Re: [PATCH] Windows: only add a no-op pthread_sigmask() when needed

From: Junio C Hamano <gitster@pobox.com>
Date: May 10, 2016, 17:57 UTC
Message-ID: <xmqqvb2lpzij.fsf@gitster.mtv.corp.google.com>
In-Reply-To: <26c2fb5560246fc7f980da24a239edc333864527.1462885167.git.johannes.schindelin@gmx.de>

Johannes Schindelin <johannes.schindelin@gmx.de> writes:

Show 15 quoted lines

> In f924b52 (Windows: add pthread_sigmask() that does nothing,
> 2016-05-01), we introduced a no-op for Windows. However, this breaks
> building Git in Git for Windows' SDK because pthread_sigmask() is
> already a no-op there, #define'd in the pthread_signal.h header in
> /mingw64/x86_64-w64-mingw32/include/.
>
> Let's guard the definition of pthread_sigmask() in #ifndef...#endif to
> make the code compile both with modern MinGW-w64 as well as with the
> previously common MinGW headers.
>
> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> ---
>
> 	This patch is obviously based on 'next' (because 'master' does not
> 	have the referenced commit yet).

One thing that makes me wonder is what would happen when /mingw64/x86_64-w64-mingw32/include/pthread_signal.h changes its mind and uses "static inline" instead of "#define". How much control does Git-for-Windows folks have over that header file?

Also, could you explain "However" part a bit? Obviously in _some_ environment other than "Git for Windows' SDK", the previous patch helped you compile. And you need to #ifdef it out, because "Git for Windows' SDK" already has its own support. What is that other environment that lacks the support (hence we need our own "static inline") and is there a way to tell "Git for Windows' SDK" from that other environment?

What I am trying to get at is:

 (1) If the answer is "we have total control", then I am perfectly
     fine with using "#ifdef pthread_sigmask" here.

 (2) If not, i.e. "they can change the implementation to 'static
     inline' themselves", then I do not think it is prudent to use
     "#ifndef pthread_sigmask" as the conditional here--using a
     symbol that lets you check for that "other" environment and
     doing "#ifdef THAT_OTHER_ONE_THAT_LACKS_SIGMASK" would be
     safer.

Also is https://lists.gnu.org/archive/html/bug-gnulib/2015-04/msg00068.html relevant? Does /mingw64/x86_64-w64-mingw32/include/ implement "macro only without function"?

Show 20 quoted lines

> Published-As: https://github.com/dscho/git/releases/tag/mingw-sigmask-v1
>  compat/win32/pthread.h | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/compat/win32/pthread.h b/compat/win32/pthread.h
> index d336451..8df702c 100644
> --- a/compat/win32/pthread.h
> +++ b/compat/win32/pthread.h
> @@ -104,9 +104,11 @@ static inline void *pthread_getspecific(pthread_key_t key)
>  	return TlsGetValue(key);
>  }
>  
> +#ifndef pthread_sigmask
>  static inline int pthread_sigmask(int how, const sigset_t *set, sigset_t *oset)
>  {
>  	return 0;
>  }
> +#endif
>  
>  #endif /* PTHREAD_H */