Re: [O] Citation syntax: Underscore MUST(?) be allowed in cite keys?

[Thomas S. Dye]

Aloha Rasmus,

Rasmus <address@hidden> writes:

> address@hidden (Thomas S. Dye) writes:
>
>> Rasmus <address@hidden> writes:
>>
>>> Nicolas Goaziou <address@hidden> writes:
>>>
>>>>> I'm asking because I haven't fully grasped uses for the shorthand.  What
>>>>> is the use case?
>>>>
>>>> More readable, I guess.
>>>
>>> I agree.  In time, "org-reftex" would insert @key if no notes are
>>> requested at the time of insertion.
>>
>> I think the OP has a valid point.  After we teach org-reftex to insert
>> @key if no notes are requested, are we going to convince all key
>> generating software to prohibit keys that end in punctuation?
>
> So just to get it straight: are you advocating for only allowing
> [cite:@key]-like constructs to allow punctuation at the end of words?
>
> Perhaps it's a can of worms, but you can also match keys against a
> "punctuation at end of word"-regexp and use the fuller cite command then.
> I'm not too happy with having the regexps used in [cite:@·] and @· diverge
> too much, though...
>
> So /given support for end-of-word punctuation/, we'd either have two
> abandon a "single" org-element--citation-key-re (yes that's not entirely
> correct) or give up short citations.
>
>> As I currently understand the problem, that seems like a tall order to
>> me.
>
> It's also a tall order to support end of word punctuation cf. above.
>
> I think another important question is how easy is it to configure the
> citation manager in question not to insert punctuation marks at the end?

I'm not an advocate at this point.  I'm just trying to be clear about
a choice that apparently needs to be made.

As I see it, the choice boils down to the relative benefit of citation
shortcuts vs. the limitation of requiring authors to configure the
citation manager so it doesn't produce a key ending in punctuation (or
your solution that uses different regexps for full citations and
shortcuts).

Nicolas guessed that the benefit of citation shortcuts is that they are
more "readable" than a full citation, and you agree with his guess.  The
shortcuts are certainly shorter, so in this sense are more readable.
However, having two different representations of the same thing, a
shortcut and a full citation, means that, for the author (and the
software) recognition is more complex and thus, less readable.  For this
reason, IMHO the readability benefit is not particularly strong.

Richard and Stefan both see keys ending in punctuation marks as corner
cases, so the burden imposed on the author to configure the citation
manager is relatively infrequent.  They know more about this than I do,
so I'm heartened by this information.  However, in the event the
citation manager has to be configured, the author faces a potentially
daunting task.  The algorithm for automatic key generation in
bibtex-mode is summarized in 18 steps, including two near the end that
allow arbitrary input!  I strongly believe Org mode shouldn't send an
author here, unless the corresponding benefits are great.

I'm not capable of forming an opinion about your solution that uses
different regexps.

At this point I think the benefit of citation shortcuts is relatively
modest and the limitation of requiring authors to ensure keys don't end
in punctuation potentially onerous.  On balance, I think strong
consideration should be given to the option of not using shortcuts.

All the best,
Tom

-- 
Thomas S. Dye
http://www.tsdye.com