[Python-Dev] 2.2a0: latest unicode change breaks test_unicodedata

[Tim Peters]

[Mark Favas]
> In a change from 2.1 (final) to 2.2a0 - test_unicodedata now fails the
> methods test:
>
> test test_unicodedata failed -- Writing:
> '00c22b40a906a1a738012676c9feaedfc6be20
> d9', expected: '6c7a7c02657b69d0fdd7a7d174f573194bba2e18'
> ...
> (Tru64 Unix, Compaq C)

Reproduced identically on Windows (guess *everything* isn't the fault of
Compaq's compiler <wink>).  I assume this has something to do with the very
recent Unicode "optimization" patch.

Mystery:  since running the test suite before committing the change would
have caught this, how did the bug get into the CVS tree?  Since it appears
test_unicodedata is skipped under Unix when building the quicktest target, is
this due to people mechanically using quicktest instead of test?  Then that's
a second optimization bug <0.6 wink>.