Building Guilemacs

Building Guilemacs

[Wilfred Hughes]

[-- Attachment #1: Type: text/plain, Size: 1350 bytes --]

Hi folks

I fancied exploring Guile with Emacs, so I've been trying to build Guile
and Emacs as documented in https://www.emacswiki.org/emacs/GuileEmacsBuild .

I've successfully built the wip-elisp branch of the main guile repo. The
wiki page discusses http://git.hcoop.net/?p=bpt/guile.git and
http://git.savannah.gnu.org/cgit/guile.git/log/?h=wip-elisp . I understand
that the latter is a rebase of the hcoop commits on Guile 2.1.

However, I've also noticed there's a wip-bpt-elisp branch in the main guile
repo. This has the newest commits, but it seems to based on an older guile
version.

Which branch should I use?

I've also failed to compile the Emacs branch that uses Guile. I'm getting
the error:

gcc -c  -Demacs  -I. -I. -I../lib -I./../lib  [many includes elided...]
-I/usr/local/include/guile/2.2   -g3 -O2  vm-limit.c
In file included from vm-limit.c:21:0:
lisp.h:33:22: fatal error: libguile.h: No such file or directory
 #include <libguile.h>
                      ^
compilation terminated.

This rather looks like I've screwed up with autoconf somehow. I've set the
environment variables suggested by the wiki page:

ACLOCAL_PATH=/home/wilfred/guile-emacs/guile/share/aclocal
PKG_CONFIG_PATH=/home/wilfred/guile-emacs/guile/lib/pkgconfig
LD_RUN_PATH=/home/wilfred/guile-emacs/guile/lib

so am I missing something?

Thanks
Wilfred

[-- Attachment #2: Type: text/html, Size: 1861 bytes --]

Re: Building Guilemacs

[Christopher Allan Webber]

Wilfred Hughes writes:

> Hi folks
>
> I fancied exploring Guile with Emacs, so I've been trying to build Guile
> and Emacs as documented in https://www.emacswiki.org/emacs/GuileEmacsBuild .

Excellent!  Happy to hear of your interest.

> I've successfully built the wip-elisp branch of the main guile repo. The
> wiki page discusses http://git.hcoop.net/?p=bpt/guile.git and
> http://git.savannah.gnu.org/cgit/guile.git/log/?h=wip-elisp . I understand
> that the latter is a rebase of the hcoop commits on Guile 2.1.
>
> However, I've also noticed there's a wip-bpt-elisp branch in the main guile
> repo. This has the newest commits, but it seems to based on an older guile
> version.
>
> Which branch should I use?

wip-bpt-elisp has commits from years ago; wip-elisp is rebased on top of
relatively-recent master, with updated ChangeLog.  This was based on the
recent work of robin's from
http://git.savannah.gnu.org/cgit/guile.git/log/?h=wip-elisp

So, you want wip-elisp in origin.  That's the most up to date.

> I've also failed to compile the Emacs branch that uses Guile. I'm getting
> the error:
>
> gcc -c  -Demacs  -I. -I. -I../lib -I./../lib  [many includes elided...]
> -I/usr/local/include/guile/2.2   -g3 -O2  vm-limit.c
> In file included from vm-limit.c:21:0:
> lisp.h:33:22: fatal error: libguile.h: No such file or directory
>  #include <libguile.h>
>                       ^
> compilation terminated.
>
> This rather looks like I've screwed up with autoconf somehow. I've set the
> environment variables suggested by the wiki page:

I don't have time to look into this part, maybe someone else can help
you out!

Re: Building Guilemacs

[Wilfred Hughes]

> I think the variables you mentioned are for building in your home
> directory only. If you build system-wide, for instance under /usr/local
> -- and your mail makes me think you do so -- you should not need them.

I wanted to build them in my home directory. I already have Guile installed
and I didn't want to get things confused.

> I have made some packaging recipes for Arch Linux, and there it works
> without specifing these variables.
>
> https://aur.archlinux.org/packages/guile-emacs-git/ (for emacs)
> https://aur.archlinux.org/packages/guile4emacs-git/ (for guile)

Oh, that's really useful, thanks. I'll have a look at those for
reference (and I've documented them on the GuileEmacsBuild wiki
build).

