Re: [wip-cite-new] Adjust punctuation around citations

[Nicolas Goaziou]

Hello,

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

> On Mon, Jun 14, 2021 at 7:45 AM Denis Maier <denismaier@mailbox.org> wrote:

>> * Note style input (=semantically strict input)
>>
>> "A quotation ending with a period." [cite: @hoel-71-whole]
>>
>> "A quotation ending without punctuation". [cite: @hoel-71-whole]
>>
>> As the input preserves the location of punctuation in the original
>> material, I'd say it should be much easier to deal with this. We
>> don't have to add information which isn't in the input, but rather
>> we'll just have to move any punctuation to after the citation
>> object. Maybe I'm missing something, but to me this looks like
>> a much simpler operation than going in the opposite direction.

This cannot be. We don't know anything about the cite after the
quotation. A bare cite could be starting out a new sentence:

  "A quotation ending with a period." [cite: @hoel-71-whole] pretends…

OTOH, we know perfectly when a citation is meant to become a footnote
(at least in basic and csl processors). And we know — almost, as you
demonstrated — where to put that footnote.

Moreover, I think the syntax you propose has another drawback: it
doesn't correspond to any desired output (note or something else). As
this looks artificial, I fear it might hinder readability of the Org
document.

  ... period." [cite:@doe21] [cite/text:@doe21] pretends…

>> Maybe we should stop talking about author date vs note style input, but 
>> rather about strict vs. non-strict input.
>
> It's definitely not author-date vs note. I see it as in-text citations
> vs note citations. As in, the former applies to other styles beyond
> author-date.

I think the current patch is purely about note citations. I mentioned
"author-date" in a docstring just because I didn't know how to express
it otherwise. So, in a way, I agree it can be considered as in-text
citations vs note citations, indeed.

> The example you are highlighlighting here was why I was earlier
> suggesting for a rule that would allow something like this input:
>
> "A quotation ending with a period." [cite: @hoel-71-whole].
>
> ... where the second would be dropped, hence getting the expected output.

This is interesting, but we might get false positives, as in the
following (far-fetched) example

  … the so-called "foobar". [cite/text: See @hoel-71-whole p. 42].

which bites us because we need to process even non-note citations to
remove the spurious punctuation while ignoring the necessity of a given
punctuation character.

As another, imperfect, workaround, I submit the following idea for
consideration:

  "A quotation ending without punctuation" [cite: @hoel-71-whole].
  "A quotation ending with a period"[cite: @hoel-71-whole].

IOW, the presence or absence of a space before the citation determines,
according to a note rule, if the punctuation should go inside or outside
the quotation. When processing non-note citations, we just need to
ensure there is at least a space after the previous element, which is
less "dangerous" than removing punctuation.

I find it a bit too subtle, and so error-prone, but so is punctuation
anyway.

WDYT?

Regards,
-- 
Nicolas Goaziou