Term spawning with multiple lines #1543

Rohan commented 2023年11月06日 06:00:07 +01:00

I been having this bug with foot when i spawn a new terminal everytime it occours

I been having this bug with foot when i spawn a new terminal everytime it occours

craigbarnes commented 2023年11月06日 09:09:09 +01:00

• How are you spawning the terminal?
• Which shell are you using?
• Does it still happen if you run foot and/or the shell with the default config? (e.g. foot -c /dev/null bash --noprofile --rcfile /etc/profile)

- How are you spawning the terminal? - Which shell are you using? - Does it still happen if you run foot and/or the shell with the default config? (e.g. `foot -c /dev/null bash --noprofile --rcfile /etc/profile`)

dnkl commented 2023年11月06日 09:11:26 +01:00

Guessing this is yet another duplicate of #195

Guessing this is yet another duplicate of https://codeberg.org/dnkl/foot/issues/195

Rohan commented 2023年11月06日 15:52:03 +01:00

@craigbarnes I'm using sway with bindsym $mod+Return exec foot, and i'm using ash from busybox. When i run foot -c /dev/null ash --noprofile --rcfile /etc/profile i get this output.

warn: config.c:3563: Arimo: font does not appear to be monospace; check your config, or disable this warning by setting [tweak].font-monospace-warn=no
warn: wayland.c:1509: fractional scaling not available
warn: wayland.c:1512: no server-side cursors available, falling back to client-side cursors

The font warning is normal i just changed it to the correct font but with or whitout this change i still get this problem

@craigbarnes I'm using sway with `bindsym $mod+Return exec foot`, and i'm using ash from busybox. When i run `foot -c /dev/null ash --noprofile --rcfile /etc/profile` i get this output. ``` warn: config.c:3563: Arimo: font does not appear to be monospace; check your config, or disable this warning by setting [tweak].font-monospace-warn=no warn: wayland.c:1509: fractional scaling not available warn: wayland.c:1512: no server-side cursors available, falling back to client-side cursors ``` The font warning is normal i just changed it to the correct font but with or whitout this change i still get this problem

craigbarnes commented 2023年11月06日 20:14:27 +01:00

I'd forgotten that the issue mentioned in #195 (and #453) has been known to cause this type of prompt behavior before, but @dnkl's suspicion about that seems to be correct. I can reproduce the same behavior as you observed if I run:

foot busybox ash

...whereas the prompt works as expected if I run either of the following:

foot sh -c 'sleep 0.1; busybox ash'
# Less added latency vs. using sleep; see https://codeberg.org/craigbarnes/waitwinch
foot waitwinch busybox ash

Or in other words, this is a duplicate of several previous issues and the solution is being tracked in #453.

I'd forgotten that the issue mentioned in #195 (and #453) has been known to cause this type of prompt behavior before, but @dnkl's suspicion about that seems to be correct. I can reproduce the same behavior as you observed if I run: ```sh foot busybox ash ``` ...whereas the prompt works as expected if I run either of the following: ```sh foot sh -c 'sleep 0.1; busybox ash' # Less added latency vs. using sleep; see https://codeberg.org/craigbarnes/waitwinch foot waitwinch busybox ash ``` Or in other words, this is a duplicate of several previous issues and the solution is being tracked in #453.

craigbarnes commented 2023年11月06日 20:36:59 +01:00

@dnkl mentioned the following point in #453:

... my guess is applications that exhibit sizing issues (like described in #723) are racy, and request the initial size before registering their SIGWINCH handler.

...which appears to be the case for busybox ash:

# Press Ctrl+D after the interactive shell starts
strace -o ash-trace.txt busybox ash; grep -E 'TIOCGWINSZ|SIGWINCH' ash-trace.txt

ioctl(0, TIOCGWINSZ, {ws_row=47, ws_col=191, ws_xpixel=1910, ws_ypixel=987}) = 0
ioctl(1, TIOCGWINSZ, {ws_row=47, ws_col=191, ws_xpixel=1910, ws_ypixel=987}) = 0
ioctl(1, TIOCGWINSZ, 0x7ffd1ce14df0) = -1 ENOTTY (Inappropriate ioctl for device)
ioctl(0, TIOCGWINSZ, {ws_row=47, ws_col=191, ws_xpixel=1910, ws_ypixel=987}) = 0
rt_sigaction(SIGWINCH, {sa_handler=0x4b6a1f, sa_mask=[], sa_flags=SA_RESTORER|SA_RESTART, sa_restorer=0x4d5bdb}, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0
rt_sigaction(SIGWINCH, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x4d5bdb}, NULL, 8) = 0

...or in other words, this specific issue could probably be avoided if ash set up its SIGWINCH handler before requesting the initial terminal size.

