bug#17629: Can't build emacs-24 branch

bug#17629: Can't build emacs-24 branch

[Jorgen Schaefer]

Building the current emacs-24 branch fails for me.

Latest commit (git ref, sorry, no bzr here):

3302b06 Do not mention individual ns options in the manual

In a fresh checkout and with a clean tree:

$ ./autogen.sh && ./configure --without-x && make clean && make

[...]

make[2]: Leaving directory `/home/forcer/Programs/Emacs/trunk/lisp'
if test "no" = "yes"; then \
          rm -f bootstrap-emacs; \
          ln temacs bootstrap-emacs; \
        else \
          ./temacs --batch --load loadup bootstrap || exit 1; \
          test "X" = X ||  -zex emacs; \
          mv -f emacs bootstrap-emacs; \
        fi
Loading loadup.el (source)...
Using load-path (/home/forcer/Programs/Emacs/trunk/lisp /home/forcer/Programs/Emacs/trunk/lisp/emacs-lisp /home/forcer/Programs/Emacs/trunk/lisp/language /home/forcer/Programs/Emacs/trunk/lisp/international /home/forcer/Programs/Emacs/trunk/lisp/textmodes /home/forcer/Programs/Emacs/trunk/lisp/vc)
Loading emacs-lisp/byte-run (compiled; note, source file is newer)...
Loading emacs-lisp/backquote...
Loading subr (compiled; note, source file is newer)...
Loading version...
Loading widget...
Loading custom...
Loading emacs-lisp/map-ynp...
Loading international/mule...
Loading international/mule-conf...
Loading env...
Loading format...
Loading bindings...
Loading cus-start (compiled; note, source file is newer)...
Note, built-in variable `report-emacs-bug-address' not bound
Loading window (compiled; note, source file is newer)...
Loading files (compiled; note, source file is newer)...
Loading emacs-lisp/macroexp (compiled; note, source file is newer)...
Loading cus-face (compiled; note, source file is newer)...
Loading faces (compiled; note, source file is newer)...
Loading button...
Loading startup (compiled; note, source file is newer)...
Loading /home/forcer/Programs/Emacs/trunk/lisp/loaddefs.el (source)...
Wrong number of arguments: (2 . 2), 1
make[1]: *** [bootstrap-emacs] Error 1
make[1]: Leaving directory `/home/forcer/Programs/Emacs/trunk/src'
make: *** [src] Error 2           


x86_64-unknown-linux-gnu
System Description:	Debian GNU/Linux 7.5 (wheezy)

bug#17629: Can't build emacs-24 branch

[Paul Eggert]

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

I cannot reproduce the problem; the same checkout (emacs-24 bzr 117169) 
works for me with that recipe on my host (Fedora 20 x86-64, GCC 4.9.0). 
  See attached build transcript and config.log.

Looking at your build snippet I'm suspicious of lines like this:

Loading emacs-lisp/byte-run (compiled; note, source file is newer)...

At this point during the bootstrap process there shouldn't be a compiled 
file byte-run.elc, because it hasn't been built yet.  In my log I see:

Loading emacs-lisp/byte-run (source)...

which is what I'd expect.  Perhaps you were doing a 'make -j' and ran 
into a parallelization problem?  Or maybe you did two 'make's, one 
interrupted?  Neither should cause this problem, but still, we need an 
explanation for why byte-run.elc was present when it should have been 
absent.

[-- Attachment #2: build-transcript.txt.gz --]
[-- Type: application/gzip, Size: 48954 bytes --]

[-- Attachment #3: config.log.gz --]
[-- Type: application/gzip, Size: 34200 bytes --]

bug#17629: Can't build emacs-24 branch

[Eli Zaretskii]

> From: Jorgen Schaefer <forcer@forcix.cx>
> Date: Thu, 29 May 2014 13:07:05 +0200
> 
> Building the current emacs-24 branch fails for me.
> 
> Latest commit (git ref, sorry, no bzr here):
> 
> 3302b06 Do not mention individual ns options in the manual
> 
> In a fresh checkout and with a clean tree:

This doesn't seem to be a clean tree:

