellman group exchange, then PuTTY will send the old message
now known as \cw{SSH2_MSG_KEX_DH_GEX_REQUEST_OLD} in place of the new
\cw{SSH2_MSG_KEX_DH_GEX_REQUEST}.

This is an SSH-2-specific bug.

\S{config-ssh-bug-hmac2} \q{Miscomputes SSH-2 HMAC keys}

Versions 2.3.0 and below of the SSH server software from
\cw{ssh.com} compute the keys for their \i{HMAC} \i{message authentication
code}s incorrectly. A typical symptom of this problem is that PuTTY
dies unexpectedly at the beginning of the session, saying
\q{Incorrect MAC received on packet}.

If this bug is detected, PuTTY will compute its HMAC keys in the
same way as the buggy server, so that communication will still be
possible. If this bug is enabled when talking to a correct server,
communication will fail.

This is an SSH-2-specific bug.

\S{config-ssh-bug-pksessid2} \q{Misuses the \i{session ID} in SSH-2 PK auth}

Versions below 2.3 of \i{OpenSSH} require SSH-2 \i{public-key authentication}
to be done slightly differently: the data to be signed by the client
contains the session ID formatted in a different way. If public-key
authentication mysteriously does not work but the Event Log (see
\k{using-eventlog}) thinks it has successfully sent a signature, it
might be worth enabling the workaround for this bug to see if it
helps.

If this bug is detected, PuTTY will sign data in the way OpenSSH
expects. If this bug is enabled when talking to a correct server,
SSH-2 public-key authentication will fail.

This is an SSH-2-specific bug.

\S{config-ssh-bug-derivekey2} \q{Miscomputes SSH-2 \i{encryption} keys}

Versions below 2.0.11 of the SSH server software from \i\cw{ssh.com}
compute the keys for the session encryption incorrectly. This
problem can cause various error messages, such as \q{Incoming packet
was garbled on decryption}, or possibly even \q{Out of memory}.

If this bug is detected, PuTTY will compute its encryption keys in
the same way as the buggy server, so that communication will still
be possible. If this bug is enabled when talking to a correct
server, communication will fail.

This is an SSH-2-specific bug.

\S{config-ssh-bug-ignore1} \q{Chokes on SSH-1 \i{ignore message}s}

An ignore message (SSH_MSG_IGNORE) is a message in the SSH protocol
which can be sent from the client to the server, or from the server
to the client, at any time. Either side is required to ignore the
message whenever it receives it. PuTTY uses ignore messages to
\I{password camouflage}hide the password packet in SSH-1, so that
a listener cannot tell the length of the user's password; it also
uses ignore messages for connection \i{keepalives} (see
\k{config-keepalive}).

If this bug is detected, PuTTY will stop using ignore messages. This
means that keepalives will stop working, and PuTTY will have to fall
back to a secondary defence against SSH-1 password-length
eavesdropping. See \k{config-ssh-bug-plainpw1}. If this bug is
enabled when talking to a correct server, the session will succeed,
but keepalives will not work and the session might be more
vulnerable to eavesdroppers than it could be.

\S{config-ssh-bug-plainpw1} \q{Refuses all SSH-1 \i{password camouflage}}

When talking to an SSH-1 server which cannot deal with ignore
messages (see \k{config-ssh-bug-ignore1}), PuTTY will attempt to
disguise the length of the user's password by sending additional
padding \e{within} the password packet. This is technically a
violation of the SSH-1 specification, and so PuTTY will only do it
when it cannot use standards-compliant ignore messages as
camouflage. In this sense, for a server to refuse to accept a padded
password packet is not really a bug, but it does make life
inconvenient if the server can also not handle ignore messages.

If this \q{bug} is detected, PuTTY will assume that neither ignore
messages nor padding are acceptable, and that it thus has no choice
but to send the user's password with no form of camouflage, so that
an eavesdropping user will be easily able to find out the exact length
of the password. If this bug is enabled when talking to a correct
server, the session will succeed, but will be more vulnerable to
eavesdroppers than it could be.

This is an SSH-1-specific bug. SSH-2 is secure against this type of
attack.

\S{config-ssh-bug-rsa1} \q{Chokes on SSH-1 \i{RSA} authentication}

Some SSH-1 servers cannot deal with RSA authentication messages at
all. If \i{Pageant} is running and contains any SSH-1 keys, PuTTY will
normally automatically try RSA authentication before falling back to
passwords, so these servers will crash when they see the RSA attempt.

