[XML-SIG] DOM test failures

[Martin v. Loewis]

>   Which brings back the idea that what we need is something new.  How
> many of us are going to be at Python 10?  (Martin?  Lars?)  Would a
> BoF or DevDay session be useful?

I'm not convinced that we need something new (although Lars seems to
think so). Something new would have all the disadvantages of the DOM,
plus
- it would come with no documentation whatsoever (not even a spec
  to refer to),
- it would have buggy implementations, and people would change it forth
  and back,
- people would complain that something they want to do with it cannot
  be done (like: knowing where exactly the entity references where in
  the original text)
- if it is changed to do these things, others will complain that it is
  too complicated to use.
- after a while, it would become unmaintained, and
- nobody new to Python and XML could relate to it.

[Please understand that I'm not talking about some specific library
 already in existence - this is idle speculation, and about the future
 :-]

I think PyXML has seen a number of such libraries in its life, the
last one that is still incorporated is qp_xml (is anybody using that?).

I'll be at Python 10, and I'd certainly go to a BoF seession on the
new Python/XML object API. Maybe we can produce a requirements list.

Regards,
Martin