citation delimiter issue

CSL Development
1

So Frank reported a bug in CSL, which if true, would be both a) really
surprising that this oversight slipped through, and b) require an
immediate fix in my view that would impact all implementers.

https://github.com/citation-style-language/schema/issues/40/

Am still not 100% sure though.

Bruce

2

Bumping this with an update; a head’s up for implementors:

So this issue is actually a bit broader than just the delimiters, and
is almost surely to result in a significant change in 1.0.1.

In particular, as I say on the ticket, I think we need add an explicit
notion of author grouping to the spec, to define grouping behavior as
default, and to add some new attribute to optionally turn it off.

This is likely to have little-to-no impact on styles, but some impact
on the test suite and related implementation code.

3

Bumping this with an update; a head’s up for implementors:

So this issue is actually a bit broader than just the delimiters, and
is almost surely to result in a significant change in 1.0.1.

In particular, as I say on the ticket, I think we need add an explicit
notion of author grouping to the spec, to define grouping behavior as
default, and to add some new attribute to optionally turn it off.

This is likely to have little-to-no impact on styles, but some impact
on the test suite and related implementation code.

This would change output significantly without any change to styles.
Are you sure that’s a good idea?

4

Here’s what I think the situation is:

There’s a fundamental bug in the spec that only gets exposed under
certain circumstances that are not all that common.

So I think the change is unlikely to lead to widespread problems, but
will reduce the sort of bug reports we’ve seen over the last week.

This is WRT to author-date styles. I need to think more about the
implications for note styles. If they don’t typically include sorting
instructions, then obviously it’s not a problem.

Bruce

5

Bumping this with an update; a head’s up for implementors:

So this issue is actually a bit broader than just the delimiters, and
is almost surely to result in a significant change in 1.0.1.

In particular, as I say on the ticket, I think we need add an explicit
notion of author grouping to the spec, to define grouping behavior as
default, and to add some new attribute to optionally turn it off.

This is likely to have little-to-no impact on styles, but some impact
on the test suite and related implementation code.

This would change output significantly without any change to styles.
Are you sure that’s a good idea?

Here’s what I think the situation is:

There’s a fundamental bug in the spec that only gets exposed under
certain circumstances that are not all that common.

So I think the change is unlikely to lead to widespread problems, but
will reduce the sort of bug reports we’ve seen over the last week.

It might be a good idea to poll Zotero and Mendeley and their user
communities before making the change. Any difference in implicit
behavior is likely to produce a reaction among users.

6

Any suggestions of how to put the poll question to very clearly lay
out the problem for non-technical users? Something like the following?

If you have a citation style where you chose to sort the citations by
date, which of the following expected results would you expect?

a) (Smith, 2000, 2010; Doe, 2005)
b) (Smith, 2000; Doe, 2005; Smith, 2010)

… with maybe an equivalent for note output?

Bruce