Testing for stabliity (was [Zope-dev] Re: large files and --pymalloc on the new release? (was Re: no more SIG11 ...))

Chris McDonough chrism@zope.com
Thu, 17 Jan 2002 07:24:56 -0500


A while back, someone was able to provide an Python Script that *always* 
crashed Zope.    That's how Evan was able to determine that there was a 
stack size computation bug in the compiler module.  I *think* he may 
have turned it in to a unit test.

The current Zope test suite is the first line of defense.  It can be run 
by using utilities/testrunner.

If you do notice problems, the second best line of defense is an ability 
to narrow the problem down to a small, reproducible test case that has 
as few "external" dependencies (database adapters, products, etc.) as 
possible.  If you can make it happen predictably, and it crashes Zope, 
we can (and will) fix it.

- C


> Is there a way to perform some tests or is there a testsuite that can be
> used to stress test these issues? I too got bitten by these bugs + assorted
> thread related errors, which has cost us very dearly in our project in terms
> of both time, additional costs and unexpected hardware investments due to
> change of deployment platform, so I *really* would like to have some way of
> being able to confirm if these bugs are dealt with or no. I am not sure
> there are any consistent tests for this, though, so hints on what to look
> out for are equally appreciated.
-- 
Chris McDonough                    Zope Corporation
http://www.zope.org             http://www.zope.com
"Killing hundreds of birds with thousands of stones"