[Python-Dev] Std tests failing, Windows: test_builtin test_charmapcodec test_pow

[Tim Peters]

[NeilS]
> Wierd. rt.bat does not run the test_pow script.

Works for me, else I never would have noticed <wink>.  Also works for me in
single-test mode:

C:\Code\python\dist\src\PCbuild>rt test_pow

C:\Code\python\dist\src\PCbuild>python ../lib/test/regrtest.py test_pow
test_pow
The actual stdout doesn't match the expected stdout.
This much did match (between asterisk lines):
**********************************************************************
test_pow
Testing integer mode...
    Testing 2-argument pow() function...
    Testing 3-argument pow() function...
Testing long integer mode...
    Testing 2-argument pow() function...
    Testing 3-argument pow() function...
Testing floating point mode...
    Testing 3-argument pow() function...
The number in both columns should match.
3 3
-5 -5
-1 -1
5 5
-3 -3
-7 -7

3L 3L
-5L -5L
-1L -1L
5L 5L
-3L -3L
-7L -7L

3.0 3.0
-5.0 -5.0
-1.0 -1.0
-7.0 -7.0

**********************************************************************
Then ...
We expected (repr): ''
But instead we got: 'Float mismatch:'
test test_pow failed -- Writing: 'Float mismatch:', expected: ''
1 test failed: test_pow

C:\Code\python\dist\src\PCbuild>

That may point to the problem, too:  the canned output file is truncated?

> If I run "regrtet test_pow" then the test fails.  It could be a
> problem with line endings (I copied the source for a Unix CVS
> checkout).

Don't understand; e.g., "copied" what, from where to where?  I'm not sure I
gave you write access to my box, and hacking into Windows machines is uncool
because it's not challenging <wink>.

> Anyhow, I found the bug.  I don't know how test_pow was passing
> under Linux.  Time to reboot again.

Cool!  BTW, Windows solves the "don't reboot enough" problem for you via
automation, sometimes on an hourly basis.

Thanks for sharing the brain cells, Neil!