| 29 Sep 2024 |
K900 | In reply to @ilya-fedin:matrix.org
Do you mean make it look at XDG_DATA_DIRS or make it to look into index.theme?
As per https://gitlab.gnome.org/GNOME/gtk/-/issues/5388 discusison it seems they have no interest in either but getting the former should be easy (basically call get_cursor_theme recursively with default as theme name before the last wl_cursor_theme_create fallback, with strcmp(name, "default") condtion).
The latter requires implementing cursor theme merging in gtk as it basically lacks it, they extracted only very minimalistic part they need from wayland-cursor.
Amazing | 14:17:51 |
K900 | Maybe we can just get some of the GNOME people to deal with this | 14:19:09 |
K900 | Because I can't | 14:19:17 |
πΌπ. π±ππ πππ π π·ππ | gtk4 also uses custom built-in fallback cursor than gtk3.
As gtk3 uses wayland-cursor, it provides the last fallback which it apparently took from xcursor, given that it has lots of functions with the same name as in libxcursor and those cursors are the same as X11 applications have in this situation (that tiny cursor I uploaded in screenshot).
gtk4, it seems, didn't copy that code from wayland-cursor, it has its own builtin fallback pixmap. It seems to be the default (left_ptr) cursor from Adwaita. It can properly scale it (or maybe almost, it's of the right size but looks blurry) which is avantage but wayland-cursor seem to have not only the default cursor but also the hand cursor at least which is disadvantage. Thus I had a trouble resizing gtk4 windows while testing it with GDK_DEBUG=portals. | 14:32:35 |
πΌπ. π±ππ πππ π π·ππ | meh, I have a very big trouble connecting to matrix and Element doesn't seem to even have proxy setitngs π | 14:33:04 |
πΌπ. π±ππ πππ π π·ππ | In reply to @k900:0upti.me Because I can't That could be really easily be solved by symlinking... I honestly don't really undertstand what's the problem with runCommand given that e.g. fonts module uses it: https://github.com/NixOS/nixpkgs/blob/master/nixos/modules/config/fonts/fontdir.nix | 14:34:23 |
πΌπ. π±ππ πππ π π·ππ | In reply to @k900:0upti.me Because I can't * That could be really easily solved by symlinking... I honestly don't really undertstand what's the problem with runCommand given that e.g. fonts module uses it: https://github.com/NixOS/nixpkgs/blob/master/nixos/modules/config/fonts/fontdir.nix | 14:36:08 |
πΌπ. π±ππ πππ π π·ππ | * That could be really easily solved by symlinking... I honestly don't really understand what's the problem with runCommand given that e.g. fonts module uses it: https://github.com/NixOS/nixpkgs/blob/master/nixos/modules/config/fonts/fontdir.nix | 14:36:21 |
K900 | In reply to @ilya-fedin:matrix.org That could be really easily solved by symlinking... I honestly don't really understand what's the problem with runCommand given that e.g. fonts module uses it: https://github.com/NixOS/nixpkgs/blob/master/nixos/modules/config/fonts/fontdir.nix But it won't be solved by symlinking | 14:54:05 |
K900 | If the path is hardcoded | 14:54:10 |
πΌπ. π±ππ πππ π π·ππ | In reply to @k900:0upti.me If the path is hardcoded the path problem is of two lines patch at least | 15:05:06 |
πΌπ. π±ππ πππ π π·ππ | diff --git a/gdk/wayland/gdkdisplay-wayland.c b/gdk/wayland/gdkdisplay-wayland.c
index 73e9af0246..13066e7cc0 100644
--- a/gdk/wayland/gdkdisplay-wayland.c
+++ b/gdk/wayland/gdkdisplay-wayland.c
@@ -1131,6 +1131,9 @@ get_cursor_theme (GdkWaylandDisplay *display_wayland,
return theme;
}
+ if (strcmp(name, "default"))
+ return get_cursor_theme (display_wayland, "default", size);
+
/* This may fall back to builtin cursors */
return wl_cursor_theme_create ("/usr/share/icons/default/cursors", size, display_wayland->shm);
}
| 17:52:09 |
| 30 Sep 2024 |
πΌπ. π±ππ πππ π π·ππ | https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/7766
https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/7767 | 09:30:03 |
πΌπ. π±ππ πππ π π·ππ | merged | 15:07:53 |
K900 | Wow those folks are fast | 15:27:37 |
πΌπ. π±ππ πππ π π·ππ | much faster than qt's gerrit xd | 15:53:28 |
| 1 Oct 2024 |
| -_o joined the room. | 20:59:57 |
| 3 Oct 2024 |
outfoxxed | @k900:0upti.me just checking in on the pr again since its been about a month now and I'm going to need it somewhat soon
| 21:14:18 |
outfoxxed | * @k900:0upti.me just checking in on the caching pr again since its been about a month now and I'm going to need it somewhat soon
| 21:14:38 |
K900 | It's on the stack | 21:15:05 |
K900 | I'm sorry | 21:15:07 |
K900 | A lot of things going on at the same time | 21:15:14 |
outfoxxed | Alright
just want to make sure it does actually happen at some point | 21:18:02 |
| 5 Oct 2024 |
K900 | @outfoxxed finally got around to the patch | 07:58:41 |
K900 | It's mostly nits in the current state | 07:58:46 |
K900 | But also | 07:58:47 |
K900 | 6.8 is going to fuck up this entire code | 07:59:06 |
K900 | 6.8 is also, uh, on Tuesday | 07:59:27 |