APA6th: et-al-use-last

I do not want to waste your time with some useless theoretical
comments on data-type design, but if you believe that this feature
could be extended at a later time I agree we should have a positive
integer to represent it.

What I do not understand is why, if we decide to use an integer
because of possible future extension, don’t we just add the extended
feature all together. That is to say, “et-al-use-last” will be an
integer indicating the number of names to be put after the ellipses
(1, 10, or 1000, as long as we have enough contributors). And that’s
it.

It is the “it is a number but it takes only two values” part I’m
against to (and the “add a number” on a request basis part too).

Andrea

OK; fair point.

Bruce

The reasons for restricting et-al-use-last to “1”/true and “0”/false were a)
we aren’t currently aware of any styles using more than one trailing author
and b) it is not immediately obvious how multiple trailing authors should be
rendered (see
http://bitbucket.org/bdarcus/csl-schema/issue/14/list-last-authors-name-in-truncated#comment-202954).

Rintze

What I do not understand is why, if we decide to use an integer
because of possible future extension, don’t we just add the extended
feature all together. That is to say, “et-al-use-last” will be an
integer indicating the number of names to be put after the ellipses
(1, 10, or 1000, as long as we have enough contributors). And that’s
it.

The reasons for restricting et-al-use-last to “1”/true and “0”/false were a)
we aren’t currently aware of any styles using more than one trailing author
and b) it is not immediately obvious how multiple trailing authors should be
rendered (see
http://bitbucket.org/bdarcus/csl-schema/issue/14/list-last-authors-name-in-truncated#comment-202954
).

That’s it in a nutshell. If a style with multiple trailing authors
eventually emerges, it might require that the last of the trailing
authors be preceded by “and”, or it might not. I’ll work with any
decision that emerges from the group, but I think we’re
over-engineering this. My personal preference would be to do the
minimum needed to cleanly resolve the known use case, and move on.
That would mean sticking with values “true” or “false” for the
present.

The reasoning behind that preference is that the time we invest in
discussing the multiple-trailing-authors case does not produce any
immediately useful results, and quite possibly will turn out to have
been wasted effort. Avoiding a distant and uncertain prospect of
doing some code refactoring down the road isn’t worth the candle, in
my view.

Anyway, here are the options and the trade-offs.

[A] Implement et-al-use-last, with values false or true
Risks: if multiple trailing authors is ever required, then …
(1) must change data type
(2) must refactor code

[B] Implement et-al-use-last, with values 0 or 1 only.
Risks:
(1) if multiple trailing authors is ever required, must refactor code
(2) documentation on multiple trailing authors encourages
emergence of such a style

[C] Implement et-al-use-last, with arbitrary numeric value
Risks: final join is a blind guess, so …
(1) if multiple trailing authors is required, might need to
refactor final joins
(2) further discussion will be needed to settle the best-guess
choice of final join
(2) if the option is never needed, development effort is wasted

[D] Implement et-al-use-last, with arbitrary numeric value, and
specify post-ellipsis join
Risks:
(1) further discussion needed to settle how to control final join
(2) if the option is never used, development effort is wasted

My own vote is [A], then [B], then [D], then [C], ranked from most to
least desirable. But I’ll follow whatever the group decides.

Frank

What I do not understand is why, if we decide to use an integer
because of possible future extension, don’t we just add the extended
feature all together. That is to say, “et-al-use-last” will be an
integer indicating the number of names to be put after the ellipses
(1, 10, or 1000, as long as we have enough contributors). And that’s
it.

The reasons for restricting et-al-use-last to “1”/true and “0”/false were a)
we aren’t currently aware of any styles using more than one trailing author
and b) it is not immediately obvious how multiple trailing authors should be
rendered (see
http://bitbucket.org/bdarcus/csl-schema/issue/14/list-last-authors-name-in-truncated#comment-202954
).

That’s it in a nutshell. If a style with multiple trailing authors
eventually emerges, it might require that the last of the trailing
authors be preceded by “and”, or it might not. I’ll work with any
decision that emerges from the group, but I think we’re
over-engineering this. My personal preference would be to do the
minimum needed to cleanly resolve the known use case, and move on.
That would mean sticking with values “true” or “false” for the
present.

The reasoning behind that preference is that the time we invest in
discussing the multiple-trailing-authors case does not produce any
immediately useful results, and quite possibly will turn out to have
been wasted effort. Avoiding a distant and uncertain prospect of
doing some code refactoring down the road isn’t worth the candle, in
my view.

Anyway, here are the options and the trade-offs.

[A] Implement et-al-use-last, with values false or true
Risks: if multiple trailing authors is ever required, then …
(1) must change data type
(2) must refactor code

These aren’t the only options with this route. The other two are:

a) won’t fix
b) invent another new attribute

[B] Implement et-al-use-last, with values 0 or 1 only.
Risks:
(1) if multiple trailing authors is ever required, must refactor code
(2) documentation on multiple trailing authors encourages
emergence of such a style

[C] Implement et-al-use-last, with arbitrary numeric value
Risks: final join is a blind guess, so …
(1) if multiple trailing authors is required, might need to
refactor final joins
(2) further discussion will be needed to settle the best-guess
choice of final join
(2) if the option is never needed, development effort is wasted

[D] Implement et-al-use-last, with arbitrary numeric value, and
specify post-ellipsis join
Risks:
(1) further discussion needed to settle how to control final join
(2) if the option is never used, development effort is wasted

My own vote is [A], then [B], then [D], then [C], ranked from most to
least desirable. But I’ll follow whatever the group decides.

I think I’m leaning towards A at this point, with the caveat that if
we go this route, I’d prefer to foreclose any future extension (e.g.
once a boolean, always a boolean). There’s a reasonable point to be
made that the APA is ridiculous to begin with, and the only reason
we’re bothering is because it’s APA.

Bruce

My vote is for A. I wouldn’t want to foreclose anything (who knows what the
exciting world of citation formatting will bring next?), but I agree that we
might only want to consider extension if one of the major styles starts
requiring it.

Rintze

Just letting you all know that I wrote the spec and schema entries for
the et-al-use-last
and name-use-last attributes. Comments welcome. I followed option A: boolean
options with an implicit ellipsis character.

http://bitbucket.org/bdarcus/csl-schema/issue/14/list-last-authors-name-in-truncated-authors#comment-308933http://bitbucket.org/bdarcus/csl-docs/issue/3/spec-language-for-delimiter-precedes-et-al#comment-241275

Rintze