Build break: GNU Emacs 21.3.50 (CVS) HEAD

Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dhruva Krishnamurthy]

Hello,
 There is a compilation error in GNU Emacs 21.3.50 (CVS) src/strftime.c
 (extra args in call to my_strftime() at #1491) on W2K using MinGW32-GCC
 3.2.3

with regards,
dhruva
-- 
Dhruva Krishnamurthy
Home: http://www32.brinkster.com/schemer/

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Benjamin Rutt]

"Dhruva Krishnamurthy" <seagull@fastmail.fm> writes:

>  There is a compilation error in GNU Emacs 21.3.50 (CVS) src/strftime.c
>  (extra args in call to my_strftime() at #1491) on W2K using MinGW32-GCC
>  3.2.3

I am seeing this as well right now, on the Solaris 2.8 sparc
platform.  Here is the exact error message:

    gcc -c  -Demacs -DHAVE_CONFIG_H -DUSE_LUCID  -I. -I/tmp/rutt/emacs/src -I/usr/openwin/include -I/usr/dt/include -g -O2 strftime.c
    strftime.c: In function `emacs_strftimeu':
    strftime.c:1491: too many arguments to function `_strftime_copytm'
    make[1]: *** [strftime.o] Error 1
    make[1]: Leaving directory `/tmp/rutt/emacs/src'
    make: *** [bootstrap] Error 2

The following patch gets rid of the compile error, but is the fix the
right one?  Thanks for any feedback.

Index: strftime.c
===================================================================
RCS file: /cvsroot/emacs/emacs/src/strftime.c,v
retrieving revision 1.29
diff -c -r1.29 strftime.c
*** strftime.c  24 Jun 2003 09:42:14 -0000      1.29
--- strftime.c  25 Jun 2003 22:52:56 -0000
***************
*** 1488,1493 ****
        const struct tm *tp;
        int ut;
  {
!   return my_strftime (s, maxsize, format, tp, ut, 0);
  }
  #endif
--- 1488,1493 ----
        const struct tm *tp;
        int ut;
  {
!   return my_strftime (s, maxsize, format, tp);
  }
  #endif


-- 
Benjamin

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Jan D.]

>>  There is a compilation error in GNU Emacs 21.3.50 (CVS) 
>> src/strftime.c
>>  (extra args in call to my_strftime() at #1491) on W2K using 
>> MinGW32-GCC
>>  3.2.3
>
> I am seeing this as well right now, on the Solaris 2.8 sparc
> platform.  Here is the exact error message:
>
>     gcc -c  -Demacs -DHAVE_CONFIG_H -DUSE_LUCID  -I. 
> -I/tmp/rutt/emacs/src -I/usr/openwin/include -I/usr/dt/include -g -O2 
> strftime.c
>     strftime.c: In function `emacs_strftimeu':
>     strftime.c:1491: too many arguments to function `_strftime_copytm'
>     make[1]: *** [strftime.o] Error 1
>     make[1]: Leaving directory `/tmp/rutt/emacs/src'
>     make: *** [bootstrap] Error 2
>
> The following patch gets rid of the compile error, but is the fix the
> right one?  Thanks for any feedback.

This has been added to configure.in, but configure was not regenerated:

#define my_strftime nstrftime	/* for strftime.c */

I regenerated configure, so if you rerun configure and build
again it should compile.  MS Windows don't use configure (I think)
but I imagine someone that knows how that platform works can do the
proper adjustments.  src/strftime.c explains it:

/* When compiling this file, GNU applications can #define my_strftime
    to a symbol (typically nstrftime) to get an extended strftime with
    extra arguments UT and NS.  */

	Jan D.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dave Love]

"Jan D." <jan.h.d@swipnet.se> writes:

> This has been added to configure.in, but configure was not regenerated:

Ah.  This should not be necessary.  I'm baffled why the Makefile rules
to re-run configure have been taken out, so Emacs no longer conforms
to that bit of the GNU build conventions.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Richard Stallman]

      I'm baffled why the Makefile rules
    to re-run configure have been taken out, so Emacs no longer conforms
    to that bit of the GNU build conventions.

It is not actually part of the GNU coding standards, as far as I can
see.  But the rule appears to be present in Emacs:

${srcdir}/configure: $(MAINT)
	cd ${srcdir} && autoconf

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dave Love]

Richard Stallman <rms@gnu.org> writes:

> It is not actually part of the GNU coding standards,

That's why I said `convention', but I don't know of any packages which
don't do that.  Everything built with automake does, and presumably
what that does is canonical.

