Build failure on MS-Windows

Build failure on MS-Windows

[Dani Moncayo]

Hi,

I've just tried to build from scratch the last trunk version of Emacs
on MS-Windows, but the build fails here:

[...]
Loading abbrev...
Loading simple...
Invalid function: with-eval-after-load
mv: cannot stat `emacs.exe': No such file or directory
Makefile:823: recipe for target `bootstrap-emacs.exe' failed
make[1]: *** [bootstrap-emacs.exe] Error 1
make[1]: Leaving directory `/c/msys/home/dani/emacs/build/src'
Makefile:381: recipe for target `src' failed
make: *** [src] Error 2

--
Dani Moncayo

Re: Build failure on MS-Windows

[Juanma Barranquero]

On Tue, Jun 18, 2013 at 12:54 PM, Dani Moncayo <dmoncayo@gmail.com> wrote:

> I've just tried to build from scratch the last trunk version of Emacs
> on MS-Windows, but the build fails here:

What does it mean "from scratch" in this case? I bootstrapped
yesterday and I've built several times today with no trouble
whatsoever.

   J

Re: Build failure on MS-Windows

[Dani Moncayo]

>> I've just tried to build from scratch the last trunk version of Emacs
>> on MS-Windows, but the build fails here:
>
> What does it mean "from scratch" in this case?

It means "starting with the bzr source code, without generated files", i.e.:
1. cd <source tree>.
2. bzr pull
3. ./autogen.sh.
4. cd <build tree>.
5. rm -fr *
6. CPPFLAGS='-DGLYPH_DEBUG=1 -I/c/usr/include' CFLAGS='-O0 -g3'
$source_dir/nt/msysconfig.sh --prefix=$prefix --enable-checking
7. make

The above is done by a shell script which I've successfully used for some weeks.

> I bootstrapped
> yesterday and I've built several times today with no trouble
> whatsoever.

Perhaps the mistake is mine, but I don't know what is.

--
Dani Moncayo

Re: Build failure on MS-Windows

[Juanma Barranquero]

On Tue, Jun 18, 2013 at 2:03 PM, Dani Moncayo <dmoncayo@gmail.com> wrote:

> It means "starting with the bzr source code, without generated files", i.e.:
> 1. cd <source tree>.
> 2. bzr pull
> 3. ./autogen.sh.
> 4. cd <build tree>.
> 5. rm -fr *
> 6. CPPFLAGS='-DGLYPH_DEBUG=1 -I/c/usr/include' CFLAGS='-O0 -g3'
> $source_dir/nt/msysconfig.sh --prefix=$prefix --enable-checking
> 7. make

I've done now something similar.

0) cd /c/Devel/emacs/repo/test
1) bzr clean-tree --ignore --unknown --detritus --force
2) ./autogen.sh
3) CPPFLAGS='...blah blah...' CFLAGS='..blah blah..'
./nt/msysconfig.sh  --prefix=/c/Devel/emacs/repo/test
--enable-checking=all,glyph
4) make bootstrap

No problem at all.

    J

Re: Build failure on MS-Windows

[Dmitry Gutov]

Dani Moncayo <dmoncayo@gmail.com> writes:
>>> I've just tried to build from scratch the last trunk version of Emacs
>>> on MS-Windows, but the build fails here:
>>
>> What does it mean "from scratch" in this case?
>
> It means "starting with the bzr source code, without generated files", i.e.:
> 1. cd <source tree>.
> 2. bzr pull
> 3. ./autogen.sh.
> 4. cd <build tree>.
> 5. rm -fr *
> 6. CPPFLAGS='-DGLYPH_DEBUG=1 -I/c/usr/include' CFLAGS='-O0 -g3'
> $source_dir/nt/msysconfig.sh --prefix=$prefix --enable-checking
> 7. make
>
> The above is done by a shell script which I've successfully used for some weeks.
>
>> I bootstrapped
>> yesterday and I've built several times today with no trouble
>> whatsoever.
>
> Perhaps the mistake is mine, but I don't know what is.

