Dear CRUX port maintainers,
as I shared in #crux-devel earlier today, the recent update to
adwaita-icon-theme dropped many of the stock icons. A number of gtk3
apps in our port collections still try to load these stock icons using
deprecated code (`grep -n ^GDK_DEPRECATED /usr/include/gtk-3.0/gtk/*.h`
will help locate some of these legacy functions). Also affected is at
least one of the ports in /usr/ports/opt (geeqie).
There's already an issue on the gtk gitlab
(https://gitlab.gnome.org/GNOME/gtk/-/issues/4790) so perhaps a fix
will soon be available from upstream. In the meantime, I've been trying
without much success to patch the problematic code in my affected ports
using basic sed substitutions. But even after replacing these function
calls with the newer (post-3.10) syntax, I still got a Gtk-WARNING about
failure to locate the requested stock icons. I think it's not enough to
change the function calls superficially, because the icons themselves
are not locatable using the same const char arguments as before. A port
maintainer would have to write a dictionary to help "carry out the
relevant name conversions", which is less viable than just pushing the
issue upstream.
If anyone else is maintaining ports affected by the icon theme upgrade,
what would you suggest as an interim measure (until the issue is
addressed upstream)? Leave a README in the port tree warning users about
a degraded UI unless adwaita-icon-theme is reverted to its previous
version? Post a new issue with the developers of our affected ports
suggesting that legacy icon-loading code be replaced by its
GTK3-3.24.10+ counterpart? In light of the last post in the XFCE mailing
list thread shared by dlcusa (http://dlcusa.net/CRUX/XFCE_Wayland.mbox)
I worry that by pursuing the latter option we dodge the issue of
recruiting caretakers for an abandoned GTK3 branch, as the GTK
project focuses more of its effort on version 4 and GNOME-specific apps.
It would seem less wasteful of effort to insert the fix directly into
the gtk codebase (in the manner suggested by Rich's response to Andrzej
in the XFCE thread, or by ChrisVine in the gtk gitlab) than to
distribute among many app developers the task of chasing a moving target
(i.e., the set of functions not yet deprecated by the current stable
release of gtk3). But aggregate amount of developer effort is not the
only relevant consideration, so I'm open to hearing counterarguments
that might persuade me to file bug reports with the authors of my
affected ports.
Thanks for your thoughts.
-- John
