[tei-council] re: module dependency

[James Cummings]

Sebastian Rahtz wrote:
> c) Module dependency does not help element-level dependency.
> Thus if <msIdentifier> has <settlement> in its content model,
> a dependency on namesdates on msdescription cannot guarentee
> a result, because <settlement> may have been removed from
> namesdates. This can be solved by implementing a full schema check
> which sorts out any dangling links (presumably possible to write this,
> but I haven't tried).
> 
> So my personal conclusion is that b) does not really
> buy us anything, since we have no current use for it,
> while a) is a simple convenience we might as well put in.
> c) _should_ be implemented, but I don't know how
> at present.

Can you explain to me again why it is a bad idea to just have element-level
dependency? I.e. the elementSpec gets a @depends so with msIdentifier it has a
@depends which includes settlement and repository (since they are unfortunately
and, I would argue wrongly, mandatory I believe).  Thus when I add msDescription
module, any @depends in that are processed and it includes settlement from
names&dates module but not necessarily the entire module itself or could
include the module but only with the needed elements included.  It doesn't get
rid of the need for modules as a grouping convenience or anything.

Back at work, but jetlagged (but not as much as if I hadn't had been upgraded to
first class!),

-James
(P.s. Some people were interested in my photos, they are at:
http://purl.org/cummings/family/images/Kyoto/ )

-- 
Dr James Cummings, Oxford Text Archive, University of Oxford
James dot Cummings at oucs dot ox dot ac dot uk