<w0rm>
I run foot in server mode and occasionally when running a command that spits out a lot of output the server just...quits? It doesn't seem to create a coredump so it's not quite a crash. Anyone seen this? Any suggestions for debugging?
<cbb>
w0rm: I haven't seen anything like that
<cbb>
is there a specific command that can reproduce it reliably?
<w0rm>
last happened when running httpd ab (benchmark) against a small gin/golang web server locally. It produced a lot of logging on the web app side (some colors etc). I haven't been able to find a reliable reproducer yet
<cbb>
if it's not crashing and it's logging directly to the tty...
<cbb>
it could just be outputting an escape sequence that makes the shell exit?
<w0rm>
cbb: if it does - why/how would it exit whole foot server?
<w0rm>
i.e. take all other footclient instances with it
<cbb>
w0rm: ah yeah, good point
<cbb>
it shouldn't do that
<cbb>
I was thinking in terms of foot rather than footclient
<w0rm>
I'll try to reproduce more reliably, I am on latest foot but I haven't checked deps (i.e. fcft or something else)
vyryls has quit [Ping timeout: 268 seconds]
<anarcat>
manio: i sm
<cbb>
w0rm: one thing to do would be to see if anything was logged before it died
<cbb>
e.g. with `journalctl --user SYSLOG_IDENTIFIER=foot`
ayushnix has quit [Remote host closed the connection]
ayushnix has joined #foot
<WhyNotHugo>
I keep thinking it would be very cool for a terminal to render the mouse cursor like TTYs did in the nineties. Just colour-invert the cell under the cursor and no pointer.
<WhyNotHugo>
I guess trying to render something like this in foot would be very non-trivial, right?
<dnkl>
w0rm: that means the compositor closed the Wayland socket
<dnkl>
either foot flooded it with requests
<dnkl>
or the compositor pushed events faster than foot could process
<w0rm>
yeah, I am on wlroots 0.15.1, going through 0.16.0 changelog to see if something got fixed
<dnkl>
did you do anything else when the crash happened?
<w0rm>
nothing special comes to mint, I was watching pretty logs scroll by
<w0rm>
I'll see about reproducing...
<dnkl>
well, if you can find a way to reproduce, you can run foot with WAYLAND_DEBUG=1.
<dnkl>
that should tell us what's happening
<dnkl>
w0rm: there's a known issue in 1.13 where interactively resizing (i.e with the mouse, typically) the window can trigger this, if a very large scrollback is being used.
<dnkl>
fixed in latest master, but not yet released
<w0rm>
I have a 5000 line scroll back and I've been trying to resize/move/select while everything else is going on - no crash ye
<w0rm>
t
<dnkl>
anarcat: I've been considering changing the default log-level to warning. Perhaps that'd be good enough?
ayushnix has joined #foot
ayushnix has quit [Changing host]
smach has joined #foot
<w0rm>
hmm, no luck reproducing but I am on a different machine with different GPU (amd vs intel) etc that might be related. Or it really is a weird coincidence...
smach has quit []
<anarcat>
dnkl: that would be awesome, want a PR? :)
<anarcat>
i've been meaning to dig into the code just for kicks
<dnkl>
anarcat: sure. might be a while before i merge though. not getting a lot of screen time right now
<anarcat>
for sure
<manio>
anarcat: nice lecture; i'
<anarcat>
thanks
<anarcat>
i' too
<anarcat>
;)
<manio>
i've got there from debian planet
<anarcat>
i can imagine, like a lot of my readership
<anarcat>
or most, actually
<manio>
anarcat: I had similar dilemmas regarding apps when I was switching from xorg :)
<manio>
but finally I also switched from debian to arch on my desktop
<manio>
i leave debian on server only
<anarcat>
but RSS feeds are fickle like that, you never know where people come from
<manio>
yeah
<anarcat>
which makes me think maybe i should a utm thing to my RSS feeds eh
<anarcat>
but whatever
<anarcat>
who cares right
<anarcat>
84% uknown, according to the stats, i'll throw that all on rss :)
<manio>
:)
<anarcat>
manio: i'd be curious to hear what your choices were when migrating from xorg!
<anarcat>
it looks like my zfs article didn't hit debian planet at all
<anarcat>
or it hit it with the wrong date, so it's buried all the way down there
<manio>
foot, fuzzel
<manio>
yes, exactly as you said
<manio>
because I tend to read it from where I stop
<manio>
and I see it under the timeline...
<anarcat>
oh dear, many comments to approve
<anarcat>
yeah
<anarcat>
oh well
<manio>
i was also using rofi but fuzzel is really fast comparing to rofi
<manio>
the only thing I'd like was the 'ssh plugin'
<manio>
but I can live without this
<manio>
anarcat: btw I am also using geeqie and haven't switched to some wayland alternative yet... :D
<anarcat>
manio: then you'll be glad to hear that they fixed the thumbnail bug! :)
<anarcat>
i can't live without the ssh plugin
<manio>
:)
<anarcat>
that's the biggest thing, the whole pass thing is totally bespoke on my side and i could totally port it to whatever dmenu replacement there is out there
<anarcat>
rofi ssh kicks ass, it does autocomplete not just based on my .ssh/known_hosts, but also whatever other known_hosts file i have in .ssh/config
<anarcat>
which i use a lot too
<manio>
yes, I loved that much
<anarcat>
it's actually a bug i filed against rofi, pretty happy with that, and many thanks to the rofi author for implementating that
<anarcat>
it's really too bad there was all that drama around rofi + wayland
<anarcat>
i hold the sway people at least partly responsible for that
<manio>
:|
cbb has quit [Ping timeout: 260 seconds]
<anarcat>
i guess i should make a modi-less version of that autocompletion
<anarcat>
it's just a thing that parses .ssh/config minimally and dumps a (sorted?) list of hostnames, after all
<dnkl>
anarcat: thanks! I believe there are a couple of places where we document the default value, that needs to be updated.
<dnkl>
i think --help, the man page, and probably shell completions
<dnkl>
(and no, commit messages can't be too long ;) )
<anarcat>
oh right, totally forgot about those, will grep aorund
<anarcat>
uh, i have to change the client too i guess
<dnkl>
👍 looks good
<anarcat>
awesome, thanks
<anarcat>
and i have to say, dang it's good to see emojis correctly :)
<dnkl>
😂 right
<anarcat>
back in xterm, i would recgularly copy-paste shit into the unicode(1) command, it was ridiculous
<anarcat>
"U+1F602 FACE WITH TEARS OF JOY"
<anarcat>
now i don't actually know how to *make* emojis here but who cares
<dnkl>
depending on compositor, you can use the system's IME engine. Or enable foot's Unicode key binding and then enter them as hexadecimal Unicode code points
<dnkl>
personally, I've set up fuzzel to list all emojis, and then copy it to the clipboard when selected
<anarcat>
i'm in sway, and i have no idea what i am doing
<anarcat>
ah right
<anarcat>
that works :)
<anarcat>
i'd love to see your fuzzel recipes :)
<dnkl>
yeah, I should publish them somewhere, but haven't yet
<anarcat>
that would be pretty awesome
<anarcat>
heck, it's your git repo, you could just put them straight in there no? :)
<dnkl>
absolutely, they could go in as examples :)
<anarcat>
that would be nice
<anarcat>
dnkl: i updated the PR
ayushnix has quit [Ping timeout: 260 seconds]
<anarcat>
some people are working on a debian package for fuzzel, i stoked that fire a bit to see what happens, and i'm building a package locally to see if it works