[tei-council] From <figure> to module dependencies
Laurent Romary
laurent.romary at loria.fr
Thu Nov 19 04:13:25 EST 2009
Le 19 nov. 09 à 10:00, Sebastian Rahtz a écrit :
>
> On 19 Nov 2009, at 08:26, Laurent Romary wrote:
>
>> The discussion and module and the "all in core" principle stated by
>> Sebastian shows that we will need rather soon than later a move
>> towards a better architecture for modules. If we had a means to
>> elicit
>> module dependencies (<moduleRef> in <moduleSpec>?), this would allow
>> us to have nice, focused modules that could serve as libraries of
>> element to others.
>
> As I was remarking to Lou and Laurent separately recently, we must
> not lose sight of the fact that we increasingly enter the more complex
> world of both writing a specification and implementing it. It would
> be easy
> (relatively) to add markup to ODD to express the problem, but the
> bigger work
> is in the implementation of if in tools for user which make sense of
> that markup.
>
> I can think of lots of ways foreword, but one might be for someone to
> write a free-standing "oddlint", which would do the same work as the
> sanity
> checker in Roma, but more besides, including module dependency.
Agree. We need to carry out the two activities (spec + dev) in
parallel. One difficulty is to find evolutionary steps which are both
conceptually sound and implementable. We need to take our time on this
before we actually publicise it widely, not only because the
implementation must be there, but also to be sure that we understand
the consequences of such a move as having, says, moduleRef in moduleSpec
Whatever happen with the ESF grant (for newcomers: we deposited a
grant to pay for a prospective workshop on "the future of the TEI" to
the European Science Foundaition), we should probably keep some time
on this at the F2F meeting in March.
More information about the tei-council
mailing list