Re: Building Guilemacs

[Wilfred Hughes]

[-- Attachment #1: Type: text/plain, Size: 2187 bytes --]

I've had a play with the AUR packages too. The guile4emacs PKGBUILD works
perfectly, but the guile-emacs build process segfaults when bootstrap-emacs
attempts to load calendar/cal-loaddefs.el:

make[2]: Leaving directory
'/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
if test "yes" = "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
make -C ../lisp autoloads EMACS="../src/bootstrap-emacs"
make[2]: Entering directory
'/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
EMACSLOADPATH= '../src/bootstrap-emacs' -batch --no-site-file
--no-site-lisp -l autoload \
   --eval "(setq generate-autoload-cookie \";;;###cal-autoload\")" \
   --eval "(setq generated-autoload-file (expand-file-name
(unmsys--file-name \"calendar/cal-loaddefs.el\")))" \
   -f batch-update-autoloads ./calendar
make[2]: *** [Makefile:466: calendar/cal-loaddefs.el] Segmentation fault
(core dumped)
make[2]: Leaving directory
'/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
make[1]: *** [Makefile:805: ../lisp/loaddefs.el] Error 2
make[1]: Leaving directory
'/home/wilfred/aur/guile-emacs-git/src/guilemacs/src'
make: *** [Makefile:376: src] Error 2
==> ERROR: A failure occurred in build().
    Aborting...

Any suggestions?

On 30 August 2016 at 20:32, Wilfred Hughes <me@wilfred.me.uk> wrote:

> > I think the variables you mentioned are for building in your home
> > directory only. If you build system-wide, for instance under /usr/local
> > -- and your mail makes me think you do so -- you should not need them.
>
> I wanted to build them in my home directory. I already have Guile installed
> and I didn't want to get things confused.
>
> > I have made some packaging recipes for Arch Linux, and there it works
> > without specifing these variables.
> >
> > https://aur.archlinux.org/packages/guile-emacs-git/ (for emacs)
> > https://aur.archlinux.org/packages/guile4emacs-git/ (for guile)
>
> Oh, that's really useful, thanks. I'll have a look at those for
> reference (and I've documented them on the GuileEmacsBuild wiki
> build).
>

[-- Attachment #2: Type: text/html, Size: 3051 bytes --]

Re: Building Guilemacs

[Stefan Husmann]

Wilfred Hughes <me@wilfred.me.uk> writes:

> I've had a play with the AUR packages too. The guile4emacs PKGBUILD works
> perfectly, but the guile-emacs build process segfaults when bootstrap-emacs
> attempts to load calendar/cal-loaddefs.el:
>
> make[2]: Leaving directory
> '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
> if test "yes" = "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
> make -C ../lisp autoloads EMACS="../src/bootstrap-emacs"
> make[2]: Entering directory
> '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
> EMACSLOADPATH= '../src/bootstrap-emacs' -batch --no-site-file
> --no-site-lisp -l autoload \
>    --eval "(setq generate-autoload-cookie \";;;###cal-autoload\")" \
>    --eval "(setq generated-autoload-file (expand-file-name
> (unmsys--file-name \"calendar/cal-loaddefs.el\")))" \
>    -f batch-update-autoloads ./calendar
> make[2]: *** [Makefile:466: calendar/cal-loaddefs.el] Segmentation fault
> (core dumped)
> make[2]: Leaving directory
> '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
> make[1]: *** [Makefile:805: ../lisp/loaddefs.el] Error 2
> make[1]: Leaving directory
> '/home/wilfred/aur/guile-emacs-git/src/guilemacs/src'
> make: *** [Makefile:376: src] Error 2
> ==> ERROR: A failure occurred in build().
>     Aborting...
>
> Any suggestions?
>
> On 30 August 2016 at 20:32, Wilfred Hughes <me@wilfred.me.uk> wrote:
>
Hello,

I can confirm this, but currently have no ideas what is going on
here. This definitely used to work, but gcc was updated since then.

Sorry for inconveniance. I will try to dig deeper into it.

Best Regards

Stefan

Re: Building Guilemacs

[Wilfred Hughes]

[-- Attachment #1: Type: text/plain, Size: 2488 bytes --]

Digging some more, I think the issue is
http://debbugs.gnu.org/cgi/bugreport.cgi?bug=24065 . Emacs trunk is
depending on undefined C behaviour, which gcc 5 has started optimising,
breaking the code.

I've updated https://www.emacswiki.org/emacs/GuileEmacsBuild to note this.
You can work around with -fno-builtin-malloc or simply compile with clang
(my current solution).

Emacs trunk has fixed this (in 4b1436b702d56eedd27a0777fc7232cdfb7ac4f6
IIUC), but I believe that commit does not exist in the guile-emacs fork.

