Primecount failures on 32-bit systems

[jdemeyer]

There are various failures of the form

sage -t --long src/sage/interfaces/primecount.pyx
**********************************************************************
File "src/sage/interfaces/primecount.pyx", line 47, in sage.interfaces.primecount.prime_pi
Failed example:
    prime_pi(1000)                     # optional - primecount
Expected:
    168
Got:
    168L
**********************************************************************

if int64_t is larger than a long.

[git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

​ad4059c

Trac #25661: Fix doctests for 32bits platform

[vklein]

Failing tests fixed.

[chapoton]

ok, this will do the job, if not elegantly

[tmonteil]

Why not using # 64-bit and # 32-bit keywords as done elsewhere (e.g. when testing hash methods) ?

[vklein]

Replying to tmonteil:

Why not using # 64-bit and # 32-bit keywords as done elsewhere (e.g. when testing hash methods) ?

Because i didn't remember/know these flags. Thanks.

[vdelecroix]

Replying to tmonteil:

Why not using # 64-bit and # 32-bit keywords as done elsewhere (e.g. when testing hash methods) ?

To make the tests pass on both Python2 and Python3?

[vklein]

True ! Something like 168L # 32-bit will fail with python3.

[tmonteil]

Is the plan to turn every explicit doctest into an indirect one, just to have them pass simultaneously on python2 and python 3 ?

[vdelecroix]

Replying to tmonteil:

Is the plan to turn every explicit doctest into an indirect one, just to have them pass simultaneously on python2 and python 3 ?

I hope not. But

• testing equality of integers is a better test than checking string representation
• I think that output flag should be avoided as much as possible