GNU bug report logs - #73854
29.4; tex-mode: unexpected and inconsistent behaviour of `tex-compile`, `tex-file`, and `tex-view`

Previous Next

Full log

Message #11 received at 73854 <at> debbugs.gnu.org (full text, mbox):

From: Stephen Berman <stephen.berman <at> gmx.net>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: Thalis Charakas <thales1 <at> protonmail.com>, 73854 <at> debbugs.gnu.org
Subject: Re: bug#73854: 29.4; tex-mode: unexpected and inconsistent
 behaviour of `tex-compile`, `tex-file`, and `tex-view`
Date: Sat, 19 Oct 2024 13:31:05 +0200

[Message part 1 (text/plain, inline)]

On Sat, 19 Oct 2024 12:54:51 +0300 Eli Zaretskii <eliz <at> gnu.org> wrote:

>> Date: Thu, 17 Oct 2024 18:36:19 +0000
>> From:  Thalis Charakas via "Bug reports for GNU Emacs,
>>  the Swiss army knife of text editors" <bug-gnu-emacs <at> gnu.org>
>> 
>> Basically, `tex-view` works, opening Evince, on a compiled .pdf only
>> after a .dvi is created.
>> 
>> Running 'emacs -Q' I evaluate the following in the scratch buffer, to
>> make it so that latex is compiled to pdf:
>> ```
>> (setq tex-print-file-extension ".pdf"
>>       tex-dvi-view-command "evince"
>>       tex-dvi-print-command "lpr")
>> ```
>> the last two lines are following the instructions on the description of
>> `tex-print-file-extension`.
>> 
>> I navigate to a simple .tex in an empty directory and press `C-c C-c`
>> which is bound by default to `tex-compile`, I press enter, as the
>> default is `pdflatex`, and the file compiles, creating a .pdf in the
>> directory. After that I press `C-c C-v`, which is bound to `tex-view`
>> but Evince does not open, instead I get the message "tex-print: No
>> appropriate ‘.dvi’ file could be found. If I press `C-c C-f`
>> (`tex-file`) after that, a .dvi file is created, and now `tex-view`
>> works, meaning it opens the .pdf with Evince.
>> 
>> The fact that `tex-view` works on the .pdf only when a .dvi is first
>> created seems like a bug.
>
> Thanks.  Can you step with Edebug through tex-print in your case and
> try to figure out what happens there that causes the error message?
> There's some logic there which is supposed to verify that the current
> buffer's print file is up-to-date:
>
>     (if (and (not (equal (current-buffer) tex-last-buffer-texed))
> 	     (buffer-file-name)
> 	     ;; Check that this buffer's printed file is up to date.
> 	     (file-newer-than-file-p
> 	      (setq test-name (tex-append (buffer-file-name)
>                                           tex-print-file-extension))
> 	      (buffer-file-name)))
> 	(setq print-file-name-dvi test-name))
>
> Maybe something goes wrong there in your case?

I debugged this and the problem is that tex-print expects the variable
tex-print-file to be set, but (as its doc string says) it is set by
tex-region, tex-buffer and tex-file, which are all not commands run by
tex-compile, which is meant to run external programs (listed by
tex-compile-commands).  A simple fix is the following patch:

[Message part 2 (text/x-patch, inline)]

diff --git a/lisp/textmodes/tex-mode.el b/lisp/textmodes/tex-mode.el
index 6fc49800018..acb5bb4b860 100644
--- a/lisp/textmodes/tex-mode.el
+++ b/lisp/textmodes/tex-mode.el
@@ -2530,6 +2530,7 @@ tex-compile
       (if (tex-shell-running)
           (tex-kill-job)
         (tex-start-shell))
+      (setq tex-print-file (expand-file-name (tex-main-file)))
       (tex-send-tex-command cmd dir))))

 (defun tex-start-tex (command file &optional dir)

[Message part 3 (text/plain, inline)]

But I'm not familiar with the tex-mode.el code and don't know if this
change might have unwanted consequences (though it seems prima facie
safe).

Steve Berman

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.