Re: [Fab-user] Current head passing unit tests?

[Jeff Forcier]

On Sat, Aug 22, 2009 at 2:48 PM, Steve Steiner
(listsin)<address@hidden> wrote:
> Jeff,
>
>        Thanks for the help, it was a nose issue.
>
>        nosetests is found in the path but is shebang'd to use the system
> Python interpreter.  None of the fabric dependencies are installed in the
> system Python, but nose was.
>
>        I actually purposely don't install _anything_ into the system Python
> so that nothing will work _unless_ I'm a virtualenv.  Nose snuck into the
> system Python and faked me out.

Glad you figured it out!

This has bitten me in the past with other packages, so I always do my
best to make sure stuff goes into the virtualenv only. I did have
problems with one package that simply refused to go into the
virtualenv no matter what, but I don't recall which one; I don't think
it was one I needed for Fabric, thankfully.

At any rate, I have high hopes that in the nearish future, either
Python proper will grow virtualenv-type support (thus rendering this
moot-ish), or that such packages will be fixed to be virtualenv-aware
/ not breaking in virtualenv.

> P.S.  Did I mention neat trick turning off the deprecation warnings for
> Paramiko?  Didn't know about that one; very handy even if it does feel like
> turning off compiler warnings which I cringe every time I do...

Yea, it's a bit on the squicky side but I felt it was worth the risk.
When we get around to migrating to 3.0 I'll just need to remember to
remove or comment out that code :)

And I can't take credit for it, saw a tweet from (I think) Simon
Willison on the issue and had the same reaction you did. I don't know
if he had the issue while using Fabric or if it was some other use of
those hash modules on 2.6, but I'm glad he brought it up.

-Jeff