Rebuilt my environments and managed to get wine-7.0-rc1/2 to compile using the proposed patch.
However here’s what I’m seeing within my 10.9VM when using winemac.drv Currently rebuilding with X11 support to see if the issue is just with winemac.drv.
The last version I’d verified (6.0.2 & 6.11) functioned correctly within this 10.9VM
Copied the Van compile onto the real install (Mojave) and winecfg launched without issue.
On Wed, Dec 15, 2021 at 9:58 PM Dean Greer gcenx83@gmail.com wrote:
I figured as much but didn’t have chance to really dig into it to verify just figured it was wine-6.17 where winemac.drv broken.
On Wed, Dec 15, 2021 at 9:55 PM Chip Davis cdavis@codeweavers.com wrote:
December 15, 2021 7:17 PM, "Keno Fischer" keno@juliacomputing.com wrote:
Hi Dean,
sorry, I missed your follow up messages to this.
Even with this patch wine-7.0-rc1 without Metal fails to compile
Hmm, that's odd. The error you're seeing is precisely the one I was attempting to address with this patch. As you can see in [1], we did successfully build wine for Darwin with this patch.
As this framework was never linked prior outside of a Metal build this patch is invalid
Ok, I don't know the history here. Getting rid of QuartzCore in the non-Metal build would of course address the issue as well. I'll drop this patch then and wait for a proper fix.
I'm afraid this is all my fault. In a5cf847aa45c38bae1500ef19a9bac38bbd8d5ba, I turned on layer-backed views in winemac.drv. In 01f027b2db3196a8fe43be3d6718227794362985, I moved to using a mask layer for window regions. These introduced a dependency on QuartzCore that was not present before. I am sorry--this was an oversight on my part. We really do want to link to QuartzCore unconditionally.
Chip
