[PATCH vkd3d 6/6] vkd3d-common: Rework debug infrastructure.

Matteo Bruni matteo.mystral at gmail.com
Thu May 12 09:33:37 CDT 2022


On Thu, May 12, 2022 at 4:12 PM Giovanni Mascellani
<gmascellani at codeweavers.com> wrote:
>
> Hi,
>

>
> As I see it, it's not very different from having a
> WINE_DEFAULT_DEBUG_CHANNEL in each file, just as we do with Wine.
> Actually, my suggestion would be to really transition to that model,
> which is more flexible (there can be more than one debug channel per
> file).

Right. I saw my current patches as a step in that direction. I could
certainly take care of that part too.

> Il 12/05/22 15:57, Henri Verbeet ha scritto:
> > I can't say I like this solution. In particular, sprinkling
> > VKD3D_DEBUG_ENV_NAME through the source seems undesirable. It should
> > be possible to do this in a way that allows defining the environment
> > variable only once per module.

The point of this series is to avoid global variables, which is the
cause of the VKD3D_SHADER_DEBUG -> VKD3D_DEBUG change in the vkd3d
version integrated into Wine, and which is something that I found
confusing (i.e. to get vkd3d-shader debug output you have to use
different env vars depending if you're using integrated vs separate
vkd3d) besides of the intention of just reducing the diff with
upstream vkd3d. Patches 2-5 are preparation for patch 6, moving all
the code calling the debug trace functions out of header files.

> > Separately, for the Wine integration, I
> > think it would make more sense to integrate with Wine's debug
> > facilities, so that e.g. WINEDEBUG=+d3d_shader and WINEDEBUG=+d3d12
> > simply do the right thing.

That's an interesting idea, I'll start hacking on it.



More information about the wine-devel mailing list