I think the sensible suggestion here is to add <trailer> to the content
model of <table>, with the specific semantics of "head but occurring at
the end of the table rather than the beginning" which at least doesn't
do violence to the meaning of either element. Maybe someone would like
to fill in a SF feature request?
I'm curious about what a circular table might look like. I can imagine a
triangular one without much difficulty, and I claim the TEI table model
is (just about) up to representing it.
I'm conscious that we still have some work to do on the content model of
<figure> though...
On 29/04/11 18:34, Paul F. Schaffner wrote:
> Responding to several messages:
>
> -- I agree that *<caption> seems an awful lot like<head> and<trailer>,
> and is therefore suplerfluous.
>
> -- I do not see how one could confuse a head or caption with a<figdesc>:
> the former represent text in the source, the latter an editor-supplied
> description. Where is the confusion?
>
> On Fri, 29 Apr 2011, Martin Holmes wrote:
>
>> I have some degree of uneasiness with a situation in which I'm forced to
>> encode the text of my document out of order (encoding the table caption
>> before the table, when it clearly comes after it) just because of a schema
>> restriction.
>
> I have to agree, especially considering all the other things
> that can appear between the beginning and end of the table,
> the position of which relative to the head might be thought
> worth preserving. Page breaks, for example. If a table
> begins on page 5 and concludes on page 10 with a caption,
> do you really want to place the caption, seemingly, on page
> 5?
>
>> This uneasiness extends to lots of other cases where the schema appears to
>> force me to encode things in a particular way, when I really feel it should
>> be different. For instance, I can't have this structure:
>>
>> <div>
>> <head></head>
>> <p></p>
>> <head></head>
>> </div>
>>
>> Why not?
>
> I'd say, because a head that appears at the end of a div is
> a trailer. => head : trailer :: opener : closer. That is why
> we added<trailer> to the table model. That resolves most
> head-in-table problems. OTOH, there are some objects where
> divtop and divbottom material really does commonly appear
> almost anywhere in the middle, e.g. music and illustrations.
> With music, we move the divtop stuff to the top, sometimes
> with ill grace; with illustrations, we simply allow head
> anywhere within figure.
>
>> On 11-04-29 01:40 AM, Lou Burnard wrote:
>>> I can't see what's wrong with this suggestion. The TEI table model
>>> explicitly rules out all sort of things about placement, layout etc:
>>> it's only about the logical structure of the table.
>
> I have always been very suspicious of the abstract table purism
> expressed by this theory, and the unreasonable fear of table
> abuse of which it is symptomatic. A table is a gridlike arrangment,
> full stop. It may have a logic that is more or less extractable (and
> thus translatable into another form), or it may not, but one has to
> capture it regardless, and if the TEI table tags are not suitable
> for this, then one would have to invent tags that were.
>
> (there are also table structures that TEI table cannot handle at all,
> especially triangular and circular tables, but that is another
> problem.)
>
>>> If you move the
>>> heading of a table from the bottom to the top, it doesn't change its
>>> meaning, so the placement is probably not something you want to encode.
>>> If you *do* (perhaps the table has two headings, one above and one
>>> below), then you might need to use @rend or (as Sebastian suggests
>>> @place) to distinguish them, but this seems a fairly marginal case to me.
>
> I can almost see the sense of this, but (1) this would seem to cut
> both ways: if the location of the head does not matter, then why
> not allow head wherever it is convenient? (as we do with heads of
> figures, as above); and (2) why not extend the same logic to<trailer>s in
> <div>s? why not get rid of trailers altogether? After all, a trailer is
> simply a head that happens to appear at the end. Why not simply
> move it to the front of the div and capture it as<head rend=
> "trailing">?
>
> pfs
> --------------------------------------------------------------------
> Paul Schaffner | [log in to unmask] | http://www.umich.edu/~pfs/
> 316-C Hatcher Library N, Univ. of Michigan, Ann Arbor MI 48109-1190
> --------------------------------------------------------------------
|
