[ox bug] batch export seems to be broken or have changed

[ox bug] batch export seems to be broken or have changed

[Rasmus]

Hi,

I used to be able to something like

    emacs -Q --file foo.org  --batch --no-init-file -f org-latex-export-to-pdf
    # tex output

And everything would work. 

Now I get something like

     emacs -Q --file foo.org  --batch --no-init-file -f org-latex-export-to-pdf
     Output file:                 # RET
     Processing LaTeX file ./.tex...
     PDF file ./.tex.pdf wasn't produced

i.e. I'm asked about output file name.  I'm using the git version of
Emacs.  When I downgrade to latest release it works.

Rasmus

-- 
A page of history is worth a volume of logic

Re: [ox bug] batch export seems to be broken or have changed

[Nicolas Goaziou]

Hello,

Rasmus <rasmus@gmx.us> writes:

> I used to be able to something like
>
>     emacs -Q --file foo.org  --batch --no-init-file -f org-latex-export-to-pdf
>     # tex output
>
> And everything would work. 
>
> Now I get something like
>
>      emacs -Q --file foo.org  --batch --no-init-file -f org-latex-export-to-pdf
>      Output file:                 # RET
>      Processing LaTeX file ./.tex...
>      PDF file ./.tex.pdf wasn't produced
>
> i.e. I'm asked about output file name.

I cannot reproduce it.


Regards,

-- 
Nicolas Goaziou

Re: [ox bug] batch export seems to be broken or have changed

[Rasmus]

Hi,

I hope you had a pleasant break from the PC.

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Rasmus <rasmus@gmx.us> writes:
>
>> I used to be able to something like
>>
>>     emacs -Q --file foo.org  --batch --no-init-file -f org-latex-export-to-pdf
>>     # tex output
>>
>> And everything would work. 
>>
>> Now I get something like
>>
>>      emacs -Q --file foo.org  --batch --no-init-file -f org-latex-export-to-pdf
>>      Output file:                 # RET
>>      Processing LaTeX file ./.tex...
>>      PDF file ./.tex.pdf wasn't produced
>>
>> i.e. I'm asked about output file name.
>
> I cannot reproduce it.

I still see the problem.  Are you tracking Emacs-git?  I'm currently
using:

     commit fa5a880f733cbbe3b0d515ed3e1cf7c6fd54cfd8
     Date:   Thu Jul 16 20:04:07 2015 +0200

My Makefile calling something similar to the above is not working on my PC
and asks for an input file.  On my Debian 7 CI machine the Makefile is
still working...  I don't have access to my other laptop ATM so I haven't
tested it there.

I tried to downgrade to Emacs 24.5, but when I batch export as above I
get:

     Symbol's function definition is void: cl-struct-define

Rasmus

-- 
In theory, practice and theory are the same. In practice they are not

Re: [ox bug] batch export seems to be broken or have changed

[Nicolas Goaziou]

Rasmus <rasmus@gmx.us> writes:

> I hope you had a pleasant break from the PC.

I did. Thank you for you work meanwhile. I will do it again in 5 days,
for another week.

> I still see the problem.  Are you tracking Emacs-git?

I'm not.

> I'm currently using:
>
>      commit fa5a880f733cbbe3b0d515ed3e1cf7c6fd54cfd8
>      Date:   Thu Jul 16 20:04:07 2015 +0200
>
> My Makefile calling something similar to the above is not working on my PC
> and asks for an input file.

`org-export-output-file-name' only asks for an input file if current
buffer is not associated to any file (i.e. when (buffer-file-name
(buffer-base-buffer)) is nil).

Has Emacs 25 changed --file option behaviour?

> I tried to downgrade to Emacs 24.5, but when I batch export as above I
> get:
>
>      Symbol's function definition is void: cl-struct-define

That doesn't ring a bell.


Regards,

Re: [ox bug] batch export seems to be broken or have changed

[Rasmus]

Hi,

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

>> I hope you had a pleasant break from the PC.
>
> I did. Thank you for you work meanwhile. I will do it again in 5 days,
> for another week.

Great.  I'll be here.

