Alexander Larsson

In general, fontconfig config files are not backwards compat, and several times new versions of fontconfig has added stuff to the default config that older libfontconfig version died on. So,...

Actually, as per the comment above. I think not calling flatpak_run_add_socket_args_environment() is a bad idea.

There really is not really a physical thing like "an extension". There are only two kind of flatpak refs, "app" and "runtime". All extensions are "runtimes", which just means "things...

`org.freedesktop.Platform.GL.debug` is a bit iffy, because `org.freedesktop.Platform.GL` is already a subdir=true extension point, which means that the names of the implementers of it is `org.freedesktop.Platform.GL.*`, which conflicts with `org.freedesktop.Platform.GL.debug`. However,...

Otoh, maybe you actually want `org.freedesktop.Platform.GL.debug.mesa` so you can use a subdir=true extension point for that too. I *think* that will work. But better test it

It sounds to me like this could be caused by the GL driver in the runtime (which might be slightly different than what is used outside of flatpaks).

In my work on flatbox I also want to be able to use parts of f-b. For instance, to be able to run the build steps from individual modules in...

I merged https://github.com/flatpak/flatpak/pull/4600, so apps installing to appdata will end up being exported to metainfo in the end. It also means that the build installing in metainfo will work, but...

This is an extremely dangerous operation. Typically the files in /app (ie. the build dir) are hardlinks into the flatpak-builder build cache, and unless you use rofiles-fuse the container can...

Ok then. I think any time we do builder_cache_checkout () with options.force_copy = FALSE, we should also create a file `.hardlinked-cache` in the checkout dir, and we should then filter...