You need to 'make bootstrap', not just 'make'.

Re: Build failure on MS-Windows

[Glenn Morris]

Dmitry Gutov wrote:

>> 4. cd <build tree>.
>> 5. rm -fr *
[...]
> You need to 'make bootstrap', not just 'make'.

To elucidate, simply deleting the build tree does not delete the
compiled Lisp files, which get compiled in the source tree.

(Yes this is ugly; IIUC, it's essentially because the .elc files are
prebuilt in release tarfiles.)

Re: Build failure on MS-Windows

[Richard Copley]

[-- Attachment #1: Type: text/plain, Size: 1279 bytes --]

On 18 June 2013 20:40, Glenn Morris <rgm@gnu.org> wrote:

> Dmitry Gutov wrote:
>
> >> 4. cd <build tree>.
> >> 5. rm -fr *
> [...]
> > You need to 'make bootstrap', not just 'make'.
>
> To elucidate, simply deleting the build tree does not delete the
> compiled Lisp files, which get compiled in the source tree.
>
> (Yes this is ugly; IIUC, it's essentially because the .elc files are
> prebuilt in release tarfiles.)
>
>
Last I heard [http://debbugs.gnu.org/cgi/bugreport.cgi?bug=14503#8], make
bootstrap is broken for an out-of-tree build, but that might have been
fixed now.

I'm pretty sure this sequence of commands will give you a clean up-to-date
branch in all situations (I'm no bzr expert, and it might be doing too much
or not enough, but it seems to work for me):

bzr clean-tree --unknown --ignored --detritus --verbose --force
bzr revert --no-backup
bzr pull --overwrite
bzr update

Then you can run autogen.sh, msyconfig.sh and make, in MSYS.

I've just rebuilt from scratch like that and it worked just fine for me.

I did have to replace "make.exe" with the pre-release version mentioned in
INSTALL.MSYS. (The default MSYS make 3.81 had been working for me, even for
a parallel build, until recently. This time it seemed to hang towards the
end of "make all".)

[-- Attachment #2: Type: text/html, Size: 2035 bytes --]

Re: Build failure on MS-Windows

[Glenn Morris]

Richard Copley wrote:

> I'm pretty sure this sequence of commands will give you a clean up-to-date
> branch in all situations (I'm no bzr expert, and it might be doing too much
> or not enough, but it seems to work for me):
>
> bzr clean-tree --unknown --ignored --detritus --verbose --force
> bzr revert --no-backup
> bzr pull --overwrite
> bzr update
>
> Then you can run autogen.sh, msyconfig.sh and make, in MSYS.

Why not

rm -rf /path/to/emacs/
bzr branch ...

It's the only way to be _really_ sure. ;)

(I'd recommend the standard, documented make bootstrap over either of
the above methods.)

Re: Build failure on MS-Windows

[Richard Copley]

[-- Attachment #1: Type: text/plain, Size: 676 bytes --]

On 18 June 2013 21:18, Glenn Morris <rgm@gnu.org> wrote:

> Why not
>
> rm -rf /path/to/emacs/
> bzr branch ...
>
> It's the only way to be _really_ sure. ;)
>

That's ok if you don't mind waiting all week. I can't see any way
clean/revert/pull/update wouldn't work: the right files (and only the right
files) will be present, with the right contents, and will all be older than
the generated files (since they haven't been generated yet). What else
could possibly go wrong? ;)


> (I'd recommend the standard, documented make bootstrap over either of
> the above methods.)
>

Well yeah, ideally, but it doesn't work. (For me, just now, it hangs after
doing "config.status".)

[-- Attachment #2: Type: text/html, Size: 1281 bytes --]

Re: Build failure on MS-Windows

[Eli Zaretskii]

> Date: Tue, 18 Jun 2013 21:27:40 +0100
> From: Richard Copley <rcopley@gmail.com>
> Cc: Juanma Barranquero <lekktu@gmail.com>,
> 	Emacs development discussions <emacs-devel@gnu.org>,
> 	Dani Moncayo <dmoncayo@gmail.com>, Dmitry Gutov <dgutov@yandex.ru>
> 
> > Why not
> >
> > rm -rf /path/to/emacs/
> > bzr branch ...
> >
> > It's the only way to be _really_ sure. ;)
> >
> 
> That's ok if you don't mind waiting all week.

