On Mon, Nov 23, 2009 at 11:20 AM, Paul Taylor <paul_t100_at_fastmail.fm> wrote:
> Tatu Saloranta wrote:
...
>> Still, such applications are fundamentally broken. Anything that puts
>> significance to XML namespace prefixes is broken by definition, with
>> just one exception: DTDs are defined in terms of prefix/local-name
>> pairs, so DTD validation unfortunately requires stable namespace
>> prefixes.
>
> In a purest sense yes, but these days with so many applications out there
> that scrape data from websites useing whatever tricks are available and so
> many applications programmed by amateurs but still fundamentally useful its
> a bit churlish to say they are fundamentally broken , if they work they work
> however ithey achieve it.
Well, that's the thing -- whether they work or not is by chance, not
by design or implementation.
They happen to work if all goes well, and only then. So while these
are latent bugs, they are bugs nonetheless.
In a way I would even prefer such slapped together tools being broken
more often, since that gives little bit more incentive for authors to
fix them.
But that's probably just me. :-)
This is just my opinion; if others feel most consistent support for
retaining implied namespace bindings is important enough to be worked
on, that's fine. I am not against adding things that are nice-to-have,
esp. if effort involved is relative to benefits.
It's just that this issue (implied semantics of namespace prefixes) is
one of my pet peeves.
-+ Tatu +-