ifreund changed the topic of #river to: river - a dynamic tiling wayland compositor || https://github.com/riverwm/river || channel logs: https://libera.irclog.whitequark.org/river/
notzmv has joined #river
Guest92 has joined #river
Guest92 has quit [Client Quit]
eShaev9z_ has joined #river
eShaev9z has quit [Ping timeout: 256 seconds]
elshize has quit [Ping timeout: 246 seconds]
aryak has joined #river
aryak has quit [Read error: Connection reset by peer]
waleee has quit [Ping timeout: 246 seconds]
aryak has joined #river
<angry_vincent> leon-p: does wlomp compile with clang?
gheist has quit [Remote host closed the connection]
<leon-p> it should 🤷
<leon-p> I do all my C stuff with gcc and should clang not compile it, I'll consider clang broken, not my code since I explicitly avoid GNU extensions
<angry_vincent> ok, great
<leon-p> Kinda sad really, because some of the GNU extensions are actually kinda nice
<angry_vincent> i requested wlomp added to FreeBSD but had no time for it, someone now provided the port but explicitly set gcc as compiler. Which is not default ( FreeBSD uses clang ). Is why asked
<leon-p> huh, it does not compile with clang
<angry_vincent> oh ok
<angry_vincent> so it make sense to fix :)
<leon-p> or just use gcc :]
<leon-p> it complains about the noop function, I'll have to look into that
<angry_vincent> thank you very much
<angry_vincent> i never like zoo of compilers unless it is really hard to avoid
<leon-p> Ah, apparently -Wstrict-prototypes is enabled by -Wall in clang but not gcc. Added -Wno-strict-prototypes and it works now
<leon-p> arguably it would be nicer and more correct to not use a noop function without a prototype, but that would require dedicated listener handlers for every event that I want to ignore and I am not doing that.
<angry_vincent> ok, so in CFLAGS in port file just add -Wno-strict-prototypes ?
<angry_vincent> it is in bz-attachment above
travankor has joined #river
<leon-p> yes
orfeas194 has joined #river
<leon-p> dammit, gimp can still crash river apparently
<leon-p> it's in the open-file dialog. I suspect there is some mouse-over popup or something that does something weird
pinpoxIRC has quit [Server closed connection]
pinpoxIRC has joined #river
fitrh has joined #river
leopoldek has quit [Remote host closed the connection]
traidare has joined #river
fitrh has quit [Ping timeout: 250 seconds]
gheist has joined #river
<angry_vincent> leon-p: adding -Wno-strict-prototypes still resulting in: https://bpa.st/46BQ
<leon-p> how did you add it? because it apparently never reached the compiler?
<angry_vincent> it is added into CFLAGS
<angry_vincent> it is reached compiler from what i can tell? there is -Wno-strict-prototypes in output
<leon-p> I think I might have use CFLAGS= instead of CFLAGS+= in the Makefile, maybe
<leon-p> I'll check later
hryx has quit [Server closed connection]
hryx has joined #river
gheist has quit [Remote host closed the connection]
<ifreund> the techinically correct thing is to only enable -Werror for development
<ifreund> the thing that makes warnings different from errors is that they aren't gauranteed to be compatible across compilers/versions
<leon-p> unfortunately having dedicated debug and release builds isn't quite as nice with Makefiles and C as it is with zig
traidare has quit [Ping timeout: 245 seconds]
traidare has joined #river
<angry_vincent> hmm, ok
kotto has joined #river
<angry_vincent> leon-p: what is strange too is that when you set -Werror=strict-prototypes, failure is exactly the same. which is strange
<angry_vincent> or maybe not :)
<ifreund> leon-p: RE the gimp crash, would you happen to have a coredump?
<ifreund> systemd should collect them automatically iirc
<leon-p> not a useful run, I builr river in ReleaseSafe
<leon-p> *one
<ifreund> leon-p: gdb should still give a good stack trace
<leon-p> lemme see
<ifreund> unless you strip debug info there release safe will give good traces
<leon-p> I am pretty sure I can reproduce the crash though, so I might just build in river in debug mode and trigger
<ifreund> whatever's easier, `coredumpctl gdb` and running the bt command is pretty low effort xD
<leon-p> it would be, but my last crash isn't river, it's my little load-monitor widget that always dies when the output config changes because I have been too lazy to properly do the layer-shell output dance :]
<ifreund> heh
<leon-p> since we are talking about gdb already, does anyone know how to make gdb cache the debug info files it downloads? pretty annoying to constantly re-download the massive llvm one...
