[lxml-dev] iterparse() reimplementation cancelled

[Stefan Behnel]

Hi all,

just a quick note regarding the planned iterparse() reimplementation. I did
not include it in 2.0 alpha5, as my attempts have shown that passing a custom
parser at the API level would imply too many semantic quirks, especially since
iterparse() can't support every option that the normal parsers can. Also,
parsing against a TreeBuilder (as ET does) would kill a lot of features such
as HTML tag soup fixing, DTD validation and default attributes.

So I now think that it's better to keep the current implementation based on
the normal parser, even if it is not completely safe to use with all working
ET code (as mentioned in the compatibility docs).

Any comments or objections to this decision?

Stefan