Include <signal.h> in rgbgfx_test.cpp

[Rangi42]

This is apparently necessary for BSD, and harmless for Linux. (It already gets pid_t from some other header, but POSIX has it in signal.h.)

@nunotexbsd This should mean that in 0.9.1 you can remove that patch file. :)

[ISSOtm]

This is surprising, since Linux + FreeBSD + OpenBSD's man pages document <spawn.h> as being sufficient for posix_spawn, which includes pid_t in its prototype. Further, Linux's man page says pid_t is defined in <sys/types.h>, and POSIX says that:

The <spawn.h> header shall define the mode_t and pid_t types as described in <sys/types.h>.

@nunotexbsd, do you know why that extra header is necessary?

[nunotexbsd]

I was to do a pull request today

@nunotexbsd, do you know why that extra header is necessary?

No

[ISSOtm]

I mean, why did you have to add that patch to the port?

[nunotexbsd]

I mean, why did you have to add that patch to the port?

build fails without it

[Rangi42]

What is the exact message that it fails with? How did you choose to add that header on that line to fix it?

[nunotexbsd]

What is the exact message that it fails with? How did you choose to add that header on that line to fix it?

build faild with missing symbols, missing header

I can send log laters when I get back to my PC

[nunotexbsd]

I took info from the man:
https://man.freebsd.org/cgi/man.cgi?query=siginfo&apropos=0&sektion=0&manpath=FreeBSD+14.2-RELEASE+and+Ports&arch=default&format=html

FAILED: test/CMakeFiles/rgbgfx_test.dir/gfx/rgbgfx_test.cpp.o
/usr/bin/c++  -I/wrkdirs/usr/ports/devel/rgbds/work/rgbds/include -I/usr/local/include/libpng16 -O2 -pipe -fstack-protector-strong -fno-strict-aliasing -O2 -pipe -fstack-protector-strong -fno-strict-aliasing   -D
NDEBUG -std=gnu++20 -Wall -pedantic -Wno-gnu-zero-variadic-macro-arguments -MD -MT test/CMakeFiles/rgbgfx_test.dir/gfx/rgbgfx_test.cpp.o -MF test/CMakeFiles/rgbgfx_test.dir/gfx/rgbgfx_test.cpp.o.d -o test/CMakeFi
les/rgbgfx_test.dir/gfx/rgbgfx_test.cpp.o -c /wrkdirs/usr/ports/devel/rgbds/work/rgbds/test/gfx/rgbgfx_test.cpp
/wrkdirs/usr/ports/devel/rgbds/work/rgbds/test/gfx/rgbgfx_test.cpp:310:2: error: unknown type name 'siginfo_t'
  310 |         siginfo_t info;
      |         ^
/wrkdirs/usr/ports/devel/rgbds/work/rgbds/test/gfx/rgbgfx_test.cpp:313:29: error: use of undeclared identifier 'CLD_EXITED'
  313 |         } else if (info.si_code != CLD_EXITED) {
      |                                    ^
/wrkdirs/usr/ports/devel/rgbds/work/rgbds/test/gfx/rgbgfx_test.cpp:319:23: error: use of undeclared identifier 'CLD_DUMPED'
  319 |                     info.si_code == CLD_DUMPED ? " (core dumped)" : "",
      |                                     ^
3 errors generated.

Full log:
rgbds-0.9.0.log

[Rangi42]

rgbgfx_test.cpp:310:2: error: unknown type name 'siginfo_t'

POSIX <sys/wait.h> marks siginfo_t as an "XSI extension".
POSIX <spawn.h> also says that "inclusion [...] may make visible symbols defined in the <sched.h>, <signal.h>, and <sys/types.h> headers", but does not guarantee it.
On the other hand, POSIX <signal.h> guarantees to define it.

rgbgfx_test.cpp:313:29: error: use of undeclared identifier 'CLD_EXITED'
rgbgfx_test.cpp:319:23: error: use of undeclared identifier 'CLD_DUMPED'

These are listed specifically in POSIX <signal.h>. Presumably that header is being made visible via one of our Linux headers, but not on BSD. So I think this PR is appropriate.

[ISSOtm]

Oh, I see: the transitive exposition was made mandatory in POSIX.1-2024! So yeah, let's do this.