@miykael - do you want to test all notebooks with both pythons? I just decided to test only with python 3, since we're trying to move to py3, but we can probably start testing both.

I'm trying to rebuild the tests. I sometimes have also timeout. I believe didn't figure out which extra parameter should have to be changed to be safe.

@djarecka - testing all notebooks with both pythons would be a good idea, but that would mean we would need an additional conda environment in the docker image (i.e. creating another 1.8GB of volume), no?

How would you change the kernel in the notebook images? Replacing the specification at the end of the notebook might work.

I think we would just have to build the second dockerfile. I would also strongly suggest people to use py3 only (that's why I'm not completely sure if we want to use default python and add additional tests)

I'm also for suggesting people to use py3. But I'm recommending to leave default. Because in the docker image, default is python3. If I'm not mistaken, it should be the neuro environment.

Now, we could also specify that it should use the neuro or an py3 environment. But this would mean, anybody who opens the notebook locally on their machine (without docker), will get a kernel error, that py3 or neuro wasn't found. This error shouldn't happen with default.

Therefore I propose to leave default in. But we could also rename neuro to py3 to make it clear for user that they should use python 3. What do you think?

Oh, I see what you mean. Sure, we can use default but I think that at the same time we can keep official docker images with py3 only, so we don't "create" a new Nipype users that use py2 instead of py3. Some people will want to use py2 anyway, but there are people that simple don't know which python they should choose when they're starting, so having a py3 as a default option might help in the future IMO.

don't have any opinion regarding env name

Ok, than we leave it at default. And yes, I think it's best to move everybody slowly to py3.

And for the people that want to use py2, there's always the option to create a new docker image that is based on the tutorial, with the additional conda environment in it. I did something similar for my last workshop.