If this bug is detected, PuTTY will go straight to password
authentication. If this bug is enabled when talking to a correct
server, the session will succeed, but of course RSA authentication
will be impossible.

This is an SSH-1-specific bug.

\H{config-psusan} The \q{Bare \cw{\i{ssh-connection}}} protocol

In addition to SSH itself, PuTTY also supports a second protocol that
is derived from SSH. It's listed in the PuTTY GUI under the name
\q{Bare \cw{ssh-connection}}.

This protocol consists of just the innermost of SSH-2's three layers: it
leaves out the cryptography layer providing network security, and it
leaves out the authentication layer where you provide a username and
prove you're allowed to log in as that user.

It is therefore \s{completely unsuited to any network connection}.
Don't try to use it over a network!

The purpose of this protocol is for various specialist circumstances
in which the \q{connection} is not over a real network, but is a pipe
or IPC channel between different processes running on the \e{same}
computer. In these contexts, the operating system will already have
guaranteed that each of the two communicating processes is owned by
the expected user (so that no authentication is necessary), and that
the communications channel cannot be tapped by a hostile user on the
same machine (so that no cryptography is necessary either). Examples
of possible uses involve communicating with a strongly separated
context such as the inside of a container, or a VM, or a different
network namespace.

Explicit support for this protocol is new in PuTTY 0.75. As of
2021-04, the only known server for the bare \cw{ssh-connection}
protocol is the Unix program \cq{\i{psusan}} that is also part of the
PuTTY tool suite.

(However, this protocol is also the same one used between instances of
PuTTY to implement connection sharing: see \k{config-ssh-sharing}. In
fact, in the Unix version of PuTTY, when a sharing upstream records
\q{Sharing this connection at [pathname]} in the Event Log, it's
possible to connect another instance of PuTTY directly to that Unix
socket, by entering its pathname in the host name box and selecting
\q{Bare \cw{ssh-connection}} as the protocol!)

Many of the options under the SSH panel also affect this protocol,
although options to do with cryptography and authentication do not,
for obvious reasons.

I repeat, \s{DON'T TRY TO USE THIS PROTOCOL FOR NETWORK CONNECTIONS!}
That's not what it's for, and it's not at all safe to do it.

\H{config-serial} The Serial panel

The \i{Serial} panel allows you to configure options that only apply
when PuTTY is connecting to a local \I{serial port}\i{serial line}.

\S{config-serial-line} Selecting a serial line to connect to

The \q{Serial line to connect to} box allows you to choose which
serial line you want PuTTY to talk to, if your computer has more
than one serial port.

On Windows, the first serial line is called \i\cw{COM1}, and if there
is a second it is called \cw{COM2}, and so on.

This configuration setting is also visible on the Session panel,
where it replaces the \q{Host Name} box (see \k{config-hostname}) if
the connection type is set to \q{Serial}.

\S{config-serial-speed} Selecting the speed of your serial line

The \q{Speed} box allows you to choose the speed (or \q{baud rate})
at which to talk to the serial line. Typical values might be 9600,
19200, 38400 or 57600. Which one you need will depend on the device
at the other end of the serial cable; consult the manual for that
device if you are in doubt.

This configuration setting is also visible on the Session panel,
where it replaces the \q{Port} box (see \k{config-hostname}) if the
connection type is set to \q{Serial}.

\S{config-serial-databits} Selecting the number of data bits

The \q{Data bits} box allows you to choose how many data bits are
transmitted in each byte sent or received through the serial line.
Typical values are 7 or 8.

\S{config-serial-stopbits} Selecting the number of stop bits

The \q{Stop bits} box allows you to choose how many stop bits are
used in the serial line protocol. Typical values are 1, 1.5 or 2.

\S{config-serial-parity} Selecting the serial parity checking scheme

The \q{Parity} box allows you to choose what type of parity checking
is used on the serial line. The settings are:

\b \q{None}: no parity bit is sent at all.

\b \q{Odd}: an extra parity bit is sent alongside each byte, and
arranged so that the total number of 1 bits is odd.

\b \q{Even}: an extra parity bit is sent alongside each byte, and
arranged so that the total number of 1 bits is even.

