Mike Blumenkrantz

Gem DPS sorting also seems to be broken with this skill.

Sorry for the delay. I can confirm that this works perfectly :thumbsup: `GTK_CHECK_VERSION()` requires 3 args though, so probably you meant to add another `, 0` in there :) thanks!

hm actually though, there is a problem here: if I set `GDK_SCALE` instead of the usual `Xft.dpi` doubling, geany doesn't pick up the increased font size for the main editor...

Hm interesting. It seems like this could be solved by making the `dev->vk` vtable const? This should enable the compiler to avoid cache misses, or so I'd imagine.

I'm still thinking on this; I like the effect, but I don't like the complexity it requires from each test case. If I can't come up with an alternative in...

Can you test with changing the declaration of `struct vk_device *dev;` to `const struct vk_device *restrict dev;` ? I think a combination of const+restrict should achieve the same value here...

I think it should be possible to do this, it's just a question of how far down the `const restrict` needs to go. Might be the case that it has...

I think refactoring there for init is going to be easier in the long run than writing/rewriting every test case to fool the compiler into generating the same code.

After looking at this in godbolt for a while, it looks like declaring `dev` as `struct vk_device *const dev` will generate the same code as this intends. I think we...

It would require some small tweaks to device creation; something like ```c static struct vk_device device; struct vk_device *const dev = &device; ``` as the variable declarations, and then changing...