Why not math.h to follow other headers?

Thanks for the patch, glad someone who can reproduce the issues picking this up.

Updating sdl is a valid workaround and probably desirable on it's own, still removing the offending -Werror might be preferable or this issue will crop up sooner or later again.

If this is your first contribution you should also add yourself to the contributors file.

In D5092#216526, @vladislavbelov wrote:

Why not math.h to follow other headers?

I have tried with headers only but I get `'cmath.h' file not found. An old issue at https://github.com/actions/runner-images/issues/2274 has some pointers but I given comments from @sera, I'll see about removing the definition instead of trying to track this down.

In D5092#216530, @jpshack wrote:

I have tried with headers only but I get `'cmath.h' file not found. An old issue at https://github.com/actions/runner-images/issues/2274 has some pointers but I given comments from @sera, I'll see about removing the definition instead of trying to track this down.

cmath is a C++ library (cstdio, climits, etc), C pair is math.h (stdio.h, limits.h, etc).

Updating sdl is a valid workaround and probably desirable on it's own, still removing the offending -Werror might be preferable or this issue will crop up sooner or later again.

For more context a similar comment is made in the original issue:

SDL adds "-Wdeclaration-after-statement -Werror=declaration-after-statement" to CFLAGS, so it's not about the C standard used but that likely this diagnostic became supported on recent Apple clang (what I suspect is used here) and so now the build fails.

Updating SDL2 certainly isn't bad but sed-ing / patching out all forced Werror should be done anyway.

Do I need to address the above for this patch to be accepted?

If you are comfortable in patching out Werror then go ahead, otherwise the patch LGTM, tho can't test it as I don't have a mac.

Also you should add yourself to binaries/data/mods/public/gui/credits/texts/programming.json

I had this same issue building 0ad.
The patch allows me to build all the necessary libraries, but I am left with the following error when i run update_workspaces.sh.

==== Building zip-lib (release) ====
==== Building lua-lib (release) ====
==== Building zlib-lib (release) ====
Creating obj/Release/zlib-lib
adler32.c
mkstemp.c
compress.c
../../contrib/libzip/mkstemp.c:76:8: error: call to undeclared function 'getpid'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
        pid = getpid();
              ^
1 error generated.
make[1]: *** [obj/Release/zip-lib/mkstemp.o] Error 1
make: *** [zip-lib] Error 2
make: *** Waiting for unfinished jobs....
crc32.c
deflate.c
gzclose.c
gzlib.c
gzread.c
gzwrite.c
infback.c
inffast.c
inflate.c
inftrees.c
trees.c
uncompr.c
zutil.c
Linking zlib-lib
ERROR: Premake build failed

In D5092#216546, @real_tabasco_sauce wrote:

The patch allows me to build all the necessary libraries, but I am left with the following error when i run update_workspaces.sh.

Yes. I am seeing exactly this as well and I have not figured it out yet. I have opened a new ticket #6847 to track this next issue.

@real_tabasco_sauce I have now resolved this error in https://code.wildfiregames.com/D5093.

See also D5093.

Would be good to have Stan's opinion on this, I believe he's the most up-to-date on the Fcollada 'port' that we essentially maintain.

The SDL upgrade looks reasonable to me, but it's one of the more 'dangerous' libraries to upgrade so will need to test this.

I think the Fcollada fix is okay, we could ask the guy maintaining the rdb fork what he thinks about it.
If @vladislavbelov thinks it's okay too I have no objections.

Regarding the SDL is this the minimum working version?
We should check whether they did not add a lot of flags to turn off new features that might interfere with the game. Also work started on SDl3