> as far as I can
> see.  But the rule appears to be present in Emacs:
>
> ${srcdir}/configure: $(MAINT)
> 	cd ${srcdir} && autoconf

but disabled because of MAINT.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Andreas Schwab]

Dave Love <d.love@dl.ac.uk> writes:

|> Richard Stallman <rms@gnu.org> writes:
|> 
|> > It is not actually part of the GNU coding standards,
|> 
|> That's why I said `convention', but I don't know of any packages which
|> don't do that.  Everything built with automake does

No, not when you use AM_MAINTAINER_MODE.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dave Love]

Andreas Schwab <schwab@suse.de> writes:

> |> That's why I said `convention', but I don't know of any packages which
> |> don't do that.  Everything built with automake does
>
> No, not when you use AM_MAINTAINER_MODE.

I guess not, but AM_MAINTAINER_MODE isn't canonical.  The other 11 GNU
packages I currently have unpacked sources for don't use it.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Andreas Schwab]

Dave Love <d.love@dl.ac.uk> writes:

|> "Jan D." <jan.h.d@swipnet.se> writes:
|> 
|> > This has been added to configure.in, but configure was not regenerated:
|> 
|> Ah.  This should not be necessary.  I'm baffled why the Makefile rules
|> to re-run configure have been taken out, so Emacs no longer conforms
|> to that bit of the GNU build conventions.

Run "make MAINT=FRC" to enable those rules.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dave Love]

Andreas Schwab <schwab@suse.de> writes:

> Run "make MAINT=FRC" to enable those rules.

But why?  configure shouldn't be rebuilt unconditionally anyway; it
should depend on configure.in.  This used to just work.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Andreas Schwab]

Dave Love <d.love@dl.ac.uk> writes:

|> Andreas Schwab <schwab@suse.de> writes:
|> 
|> > Run "make MAINT=FRC" to enable those rules.
|> 
|> But why?

That's poor-man's AM_MAINTAINER_MODE.

|> configure shouldn't be rebuilt unconditionally anyway; it
|> should depend on configure.in.  This used to just work.

That creates too many problems in released tarballs or pretests for people
not having the right autotools.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Deutschherrnstr. 15-19, D-90429 Nürnberg
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dave Love]

Andreas Schwab <schwab@suse.de> writes:

> That's poor-man's AM_MAINTAINER_MODE.

But it doesn't do the same thing.  If you want AM_MAINTAINER_MODE, why
not actually use it?  At least that means you can configure it to
DTRT and leave it as long as you keep config.status intact.

> |> configure shouldn't be rebuilt unconditionally anyway; it
> |> should depend on configure.in.  This used to just work.
>
> That creates too many problems in released tarballs or pretests for people
> not having the right autotools.

How come?  I can only think that means the tarballs are built
incorrectly regarding timestamps.  If so, it could be a problem with
other things, such as .info files depending on .texi.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dave Love]

Benjamin Rutt <rutt+news@cis.ohio-state.edu> writes:

> The following patch gets rid of the compile error, but is the fix the
> right one?  Thanks for any feedback.

No.  configure makes the correct definition on all the systems I've
tried.

2003-06-23  Dave Love  <fx@gnu.org>

	* configure.in: Check for sys/_mbstate_t.h.  Test
	XRegisterIMInstantiateCallback prototype.
	(AH_BOTTOM): Define DO_BLOCK_INPUT, my_strftime.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Jan D.]

> No.  configure makes the correct definition on all the systems I've
> tried.
>
> 2003-06-23  Dave Love  <fx@gnu.org>
>
> 	* configure.in: Check for sys/_mbstate_t.h.  Test
> 	XRegisterIMInstantiateCallback prototype.
> 	(AH_BOTTOM): Define DO_BLOCK_INPUT, my_strftime.

DO_BLOCK_INPUT is defined by src/Makefile.c so why do you need it in
configure.in?

	Jan D.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dave Love]

"Jan D." <jan.h.d@swipnet.se> writes:

> DO_BLOCK_INPUT is defined by src/Makefile.c so why do you need it in
> configure.in?

I guess I was confused by the build process again, but I think that's
where the definition should be.  (I was actually going to use it
elsewhere.)  The relevant comments in and around alloca now appear to
be wrong.  I'll tidy it up sometime.

Re: Build break: GNU Emacs 21.3.50 (CVS) HEAD

[Dave Love]

"Dhruva Krishnamurthy" <seagull@fastmail.fm> writes:

> Hello,
>  There is a compilation error in GNU Emacs 21.3.50 (CVS) src/strftime.c
>  (extra args in call to my_strftime() at #1491) on W2K using MinGW32-GCC
>  3.2.3

Someone needs to sync with the change to config.in.  I don't know in
which file that should be done out of config.nt, ms-w32.h (at least)
and I can't test it.