> Loading emacs-lisp/byte-run (compiled; note, source file is newer)...
> Loading emacs-lisp/backquote...  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> Loading subr (compiled; note, source file is newer)...
> Loading version... ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> Loading widget...
> Loading custom...
> Loading emacs-lisp/map-ynp...
> Loading international/mule...
> Loading international/mule-conf...
> Loading env...
> Loading format...
> Loading bindings...
> Loading cus-start (compiled; note, source file is newer)...
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> Note, built-in variable `report-emacs-bug-address' not bound
> Loading window (compiled; note, source file is newer)...
> Loading files (compiled; note, source file is newer)...
> Loading emacs-lisp/macroexp (compiled; note, source file is newer)...
> Loading cus-face (compiled; note, source file is newer)...
> Loading faces (compiled; note, source file is newer)...
> Loading button...
> Loading startup (compiled; note, source file is newer)...
> Loading /home/forcer/Programs/Emacs/trunk/lisp/loaddefs.el (source)...
> Wrong number of arguments: (2 . 2), 1
> make[1]: *** [bootstrap-emacs] Error 1
> make[1]: Leaving directory `/home/forcer/Programs/Emacs/trunk/src'
> make: *** [src] Error 2           

Did you switch to the emacs-24 branch in the same git repository, and
then typed the command above?  If so, either "make bootstrap" or clone
a separate repository, switch to emacs-24 there, and then your
commands should work.

In Emacs development, the trunk and the release branch diverge very
quickly, so "make clean" is not enough to give you a "clean tree".

bug#17629: Can't build emacs-24 branch

[Glenn Morris]

Eli Zaretskii wrote:

>> In a fresh checkout and with a clean tree:
>
> This doesn't seem to be a clean tree:
[...]
> Did you switch to the emacs-24 branch in the same git repository, and
> then typed the command above?  If so, either "make bootstrap" or clone
> a separate repository, switch to emacs-24 there, and then your
> commands should work.
>
> In Emacs development, the trunk and the release branch diverge very
> quickly, so "make clean" is not enough to give you a "clean tree".


The main point is that `make clean' does not delete the compiled lisp
files (and cannot, since they are prebuilt in lisp tarfiles).
`make maintainer-clean', or `make bootstrap' does.
Always use `make bootstrap' if CPU time is not a factor, even when
building the same branch.

(Presumably this issue is going to occur more in the git-era, since it
doesn't have shared repos.)

bug#17629: Can't build emacs-24 branch

[Jorgen Schaefer]

On Thu, 29 May 2014 18:04:15 +0300
Eli Zaretskii <eliz@gnu.org> wrote:

> > From: Jorgen Schaefer <forcer@forcix.cx>
> > Date: Thu, 29 May 2014 13:07:05 +0200
> > 
> > Building the current emacs-24 branch fails for me.
> > 
> > Latest commit (git ref, sorry, no bzr here):
> > 
> > 3302b06 Do not mention individual ns options in the manual
> > 
> > In a fresh checkout and with a clean tree:
> 
> This doesn't seem to be a clean tree:

Indeed, this was a simple user error. Sorry for the noise.

> Did you switch to the emacs-24 branch in the same git repository, and
> then typed the command above?  If so, either "make bootstrap" or clone
> a separate repository, switch to emacs-24 there, and then your
> commands should work.

I did a git reset --hard and make distclean, but not a make bootstrap.
git st told me the repo was clean, but that obviously does not check
files in .gitignore. I should have used git clean. Meh.

Thanks for the quick response.

Regards,
Jorgen

bug#17629: Can't build emacs-24 branch

[Glenn Morris]

This make me think, should this from src/Makefile.in:

  $(RUN_TEMACS) --batch --load loadup bootstrap || exit 1; \

use (setq load-suffixes '(".el"))
to try and avoid this kind of problem?

bug#17629: Can't build emacs-24 branch

[Stefan Monnier]

> use (setq load-suffixes '(".el"))
> to try and avoid this kind of problem?

Wouldn't load-prefer-newer be a better choice?


        Stefan

bug#17629: Can't build emacs-24 branch

[Glenn Morris]

Stefan Monnier wrote:

>> use (setq load-suffixes '(".el"))
>> to try and avoid this kind of problem?
>
> Wouldn't load-prefer-newer be a better choice?

I guess it wouldn't do any harm to have a bootstrap-emacs that was
formed from a mix of compiled and uncompiled files...

The easiest (and IMO sanest) way to do this is to make load-prefer-newer
non-nil by default.
Fancy opening that can of worms...?

bug#17629: Can't build emacs-24 branch

[Stefan Monnier]

> The easiest (and IMO sanest) way to do this is to make load-prefer-newer
> non-nil by default.
> Fancy opening that can of worms...?

Try it on trunk?


        Stefan