\b \q{Mark}: an extra parity bit is sent alongside each byte, and
always set to 1.

\b \q{Space}: an extra parity bit is sent alongside each byte, and
always set to 0.

\S{config-serial-flow} Selecting the serial flow control scheme

The \q{Flow control} box allows you to choose what type of flow
control checking is used on the serial line. The settings are:

\b \q{None}: no flow control is done. Data may be lost if either
side attempts to send faster than the serial line permits.

\b \q{XON/XOFF}: flow control is done by sending XON and XOFF
characters within the data stream.

\b \q{RTS/CTS}: flow control is done using the RTS and CTS wires on
the serial line.

\b \q{DSR/DTR}: flow control is done using the DSR and DTR wires on
the serial line.

\H{config-telnet} The \i{Telnet} panel

The Telnet panel allows you to configure options that only apply to
Telnet sessions.

\S{config-oldenviron} \q{Handling of OLD_ENVIRON ambiguity}

The original Telnet mechanism for passing \i{environment variables} was
badly specified. At the time the standard (RFC 1408) was written,
BSD telnet implementations were already supporting the feature, and
the intention of the standard was to describe the behaviour the BSD
implementations were already using.

Sadly there was a typing error in the standard when it was issued,
and two vital function codes were specified the wrong way round. BSD
implementations did not change, and the standard was not corrected.
Therefore, it's possible you might find either \i{BSD} or \i{RFC}-compliant
implementations out there. This switch allows you to choose which
one PuTTY claims to be.

The problem was solved by issuing a second standard, defining a new
Telnet mechanism called \i\cw{NEW_ENVIRON}, which behaved exactly like
the original \i\cw{OLD_ENVIRON} but was not encumbered by existing
implementations. Most Telnet servers now support this, and it's
unambiguous. This feature should only be needed if you have trouble
passing environment variables to quite an old server.

\S{config-ptelnet} Passive and active \i{Telnet negotiation} modes

In a Telnet connection, there are two types of data passed between
the client and the server: actual text, and \e{negotiations} about
which Telnet extra features to use.

PuTTY can use two different strategies for negotiation:

\b In \I{active Telnet negotiation}\e{active} mode, PuTTY starts to send
negotiations as soon as the connection is opened.

\b In \I{passive Telnet negotiation}\e{passive} mode, PuTTY will wait to
negotiate until it sees a negotiation from the server.

The obvious disadvantage of passive mode is that if the server is
also operating in a passive mode, then negotiation will never begin
at all. For this reason PuTTY defaults to active mode.

However, sometimes passive mode is required in order to successfully
get through certain types of firewall and \i{Telnet proxy} server. If
you have confusing trouble with a \i{firewall}, you could try enabling
passive mode to see if it helps.

\S{config-telnetkey} \q{Keyboard sends \i{Telnet special commands}}

If this box is checked, several key sequences will have their normal
actions modified:

\b the Backspace key on the keyboard will send the \I{Erase Character,
Telnet special command}Telnet special backspace code;

\b Control-C will send the Telnet special \I{Interrupt Process, Telnet
special command}Interrupt Process code;

\b Control-Z will send the Telnet special \I{Suspend Process, Telnet
special command}Suspend Process code.

You probably shouldn't enable this
unless you know what you're doing.

\S{config-telnetnl} \q{Return key sends \i{Telnet New Line} instead of ^M}

Unlike most other remote login protocols, the Telnet protocol has a
special \q{\i{new line}} code that is not the same as the usual line
endings of Control-M or Control-J. By default, PuTTY sends the
Telnet New Line code when you press Return, instead of sending
Control-M as it does in most other protocols.

Most Unix-style Telnet servers don't mind whether they receive
Telnet New Line or Control-M; some servers do expect New Line, and
some servers prefer to see ^M. If you are seeing surprising
behaviour when you press Return in a Telnet session, you might try
turning this option off to see if it helps.

\H{config-rlogin} The Rlogin panel

The \i{Rlogin} panel allows you to configure options that only apply to
Rlogin sessions.

\S{config-rlogin-localuser} \I{local username in Rlogin}\q{Local username}

Rlogin allows an automated (password-free) form of login by means of
a file called \i\c{.rhosts} on the server. You put a line in your
\c{.rhosts} file saying something like \c{jbloggs@pc1.example.com},
and then when you make an Rlogin connection the client transmits the
username of the user running the Rlogin client. The server checks
the username and hostname against \c{.rhosts}, and if they match it
\I{passwordless login}does not ask for a password.

