On Sun, 2012-12-09 at 23:03 +0000, Sebastian Rahtz wrote:
> Just to follow up on myself, I took some of the teitohtml XSL apart over the weekend, and rearranged it 
> so that <p> in TEI converts to an HTML <p> all the time; if there is an unacceptable block-level
> object inside the <p> (a list, a quote etc), the <p> is termined, the block object
> inserted, and the <p> restarted. This seems like a cleaner solution.

I think it is cleaner, but I've noticed that in the bargain we've gained
quotes. If I use the test I initially created (before I decided that
maybe my understanding was confused) and diff the before and after, I

$ diff old/test1-fmt.html test1-fmt.html
<       <div class="p">Blah <span class="quote_inline">bluh</span>.</div>
>       <p>Blah <span class="quote">&#x2018;bluh&#x2019;</span>.</p>

(The above excludes irrelevant changes, like timestamps.)

There may be a very good reason to have the default profile add quotes
(like for instance consistency with other output formats). However, the
upshot for someone who's been working with the previous release of TEI
is a new release that breaks what used to work. A user who wanted quotes
around the word "bluh" in the example above would have them in their XML
or would have had to add them in a subsequent transformation. If this
user tries to produce a document with the changed code base, the
produced document will have extraneous quotes.

> I hope this deals with the problematic CSS issue noted, and avoids the clumsy
> use of <div class="p">
> I don't expect this to be the end of the story, though, there are so many combinations
> of elements one can come across in a TEI text.

The way <p> is now handled works for my use, but looking at just about
any philosophy book I can pull off my shelves, I see lists and quotes
which look to me to be embedded in paragraphs. This I infer by the lack
of indentation at the start of the block of text appearing just after
the quote or list. So it does seem that somewhere down the road someone
is going to be unhappy with interrupted paragraphs. 

Reflecting out loud now (so to speak), I'm inclined to think now that
boiling down block elements to div --- as mentioned by other list
members --- would be the method least likely to create issues down the
road. When I wrote my initial email, I flashed back briefly to xsl:fo
and thought that a solution for html could emulate the block and inline
structure of xsl:fo but that bit of reflexion did not make it into my
final email.