2013/5/28 John Ralls <jralls@ceridwen.us>


You might find the discussion in https://github.com/FamilySearch/gedcomx/pull/239
interesting as well. Did you notice the citation-template spec [1] that it
created? Unfortunately it lays out a storage spec of the sort that Tom Wetmore
and I objected to in #80 [2].

If you're concerned about transmitting citations in a way that won't mess up
another user's database, keep the citations as simple strings that use a
standard citation style. If you don't want to use Mills's, that's OK. Just
use something that others can look up in a book or online. As Mills says,
"Citation is an art, not a science". There's no good reason to invent your
own. BTW, "folio, page, line, volume" are possible parts of *one field* of a

Just my 2 cents as somebody who published.

1. as a researcher you only want to store fields, and be done with it. So folio, page, author, ...
2. hence, you need to define fields connected to a source type
3. style is important for publishing. You want many possibilities. This is fixed with eg bibtex a long time ago. If the fields are well designed, a publisher can change styles on the fly. So, in academics, we send the fields to a publisher, who uses his own style files to handle them.
4. The problem of "sending citation styles" is a non-issue if the possible fields are agreed upon, and the source types are agreed upon.

Bibtex source types are inadequate for this, the only ones defined are:

Unfortunately, Mills already defines 170 entry types for genealogy, and as Enno says, they don't cover the entire world.
Something like GEDCOMX or other standardization efforts, should work on defining some sort of common base for source entry types, and possible fields which styles then can work with. Something like max 30 root types. We can then create extra types, but if they fit in a root type people agree upon, others will be able to put in nice reference style everywhere.

Yates wrote me that already 3 programs used his templates, we might do so to now. So we will all have our own internal 'definition' with no standardization whatsoever. A pity, but software producers want to ship products with improvements, so understandable.