[tei-council] Re: figurin' about <figure>

Syd Bauman Syd_Bauman at Brown.edu
Wed Jul 25 11:37:38 EDT 2007


SB> [need <text> or <floatingText> in <figure>]
LB> certainly not <text>in any case

OK, then <floatingText> it is.


SB> Do you want to defend permitting
SB>    <figure>
SB>      <p/>
SB>      <graphic/>
SB>      <figDesc/>
SB>      <head/>
SB>    </figure>
SB> ?
LB> I think this is preferable to complicating the model.

As always, I ask for whom is it preferable? I don't think that making
it easier on us to maintain the content model is a reason for having
bad content models. 


SB> ... use <ab> where the semantics of <p> aren't really right
LB> I agree that <ab> might be more appropriate than <p> for random
LB> biyts of texts floating about inside an image.
AC> Agree

I think I'm glad to hear that, but I want to double-check that we're
on the same page. I wasn't talking about text *inside* the image,
which heretofore has been encoded inside a <text>, and it seems
hereafter will be encoded inside a <floatingText>. Rather, I was
talking about bits of text that orbit the image, as it were.


LB> Not sure what the difference between a caption and a head is
LB> within a figure
SR> indeed. if <head> isnt a <caption>, I am gobsmacked. seems like
SR> angels on a pin to me.
AC> An image may have its own <head> in the source that doesn't
AC> necessarily correspond with what we call a caption (more
AC> descriptive).

I think Arianna's got it right, here. While I agree that there are
many cases where only one or the other is present, and I am sure that
there are cases where it is hard to distinguish, I think that it is
very often quite clear which is which. 

I just opened Knuth _The Art of Computer Programming_ volume 3
_Sorting and Searching_, and thumbed through a bit, stopping at a
figure that intrigued me because it looked a bit like a typical tree
representation of XML. Associated with the image in bold typeface is
"Fig. 24.", and then, in roman typeface, "RANK fields, used for
searching by position".

I think most of us would be inclined to encode
   <head>Fig. 24.</head>
   <p>RANK fields, used for searching by position</p>
in P4. I think very few encoders would want to call them both
<head>s. I do think some folks would think of "Fig. 24." as a label,
not a heading, though.


AC> However, do we need an <ab type="caption> or could <figDesc> take
AC> that role?

I don't know that we need <ab type="caption">; we may conclude that
captions are a flavor of <label>, e.g., instead. But <figDesc> serves
a completely different role (describing the image for
search-and-retrieval and for those situations where the image cannot
be viewed), and I do not think we should overload it.


LB> I thought there was an example of this: it's for things like
LB> complex figures with subfigures in them.

I didn't find any case of //teix:figure//teix:figure in the
Guidelines. 

AC> I think I had given a very abstract example in TEI L long time
AC> ago, but I suspect we need something more concrete.

Do you mean the one at
http://listserv.brown.edu/?A2=ind0606&L=TEI-L&P=R7213?
That looks like you were interested in dealing with a whole image and
some fragments thereof. Lou, does that match what you're talking
about? 


LB> I don't think the added complication of this model is worth the
LB> effort, frankly.

Well, in the "nested <figure>s go at the end" case, it's a pretty
simple content model, so I actually think it is worth the effort,
because it isn't much effort.

In the "nested <figure>s can go anywhere" case, I still prefer
the structure over repeated OR groups, but I'm less convinced it is
worth the effort at this point.


SB> I also don't wonder whether <figure> should permit model.egLike
SB> in addition to model.graphicLike. One can imagine wanting to
SB> associate a heading or a coption with an example.

LB> This could work. Or could make model.egLike a subclass of
LB> model.graphicLike

You mean make model.egLike a member of model.graphicLike? This would
have the consequence that <char> and <glyph> could have <eg> or
<egXML> instead of <graphic>, <binaryObject>, or <formula>. Is that
OK? 

We'd have to remove model.egLike from model.phrase, of course.


SR> or add <head> to <exemplum>

My gut instinct is that we should do this whether or not we add
model.egLike to <figure> somehow.




More information about the tei-council mailing list