bug#58535: Build failure on old GNU/Linux system

bug#58535: Build failure on old GNU/Linux system

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Since the last gnulib merge, I've been getting

make -C lib all
  CC       fingerprint.o
make[1]: Entering directory `/home/oldosfan/emacs/lib'
In file included from ../src/config.h:2877,
                 from fingerprint.c:20:
../src/conf_post.h:61: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘bool_bf’
make[1]: *** [fingerprint.o] Error 1

on an old GNU/Linux system (Fedora 12).  Paul, could you please fix
this?  Thanks in advance.

bug#58535: Build failure on old GNU/Linux system

[Paul Eggert]

On 2022-10-14 19:52, Po Lu wrote:
> make -C lib all
>    CC       fingerprint.o
> make[1]: Entering directory `/home/oldosfan/emacs/lib'
> In file included from ../src/config.h:2877,
>                   from fingerprint.c:20:
> ../src/conf_post.h:61: error: expected ‘=’, ‘,’, ‘;’, ‘asm’ or ‘__attribute__’ before ‘bool_bf’
> make[1]: *** [fingerprint.o] Error 1

Odd; it should work. Have you done a "make bootstrap", or built from a 
fresh Git checkout?

If "make bootstrap" doesn't work, it'd be helpful to see config.log and 
config.h and the output of './configure'.

bug#58535: Build failure on old GNU/Linux system

[Paul Eggert]

On 2022-10-15 01:49, Po Lu wrote:

> I built from "make bootstrap"; here's config.log and config.h:

Both are quite wrong: config.h doesn't mention HAVE_C_ALIGNASOF, 
HAVE_C_BOOL, HAVE_C_STATIC_ASSERT, etc., not even commented out.

I suspect your autom4te.cache is busted and that this confused 
autoreconf; try removing that cache directory and doing a "make 
boostrap". (I'm surprised that 'make bootstrap' doesn't remove that 
directory first thing; it should.) If that doesn't work, try doing a 
completely fresh Git checkout, ./autogen.sh; ./configure; make.

bug#58535: Build failure on old GNU/Linux system

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Paul Eggert <eggert@cs.ucla.edu> writes:

> On 2022-10-15 01:49, Po Lu wrote:
>
>> I built from "make bootstrap"; here's config.log and config.h:
>
> Both are quite wrong: config.h doesn't mention HAVE_C_ALIGNASOF,
> HAVE_C_BOOL, HAVE_C_STATIC_ASSERT, etc., not even commented out.
>
> I suspect your autom4te.cache is busted and that this confused
> autoreconf; try removing that cache directory and doing a "make
> boostrap". (I'm surprised that 'make bootstrap' doesn't remove that
> directory first thing; it should.) If that doesn't work, try doing a
> completely fresh Git checkout, ./autogen.sh; ./configure; make.

Right, that seems to have fixed the problem.  I'm closing this bug then,
sorry for the noise.

bug#58535: Build failure on old GNU/Linux system

[Paul Eggert]

[-- Attachment #1: Type: text/plain, Size: 267 bytes --]

On 2022-10-15 04:24, Po Lu wrote:
> Right, that seems to have fixed the problem.  I'm closing this bug then,
> sorry for the noise.

No problem. On the contrary I vaguely recall running into this hassle 
too, and I should have fixed it then. I installed the attached.

[-- Attachment #2: 0001-Fix-make-bootstrap-when-the-.m4-files-change.patch --]
[-- Type: text/x-patch, Size: 1618 bytes --]

From 0ff389c0c17b0b2938e79640e86b594344f20e55 Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Sat, 15 Oct 2022 11:59:11 -0700
Subject: [PATCH] =?UTF-8?q?Fix=20=E2=80=98make=20bootstrap=E2=80=99=20when?=
 =?UTF-8?q?=20the=20*.m4=20files=20change?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

This should help avoid problems like Bug#58535.
* Makefile.in (bootstrap-clean): Also remove autom4te.cache.
---
 Makefile.in | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/Makefile.in b/Makefile.in
index 2d617e2294..45b4a59e3d 100644
--- a/Makefile.in
+++ b/Makefile.in
@@ -52,14 +52,14 @@
 # make bootstrap
 #      Removes all the compiled files to force a new bootstrap from a
 #      clean slate, and then build in the normal way.  If the FAST Make
-#      variable is set, then the config.cache file isn't removed.  This
-#      allows you to say
+#      variable is set, then the autom4te.cache directory and the
+#      config.cache file are not removed.  This lets you say
 #
 #      ./configure -C
 #      make FAST=true bootstrap
 #
 #      and use the cached results from the configure run, which is much
-#      faster.
+#      faster though it does not work in general.
 #
 # make docs
 #      Make Emacs documentation files from their sources; requires makeinfo.
@@ -1040,7 +1040,7 @@ bootstrap-clean: $(distclean_dirs:
 	rm -f ${srcdir}/etc/refcards/emacsver.tex
 	rm -rf native-lisp/ lisp/leim/ja-dic/
 ifndef FAST
-	rm -f config.cache
+	rm -fr autom4te.cache config.cache
 endif
 	${top_bootclean}
 
-- 
2.34.1

bug#58535: Build failure on old GNU/Linux system

[Eli Zaretskii]

> Cc: 58535@debbugs.gnu.org
> Date: Sat, 15 Oct 2022 12:01:37 -0700
> From: Paul Eggert <eggert@cs.ucla.edu>
> 
> On 2022-10-15 04:24, Po Lu wrote:
> > Right, that seems to have fixed the problem.  I'm closing this bug then,
> > sorry for the noise.
> 
> No problem. On the contrary I vaguely recall running into this hassle 
> too, and I should have fixed it then. I installed the attached.

I think we decided not to remove the cache as part of a bootstrap,
since doing that slows down bisecting and clean builds in general,
when the configuration and the platform didn't change.

bug#58535: Build failure on old GNU/Linux system

[Eli Zaretskii]

> Cc: luangruo@yahoo.com, 58535@debbugs.gnu.org
> Date: Sat, 15 Oct 2022 22:04:50 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > Cc: 58535@debbugs.gnu.org
> > Date: Sat, 15 Oct 2022 12:01:37 -0700
> > From: Paul Eggert <eggert@cs.ucla.edu>
> > 
> > On 2022-10-15 04:24, Po Lu wrote:
> > > Right, that seems to have fixed the problem.  I'm closing this bug then,
> > > sorry for the noise.
> > 
> > No problem. On the contrary I vaguely recall running into this hassle 
> > too, and I should have fixed it then. I installed the attached.
> 
> I think we decided not to remove the cache as part of a bootstrap,
> since doing that slows down bisecting and clean builds in general,
> when the configuration and the platform didn't change.

Ignore me: that was for "FAST" bootstrap.