This only works because Unix systems contain a safeguard to stop a
user from pretending to be another user in an Rlogin connection.
Rlogin connections have to come from \I{privileged port}port numbers below
1024, and Unix systems prohibit this to unprivileged processes; so when the
server sees a connection from a low-numbered port, it assumes the
client end of the connection is held by a privileged (and therefore
trusted) process, so it believes the claim of who the user is.

Windows does not have this restriction: \e{any} user can initiate an
outgoing connection from a low-numbered port. Hence, the Rlogin
\c{.rhosts} mechanism is completely useless for securely
distinguishing several different users on a Windows machine. If you
have a \c{.rhosts} entry pointing at a Windows PC, you should assume
that \e{anyone} using that PC can \i{spoof} your username in
an Rlogin connection and access your account on the server.

The \q{Local username} control allows you to specify what user name
PuTTY should claim you have, in case it doesn't match your \i{Windows
user name} (or in case you didn't bother to set up a Windows user
name).

\H{config-supdup} The \i{SUPDUP} panel

The SUPDUP panel allows you to configure options that only apply
to SUPDUP sessions. See \k{using-supdup} for more about the SUPDUP
protocol.

\S{supdup-location} \q{Location string}

In SUPDUP, the client sends a piece of text of its choice to the
server giving the user's location. This is typically displayed in
lists of logged-in users.

By default, PuTTY just defaults this to "The Internet". If you want
your location to show up as something more specific, you can configure
it here.

\S{supdup-ascii} \q{Extended ASCII Character set}

This declares what kind of character set extension your terminal
supports. If the server supports it, it will send text using that
character set. \q{None} means the standard 95 printable ASCII
characters. \q{ITS} means ASCII extended with printable characters in
the control character range. This character set is documented in the
SUPDUP protocol definition. \q{WAITS} is similar to \q{ITS} but uses
some alternative characters in the extended set: most prominently, it
will display arrows instead of \c{^} and \c{_}, and \c{\}} instead of
\c{~}.  \q{ITS} extended ASCII is used by ITS and Lisp machines,
whilst \q{WAITS} is only used by the WAITS operating system from the
Stanford AI Laboratory.

\S{supdup-more} \q{**MORE** processing}

When **MORE** processing is enabled, the server causes output to pause
at the bottom of the screen, until a space is typed.

\S{supdup-scroll} \q{Terminal scrolling}

This controls whether the terminal will perform scrolling when the
cursor goes below the last line, or if the cursor will return to the
first line.

\H{config-file} \ii{Storing configuration in a file}

PuTTY does not currently support storing its configuration in a file
instead of the \i{Registry}. However, you can work around this with a
couple of \i{batch file}s.

You will need a file called (say) \c{PUTTY.BAT} which imports the
contents of a file into the Registry, then runs PuTTY, exports the
contents of the Registry back into the file, and deletes the
Registry entries. This can all be done using the Regedit command
line options, so it's all automatic. Here is what you need in
\c{PUTTY.BAT}:

\c @ECHO OFF
\c regedit /s putty.reg
\c regedit /s puttyrnd.reg
\c start /w putty.exe
\c regedit /ea new.reg HKEY_CURRENT_USER\Software\SimonTatham\PuTTY
\c copy new.reg putty.reg
\c del new.reg
\c regedit /s puttydel.reg

This batch file needs two auxiliary files: \c{PUTTYRND.REG} which
sets up an initial safe location for the \c{PUTTY.RND} random seed
file, and \c{PUTTYDEL.REG} which destroys everything in the Registry
once it's been successfully saved back to the file.

Here is \c{PUTTYDEL.REG}:

\c REGEDIT4
\c
\c [-HKEY_CURRENT_USER\Software\SimonTatham\PuTTY]

Here is an example \c{PUTTYRND.REG} file:

\c REGEDIT4
\c
\c [HKEY_CURRENT_USER\Software\SimonTatham\PuTTY]
\c "RandSeedFile"="a:\\putty.rnd"

You should replace \c{a:\\putty.rnd} with the location where you
want to store your random number data. If the aim is to carry around
PuTTY and its settings on one USB stick, you probably want to store it
on the USB stick.