| 11 May 2025 |
 carlossless | Any ideas on if why/if the recent SDL2 changes could have introduced failures to create new SDL windows? I am experiencing the issue now with q2pro and aarach64-darwin | 07:38:48 |
 K900 | Could be an SDL2_compat bug | 07:42:15 |
| K900 | Could be platform weirdness | 07:42:21 |
| K900 | Do you have any more logs than "failures to create new windows"? | 07:42:37 |
| carlossless | K900: not at the moment, no. https://gist.github.com/carlossless/5a549302148f0ecb5448d47bc6e61588 | 07:43:57 |
| K900 | Try running it with SDL_DEBUG=1 | 07:44:57 |
| carlossless | I thought that SDL2 would still refer to SDL2_classic though | 07:45:11 |
| K900 | No, that was the entire point | 07:45:21 |
| carlossless | No, additional logs with SDL_DEBUG=1, I guess the app doesn't respect it | 07:46:46 |
| carlossless | * No additional logs with SDL_DEBUG=1, I guess the app doesn't respect it | 07:49:26 |
| carlossless | I'll try a debug build | 07:49:39 |
| K900 | Do other SDL applications work? | 07:54:54 |
| carlossless | Well I just confirmed that switching back to SDL2_classic makes it work again. So it's definitely SDL that's the issue here | 07:58:50 |
| carlossless | need to look up what uses SDL2 and has darwin as one of the supported platforms | 08:00:57 |
| carlossless | K900: just tried bzflag - same issue | 08:06:41 |
| carlossless | hm, blobwars does work though | 08:10:03 |
| carlossless | Redacted or Malformed Event | 08:13:57 |
| carlossless | Here's the SDL_CreateWindow calls for all three apps:
bzflag - https://github.com/BZFlag-Dev/bzflag/blob/f6f647cdc1bc4921bf8e58c8ec6e8b518f9e818c/src/platform/SDL2Window.cxx#L264-L272
q2pro - https://github.com/skullernet/q2pro/blob/aba81ef8bc277e9a4e11733a449a29d07ea28c7a/src/unix/video/sdl.c#L258-L259
blobwars - https://github.com/perpendicular-dimensions/blobwars/blob/1622ca7486227c8993dab2657837de51f09bad94/src/init.cpp#L272 | 08:15:10 |
| carlossless | I'm going to guess it's probably the opengl option that makes the difference here | 08:16:10 |
| K900 | I wonder if it's failing to load libGL for some reason | 08:21:45 |
| carlossless | Seems like the correct env var for logging is SDL_LOGGING, but even though setting SDL_LOGGING="*=trace" produces more output, there's no info on why it couldn't create a new window | 08:31:55 |
| carlossless | hm, the previous version which works for me was already linked against sdl2-compat, the only difference between the two links is the patch bump 2.32.52 -> 2.32.54 | 08:44:31 |
| K900 | We do have 2.32.56 in staging-next | 08:46:06 |
| K900 | You might want to try that | 08:46:08 |
| carlossless | hm, no 2.32.52 still fails | 08:56:25 |
| carlossless | although I took the last commit before the 54 upgrade, seems like there were some things changes that could've affected darwin | 09:02:56 |
| carlossless | ah no, that's wrong, I failed in my process, the commit before the 54 wasn't broken. Still confirming all of this though | 09:24:31 |
| carlossless | welp, apparently I need to rebuild everything under the sun just to get that version to build now... | 11:28:29 |
| carlossless | okay, it built, and apparently it wasn't that... building and running it from f64a287 works just fine... | 13:22:40 |
| carlossless | can't easily try 2.32.56 there's other failures in staging-next | 14:49:03 |
