Re: NVDL?

2008/9/24 Jirka Kosek wrote:

  Hi

> By "handle all of NVDL" you mean if it is possible to express by using
> XProc construct same validation constraints as by using NVDL?
> I will have to check carefully, it should be possible may be without few
> small bits. But anyway such XProc pipeline will rather verbose compared
> to NVDL schema

  I think the Norman's point is that maybe an NVDL implementation is
possible in XProc itself.  In this case, it would be possible to focus
on other issues and keep NVDL to be implemented after the REC has been
issued, in a standard XProc module.

  I don't know oNVDL internals well enough, but I know George has
implemented the dispatching mechanism in XSLT 2.0.  I am not sure the
result can be validated by standard XProc steps for schema validation.

  Of course, efficiency can be an issue too (you have to apply a
meta-stylesheet to the NVDL schema to get a stylesheet, you have to
transform the instance to annotate it with sections, you need to
transform the later with the generated stylesheet, then you have to
validate the different parts in this dispatching document.)

  More generally, I feel XProc is now a little bit like XSLT was 10
years back.  We do know it will be useful, and we need it, but we do
not have experience with it yet...  So I think the more important is
to have the core, the processing model well-defined.  For the step
library, we won't think about everything anyway, and I am sure
something like EXProc will emerge as we gain feedback from experience
from the real world.

  But for NVDL, sure, I think this is important to have it ;-)

  Regards,

-- 
Florent Georges
http://www.fgeorges.org/

Received on Wednesday, 24 September 2008 08:59:16 UTC