On Wed, 3 Dec 2003, Tobias Rischer wrote:
> Which leads to "rend", which as an attribute was supposed to describe
> the rendition found in the source, rather than the rendition intended
> in the output, at least in my encoding-centered view.  If a reading
> text form shall be given in a <choice> and the other forms are also
> considered worth preserving, I'd prefer a more declarative encoding
> than a <rend>, if that comes down to <use-this-and-forget-the-rest>;
> but I may have misunderstood the intention behind <rend>.

I agree that I'm uncomfortable with <rend>, probably for a very similar
reason.  the global @rend attribute is intended to indicate "how the
element in question was rendered or presented in the source text", and
that is how I've used it in primary source transcription.  In authoring
web pages I've used (as the TEI website does) <hi rend="bold">, but
in this case the original document is the web page itself.  It is this
use to indicate the resulting presentational display that makes me
nervous of <rend> as an element.  That said, I can't really think
of something better unless <choice> is to have a <default> child.
(But that then sounds even more procedural.)

With Lou's example of:

<choice>
  <sic>dictu</sic>
  <norm>dictum</norm>
  <rend>dictu[m]</rend>
</choice>

I disagree with the idea someone suggested that <rend> (or something)
should not exist and that the processing (xslt, etc.) should somehow
calculate this based on the difference between <sic> and <norm>  (or <corr>
since <corr> and <norm> are different).  That seems to move an extreme amount
of work to the processor, and more importantly would be a barrier to
TEI users less competent with their transformation languages (e.g. XSLT).
Of course, I'm assuming that <choice> children are defined in RELAX NG
as <oneOrMore> so that anyone who wanted to use xslt to calculate <rend>
based on <sic> and <norm> would be able to. Though really, having a
choice automatically implies two things...doesn't it?


I'm curious What other phrase level elements will be allowed as children of
<choice>?

A bit nervous about some of the element names, but I like the model.

-James

---
Dr James Cummings, Oxford Text Archive, [log in to unmask]