Re: [Emacs-diffs] master 085c7f6 2/2: Test format-time-string with zone arg

[Eli Zaretskii]

> Cc: address@hidden, address@hidden, Bruno Haible <address@hidden>,
>  Gnulib bugs <address@hidden>
> From: Paul Eggert <address@hidden>
> Date: Tue, 2 May 2017 00:52:03 -0700
> 
> > the utimens.c file you've brought from Gnulib has 2 problems:
> >
> >   . it defines WIN32_LEAN_AND_MEAN unconditionally, which conflicts
> >     with our own definition in nt/inc/ms-w32.h, which is processed
> >     earlier;
> >
> >   . it includes msvc-nothrow.h, which is absent in the repository, it
> >     should have been imported from Gnulib together with utimens.c
> >
> > I fixed the first of these temporarily, in a way that at least GCC
> > lets utimens.c compile, but I think in general utimens.c should do
> >
> >   #ifndef WIN32_LEAN_AND_MEAN
> >   # define WIN32_LEAN_AND_MEAN
> >   #endif
> >
> > to avoid possible clashes with definitions elsewhere in the sources.
> 
> I'd like Bruno's opinion on the macro clash, and on the attached patch for 
> the 
> msvc-nothrow.h issue, a patch that I installed into Gnulib and into Emacs 
> master 
> to try to get the Emacs master build working again.

Thanks.

> -#if (defined _WIN32 || defined __WIN32__) && ! defined __CYGWIN__
> +/* On native Windows, use SetFileTime; but avoid this when compiling
> +   GNU Emacs, which arranges for this in some other way and which
> +   defines WIN32_LEAN_AND_MEAN itself.  */

That "other way" is the implementation of fdutimens in Emacs's w32.c.
Perhaps Bruno could look at that implementation and comment on its
merits and demerits vs the Gnulib implementation, and we could then
decide whether to start using the Gnulib implementation, augmenting
ours, or something else.