Wilfred

On 4 September 2016 at 04:48, Stefan Husmann <stefan-husmann@t-online.de>
wrote:

> Wilfred Hughes <me@wilfred.me.uk> writes:
>
> > I've had a play with the AUR packages too. The guile4emacs PKGBUILD works
> > perfectly, but the guile-emacs build process segfaults when
> bootstrap-emacs
> > attempts to load calendar/cal-loaddefs.el:
> >
> > make[2]: Leaving directory
> > '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
> > if test "yes" = "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
> > make -C ../lisp autoloads EMACS="../src/bootstrap-emacs"
> > make[2]: Entering directory
> > '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
> > EMACSLOADPATH= '../src/bootstrap-emacs' -batch --no-site-file
> > --no-site-lisp -l autoload \
> >    --eval "(setq generate-autoload-cookie \";;;###cal-autoload\")" \
> >    --eval "(setq generated-autoload-file (expand-file-name
> > (unmsys--file-name \"calendar/cal-loaddefs.el\")))" \
> >    -f batch-update-autoloads ./calendar
> > make[2]: *** [Makefile:466: calendar/cal-loaddefs.el] Segmentation fault
> > (core dumped)
> > make[2]: Leaving directory
> > '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
> > make[1]: *** [Makefile:805: ../lisp/loaddefs.el] Error 2
> > make[1]: Leaving directory
> > '/home/wilfred/aur/guile-emacs-git/src/guilemacs/src'
> > make: *** [Makefile:376: src] Error 2
> > ==> ERROR: A failure occurred in build().
> >     Aborting...
> >
> > Any suggestions?
> >
> > On 30 August 2016 at 20:32, Wilfred Hughes <me@wilfred.me.uk> wrote:
> >
> Hello,
>
> I can confirm this, but currently have no ideas what is going on
> here. This definitely used to work, but gcc was updated since then.
>
> Sorry for inconveniance. I will try to dig deeper into it.
>
> Best Regards
>
> Stefan
>

[-- Attachment #2: Type: text/html, Size: 3605 bytes --]

Re: Building Guilemacs

[Christopher Allan Webber]

If someone wanted to rebase the emacs side of the guile-emacs fork...
I think that would be great :)

Wilfred Hughes writes:

> Digging some more, I think the issue is
> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=24065 . Emacs trunk is
> depending on undefined C behaviour, which gcc 5 has started optimising,
> breaking the code.
>
> I've updated https://www.emacswiki.org/emacs/GuileEmacsBuild to note this.
> You can work around with -fno-builtin-malloc or simply compile with clang
> (my current solution).
>
> Emacs trunk has fixed this (in 4b1436b702d56eedd27a0777fc7232cdfb7ac4f6
> IIUC), but I believe that commit does not exist in the guile-emacs fork.
>
> Wilfred
>
> On 4 September 2016 at 04:48, Stefan Husmann <stefan-husmann@t-online.de>
> wrote:
>
>> Wilfred Hughes <me@wilfred.me.uk> writes:
>>
>> > I've had a play with the AUR packages too. The guile4emacs PKGBUILD works
>> > perfectly, but the guile-emacs build process segfaults when
>> bootstrap-emacs
>> > attempts to load calendar/cal-loaddefs.el:
>> >
>> > make[2]: Leaving directory
>> > '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
>> > if test "yes" = "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
>> > make -C ../lisp autoloads EMACS="../src/bootstrap-emacs"
>> > make[2]: Entering directory
>> > '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
>> > EMACSLOADPATH= '../src/bootstrap-emacs' -batch --no-site-file
>> > --no-site-lisp -l autoload \
>> >    --eval "(setq generate-autoload-cookie \";;;###cal-autoload\")" \
>> >    --eval "(setq generated-autoload-file (expand-file-name
>> > (unmsys--file-name \"calendar/cal-loaddefs.el\")))" \
>> >    -f batch-update-autoloads ./calendar
>> > make[2]: *** [Makefile:466: calendar/cal-loaddefs.el] Segmentation fault
>> > (core dumped)
>> > make[2]: Leaving directory
>> > '/home/wilfred/aur/guile-emacs-git/src/guilemacs/lisp'
>> > make[1]: *** [Makefile:805: ../lisp/loaddefs.el] Error 2
>> > make[1]: Leaving directory
>> > '/home/wilfred/aur/guile-emacs-git/src/guilemacs/src'
>> > make: *** [Makefile:376: src] Error 2
>> > ==> ERROR: A failure occurred in build().
>> >     Aborting...
>> >
>> > Any suggestions?
>> >
>> > On 30 August 2016 at 20:32, Wilfred Hughes <me@wilfred.me.uk> wrote:
>> >
>> Hello,
>>
>> I can confirm this, but currently have no ideas what is going on
>> here. This definitely used to work, but gcc was updated since then.
>>
>> Sorry for inconveniance. I will try to dig deeper into it.
>>
>> Best Regards
>>
>> Stefan
>>