[Python-Dev] Test failures under Windows?

[Hirokazu Yamamoto]

David Bolen wrote:
> I don't know why they are happening so frequently now when there was a
> reasonable period when they weren't an issue (something about new I/O
> support in 3.x perhaps?), but without preventing them it seems the
> Windows build slaves are going to become (if not already) quite a bit
> less useful.  Don't know about anyone else's but I can't watch mine
> 7x24.
> 
> -- David

CRT Assertion was totally disabled before, but recently was enabled,
and workarounds were patched for problematic functions. (ex: fdopen and 
dup) Probably this *patch* is not perfect. See 
http://bugs.python.org/issue4804

I'm now +3/4 for the idea disabling assertion by default,
and enabling by startup option or environment variable. (Or enabling
by default and disabling by environment variable?)