gmpy moving to code.google.com

[Alex Martelli]

On 2/26/07, Daniel Nogradi <nogradi at gmail.com> wrote:
   ...
> Svn checkout, compilation and installation went okay but some tests
> failed, this is the output of 'python test_gmpy.py':

Thanks for the report!  That test is trying to check that a previously
existing memory leak went away, and unfortunately this requires
measuring the memory used by the process which is kind of flaky... in
particular:

> Failed example:
>     _memsize()-_siz
> Expected:
>     0
> Got:
>     -4

...this means that (after a loop of the operation that used to leak)
your process is consuming 4KB _less_ than it was before the loop (?!).
 Guess I'll have to make that test more "tolerant" of such small
flakiness... [done -- it's now in revision 10 of the SVN repository].

> 1 items had failures:
>    2 of 132 in gmpy_test_mpz.__test__.number
> ***Test Failed*** 2 failures.

That's actually just the one failure (all tests are repeated with and
without caching enabled).


Thanks again for your report!


Alex