Hello Malthe, Tiny problem is that the doctests seem to have the attributes in lxml-sane order. That means just by removing lxml output-sanitization even zpt output does not match the doctest. Also, lxml is used to do some xpath queries. Saturday, May 23, 2009, 10:23:16 AM, you wrote: MB> 2009/5/23 Adam GROSZER <agroszer@gmail.com>:
The problem that I see here with lxml is that it is used for output checking. Even worse z3c.form requires at least 2.1.1 of lxml, where KGS 3.4 has lxml nailed at 1.3.6.
MB> It might be possible to shed this testing dependency; ``lxml`` is used MB> because of its doctest-mode, but since then, Chameleon has been MB> equipped to render attributes in the ZPT order (instead of more or MB> less random). This was the raison d'etre to use lxml in tests.
This burpes already on buildout. Now even if I would ignore this requirement for testing, (and testing) how could I be sure that tests pass (with KGS 3.4)?
MB> Yes, I agree with that observation. MB> \malthe MB> _______________________________________________ MB> Zope-Dev maillist - Zope-Dev@zope.org MB> http://mail.zope.org/mailman/listinfo/zope-dev MB> ** No cross posts or HTML encoding! ** MB> (Related lists - MB> http://mail.zope.org/mailman/listinfo/zope-announce MB> http://mail.zope.org/mailman/listinfo/zope ) -- Best regards, Adam GROSZER mailto:agroszer@gmail.com -- Quote of the day: What reason have atheists for saying that we cannot rise again? Which is more difficult: to be born, or to rise again? That what has never been should be, or that what has been should be again? Is it more difficult to come into being than to return to it? - Blaise Pascal