>> I'm currently using:
>>
>>      commit fa5a880f733cbbe3b0d515ed3e1cf7c6fd54cfd8
>>      Date:   Thu Jul 16 20:04:07 2015 +0200
>>
>> My Makefile calling something similar to the above is not working on my PC
>> and asks for an input file.
>
> `org-export-output-file-name' only asks for an input file if current
> buffer is not associated to any file (i.e. when (buffer-file-name
> (buffer-base-buffer)) is nil).
>
> Has Emacs 25 changed --file option behaviour?

That's my first guess.  But I don't know how to debug it since it's batch.
And I don't know if it's a an Emacs bug or Org needing to adapt to some
new behavior.

I could try to print the filename input to the console or something...

>> I tried to downgrade to Emacs 24.5, but when I batch export as above I
>> get:
>>
>>      Symbol's function definition is void: cl-struct-define
>
> That doesn't ring a bell.

Weird.  Cl-struct is what is used for holding backends, right?

Rasmus

-- 
Hvor meget poesi tror De kommer ud af et glas isvand?

Re: [ox bug] batch export seems to be broken or have changed

[Nicolas Goaziou]

Rasmus <rasmus@gmx.us> writes:

> That's my first guess.  But I don't know how to debug it since it's batch.
> And I don't know if it's a an Emacs bug or Org needing to adapt to some
> new behavior.
>
> I could try to print the filename input to the console or something...

You may write the output of (buffer-file-name) to a temporary file,
e.g.,

emacs --batch --file foo.org --eval="(let ((name (buffer-file-name))) (with-temp-file \"report.txt\" (insert (message \"%S\n\" name))))"

>>> I tried to downgrade to Emacs 24.5, but when I batch export as above I
>>> get:
>>>
>>>      Symbol's function definition is void: cl-struct-define
>>
>> That doesn't ring a bell.
>
> Weird.  Cl-struct is what is used for holding backends, right?

Correct. But AFAICT, `cl-struct-define' doesn't exist in Emacs 24.

Regards,

Re: [ox bug] batch export seems to be broken or have changed

[Rasmus]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> You may write the output of (buffer-file-name) to a temporary file,
> e.g.,
>
> emacs --batch --file foo.org --eval="(let ((name (buffer-file-name))) (with-temp-file \"report.txt\" (insert (message \"%S\n\" name))))"


$ emacs --batch --file foo.org --eval="(let ((name (buffer-file-name))) (with-temp-file \"report.txt\" (insert (message \"%S\n\" name))))"
$ cat report.txt
  nil

C-h v mentions no change.  Fortunately there are not too many changes to
buffer.c

$ git log --since=emacs-24.5 -- buffer.c
commit 3213d7707026573ca425ba1c865b7fa1a8b46639
Author: Paul Eggert <eggert@cs.ucla.edu>
Date:   Tue Jun 30 15:06:31 2015 -0700

    In strings, prefer plain ` and ' to \` and \'
    
    * lisp/allout.el (allout-insert-listified):
    * lisp/emacs-lisp/easy-mmode.el (define-globalized-minor-mode):
    * lisp/ls-lisp.el (ls-lisp-UCA-like-collation)
    (ls-lisp-string-lessp):
    * lisp/menu-bar.el (menu-bar-open):
    * lisp/obsolete/otodo-mode.el (todo-top-priorities):
    * lisp/progmodes/compile.el (compile):
    * lisp/progmodes/etags.el (tags-loop-scan):
    * lisp/progmodes/make-mode.el (makefile-browser-insert-continuation):
    * lisp/subr.el (posn-actual-col-row):
    * lisp/term/pc-win.el (x-list-fonts):
    * lisp/textmodes/texinfmt.el (texinfmt-version):
    * lisp/textmodes/texnfo-upd.el (texinfo-master-menu):
    * lisp/time.el (display-time-world-list):
    * lisp/tmm.el (tmm-menubar):
    * src/buffer.c (syms_of_buffer):
    * src/fileio.c (syms_of_fileio):
    Omit unnecessary and confusing backslash before quote.
    * lisp/erc/erc.el (erc-cmd-LASTLOG):
    * lisp/progmodes/flymake.el (flymake-fix-file-name):
    * lisp/progmodes/vhdl-mode.el (vhdl-in-extended-identifier-p):
    Fix string that was intended to escape a backslash and not a quote.

commit f8e26cd3af6b48428c3199f3f9c2a08c4f359234
Author: Glenn Morris <rgm@gnu.org>
Date:   Fri Jun 12 18:07:48 2015 -0400

    * src/buffer.c (init_buffer): Add final newline to message.

commit ebbc6a4782c279527c52d6b1d8b379517aeec2d5
Author: Glenn Morris <rgm@gnu.org>
Date:   Thu Jun 11 20:34:54 2015 -0400

    Some progress towards starting with PWD deleted.  (Bug#18851)
    
    * src/buffer.c (init_buffer): Handle get_current_dir_name failures.
    
