[XML-SIG] pyexpatreader default namespace feature setting

[Lars Marius Garshol]

* Lars Marius Garshol
|
| Yes. It's wrong. I've commented on it before, and tried to start a
| discussion on it, but nobody would bite.

* Alexandre Fayolle
| 
| Sorry I missed the oportunity to bite at that time. Now I have ;o)

Good! :-)  We need to get this thrashed out and agreed on, once and
for all.
 
* Lars Marius Garshol
|
| Not since revision 1.11, where I corrected that. What revision are you
| looking at?
 
* Alexandre Fayolle
|
| I was looking at the one that shipped with PyXML-0.6.6. 

Strange, because I fixed this two months before that version went out.
Revision 1.11 was made on 2001-06-07, while 0.6.6 wasn't released
until 2001-08-05.  Surely we didn't start a release branch two months
before the release?

| Please note that changing this will break the same fraction of
| application using a validating parser than changing pyexpat with
| application using a non validating parser.

Actually, that's unlikely to be true. I guess the applications that
use validation don't use namespaces, so they probably turn it off
explicitly. In any case, the number of applications using validation
is much lower that the number of applications that don't, so in
absolute numbers the breakage is lower this way.

--Lars M.