[O] Question to the list about csquotes

[Nick Dokos]

There's been a recent discussion about using csquotes in the latex
exporter[fn:1] and Tom Dye came up with a good idea of how to implement
it. We have been discussing it off list but there is a point that might
benefit from more general airing, so I offered to solicit opinions on
the list.

The question is about the following bit of code in
org-export-latex-quotation-marks:

,----
|   ...
|   (let* ((lang (plist-get org-export-latex-options-plist :language))
|        (quote-rpl (if (equal lang "fr")
|                       '(("\\(\\s-\\)\"" "«~")
|                         ("\\(\\S-\\)\"" "~»")
|                         ("\\(\\s-\\)'" "`"))
|                     '(("\\(\\s-\\|[[(]\\)\"" "``")
|                       ("\\(\\S-\\)\"" "''")
|                       ("\\(\\s-\\|(\\)'" "`")))))
|   ...
`----

The question is what to do about the lang = "fr" case (which I think we
all agree is a hack):

o leave it alone and implement the csquotes mechanism on top of it?

o get rid of it and depend on the general csquotes mechanism to replace it?

The first solution perpetuates the hack but leaves existing org files that
use it unaffected.

The second cleans up the hack but at the cost of some backward inompatibility:
files that use the hack will not get guillemets unless some (one-time)
customizations are done: the LaTeX preamble has to be expanded to pull
in the appropriate packages (babel and csquotes) and a couple of string
variables have to be set to the right values.

So we turn to the collective wisdom of the list: how important is
backward compatibility in this case (i.e. are there lots of files out
there that use the hack)? And if it is important, how would you weigh
the inconvenience of the required customizations against the code
cleanup?  Presumably, french speakers have been the heaviest users of
this, so we'd like to hear from you - but opinions are welcome.

Thanks,
Nick

Footnotes:

[fn:1] http://thread.gmane.org/gmane.emacs.orgmode/43689