Hi,

On Wed, 2005-09-07 at 12:08 +0200, Kasimier Buchcik wrote:
> Hi,
> 
> On Wed, 2005-09-07 at 10:30 +0100, Henry S. Thompson wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> > 
> > 1) I think it's a bug that XSV allows an import of the same (=none in
> > this case) namespace as that of the enclosing schema doc.
> 
> OK. Would a schema processor be considered as not working correctly
> if ruling out such imports, because Henry said he thinks it's a bug
> otherwise? ;-)

OK. I'll disallow such an import in Libxml2.

Additional tests with Saxon 8.5.1 and IBM's SQC yielded that
none of the 4 schema processors I used for testing thinks that
this is invalid. So please bark loud if you change your mind.

> > 2) If it were an include (and it looks like XSV is treating it as
> > such) then yes, the results would be eligible for chameleon renaming.
> > 
> > 3) If it were a legal import, e.g. of a non-null namespace, then the
> > results would, obviously, not be eligible.
> 
> > 4) The whole logic of import/(chameleon) include/redefine needs to be,
> > and will be, cleaned up in XML Schema 1.1.
> 
> Good. I refer here to XML Schema 1.0 and hope that the
> schema people don't get tired to try to clarify such cases for XML
> Schema 1.0 to the most possible extent. 1.0 will accompany us for
> a long time until 1.1 becomes a recomendation; it will
> take time for schema authors to migrate to 1.1; plus not all
> schemas will be rewritten for 1.1.
> 
> Woudn't it be better to clean this up in XML Schema 1.0 as an
> erratum and then just take over to XML Schema 1.1? Why leaving
> 1.0 here in a diffuse state?

Regards,

Kasimier

From adi@v... Thu Oct 06 10:11:53 2005
Received: from wiggum.w3.org ([128.30.5