nmm1 at cus
Apr 12, 2000, 4:48 AM
Post #1 of 3
(65 views)
Permalink
|
|
Problems in Python 1.6 alpha 1 (PR#291)
|
|
I understood the comments in the release notes to say post these to
the newsgroup, but I am not sure that is correct. Anyway, here is
a slightly corrected report on the building process.
Here are some problems with building and running the tests of the
1.6 Alpha release:
1) The Modules/sre* files have control-M at the end.
2) The 'unsigned char *' on lines 596 and 631 of mmapmodule.c should
be 'char *'. There is also a pointless check '(where >= 0)' for an
unsigned value on line 408.
3) _sre.c lines 222 and 225 and mmapmodule.c lines 715 compare
pointers to object types and pointers to void - this is not allowed,
but I don't think that it is particularly unsafe. But it is a hard
error in a few compilers, so should be fixed.
4) If the call to mmap in mmapmodule.c fails, the diagnostic is
"Invalid argument" - which can be downright wrong (as it was in this
case.)
5) There is something very odd with the test harness, though I can't
see the problem in regrtest.py. 'make test' says test_pyexpat crashed,
but running the single test says that the module is not configured. This
happens even under Irix.
6) test_unicode fails on Hitachi systems -- Writing: '*', expected:
'T'. Again, I failed to tie it down, and it looks like a test harness
problem.
7) configure is still buggy and gets things wrong :-(
8) A few other failures weren't Python's fault, and I have
reported them to the vendors.
Regards,
Nick Maclaren,
University of Cambridge Computing Service,
New Museums Site, Pembroke Street, Cambridge CB2 3QG, England.
Email: nmm1 [at] cam
Tel.: +44 1223 334761 Fax: +44 1223 334679
|
