On Mon, 27 Jan 2020 at 18:27, Jeff Smith whydoubt@gmail.com wrote:
On Mon, Jan 27, 2020 at 5:47 AM Henri Verbeet hverbeet@gmail.com wrote:
I don't think any testbot machine is running VMware SVGA. In particular, newtb-wxppro is running "vga.dll". There are test reports on test.winehq.org using VMware SVGA (e.g. fg-winxp-lusr), but those aren't from the testbot.
Ah yes, I wasn't think about that box being on test but not testbot. So I take it you do not wish to accommodate for machines that are in "test" but not "testbot"?
It's not quite that black and white. We'd certainly accommodate common AMD/NVIDIA/Intel hardware. Even for VMware SVGA, if someone cared enough about it, it wouldn't necessarily be against it. (But you'd probably have to do something involving ddraw_is_vmware() instead.) Nevertheless, results on VMware SVGA have very little relation to what ddraw applications written at the time may or may not expect drivers to do; it's just not representative. And ultimately the goal of running the tests on Windows is to find out how applications expect the API to behave.
As an aside on the Jan 23 test, it appears that the "VBoxDisp.dll":"VirtualBox Graphics Adapter" driver exhibits the same failures (as well as others).
Similar issue.
Also, the "igxprd32.dll":"Mobile Intel(R) 945 Express Chipset Family" has a very different set of flags and fails several test_caps checks. It makes me wonder if test_caps, needs more significant help.
Stefan was looking at test_caps() recently and may be able to help.