Re: [xslt] Patch for 135938

["Brett Kail" <junkaddr hotmail com>]

Hello,

> However, the "proper" fix is to achieve the objective (change the
> pointer from pointing to a dictionary entry which includes the
> namespace, over to a dictionary entry which has only the NCName).
> I'm working on that one (rather slowly, I'm afraid, because of some
> other unrelated work), and it will be fixed before the next release.
>  If you would like to propose an enhanced patch to achieve what I've
> described, that would (of course) be welcome!  It would also be wise
> to double-check whether a similar problem appears anywhere else in
> the library.

I tried this, but xsltGetQNameURI modifies the string it's given in order to
look up the namespace.  I suppose it could do:

    save = qname[len];
    qname[len] = 0;
    ns = xmlSearchNs(node->doc, node, qname);
    qname[len] = save;

.but that didn't seem all for appealing.  Alternatively, an
xmlSearchNsForPrefix that is given the number of characters to use from
qname could be added to libxml.  I can pursue this further if you like, but
which approach do you think is best?

> On your final question, it's probably better to post patches which
> are directly related to an open bug onto bugzilla, and (if
> necessary) add yourself to the notification list for the bug.  My
> thinking here is that there are probably quite a few list
> subscribers who don't really care to receive that sort of
> information unless it would be critical to normal usage.  Anything
> posted to bugzilla is automatically forwarded to both Daniel and
> myself, so there is no particular advantage to post to the list as
> well.

Ok, I've no problem with conversing via bugzilla.

When I reported this bug, I posted it to the list first to make sure the bug
was legitimate.  Is this preferred, or should new bugs just be opened and
then closed if they're invalid?

Thanks!
-Brett