??? "bzr branch" is a local operation in this case, which takes about
20 seconds on my XP machine.

Re: Build failure on MS-Windows

[Richard Copley]

[-- Attachment #1: Type: text/plain, Size: 809 bytes --]

On 19 June 2013 03:54, Eli Zaretskii <eliz@gnu.org> wrote:

> > Date: Tue, 18 Jun 2013 21:27:40 +0100
> > From: Richard Copley <rcopley@gmail.com>
> > Cc: Juanma Barranquero <lekktu@gmail.com>,
> >       Emacs development discussions <emacs-devel@gnu.org>,
> >       Dani Moncayo <dmoncayo@gmail.com>, Dmitry Gutov <dgutov@yandex.ru>
> >
> > > Why not
> > >
> > > rm -rf /path/to/emacs/
> > > bzr branch ...
> > >
> > > It's the only way to be _really_ sure. ;)
> > >
> >
> > That's ok if you don't mind waiting all week.
>
> ??? "bzr branch" is a local operation in this case, which takes about
> 20 seconds on my XP machine.
>

"rm -rf /path/to/emacs" deletes my entire local repository, because I've
been doing it all wrong, as recommended here:
  http://www.emacswiki.org/emacs/BzrForEmacsCasualDevs#toc2

[-- Attachment #2: Type: text/html, Size: 1703 bytes --]

Re: Build failure on MS-Windows

[Eli Zaretskii]

> Date: Wed, 19 Jun 2013 07:33:58 +0100
> From: Richard Copley <rcopley@gmail.com>
> Cc: Glenn Morris <rgm@gnu.org>, Juanma Barranquero <lekktu@gmail.com>, 
> 	Emacs Development <emacs-devel@gnu.org>, Dani Moncayo Melgar <dmoncayo@gmail.com>, 
> 	Brief Busters <dgutov@yandex.ru>
> 
> > > > rm -rf /path/to/emacs/
> > > > bzr branch ...
> > > >
> > > > It's the only way to be _really_ sure. ;)
> > > >
> > >
> > > That's ok if you don't mind waiting all week.
> >
> > ??? "bzr branch" is a local operation in this case, which takes about
> > 20 seconds on my XP machine.
> >
> 
> "rm -rf /path/to/emacs" deletes my entire local repository, because I've
> been doing it all wrong, as recommended here:
>   http://www.emacswiki.org/emacs/BzrForEmacsCasualDevs#toc2

I hope you are using http://www.emacswiki.org/emacs/BzrForEmacsDevs
now, or will from now on.

Re: Build failure on MS-Windows

[Eli Zaretskii]

> Date: Tue, 18 Jun 2013 12:54:41 +0200
> From: Dani Moncayo <dmoncayo@gmail.com>
> 
> I've just tried to build from scratch the last trunk version of Emacs
> on MS-Windows, but the build fails here:
> 
> [...]
> Loading abbrev...
> Loading simple...
> Invalid function: with-eval-after-load
> mv: cannot stat `emacs.exe': No such file or directory
> Makefile:823: recipe for target `bootstrap-emacs.exe' failed
> make[1]: *** [bootstrap-emacs.exe] Error 1
> make[1]: Leaving directory `/c/msys/home/dani/emacs/build/src'
> Makefile:381: recipe for target `src' failed
> make: *** [src] Error 2

with-eval-after-load is a macro defined in subr.el, which is loaded by
loadup before simple.el.  Therefore, this problem should not happen.
Perhaps you have a stale subr.elc lurking somewhere, or maybe that
"bzr pull" didn't complete normally, and left an old version of
subr.el, or you have merge conflicts in subr.el (what does "bzr st"
say?).