Re: master f6fb2b9e1f: Port better to C23 bool+true+false keywords

Re: master f6fb2b9e1f: Port better to C23 bool+true+false keywords

[Po Lu]

One of the gnulib changes results in this on MS-DOS:

In file included from ./stdlib.h:100,
                 from ./stdlib.h:36,
                 from getopt.c:27:
./unistd.h:663:3: #error "Please include config.h first."
In file included from ./stdlib.h:100,
                 from ./stdlib.h:36,
                 from getopt.c:27:
./unistd.h:1058: error: syntax error before "extern"
In file included from ./stdlib.h:36,
                 from getopt.c:27:
./stdlib.h:804: error: syntax error before "extern"
getopt.c:381: error: syntax error before "int"
getopt.c: In function `_getopt_initialize':
getopt.c:388: error: `d' undeclared (first use in this function)
getopt.c:388: error: (Each undeclared identifier is reported only once
getopt.c:388: error: for each function it appears in.)
getopt.c:395: error: `optstring' undeclared (first use in this function)
getopt.c:405: error: `posixly_correct' undeclared (first use in this function)

that is after I added

/^ASSERT_H *=/s/@[^@\n]*@/alloca.h/

and

s/= @GL_GENERATE_ASSERT_H_CONDITION@/= 0/

to sedlibmk.inp, and

/^#undef HAVE_STDBOOL_H/#define HAVE_STDBOOL_H/

to sed2v2.inp.  AFAICT, both headers are present in DJGPP.

_LIBC is also not defined, so getopt is really including config.h.  Any
ideas?

Re: master f6fb2b9e1f: Port better to C23 bool+true+false keywords

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: Paul Eggert <eggert@cs.ucla.edu>
> Date: Wed, 28 Sep 2022 19:15:20 +0800
> 
> that is after I added
> 
> /^ASSERT_H *=/s/@[^@\n]*@/alloca.h/
                            ^^^^^^^^
Why alloca.h? shouldn't it be assert.h instead?

Re: master f6fb2b9e1f: Port better to C23 bool+true+false keywords

[Po Lu]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Po Lu <luangruo@yahoo.com>
>> Cc: Paul Eggert <eggert@cs.ucla.edu>
>> Date: Wed, 28 Sep 2022 19:15:20 +0800
>> 
>> that is after I added
>> 
>> /^ASSERT_H *=/s/@[^@\n]*@/alloca.h/
>                             ^^^^^^^^
> Why alloca.h? shouldn't it be assert.h instead?

Yeah, sorry, that was a typo in the email (not the actual code I wrote.)

Re: master f6fb2b9e1f: Port better to C23 bool+true+false keywords

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: emacs-devel@gnu.org,  eggert@cs.ucla.edu
> Date: Thu, 29 Sep 2022 08:17:51 +0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> From: Po Lu <luangruo@yahoo.com>
> >> Cc: Paul Eggert <eggert@cs.ucla.edu>
> >> Date: Wed, 28 Sep 2022 19:15:20 +0800
> >> 
> >> that is after I added
> >> 
> >> /^ASSERT_H *=/s/@[^@\n]*@/alloca.h/
> >                             ^^^^^^^^
> > Why alloca.h? shouldn't it be assert.h instead?
> 
> Yeah, sorry, that was a typo in the email (not the actual code I wrote.)

If the above doesn't fix the compilation warnings/errors, please post
the expanded source around the problematic location(s).  The messages
alone cannot be interpreted without seeing the source which triggered
them.

Re: master f6fb2b9e1f: Port better to C23 bool+true+false keywords

[Paul Eggert]

On 9/28/22 04:15, Po Lu wrote:
> _LIBC is also not defined, so getopt is really including config.h.  Any
> ideas?

This part looks OK, as _LIBC is not supposed to be defined, and 
lib/getopt.c should include config.h in an Emacs build. (_LIBC is 
defined only when compiling glibc itself.)

I think Eli addressed the assert.h issue.