See the documentation of each class for details.

It is a rather big commit, however seperating it is non-trivial. The tests
for all of these features all rely on using each other, so there's a
circular dependency in the tests and I wanted to commit the tests and
features at the same time. (However, the non-test-code does not have a circular
dependency.)

A method for cloning nodes. I expect this one to work on all descandants, but
it can be overriden if a node has special needs. It seems natural to put
such core functionality in the node classes rather than in a visitor.

There was an option in VisitorTransform for this but it was way too obscure,
better to have a seperate class.

This facilitates using the parser and compiler with runtime sources (such as
strings), while still being able to provide context for error messages/C debugging comments.

put the annotated .html file next to the .c file to fix annotation when the .c file is not in the same directory as the .pyx file

We ignore *.pyc, vim swap files, build results under BUILD/ and Lexicon.pickle

This is handy, because otherwise, say after runtest.py run, 'hg status' shows
lots of unrelated info, thus lowering signal-to-noise ratio.

Works with test cases for a single class, have not dealt with subclass issues
yet.

make test for new style exceptions run on all Python versions by not actually testing for new style exceptions :)

another Main.parse() cleanup to make sure a unicode error really comes from the parser, not from the filename decoding

--HG--
rename : tests/broken/getattr3call.pyx => tests/run/getattr3call.pyx

I didn't realize how unclean the seperation between the two was, hopefully Dag's project will clean this up soon.