PIN: Update nipype

[effigies]

Includes nipy/nipype#2198, nipy/nipype#2200, nipy/nipype#2205

Pending:

• FMRIPREP test
• MRIQC test

Edit: Added a further pin. If that causes issues, I'll drop it.

[effigies]

@oesteban Do we need to include nipy/nipype#2209 to fix MRIQC?

[oesteban]

No, I have to fix MRIQC first. Let me push some commits.

[effigies]

Okay. If it works for you, I'll push a commit here with the latest nipype PR and you can push your MRIQC fixes and pin in one shot.

[oesteban]

It works :)

[oesteban]

Sorry, I misread the thread. No, nipy/nipype#2209 is not necessary. Since it is failing to build we better hold off pinning to it.

[effigies]

Okay, I collapsed the pin to include all the updates. You can pin MRIQC to dafbfda.

[oesteban]

Sweet thanks

[effigies]

Updated the fmriprep test in the checklist. If you want to do the same when you push an mriqc branch, I'll merge when they both pass.

[oesteban]

Done

[oesteban]

Let's see if MRIQC passes :)

[effigies]

Not causing a failure but it is a bit unsightly:

Error in atexit._run_exitfuncs:

Traceback (most recent call last):
  File "/usr/local/miniconda/lib/python3.6/site-packages/niworkflows/nipype/utils/config.py", line 327, in free_display
    config.stop_display()
AttributeError: 'NipypeConfig' object has no attribute 'stop_display'

From: https://circleci.com/api/v1.1/project/github/effigies/fmriprep/861/output/31/0?file=true

I want to go ahead and merge to move forward, if MRIQC passes, but just so you're aware of the issue.

[oesteban]

That is because of the absolute import of nipype in the atexit function (it should try niworkflows.nipype instead). Not sure if that import could be relative, I guess so.

[effigies]

Yup. I'm submitting a patch. Also dropping the verbosity of multi-proc again...