[tz] make error

[Guy Harris]

On Nov 16, 2023, at 1:33 PM, Brian Inglis via tz <tz at iana.org> wrote:

> Nobody else admits to running such old compilers and systems! ;^>
> 
> I would reduce that test to:
> 
> #if __STDC_VERSION__ < 199901
> # define restrict
> #endif

Or perhaps

#if __STDC_VERSION__ < 199901 && !defined restrict
# define restrict
#endif

so as not to redefine it.

If that doesn't work, what is the environment Alois is using defining restrict as?

> In general for standards feature tests:
> 
> #if defined(__STDC__)
> # define C89
> # if defined(__STDC_VERSION__)
> #  define C90
> #  if (__STDC_VERSION__ >= 199409L)
> #   define C94
> #  endif

	...

> #  if (__STDC_VERSION__ >= 202000L) /* TBD */
> #   define C2X
> #  endif
> # endif
> #endif

...and hope that the compiler doesn't define __STDC_VERSION__ for a particular version of the standard without implementing *all* of the compiler-implemented parts of the standard.

The *other* parts, implemented by header files and libraries, might also be a problem, e.g. installing a Shiny New GCC on a system where 1) the C library is provided by the operating system rather than the compiler and 2) the OS has an old library that, for example, doesn't support all the Shiny New Features of *printf formats.  If the compiler claims Cnn support but the library only offers Cmm support, where mm is an earlier year than nn, relying on __STDC_VERSION__ won't work as a test for missing library features.

> For updates see:
> 
> $ info cpp macros predef standard

	$ info cpp macros predef standard
	-bash: info: command not found