PyPy test status

Bryan O'Sullivan bos at serpentine.com
Wed Dec 23 07:02:39 UTC 2015


I just dropped a pile of patches that fix up some problems with running the
test suite under PyPy 4.0.1.

There remain 18 test failures.

Of those, 15 are all down to a single bug that causes all histedit tests to
fail. The cause relates to histedit's use of cPickle to save and load
state. You can see a backtrace here, based off fe376159a58d:
http://pastebin.com/51fMquyX

It seems plausible to me that this could be a bug in PyPy's
reimplementation of cPickle, but I haven't yet investigated.

For laughs, I hacked out a patch to replace use of the cPickle module with
json, and now all 15 of the failing histedit tests pass.

In summary, with my patches and other recent work, there could be as few as
4 real spots where test results for PyPy are different from CPython. That's
quite encouraging!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20151222/7e7726ed/attachment.html>


More information about the Mercurial-devel mailing list