[TransWarp] Test suite fails

[Phillip J. Eby]

At 10:26 AM 5/14/03 -0700, Vineet Jain wrote:
>Thanks Ulrich.
>
>I tried the unit test cases and the example at work (on a different
>computer) with the python distribution from python.org and both of them
>crashed the interpreter.

That is utterly baffling to me.  Unfortunately, it sounds like there is 
something that's independent of which Python version you use.  Perhaps it 
is something else, then, like the OS version.  I work exclusively on 
Windows 98 SE; what OS are you on?  Also, perhaps you could try running 
individual test suites, i.e.

peak.binding.tests.test_suite
peak.config.tests.test_suite
peak.model.tests.test_suite
peak.naming.tests.test_suite
peak.util.tests.test_suite

and see which ones crash.  It also seems bizarre to me that you're not 
getting any test output.  You should see a line of dots go by, followed by 
a "126 tests run in blah blah seconds".

Also, if there's anybody else here with Windows who can confirm or deny 
whether the binary installer version of 0.5a1 works for them, and if so, on 
what operating system, then that would be helpful.  Meanwhile, I'll try 
installing the binary version myself on a relatively fresh PC (also 98SE 
I'm afraid) and see what happens.  But I won't be able to do that until 
tomorrow or so.



>I then downloaded the latest code from cvs and tried to rebuild it but
>don't have cl.exe (microsoft's c compiler) and it fails on building
>kjbuckets.

There are instructions in the Python manual that's called "Installing 
Python Modules" for how to use the free MinGW tools to build 
extensions.  But I expect that won't help you since the MinGW-built modules 
are what I distributed with the binary.  Unless it's a OS or libraries 
versioning issue, in which case it might help.


>Is it not possible to have the default distribution contain only python
>code and if optimization is enabled then use the c code?

No.  The persistence code from ZODB is in C, so you couldn't use PEAK for 
much then.