[XML-SIG] Worth keeping xml-ref?

[Martin v. Loewis]

Andrew Kuchling <akuchlin@mems-exchange.org> writes:

> On that note, while making the list of modules (there are something
> like 277 *.py files under xml/ !), I noticed a few things that could
> be deprecated.  
> 
> * xml.schema.trex : Trex effort is now going into RELAX NG.

Until RELAX NG is integrated, I'd like to keep trex. I'm sure anybody
using the trex module will be aware of its status.

> * Should the SAX1 classes be deprecated?

Sure, although I'm not so sure how to actually implement the
deprecation: We want to support Python 2.0, and this doesn't have the
warning framework.

Also, people still ask for PyXML 0.5 features from time to time
(because of code they found on the net); I expect that the SAX1-based
applications will stay around for much longer.

Since the code does not introduce maintainance burden, I'd keep it.

> * qp_xml: does it provide anything that pulldom doesn't?

That's a good question - I've never heard of any qp_xml user.

> * xml.dom.ext.Visitor: is this still needed with the addition of 
>   DOM Level 2's TreeWalker?

Probably not - although I believe that there are a number of bugs
hiding in the traversal-and-ranges part of 4DOM. We really should pass
a third-party test suite before telling people to migrate to it.

> Deprecation would have to be gradual, of course, following Python's
> approach of "remove docs in first version; add warnings in second
> version; remove code in third version".  

I think this could be done at different paces for different
modules. I.e. if no user indicates qp_xml usage, I'd happily outright
remove it from 0.8 - if somebody the complains, releasing 0.8.1 that
adds it back in (or, put an ImportError at the top).

For SAX1, I'd be much more hesitant, like keeping it until after PyXML
1.0.

Regards,
Martin