[Zope3-dev] Test Failures when running tests with "-N 2"

[Tim Peters]

...

[Tim Peters]
>> Really?  You don't see the new failure I reported here yesterday?
>> That's got nothing to do with -N, though:
>>
>>     http://mail.zope.org/pipermail/zope3-dev/2005-April/014158.html

[Stephan Richter]
> I had originally deleted this last test, so they would pass (it was not only
> on Windows), so a simple "svn up" should have sufficed.

Well, the test ran on my box, and it wasn't a question of not doing
"svn up" here -- I do "svn up" so frequently my fingers hurt <wink>.

> However, I just reenabled the test and fixed it for good.

It works here now -- thanks!

[problems testing deprecation warnings across multiple runs] 

> Right, I know about this. Mmh, in fact, it surprises me now that my
> deprecation module doctests do not complain in the second run. It is probably
> one of the advantages of doc file tests, since their module gets regenerated
> for every test run, so that the deprecation framework does not detect the
> warning duplication.

That probably accounts for it.  warnings.py does a truly horrid thing
to detect duplicates:  when a warning is first raised in module M, it
creates a "secret" M.__warningregistry__ attribute, to record all
warnings (text, category, and line number) ever raised in M.  If M
goes away, so does the record of the warnings it raised.