    * lisp/startup.el (normal-top-level, command-line-1):
    * lisp/minibuffer.el (read-file-name-default):
    Handle default-directory being nil.

commit 2c656f7deb3b192a5063151368782c2715e81051
Author: Paul Eggert <eggert@cs.ucla.edu>
Date:   Sun May 31 23:52:09 2015 -0700

    Avoid grave accent quoting in stderr diagnostics
    
    A few Emacs diagnostics go directly to stderr, and so can't easily
    contain curved quotes (as non-UTF-8 locales might mishandle them).
    Instead of bothering to add support for this rarity, reword the
    diagnostics so that they don't use grave accent to quote.
    * src/alloc.c (mark_memory): Fix comment.
    * src/buffer.c (init_buffer):
    * src/dispnew.c (init_display):
    * src/emacs.c (main, sort_args):
    * src/lread.c (dir_warning):
    * src/term.c (init_tty):
    * src/unexmacosx.c (unexec):
    * src/xfns.c (select_visual):
    * src/xterm.c (cvt_string_to_pixel, x_io_error_quitter):
    Reword stderr diagnostics to avoid quoting `like this'.
    * src/unexmacosx.c: Include errno.h.
    * src/xfns.c (select_visual): Encode value for locale.

commit d90a3b186d57ac700a32988b720f76f289740ece
Author: Paul Eggert <eggert@cs.ucla.edu>
Date:   Sat May 30 15:29:41 2015 -0700

    Remove format2
    
    * src/editfns.c, src/lisp.h (format2): Remove.
    It is more trouble than it's worth, now that we have CALLN.
    This is just a minor refactoring.
    * src/buffer.c (Fkill_buffer):
    * src/dbusbind.c (XD_OBJECT_TO_STRING):
    * src/fileio.c (barf_or_query_if_file_exists):
    Adjust to format2 going away.


>> Weird.  Cl-struct is what is used for holding backends, right?
>
> Correct. But AFAICT, `cl-struct-define' doesn't exist in Emacs 24.

The stable Emacs version I've got is 24.5. Why is Org git trying to pull
it in when exporting if it doesn't exist?  Probably I missed something.

Rasmus

-- 
May the Force be with you

Re: [ox bug] batch export seems to be broken or have changed

[Nicolas Goaziou]

Rasmus <rasmus@gmx.us> writes:

> $ emacs --batch --file foo.org --eval="(let ((name (buffer-file-name))) (with-temp-file \"report.txt\" (insert (message \"%S\n\" name))))"
> $ cat report.txt
>   nil

OK, so this changed.

> C-h v mentions no change.  Fortunately there are not too many changes to
> buffer.c

It might be worth to make a bug report about it. It isn't clear that
this change was intended.

> The stable Emacs version I've got is 24.5. Why is Org git trying to pull
> it in when exporting if it doesn't exist?  Probably I missed
> something.

Shot in the dark: could it be related to asynchronous export, calling
a different Emacs executable (Emacs 25 in this case)?


Regards,

Re: [ox bug] batch export seems to be broken or have changed

[Rasmus]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

> Shot in the dark: could it be related to asynchronous export, calling
> a different Emacs executable (Emacs 25 in this case)?

No, 'cause I've only got on Emacs installed (/usr/bin/emacs) at the time
and I install org to /usr/share/emacs/site-lisp/ (but it's stable).  Maybe
the .elc files need to cleaned out.

Rasmus

-- 
Summon the Mothership!

Re: [ox bug] batch export seems to be broken or have changed

[Rasmus]

Nicolas Goaziou <mail@nicolasgoaziou.fr> writes:

>> C-h v mentions no change.  Fortunately there are not too many changes to
>> buffer.c
>
> It might be worth to make a bug report about it. It isn't clear that
> this change was intended.

Reported here:

         http://debbugs.gnu.org/cgi/bugreport.cgi?bug=21095

-- 
Send from my Emacs