On Apr 16, 2010, at 9:48 AM, Charlie Clark wrote:
> Am 16.04.2010, 17:42 Uhr, schrieb Casey Duncan <ca...@pandora.com>:
>> activate is a bit of a kludge, though it seems easy enough to just have
>> multiple shells open if activate screws with certain tasks. I'm curious
>> though, how do you switch virtualenvs? I can envision creating wrapper
>> scripts that only pollute the environment for a single process (running
>> it in a subshell or whatever), is that what you are doing?
>> I have to admit I haven't used virtualenv a whole lot, so I'm interested
>> in hearing what you seasoned old timers do ;^)
> I don't know if I count as an old timer but I've been seriously tripped up
> by "activate". virtualenv just sets up local Python environment for a
> project. All you then need to remember is to run "bin/pip", as opposed to
> "pip" to install stuff. So, you don't need to worry about closing the
> shell or wondering about any side-effects when you do something in the
> same shell a few days later. "activate" provides minimal additional
> convenience at the risk of considerable confusion.
I guess there are also two sort of usage cases too.
One is purely isolating environments for installing and running existing
"released" things. In this case activate seems entirely unnecessary. When you
run the things in an environment, you can just run them from the environment
The other sort of usage case is development. In that case you are installing a
bunch of packages as well as developing some new things. I imagine in that case
activate isn't so useful either, assuming you can run nosetests from the
virtualenv directory. Running the stuff for functional testing (automated or
manual) is really the same as above.
My problem is that I'm working on a library for Python 2 and 3. So for the
latter, I have to settle for prehistoric development strategies. Though it
looks like distribute is working in python 3, so maybe I should see if
virtualenv can be coaxed into the fold.
Repoze-dev mailing list