Re: [AUCTeX-devel] preview-latex coding system problem with Japanese LaTeX

[Ikumi Keita]

Hi Mosè, thank you for your response.

> Regarding A1, your proposed approach looks good to me (always with the
> caveat I'm not an expert on preview).  My understanding is that your
> patch would completely fix this bug, right?

I'm afraid that it wouldn't completely.  Preview-latex sets the coding
system for the messages sent from latex process by the lines
          (when (featurep 'mule)
            (setq preview-coding-system
                  (or (and (boundp 'TeX-japanese-process-output-coding-system)
                           TeX-japanese-process-output-coding-system)
                      (with-current-buffer commandbuff
                        buffer-file-coding-system)))
            (when preview-coding-system
              (setq preview-coding-system
                    (preview-buffer-recode-system
                     (coding-system-base preview-coding-system))))
            (set-process-coding-system
             process preview-coding-system))
.  Thus the recieved texts are not decoded correctly if
`preview-coding-system' (which is basically the same as
`buffer-file-coding-system') does not match the actual coding system
that the latex outputs.  (The Japanese tex engine converts internally
the coding systems of Japanese text, so the coding systems of the tex
file and its output sometimes differ.)

It is hard to resolve this problem as long as preview-latex and
tex-jp.el set the process coding system independently.  I'd like AUCTeX
to have some framework that the two coordinate to determine the
appropriate coding system.

> Regarding A2, I don't think I can help you.  I hope David can give
> some comments on this.

Thanks, I hope him to do that, too.

> In any case, a better cooperation between preview and core AUCTeX is
> desirable, as you suggested in point C.  If I remember correctly, some
> times ago it was suggested to combine their error parsing frameworks,
> as there is some overlap.  I guess the main problem is the lack of
> active manpower with good knowledge of both packages.

Indeed.

Best regards,
Ikumi Keita