[PyPy-issue] [issue545] Rare bug in optimizeopt

[Armin Rigo]

Armin Rigo <arigo+pypyissues at codespeak.net> added the comment:

Forget it, by digging irc logs I found
http://buildbot.twistedmatrix.com/builders/lucid32-pypy/builds/5/steps/trial/logs/stdio
.  This issue is first of all a bug in compile_done_with_this_frame() which
should be prepared to receive None from compile_new_bridge().

I agree that what you're investigating looks a bit strange too, but that's more
likely not a bug in optimizeopt at all, but an issue with the fact that calling
several times W_TypeObject._pure_lookup_where_with_method_cache() with the same
arguments does not always return the same thing (which is strange).

_______________________________________________________
PyPy development tracker <pypy-dev-issue at codespeak.net>
<https://codespeak.net/issue/pypy-dev/issue545>
_______________________________________________________