Getting a Segmentation fault when running all tests 
against cvs code (also for 0.4.1).

oliver@garp roundup $ ./run_tests 
bsddb3 module not found, skipping bsddb3 DBTestCase 
.....EEEEEEEEEEEEEEEEEEEEEEEEEEEEEE..........
........................................Segmentation fault

I changed test/__init__.py to use verbosity=2 and it 
showed this:

testChanges (test.test_db.anydbmDBTestCase) ... 
Segmentation fault

Within test_db.py, it looks like the failure is coming at 
(or about) line 31 (status.create(name='unread')). I added 
print '<line num>' throughout the file, and that was the 
last one to get printed.

After a little more searching, it looks like it might be 
dying on line 374:

newid = self.db.newid(self.classname)

I'm not sure why -- I'm not familiar with the dbm 
modules. Any tips or suggestions as to how to pursue 
this would be appreciated!

Logged In: YES 
user_id=6405

We need to find out what dbm is actually being used. Simplest 
way is to import anydbm and print anydbm._defaultmod. 
 
Then we can start worrying about the locking that's done in 
newid(). 
 

Logged In: YES 
user_id=29957

Also, it would be very useful to know what operating system
and what version of the OS. 

I hope it's not the bsddb.hashopen() rubbish again.

Logged In: NO 

Python 2.2.1 (#1, May  7 2002, 15:26:24) 
[GCC 2.95.3 20010315 (release)] on linux2 
Type "help", "copyright", "credits" or "license" for more 
information. 
>>> import anydbm 
>>> anydbm._mod 
<module 'dumbdbm' from '/usr/lib/python2.2/dumbdbm.py'> 
>>> 
 
OS: Gentoo Linux (v 1.1a) w/a 2.4.19 kernel. 

Logged In: YES 
user_id=6405

Hrm - you say you get the same problem with 0.4.1? Could  
you please trace the problem there? dumbdbm should be safe, 
unless there's some strange locking problems going on (but 
there was no locking in 0.4.1). Also, could you please try the 
tests with strace to see with more detail what it's doing? 
 

Logged In: YES 
user_id=265596

Looking into this, I found that Python was also Seg Faulting 
when running 'make test', so it seemed that it was a library 
problem w/Gentoo. I submitted a bug with them which was 
fixed yesterday, and now neither roundup nor python seg fault 
when running the tests anymore.