Re: [xsl] Re: Re: Creating a XSL to return the input

[Daniel Veillard <daniel@xxxxxxxxxxxx>]

On Wed, Mar 05, 2003 at 05:21:34AM -0800, Dimitre Novatchev wrote:
> > > Everybody can and will make their own conclusion about the real 
> > > level of testing of an XSLT processor that does not implement 
> > > correctly something so basic as the child::node() node test. 
> > 
> >   <xsl:template match="@*|node()">
> > 
> > is not child::node() node test. 
> 
> node() is an abbreviation for child::node(), as per spec:

  The select expression of a template is not evaluated as an XPath
expression on the document.  It's a separate code path which from
a given node finds back the templates which applies. At least Saxon
operates in the same way accordingly to a document published on the
IBM developer site a couple of years ago.
  The bug was NOT in the node() XPath evaluation code.

> > You tend to jump very hastilly on 
> > conclusions... which often happens when people didn't go down to
> > actually implementing the stuff :-)
> 
> There are great people, who are actually implementing the stuff and
> implementing it *correctly*. 

I would actually like to see a programmer claiming "I don't make bugs
or errors". Even Knuth after decennies of use of TeX admit he may
have bugs left !

> >   I still don't know what aim you're aiming at but this looks less
> > and less trying to be useful the community on this specific thread.
> 
> Right. Usually such threads are not created at all, because the
> developers are grateful to the people, who have found some bug and thus
> contributed to making their product better.

  You're more than irritating... Have you *EVER* reported a bug
against some of my code ? That bug *IS* fixed in CVS. I fixed it once it
got entered in bugzilla and I got some time to handle it. You don't like
my code, fine ... You don't want to use it, fine ... I have never
forced you to use it or comment on it. Now whining because there is
a bug (that you didn't reported) on software you didn't had to use, or
pay for, is a pretty strange attitude. Blaming me on lack of testing
of my code is nearly funny, I'm among the people who promote the use
of testsuite and regression tests, having myself helped building such
testsuites within W3C and maintaining tools and regressions tests to
check my code against such testsuites. People from Zvon apparently have
run their tests against my code, and I fixed the errors they reported
of course. The only widely advertized XSLT test suite is the OASIS one
but apparently it never got released publicly, apparently you're on
that comitee as http://www.oasis-open.org/committees/xslt/ seems to
point out. Where is the test suite download link ? Maybe it's a 
lack of communication but I don't see any testsuite available, and
I have been waiting for a couple year now for it !

  What I reacted to is your attitude "there is a bug, hence the code
is not worth using" and the "identity transformation" had to be the
Identity while it's not, it's just the identity for the XPath datamodel
subset. Equivalence of the XPath data model doesn't mean equivalence
of the documents by far. It wipes out a huge amount of informations.
Patronizing comments like:
  "Identity is identity -- it must copy everything."
  or
  "node() is an abbreviation for child::node()"

 are acceptable when they are true, and really unpleasant when wrong.
Your identity does NOT copy everything, unless taking a very narrow
view so such a general statement ought to be used with more care.

Daniel

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

 XSL-List info and archive:  http://www.mulberrytech.com/xsl/xsl-list