Guest10 has joined #river
<leon-p> not sure how helpful that is
<leon-p> I am like 85% certain this is caused by some weird redirect override window
<ifreund> I mean, it's better than nothing xD
<leon-p> I have heard that the gimp 3.0 feature freeze is imminent. Will be nice to finally have it use GTK3 and also have good drawing tablet support
<leon-p> I'll try to reproduce once my session isn't as valuable :)
<ifreund> well, I still need to implement drawing tablet support in river xD
<leon-p> I am sorry for you and happy for myself that you stepped forward and claimed that issue :]
<ifreund> I don't think it'll be that hard really, just a bit of reading wlroots code and putting the right glue in the right places
<ifreund> #807 is much scarier
waleee has joined #river
<leon-p> welp, at least this inspired me to fix my widgets
<leon-p> would be kinda nice if a server could "migrate" a layer surface from one output to another automatically (maybe only if created with NULL as output parameter)
elshize has joined #river
kotto has quit [Quit: WeeChat 4.1.1]
notzmv has quit [Ping timeout: 256 seconds]
<leon-p> ifreund: btw, regarding the waylock issue: maybe you can't reproduce because your cursor hides on keyboard input. there is a setting in river and foot also has one
<ifreund> leon-p: hmm, I don't think I have any of the hide-cursor setting enabled but perhaps the reported does?
<ifreund> can you reproduce reliably?
<ifreund> could also be a race and I need to try my slow laptop
Misthios has quit [Quit: Ping timeout (120 seconds)]
haliucinas has quit [Read error: Connection reset by peer]
haliucinas4 has joined #river
peelz has quit [Server closed connection]
peelz has joined #river
peelz has quit [Changing host]
peelz has joined #river
Misthios has joined #river
waleee has quit [Ping timeout: 255 seconds]
elshize has quit [Ping timeout: 276 seconds]
LemanR has joined #river
LemanR has quit [Client Quit]
<mainiomano> ifreund: On my machine I only see river send a pointer enter to swaylock after I move the cursor, and it happens reliably
<ifreund> mainiomano: with a recent river master commit?
<mainiomano> From a few weeks ago maybe
<mainiomano> Seems to happen on master too
<mainiomano> I don't remember the details exactly, but for my own compositor's session locking I had to use the surface's commit event instead of the map event because the scene surface was not visible to my map hander
<mainiomano> might be relevant here too
<ifreund> urgh, this could be a signal ordering thing internal to wlroots
<ifreund> I can reproduce on my slow laptop by the way
notzmv has joined #river
Cornelius-Figgle has quit [Server closed connection]
traidare has quit [Ping timeout: 276 seconds]
Cornelius-Figgle has joined #river
<ifreund> so I think what happens here is: 1. a lock surface is created, the wlroots session lock implementation sets up a commit handler for the surface
<ifreund> 2. river is notified of the surface with the new_surface event and creates a scene surface, which also sets up a commit handler
<ifreund> 3. the client attaches and commits the first buffer
<ifreund> 4. the wlroots session lock commit handler is called since it was set up first, it emits the map event for the session lock surface
<ifreund> 5. the scene surface's commit handler is called. It modifies state to add the newly committed buffer to the scene
<ifreund> before 5 is done, the scene node will have 0 size since there is no buffer, meaning that the scene_node_at() call used to find the surface to send a pointer enter to will not find anything
<ifreund> this probably explains why using a surface commit listener instead works for mainiomano, assuming it is set up after the scene surface is created
Guest10 has quit [Quit: Client closed]
<ifreund> by the way, the reason my "fast" computer doesn't reproduce this is not because it is faster, but because it has 2 monitors xD
<ifreund> which means that it works on one of the monitors but not the other
elshize has joined #river
lbia has quit [Quit: lbia]
lbia has joined #river
leopoldek has joined #river
notzmv has quit [Ping timeout: 255 seconds]
orfeas194 has quit [Ping timeout: 246 seconds]
notzmv has joined #river
waleee has joined #river
orfeas194 has joined #river
notzmv has quit [Ping timeout: 255 seconds]
wuyoli has joined #river
traidare has joined #river
traidare has quit [Ping timeout: 255 seconds]
Guest94 has joined #river
Guest94 has quit [Quit: Client closed]
orfeas194 has quit [Quit: orfeas194]
notzmv has joined #river
ayushnix has quit [Server closed connection]
ayushnix has joined #river