On 1998-05-01T18:06:54-04 Espen asked about what to do with
the dot (sometimes called "period") that usually separates
the speaker from the spoken speech in his drama:
> <sp><speaker>FILCH.</speaker><p>Sir, Black Moll ...
but occasionally separates a stage direction from the
> <sp><speaker>NORA<stage>something</stage>.</speaker><p> ...
and laments that STAGE is not permitted inside SPEAKER. Lou
makes a case for allowing STAGE inside SPEAKER.
While I have no disagreement with Lou's suggestion, I do
not think that the position of the dot alone is sufficient
information upon which to determine whether the structure
Other clues, like use of whitespace, typeface, typesize, etc.,
should be considered. (BTW, I'm not implying that Lou thinks
the dot is sufficient.)
At the WWP, the dot is considered renditional markup, not data, and thus
is encoded on the rend= attribute. This allows our vaporware processing
software to either pay attention to the information on the rend=
attribute and thus output something that looks similar to the original,
or to ignore the rend= attributes and output something that looks modern
It would be encoded on SPEAKER or STAGE, depending on where it
<sp><speaker rend="case(allcaps) post(.)">Filch</speaker>
<p>Sir, Black Moll ...
<stage rend="post(.)">something</stage><p> ...
 Something Peter Flynn said a few months back bears repeating.
In response to Sue Fisher's question:
if your SGML tools cannot do anything meaningful with [a valid
encoding], then how meaningful is it to [encode said feature]
properly in the first place? OK, so that question was
rhetorical, but how ... do other projects manage this problem
from the data-input and data-retieval end of things?
Just because the current crop of tools is broken doesn't mean
we should penalize the next generation of researchers. If
SGML-encoded text is going to last forever :-) the least we
can do is make it reasonably bullet-proof.
With which I couldn't agree more. At the WWP, in general, we manage
the data-input end of things by trying to be painstakingly correct,
even if current software can't handle it.
[a] This was really supposed to be a reply to Espen's posting of
1998-05-18T12:44:34-04, itself a reply to Lou's posting of
1998-05-15T17:43:06-04, but Brown's news server (which holds
mailing lists, too) seems to have lost them.