Message144325
| Author |
ncoghlan |
| Recipients |
Yury.Selivanov, alex, asvetlov, eric.araujo, gcewing, ncoghlan, rfk, rhettinger, rndblnch, zbysz |
| Date |
2011年09月20日.12:29:29 |
| SpamBayes Score |
6.499418e-11 |
| Marked as misclassified |
No |
| Message-id |
<1316521770.85.0.386296695265.issue11682@psf.upfronthosting.co.za> |
| In-reply-to |
| Content |
I have updated the bitbucket repo with changes to address most of Benjamin's review comments.
A few points of note:
- I agree in principle with the idea of splitting Yield and YieldFrom into distinct AST nodes, but I'd prefer to focus on getting the current implementation into the main repo first
- in cleaning up the handling of missing send/close/throw attributes on subiterators I discovered that the PyObject_CallMethod* APIs were discarding exception information and coercing everything to a terse AttributeError. The branch now changes them to allow the error reported by the underlying call to PyObject_GetAttr to pass through unmodified.
- the generator close() method treats an AttributeError as expected when looking for a close() method on the subiterator, but uses WriteUnraisable to deal with anything else.
- I share Benjamin's suspicion that some of the INCREF/DECREF pairs in genobject.c aren't actually necessary, but I don't think it's worth messing with them at this stage.
I haven't looked at Zbyszek's proposed doc changes at this point - the pull request has a lot of irrelevant line wrapping changes in it, so it makes it hard to review. |
|