[tei-council] layers again

[Brett Barney]

On 15 October 2011 at 05:09:17 AM Lou Burnard wrote:

> First, the name of the elephant (the element which contains a genetic
> description) : the solution so far preferred is <sourceDocument> or
> <sourceDoc> I prefer the latter because we use "doc" in the same way in
> <docStatus>, <docAuthor> etc. so that's what I propose to implement.

Makes sense to me.

> Second, the name of the element used to describe a set of changes, and
> of the attribute used to point to them (stage, layer, changeSet...)
>
> There are problems with all of these. Following discussion with
> geneticians and others here at the TEI MM, I now have a new proposal: we
> use the existing <revisionDesc>, No, hear me out.

. . .

> So the proposal is
>
> a) make some minor changes to the wording of the existing <change> and
> <revisionDesc> elements and provide some new examples
>
> b) change the content model of <creation> to include <revisionDesc>
>
> c) add a new @change attribute, which can point to a <change> element
>
> I like this idea a lot, and so does everyone I've discussed it with so
> far. It involves very little disruption and gets us a lot further
forward.

Hmmm. Very interesting. I don't have objections to the general proposal,
and if it does in fact offer a sort of fast, inside track for getting to a
real implementation of this element that's a big merit. One thing that's
not clear to me yet is the attribute set that is being proposed for
<change>. I did a quick comparison of those available for the current
<change> element and those listed in the draft spec for <layer> and they're
pretty different. I know, though, that the formal specification isn't the
best developed part of the draft and so the content model that's outlined
for there for <layer> might not represent very well the content model that
folks have been working toward. For one thing, one of the examples in the
narrative part of the draft uses @target, which isn't listed in the spec.
But @target isn't available for <change> either. So we'll need to do some
work reconciling the attribute sets, it seems. Any thoughts about how much
reconciling will be called for?

Brett