10:54 <gst> hi, when I foot, ssh, term=foot vim, sometimes there is some garbage inserted in vim; white lines and "6 Mod key". I tried without my vimrc and still the same. This does not seem to happen with vi and nano. Any ideas?

10:56 <j`ey> does the machine youre ssh-ing into have the foot terminfo?

10:56 <gst> yes, I also tried term=foot-extra, same

10:57 <gst> and yes it is TERM not term

11:03 <gst> my fault, it is related to vim config (still searching for it). not foot issue. sorry for the noise.

11:20 <gst> it was caused by ttimeoutlen=0, that I used to fix an issue with my old setup urxvt + ssh + tmux + vim. I reverted to default timeout and there is no more garbage.

12:18 <rjarry> dnkl: with the kitty-is-text branch, my altgr modified keys do not produce any text in kitty enabled applications

12:18 <rjarry> I have a fr keymap

12:19 <rjarry> foot version: 1.16.2-78-gdf958cbb77a6 (Mar 01 2024, branch 'kitty-is-text') -pgo +ime +graphemes +assertions

12:39 <dnkl> rjarry: works for me. maybe try with a "standard" keymap instead of your custom one?

12:40 <rjarry> dnkl: what keymap are you using?

12:45 <dnkl> standard "us" and "se"

12:54 <rjarry> dnkl: after `swaymsg input type:keyboard xkb_layout fr`, nvim does not register any keypresses :(

12:55 <rjarry> dbg: input.c:1569: released: i (105/0x69), seat=seat0, term=0x158d2b0, serial=23191, mods=<none> (0x00000000), consumed=Shift+Lock+Mod5 (0x00000083), locked=Mod2 (0x00000010), repeats=0

12:55 <rjarry> sorry,

12:55 <rjarry> dbg: input.c:1569: pressed: i (105/0x69), seat=seat0, term=0x158d2b0, serial=23190, mods=<none> (0x00000000), consumed=Shift+Lock+Mod5 (0x00000083), locked=Mod2 (0x00000010), repeats=1

12:55 <rjarry> this is not registered by nvim

12:56 <rjarry> $ printf "\e[>1u" && cat -v

12:56 <rjarry> ^[[105;129u <----------- pressing the i key

12:57 <dnkl> your commit hash, from version output, doesn't match any of the commits in the kitty branch

12:57 <dnkl> make sure you're up to date with that branch

12:58 <dnkl> looks like you're on an interim commit that emitted CSIu whenever a locked modifier was enabled (NumLock)

12:58 <rjarry> oh you updated the branch again

12:58 <dnkl> the specification was then changed

12:58 <rjarry> changed again ?

12:59 <rjarry> ok, I tested with the latest kitty branch

12:59 <dnkl> first time the specification was updated, was to clarify all enabled modifiers, locked and normal, are to be included in the key event

12:59 <dnkl> that meant all keys turned into CSIu when NumLock is enabled

12:59 <dnkl> the specification was then updated again

13:00 <dnkl> to say that locked modifiers do not cause text to become CSIu

13:00 <rjarry> -_-

13:00 <rjarry> ok

13:00 <rjarry> shift+space still cause CSIu

13:00 <dnkl> i.e locked modifiers are only included if emitting a CSIu for other reasons

13:00 <rjarry> maybe because of my keymap

13:01 <dnkl> still believe that's because your shift isn't building

13:01 <dnkl> consumed

13:01 <dnkl> phone...

15:41 <rjarry> dnkl: I just tested with a "standard" keymap https://0x0.st/H7Bd.txt

15:41 <rjarry> same issue

15:57 <dnkl> well, the issue is the same. The keymap doesn't consume shift. I can't explain why. Maybe you have a different version of XKB. maybe there's something else on your system that changes the keymap somehow. maybe your distro has changed something...

16:01 <rjarry> T_T

20:14 <mvdan> hi all! back after some time, out of the loop on the whole matrix bridge stuff not working anymore

20:31 <rjarry> out of curiosity, what distro are you on?

20:34 <rjarry> I have libxkbcommon-1.6.0-1.fc39.x86_64

20:45 <rjarry> dnkl: I tested with an us keymap and numlock disabled

20:45 <rjarry> same behaviour https://0x0.st/H7uk.txt

20:45 <rjarry> shift is not consumed

20:46 <rjarry> for the record, kitty does see a real printable space character when I do the same thing, same keymap

20:49 <rjarry> here is a kitty debug output for the same key presses: https://0x0.st/H7un.txt

21:39 <qyliss> dnkl: for --pty autocomplete, we could either complete paths where PTY files usually live (/dev/pts/*, /dev/pty*, etc.), or we could just complete all paths. Do you have a preference?