[XML-SIG] memory related issues in Python - bug??

[Fred L. Drake, Jr.]

Andrew Clover writes:
 > Indeed. It's known that at least in the minidoms supplied with PyXML 0.6
 > and Python 2.0-2.2, parsing is unlikely to generate a normalised minidom
 > tree. This is listed as issue 22 here:
 > 
 >   http://pyxml.sourceforge.net/topics/compliance.html

A great piece of work, BTW!  When I finally had a chance to look at
that, I found it really nice to see an overview of how we're doing.

As for PyXML 0.6, I don't think I even remember things that far back;
it's certainly true that older versions produced non-normalized trees;
much of the work put into creating normalized trees was to improve
tree-building times by avoiding node creation.

 > I don't know whether this is fully "fixed" in later versions as AFAIK
 > minidom has never promised to deliver a tree in normal form (only DOM Level
 > 3 Load really makes that assertation explicit). However, I have always
 > received normal-form trees in PyXML 0.7+ minidoms.

And if you find one that isn't, please provide a test case; I'd
consider it a bug, even if the API doesn't always clearly guarantee a
normalized tree.  The typical user expectation certainly seems to be
that a new tree is normalized, and that's not unreasonable.


  -Fred

-- 
Fred L. Drake, Jr.  <fdrake at acm.org>
PythonLabs at Zope Corporation