[PEAK] Trellis: undetected circularity via optional rules

[Grant Baillie]

On 16 Apr, 2009, at 16:13, P.J. Eby wrote:

> ...
> At 12:37 AM 4/17/2009 +0300, Sergey Schetinin wrote:
>> ...
>> Using WinXP, Py2.5.
>
> Just as a datapoint, I don't get any test failures or hangs on my  
> own machine.

I get similar failures to Sergey on Ubuntu 8.10, which has Python 2.5.2.

I had thought my Mac OS X 10.5 failures might have been a result of  
the crufty oldish wxPython/twisted that are installed as part of the  
OS, but on Ubuntu I don't have either of those installed.

> I'm particularly confused by the set-related failures.

It looked to me as if some prior tests might have left some globals in  
a crufty state (since those tests pass when you run test_sets.py by  
itself), but I've been unable to figure out yet which those prior  
tests might be.

--Grant