[IPython-dev] Test failures and errors in trunk

[Brian Granger]

Fernando,

Ondrej has just tried iptest on python 2.5 and gets the following
semi-interesting result:

http://paste.debian.net/30957/

If you get a chance, at least have a look at the message, as they
might show us what is going on with this issue.

Brian

On Wed, Mar 18, 2009 at 8:41 PM, Brian Granger <ellisonbg.net at gmail.com> wrote:
> OK, I found the commit that introduced the error.  It first appears in
> this revision:
>
> http://bazaar.launchpad.net/~fdo.perez/ipython/trunk-dev/revision/1155
>
> The previous revision passes all tests just fine.  You can get the
> problematic revision by doing:
>
> bzr branch lp:~fdo.perez/ipython/trunk-dev -r 1155
>
> It is worth checking that the previous revision is OK on your system as well.
>
> But, this one could be a challenge to figure out.  In this revision,
> you were only touching non-twisted parts of the test suite - basically
> getting the ipdoctest stuff working right.  Hopefully between the two
> of us we can figure this out.  I know the Twisted stuff, you know the
> ip testing stuff.
>
> I am going to switch gears to help Ondrej with some things.  If you
> end up looking at this and have questions on the Twisted side of
> things let me know.
>
>> Wow, that would be fantastic.  I didn't think I was touching anything
>> twisted-related, and since I'd seen this problem before, I thought it
>> was the same thing.
>
> Looks pretty similar, but I think it is somewhat different.  I have
> never seen the double call to reactor.run() before.
>
> But
>
>> I'll know better from now on never to ignore a twisted problem, even
>> if it's a bit of a Heisenbug.
>>
>> In the meantime, I'm working up some utilities to ease the virtualenv
>> workflow, based on some of the scripts that are out there from Doug
>> Hellman and Chris Burns.  This will let us do tests in 'naked python'
>> virtualenvs more easily, so we can really have a test suite that at
>> least runs gracefully (skips but no errors) on nothing but
>> python(stdlib)+nose+ipython.
>
> Yes, this seems like it is pretty important.  I ended up doing that
> when we did the last release and it really helped.
>
> Cheers,
>
> Brian
>
>> Stay tuned.
>>
>> f
>>
>