freeswitch: fix build error for undeclared identifier 'NSIG'

[cho-m]

• Have you followed the guidelines for contributing?
• Have you checked that there aren't other open pull requests for the same formula update/change?
• Have you built your formula locally with brew install --build-from-source <formula>, where <formula> is the name of the formula you're submitting?
• Is your test running fine brew test <formula>, where <formula> is the name of the formula you're submitting?
• Does your build pass brew audit --strict <formula> (after doing brew install <formula>)?

---

Try to fix the freeswitch error seen in PR #76780.

I confirmed error on local --build-from-source:

In file included from ./include/apr_general.h:33:
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/signal.h:69:42: error: use of undeclared identifier 'NSIG'
extern __const char *__const sys_signame[NSIG];
                                         ^
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/signal.h:70:42: error: use of undeclared identifier 'NSIG'
extern __const char *__const sys_siglist[NSIG];
                                         ^
2 errors generated.

The condition to define NSIG is:

#if !defined(_ANSI_SOURCE) && (!defined(_POSIX_C_SOURCE) || defined(_DARWIN_C_SOURCE))
#define NSIG	__DARWIN_NSIG
#endif

[SMillerDev]

ARM says:

 ==> Installing freeswitch dependency: libevent
Error: SystemCommand::ProcessTerminatedInterrupt
/opt/homebrew/Library/Homebrew/system_command.rb:211:in `block in each_output_line'

I'll just re-run it

[cho-m]

May need someone to take a deeper look into why the existing Formula ends up failing in #76780 (and #75728, which is for next version, but is probably due to same reason).

The condition that is causing the failure is also the same condition that is used to define NSIG, which should have been processed due to #include <sys/signal.h>.

This PR essentially forcefully set the value _DARWIN_C_SOURCE to guarantee NSIG is defined, but it would be a good idea to see if there is a nicer alternative (and perhaps check if the failures are due to Apple or Homebrew given that the Formula hasn't changed since originally bottled).

---

Related headers...

/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/signal.h

...
#include <sys/signal.h>

#include <sys/_pthread/_pthread_types.h>
#include <sys/_pthread/_pthread_t.h>

#if !defined(_ANSI_SOURCE) && (!defined(_POSIX_C_SOURCE) || defined(_DARWIN_C_SOURCE))
extern __const char *__const sys_signame[NSIG];
extern __const char *__const sys_siglist[NSIG];
#endif

/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/sys/signal.h

...
#define __DARWIN_NSIG   32      /* counting 0; could be 33 (mask is 1-32) */

#if !defined(_ANSI_SOURCE) && (!defined(_POSIX_C_SOURCE) || defined(_DARWIN_C_SOURCE))
#define NSIG    __DARWIN_NSIG
#endif

[carlocab]

Could we have a comment here explaining what this is for?

[cho-m]

Would something like the headline message along with related upstream issue (signalwire/freeswitch#1145) be sufficient?

---

I'm actually not sure of exact reason why this is required yet.

If I had to guess, I think it is due to _XOPEN_SOURCE being set by freeswitch in certain places, which changes value of _POSIX_C_SOURCE.

Perhaps due to order of includes, we hit the odd behavior where the same condition to set NSIG and access NSIG are not equal.

[carlocab]

Ah, macro voodoo. very nice. I guess at minimum a comment should give some indication of when a workaround can be removed, so a link to the upstream issue would help.

[cho-m]

Okay for the comment.

For actual fix, another alternative may be to set _XOPEN_SOURCE, which would do the opposite of _DARWIN_C_SOURCE, i.e. darwin macro would probably allow platform-specific behavior while open-source macro would use portable functions instead.