Remove ambiguity from two-field names that have only a family value

[fbennett]

Some users have been confused by the handling of names in Zotero 2.1. citeproc-js discriminates between personal and institutional names, and will format these correctly when authorship is by a mixture of persons and the institutions with which they are affiliated. It's not a bad feature, but the interface needs a small adjustment to make it more intuitive.

Currently (Zotero 2.1.5, citeproc-js 1.0.144), any name with a value only in the last name field is treated as an institution, in both Zotero field modes. With the attached patch and citeproc-js 1.0.145 or higher, names in two-field mode will always be treated as personal names. With this change, the UI becomes smoothly intuitive (one-field mode === institution, two-field mode === person). It should pretty well eliminate this category of questions/complaints from the forums.

Looking back, this behavior really should have been documented in the processor manual, so it could be picked up and discussed. But the last small glitches in behavior have been ferreted out in response to user feedback, and if this UI change can be applied, it should work out quite smoothly.

(Note that the patch and the processor upgrade do not absolutely need to be introduced together, although that would be best. If the processor is upgraded without this patch, all names will be treated as personal names (Zotero 2.0.9 behavior). If the patch is applied without a processor upgrade, current Zotero 2.1 behavior will remain the same.)

[dstillman]

We're not prepared to introduce this patch without additional UI support, since single-field mode was never meant to be equivalent to institutional creator mode (which we always intended to introduce as a separate field mode).

A couple options:

1) Add a new fieldMode for institutional creators. This would be the most explicit and would require no conversions by users of existing data, but it would require a decent number of code changes.

2) Switch to your proposed data model, but add additional UI support such that, to the user, it looked like there was a new institutional mode. All existing single-field values would become institutional creators, but at least it'd be clear in the UI.

We're inclined to go with #1, but what are the practical formatting consequences here? What would happen, formatting-wise, to a "Mao Zedong" entered in single-field mode?

[fbennett]

That's fair enough. Behavior should return to that of 2.0.9 with the processor upgrade.

A "Mao Zedong" coming in on "family" without an institutional hint will render as a personal name, rendered without modification. If single-field mode were co-opted for institutions, that effect would be achieved by entering "Mao Zedong" in the "family" subfield and leaving the "given" subfield empty.

[fbennett]

Attaching the original proposal that covers the formatting of personal and institutional names, which has gone away from its original home at Google Groups.

[dstillman]

OK, but what would happen to "Mao Zedong" entered in single-field mode if treated as an institution? As in, if your patch went through (regardless of any UI changes we might make) and a user didn't change the name to two-field mode with the given name empty?

[fbennett]

With the patch, in that case it would render normally, if as a single name. The unexpected behavior kicks in with multiple names, in changes to the delimiters (i.e. "John Smith [2-field] & Mao Zedong [1-field]" would be misrendered as:

"John Smith, Mao Zedong"

"Joe Brown, Mao Zedong [1-field] & John Smith [2-field]" would be misrendered as:

"John Smith, with Joe Brown, Mao Zedong".

(There are a number of other more complex examples in the proposal document.)

The renderings make sense if the name at "Mao Zedong" position is a true institution, if accidentally applied to personal names, they certainly could alarm a user who doesn't understand what's driving the behavior.