This is the mail archive of the docbook-apps@lists.oasis-open.org mailing list .


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: libxslt/xsltproc and XML catalogs


On Sun, Nov 25, 2001 at 11:32:11AM +0000, Dave Pawson wrote:
> At 05:55 25/11/2001 -0500, Daniel Veillard wrote:
> > > I'm a bit confused, does this mean that I will need to put both
> >
> >   Me too about the where the distinction between URI and System really is,
> >so far I didn't got a good answer :-)
> >
> >   http://www.oasis-open.org/committees/entity/issues-15.html
> >   Issue 30
> >   "because they are different" ... ah !
> 
> +1 on the confusion. I said it badly, but I commented to Norm the other day
> that I found two items quite 'similar' and confusing.
> 
> How about renaming them system and web?

  The name ain't the problem. I don't see why one should use 2 different
mechanism to lookup a catalog depending on the place in the source where
the URI-Reference was made !

  The XML spec is 100% clear on this, the SYSTEM ID is an URI-Reference.
    http://www.w3.org/TR/REC-xml#dt-sysid

It's just in the case that an Public ID is provided that a different
URI-Reference may be used to lookup the entity. This suggest to me that
a DOCTYPE without a Public ID should be resolved in the exact same way
as if that reference had been done for other purposes (like an XInclude
text inclusion of that DTD or entity for example).

XML Catalog provides support for System lookup with no Public ID, I still
didn't have seen a single example why this can't be applied in a generialized
way for URI-References outside of entities lookup, and why this artificial
duplication URI/System (or Web/System) is needed or is in any way correct 
w.r.t. the whole XML and URI framework.

  But currently we are stuck with:
    - applications having to bet whether the URI reference they got in their
      resolver layer came from a DOCTYPE, ENTITY or something else (why on
      eath should they had to know ???)
    - application having to expect that catalog may be built with System
      construct while URI should be used in those case
    - XML Catalog tools having to guess whether the URI/path which get added
      might be used for either URI/Web or DOCTYPE/ENTITY and ending up
      duplicating every entry in the catalogs to avoid problems
 
  This distinction is artificial, forces to carry or guess scope informations
at levels which really shouldn't need/know it, and I still didn't see a single
occurence of where and why this might be useful.
  Hard to justify on architectural grounds IMHO, and a PITA w.r.t user's and
tools support in practice.

  Again, I sill appreciate a lot everything else in the XML Catalogs
specification !

Daniel

-- 
Daniel Veillard      | Red Hat Network https://rhn.redhat.com/
veillard@redhat.com  | libxml Gnome XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]