#473 add @type to desc

[Martin de la Iglesia]

<desc> is a repeatable element, so @type should be allowed to distinguish multiple <desc> elements within the same parent element.

Example:

<graphic>
  <desc type="publication_status">not published</desc>
  <desc type="availability">not public</desc>
  <desc type="camera">Canon XYZ</desc> 
</graphic>

See also TEI-L: http://listserv.brown.edu/archives/cgi-bin/wa?A2=TEI-L;acc92b4.1309

[Sebastian Rahtz]

I can see no reason not to allow this, unless the reason it is repeatable is to allow for the same in multiple languages. If we really believe you can only have one , semantically, then we should say so explicitly.

[Lou Burnard]

I can see two reasons for not doing this. The first is that we only allowed desc to repeat because we want to permit descriptiopns in multiple languages. There is an implicit assumption that sibling descs are to be taken as alternatives, rather than combining them into some higher level multi-faceted description. The second is that @type if used in this way should be saying something about the type of description (e.g. its length or target audience) not about its subject matter. (Yes, this is the same argument as the one I lost about using @type on persName to specify a person's role rather than the kind of name it is). I've no objection to proposing a new attribute e.g. @scope or @role for desc if it is to be used as a generic something-we-dont-have-a-better-tag-for element, though the first problem remains.

[Sebastian Rahtz]

I don't really understand the difference between type of description as opposed to subject matter, I am afraid.

The first objection is more important. Saying "implicit assumption" has to raise alarm bells, of course. We may need a new notation to say "only one, but language variants are allowed", even if the derived scheme ends up the same. The question is whether we do want to support the idea of multiple cumulative . I am amused to see that is valid within , whatever that might mean.

[Martin de la Iglesia]

How about this:

<graphic>
  <desc type="short">Canon XYZ</desc>
  <desc type="long">This picture was taken with a Canon camera of the "XYZ" model built in 2010.</desc>
</graphic>

[Lou Burnard]

Yes, short/long seems much more appropriate as a taxonomy for desc/@type than "camera/availability". However is that what anyone wants? Surely specifying that your multiple descs are actually related to multiple topics would be much more useful. We would need to introduce a new description-grouping element however to address the multi-lingual issue I guess.

[Martin Holmes]

Can't the multilingual issue be handled with @sameAs, as suggested on the TEI-L discussion?

[Laurent Romary]

Well multilinguality in has since ages been dealt with (e.g. in the guidelines) by means of a difference in xml:lang
I would not complicate the mechanism superfluously.

[Martin de la Iglesia]

Maybe the real problem lies somewhere else: isn't there quite a discrepancy between the prose description of <desc> in http://www.tei-c.org/release/doc/tei-p5-doc/en/html/ref-desc.html and http://www.tei-c.org/release/doc/tei-p5-doc/en/html/TD.html#TDcrystalsCEdc on the one hand, and the many parent elements it's allowed in on the other? The former suggests the use of <desc> in ODDs only, whereas the latter encourages usages such as the one in Gioele Barabucci's example. If <desc> is intended for documentation purposes only, I don't see the need for @type either.

[Sebastian Rahtz]

Martin is right, <desc> has suffered from feature creep, and we need to rein it back or change its documentation. I guess it would be sensible to leave this for November Council meeting, as there isn't an obvious outcome at the moment.

[Martin Holmes]

@type if used in this way should be saying something about the type of description (e.g. its length or target audience) not about its subject matter.

I think this is another situation (along with @resp and @corresp) where Lou's view of the scope of an attribute is much more constrained than what I believe to be the more common view. I'm not saying you're wrong, Lou, but I think we're going to keep arguing to a standstill about these attributes because of this schism. When we discuss this in November, I think we should look at all three of these attributes and decide whether the strict or loose interpretation is the one we want to go for, and then revise the Guidelines accordingly. Decisions on whether they can be used for the purposes suggested in the tickets will follow from that.

[Lou Burnard]

I don't think my opinions on the use of @resp are notably different from anyone else's, and I don't think (even if they are) that conflating discussion of three quite distinct issues in this way is likely to be at at all helpful.

[James Cummings]

I see no problem in desc being repeatable for sake of different languages and different form of classifications. I'm convinced by the possibility to both repeat and classify it, therefore agree it needs att.typed. I can see something with short/long descriptions as suggested with those also being in various languages. Yes, desc has changed, it is now available in all sorts of elements. This outcome was always an extension of the war on free-text attributes, so desc can now be used to describe a wide variety of elements in situ, it is not just for ODDs any more.

[James Cummings]

Ticket closed at 2013-11 face-to-face; implemented.