Re: [xml] Strange xml:base/CWD problem

[Daniel Veillard <veillard redhat com>]

On Tue, Dec 21, 2004 at 11:07:32PM +0000, Frans Englich wrote:
> On Sunday 19 December 2004 20:37, Daniel Veillard wrote:
> > Part of XInclude xml:base fixup. libxml2 will add the base attribute only
> > if necessary, i.e. if URI references mades from the document resulting from
> > the XInclude processing may get wrong. if everything is in the same
> > directory then those are not needed.
>
> Ok, it is inevitable. From what I can tell, this means that in practice must 
> _any_ XML format where its users may want to use XInclusions must add support 
> for attributes from the xmlns ns(the lang attribute is need to, AFAICT). Is 
> it just me who find that a major setback for XInclusions' usefulness? 

  Generating broken URI-References was considered a far more dramatic
problem. It affects the usefulness of the resulting instances instead
of the - easier to change - framework.

> Afterall, the schema must be changed for the sole reason of supporting 
> XInclude.

  It must be changed too if you want to validate the input documents too.
We are well aware of that, DocBook DTDs had to be changed.

> I'm pondering, is it at all possible to make the implementation not break the 
> validation? Not AFAICT, since the spec says to add the attributes to the 
> result infoset, and that the validation must be done on the result info set, 
> not the source info sets..

  Weakening the validation to avoid taking into account a different processing
model doesn't sounds the right approach to me. It also depends on the 
kind of validation you are expecting ...

Daniel

-- 
Daniel Veillard      | Red Hat Desktop team http://redhat.com/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/