Re: [xslt] Re: [xml] Some words about libxslt

[Daniel Veillard <veillard redhat com>]

On Tue, Mar 13, 2001 at 10:17:44PM +0100, Tom . wrote:
> 
> Le 05/03/01 14:35:47, Tom . a écrit :
> > I'll try to provide some patch soon...
> 
> Here it is. Not deeply tested but the tests run well.

  Well thanks for the patch but I'm afraid I don't come up with the same
conclusion ...

> This completes the document() (except for the XPointer part) and key()
> functions as well as system-property() and function-available()

  Hum, I'm afraid the way you use xmlSplitQName2() is wrong ...
You don't handle properly the case where the name isn't qualified.
This did show up very easilly since tests/REC/test-12.2-1.xml fails
now (it was segfaulting due to an error in calling fprintf but even
with this removed it simply fails to find the key anymore).
  Also there is a framework for debugging messages, use it instead of
fprintf. BTW, I still don't understand how this managed to pass the tests
considering the call to fprintf("static string", ...).

> I still think system properties should be set in an application-wide hash
> table with local-name and namespace-URI as keys. I'll do this in another
> patch.

  Hum, currently libxslt is stateless i.e. all state is stored in dynamic
data, adding an application wide hash table breaks this nice property, and
I don't see any compelling reason to break this just to store 3 static
(name, values) tupples. So unless you can provide a compelling reason to
do so i won't accept such a patch.

   thanks anyway I will see if I can correct the QName handling in your
patch.

Daniel

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