XTMODKEYS Set Key Modifier Options

Choose how keys pressed with Shift, Control or Alt are sent, and ask what is chosen.

CSI > Pp ; Pv m

xterm: Functions using CSI

A key pressed with Shift, Control, Alt or Meta has to be sent somehow, and xterm has several ways. CSI > Pp ; Pv m sets resource Pp to Pv; CSI ? Pp m, XTQMODKEYS, asks for its value, and is answered as CSI > Pp ; Pv m. The > is what separates it from SGR.

`Pp`	Resource	Says how to send	Default
`0`	`modifyKeyboard`	the keyboard as a whole	`0`
`1`	`modifyCursorKeys`	the arrow keys, Home and End	`2`
`2`	`modifyFunctionKeys`	the function keys	`2`
`3`	`modifyKeypadKeys`	the keypad	`0`
`4`	`modifyOtherKeys`	everything else, such as Control and `,`	`0`
`6`	`modifyModifierKeys`	the modifier keys themselves	`0`
`7`	`modifySpecialKeys`	other special keys	`0`

For the cursor and function keys the value says where the modifier goes, -1 leaving it out and 0 to 3 putting it in a progressively more explicit place, so that Shift and Right arrow is CSI C, CSI 2 C, CSI 2 C, CSI 1 ; 2 C or CSI > 1 ; 2 C. modifyOtherKeys of 1 or 2 makes keys that normally send an ordinary character, or nothing, send CSI 27 ; modifier ; code ~ instead, so that a program can tell Control and , from ,; 2 does so for more keys than 1. Editors and shells set it to read key combinations a terminal otherwise loses (ptyx.h, modifyCursorKey).

The forms, in a default xterm (CASE_SET_MOD_FKEYS, set_mod_fkeys, report_mod_fkeys):

CSI > Pp ; Pv m sets one; CSI > Pp m, with no value, puts it back as it started; CSI > m puts back 1 to 5.
CSI > Pp n turns one off, setting it to -1; CSI > n turns off the function keys’.
CSI ? Pp m asks; several at once, CSI ? 1 ; 2 m, get an answer each. A value of -1 is reported as 65535, its sixteen bits read as unsigned, and so is a resource xterm does not have.
DECRQSS asks as well: DCS $ q > Pp m ST is answered DCS 1 $ r > Pp ; Pv m ST (misc.c).
DECSTR and RIS put them all back as they started (ReallyReset).

What a key sends cannot be seen without pressing it, but XTGETTCAP answers for a key by pretending to press it, so a case can ask it what Shift and Right arrow sends now.

libghostty-vt keeps only one thing from all this: whether modifyOtherKeys is 2, which its key encoder uses. It accepts CSI > 4 ; 2 m and turns that off on most other XTMODKEYS and on CSI > n, and ignores the rest. It does not answer XTQMODKEYS or the DECRQSS form. The cases follow a default xterm, so those that ask are known differences.

Validation

XTMODKEYS-1: The defaults

\e[?0;1;2;3;4m

|______|
reply \e[>0;0m\e[>1;2m\e[>2;2m\e[>3;0m\e[>4;0m

XTMODKEYS-2: Set modifyOtherKeys

\e[>4;2m
\e[?4m

|______|
reply \e[>4;2m

XTMODKEYS-3: Put it back

\e[>4;2m
\e[>4m
\e[?4m

|______|
reply \e[>4;0m

XTMODKEYS-4: Turned off

\e[>n     # the function keys
\e[?2m

|______|
reply \e[>2;65535m

XTMODKEYS-5: What Shift and Right arrow sends

\e[>1;3m
\eP+q6B524954\e\\  # kRIT: CSI > 1 ; 2 C

|______|
reply \eP1+r6B524954=1B5B3E313B3243\e\\

XTMODKEYS-6: Through DECRQSS

\eP$q>1m\e\\

|______|
reply \eP1$r>1;2m\e\\

XTMODKEYS-7: Not SGR

\e[>4;2m
A

|A_____|
attr 1,1 plain
reply none