[PATCH] mscoree: Use builtin XNA implementation by default.

[Esme Povirk]

Wine-bug: https://bugs.winehq.org/show_bug.cgi?id=32877
Signed-off-by: Esme Povirk 
---
Proton has been using this setting for a while, and I think it's
shown that FNA works very well as an XNA replacement. This is
clearly the best option for users, and it can still be
overridden with an environment variable or registry key if a
developer wants to work on supporting the XNA redist.

dlls/mscoree/metahost.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/dlls/mscoree/metahost.c b/dlls/mscoree/metahost.c
index 77029942528..e854ec2bdfa 100644
--- a/dlls/mscoree/metahost.c
+++ b/dlls/mscoree/metahost.c
@@ -1502,6 +1502,10 @@ static DWORD get_basename_search_flags(const char *basename, MonoAssemblyName *a
         return reg_entry.flags;
     }

+    if (strcmp(basename, "Microsoft.Xna.Framework.*") == 0)
+        /* XNA redist is broken in Wine Mono, use FNA instead. */
+        return 0;
+
     return ASSEMBLY_SEARCH_UNDEFINED;
 }

[Marvin]

Hi,

While running your changed tests, I think I found new failures.
Being a bot and all I'm not very good at pattern recognition, so I might be
wrong, but could you please double-check?

Full results can be found at:
https://testbot.winehq.org/JobDetails.pl?Key=76942

Your paranoid android.

=== debiant (64 bit WoW report) ===

mscoree:
comtest.c:318: Test failed: Timed out waiting for the child process
comtest.c:294: Test failed: DeleteFileA failed: 5
comtest.c:139: Test succeeded inside todo block: Got 0
mscoree: Timeout

[Esme Povirk (they/them)]

Not sure what happened here. I uploaded the patch manually to testbot
and didn't get any failures:
https://testbot.winehq.org/JobDetails.pl?Key=76997

[Francois Gouget]

On Wed, 12 Aug 2020, Esme Povirk (they/them) wrote:
> Not sure what happened here. I uploaded the patch manually to testbot
> and didn't get any failures:
> https://testbot.winehq.org/JobDetails.pl?Key=76997

[...]
> comtest.c:318: Test failed: Timed out waiting for the child process
> comtest.c:294: Test failed: DeleteFileA failed: 5
> comtest.c:139: Test succeeded inside todo block: Got 0
> mscoree: Timeout

These are intermittent failures. So far they have only happened twice, 
once in the debian10-wow32 configuration and another in the 
debian10-win32-fr-FR one. One might have thought that they were 
specific to Debian 10 but apparently not.

Note that this wait has a 30 second timeout so it does not look like a 
minor execution delay. Also, doesn't the permission denied error for 
DeleteFile() mean the dll is still in use? In that case the issue would 
be with why run_test() did not complete.