Re: [wip-cite-new] Initial implementation of `csl' citation processor

[Nicolas Goaziou]

Hello,

"Bruce D'Arcus" <bdarcus@gmail.com> writes:

> On Fri, May 28, 2021 at 4:31 PM András Simonyi <andras.simonyi@gmail.com> 
> wrote:
>
>> Maybe instead of a full alist mapping backends to citation processors
>> we could have only options to declare a separate processor for
>> latex-based backends and another for non-latex ones?
>
> This would go a long way, and is probably all that's necessary.
>
> Really "latex" is the unique output mode here.

But one may want to use a different processor for, say, beamer and
regular latex. Both are "latex" based. Worse, all custom back-ends
derived from "latex" are bound to use the same processor.

Here's another proposal:

`org-cite-export-processor' is now an alist, where keys are export
back-ends or t, which is the default key.

  '((latex biblatex bibstyle citestyle)
    (beamer natbib nil nil)
    (my-latex natbib bibstyle)
    (t csl nil nil))

The selected processor is the one associated to the back-ends closest to
the current one used for export, by `org-export-derived-backend-p'
order. So if `my-other-latex' is derived from beamer, it will use
(natbib nil nil).

OTOH, I suggest to stick to a single "cite_export" keyword, which
overrides any selected processor above. IOW

   #+cite_export: basic

will use basic whatever the current export back-end is.

In practice, I think it is sufficient. The only case where it may be
limiting is if you need to export with two different back-ends with two
processors different from those set in `org-cite-export-processor'. But
in that situation, I think swapping the cite_export keyword is
acceptable.

So overall, I think it is a good compromise between simplicity and
power.

WDYT?

Regards,
-- 
Nicolas Goaziou