Batch of warning fixes #18343
Batch of warning fixes #18343
Conversation
|
Hm, I'm not a super huge fan of changing code in deps/ unless it's really necessary. Right now the SPIRV-Cross code happens to exactly match a particular commit from the upstream repo and keeping it that way will make future updates easier. I feel similarly about the rest of the code in deps/; if it's necessary to get something working then yeah that makes sense (and hopefully get that upstreamed) but if it's just to satisfy a nitpicky compiler then it makes me sad. Not sure about the one in glslang, whether that fixes a legitimate bug or not. If it does, that sounds good to me (along with an upstream PR). |
|
Yeah, I need to revisit this since it also seems to have caused some regressions. |
Except for deps and the parts that are specifically pipewire includes, this compiles under pedantic and C89 for me on Linux. Mostly this was fixing issues, but in pipewire's case it involved ignoring some errors.
JoeOsborn
force-pushed
the
fix-warnings-10-30-25
branch
from
2fe84aa to
ba145c5
Compare
|
I fixed the issue (a wrong hex constant in runloop.h) and removed deps/ paths from the changes. |
|
I always worry about commits like this with the way that RetroArch has some 'features' which depend on bad string hygiene. For instance, did you know that in spite of RetroArch not having a way to edit a manual scanned playlist search (just refreshing the playlist\running the old search), there is a way? Just refresh playlist once, then go to the manual scanner. The fields will be filled. One day, one of these gcc warning commits will fix this and I will be sad. Of course, in this case, the actual remedy is to add a "edit search" button in manage playlists, but I'm sure there are more usecases i dont know about. Press capslock to turn on the fan indeed. |
|
Can you check if this unintended functionality you can get with playlists still works with this PR? If so, the road is clear to merge this, right? |
|
this commit doesn't affect the manual scanner so it doesn't matter for that. |
|
What will these |
|
Clang seems to ignore them or honor them depending on which pragma specifically. Are any of the older/weirder compilers tested through GitHub CI? |
|
You'll need to rebase this PR now |
|
0a253e3 |
|
This needs a rebase still |
|
I will, work is busy lately but I’ll try and find some time next week or in December. This could be closed for now if you want it off your queue. |
4 participants
This fixes a number of format, initialization, unused variable, etc warnings under both emscripten and GCC.