@dnkl [mentioned](https://codeberg.org/dnkl/foot/issues/453#issuecomment-266314) the following point in #453: > ... my guess is applications that exhibit sizing issues (like described in #723) are racy, and request the initial size before registering their SIGWINCH handler. ...which appears to be the case for `busybox ash`: ```sh # Press Ctrl+D after the interactive shell starts strace -o ash-trace.txt busybox ash; grep -E 'TIOCGWINSZ|SIGWINCH' ash-trace.txt ``` ``` ioctl(0, TIOCGWINSZ, {ws_row=47, ws_col=191, ws_xpixel=1910, ws_ypixel=987}) = 0 ioctl(1, TIOCGWINSZ, {ws_row=47, ws_col=191, ws_xpixel=1910, ws_ypixel=987}) = 0 ioctl(1, TIOCGWINSZ, 0x7ffd1ce14df0) = -1 ENOTTY (Inappropriate ioctl for device) ioctl(0, TIOCGWINSZ, {ws_row=47, ws_col=191, ws_xpixel=1910, ws_ypixel=987}) = 0 rt_sigaction(SIGWINCH, {sa_handler=0x4b6a1f, sa_mask=[], sa_flags=SA_RESTORER|SA_RESTART, sa_restorer=0x4d5bdb}, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, 8) = 0 rt_sigaction(SIGWINCH, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=SA_RESTORER, sa_restorer=0x4d5bdb}, NULL, 8) = 0 ``` ...or in other words, this specific issue could probably be avoided if `ash` set up its SIGWINCH handler *before* requesting the initial terminal size.

Rohan commented 2023年11月06日 22:13:10 +01:00

@craigbarnes your solution for time waiting just work for the new term that are spawned, the previous term still get a new line everytime i spawn a new terminal.

@craigbarnes your solution for time waiting just work for the new term that are spawned, the previous term still get a new line everytime i spawn a new terminal.

craigbarnes commented 2023年11月07日 00:01:08 +01:00

It's not clear what distinction you're making. What do you mean by "the previous term"?

If you're still spawning foot via bindsym $mod+Return exec foot, you'll continue to see the same problem, for the reasons outlined above (and described in #453).

It's not clear what distinction you're making. What do you mean by "the previous term"? If you're still spawning foot via `bindsym $mod+Return exec foot`, you'll continue to see the same problem, for the reasons outlined above (and described in #453).

Rohan commented 2023年11月07日 01:06:26 +01:00

@craigbarnes if you see the image i sended you can see that every terminal is like one more line than the previous, that's what i mean. Try to exec a new term and you see that the last term get a new break line a "enter command". You can see again in this image

@craigbarnes if you see the image i sended you can see that every terminal is like one more line than the previous, that's what i mean. Try to exec a new term and you see that the last term get a new break line a "enter command". You can see again in this [image](https://0x0.st/HtOu.png)

craigbarnes commented 2023年11月08日 08:34:18 +01:00

Try to exec a new term ...

Do you mean with Ctrl+Shift+N? If so, that just launches your default shell, presumably without the sleep workaround. When I mentioned the example of running:

foot sh -c 'sleep 0.1; busybox ash'

...it was just for demonstration purposes, to highlight that the bug is in ash (and/or sway) and not foot. If you want to use that same workaround for terminals spawned via Ctrl+Shift+N you need to add something like the following to ~/.config/foot/foot.ini:

shell = sh -c 'sleep 0.1; exec busybox ash'

Note that this is just an example workaround and not necessarily a suggestion. My suggestion would be to use a different shell.

> Try to exec a new term ... Do you mean with Ctrl+Shift+N? If so, that just launches your [default shell](https://codeberg.org/dnkl/foot/src/branch/master/doc/foot.ini.5.scd#L36-L40), presumably *without* the `sleep` workaround. When I mentioned the example of running: ```sh foot sh -c 'sleep 0.1; busybox ash' ```` ...it was just for demonstration purposes, to highlight that the bug is in `ash` (and/or `sway`) and not foot. If you want to use that same workaround for terminals spawned via Ctrl+Shift+N you need to add something like the following to `~/.config/foot/foot.ini`: ```ini shell = sh -c 'sleep 0.1; exec busybox ash' ``` Note that this is just an *example* workaround and **not** necessarily a suggestion. My suggestion would be to use a different shell.