[Zope-Coders] Re: test failures in TAL

[Martijn Faassen]

Evan Simpson wrote:
> Martijn Faassen wrote:
> > Yes, but that is true for any Zope product, so this is no explanation.
> 
> It is especially true for Zope Products that aren't included in the core. 
> It would, of course, be easy to declare that future versions of these 
> Products are Zope 2.4+ only. Since that drastically lowers the number of 
> users of the new release, at the moment, we're reluctant to do that yet.

But ParsedXML is one of the products where this makes the least sense of
all, because as far as I'm aware the number of ParsedXML users currently
is very low.  It's currently so slow it's next to useless to use directly
from within Zope, for one.

Those users that do use the DOM from Python can continue using the old
version. It's not like they're screaming for upgrades, as the last
release was back in april (with a windows release of the same version in
may).  

Add to this the whole unicode issue that has rather special impact on
XML and which is hard to support in a unicode-enabled version as well
as a no-unicode version, and the need to carry around an extra C
module (expat parser) that is also included with Python 2.1 (though
perhaps in a version insufficient for our needs.. nobody has spelled
this out yet)...

The tradeoffs seem to weigh in favor of dumping 1.5.2 support as quickly
as possible.

It seems decidedly weird that we're continuing to 'support' ParsedXML
in a 1.5.2 version which is used by so few people in the world, 
while Zope itself requires now Python 2.1.. I would understand better
if ParsedXML were a product in which Python 1.5.2 is easy to support and
if it had lots of users.. I'm developing the thing (the last months I
seem to be the sole person to) as it's not ready for *my* use yet!
(multiple hundreds of unit tests are failing with Python 2.1, for one
thing)

Anyway, of course I'll try to support 1.5.2 if this is a requirement,
but it would be nice to hear some feedback to these arguments.

Regards,

Martijn