Notebooks not being run in CI or not reporting errors

[jhlegarreta]

Notebooks are not being run in CI checks or they are not reporting errors: the fMRI BOLD run notebook requires data to be downloaded from g-node, and that step is not being carried out prior to
https://github.com/nipreps/nifreeze/blob/main/tox.ini#L73

The script in PR #93 will allow downloading the data.

If running the notebook is too time-consuming for every PR, a weekly cron job will come in handy to run the notebooks on a regular basis.

[jhlegarreta]

Notebooks are being run and reporting errors now:
https://github.com/jhlegarreta/nifreeze/actions/runs/14813328489/job/41590770278#step:9:31

Some notes:

• bold_realignment.ipynb is excluded from the notebooks collected.
• dmri_covariance.ipynb is failing: TST: Download data from G-Node GIN in notebook testing GHA workflow #111 (comment)

Testing bold_realignment.ipynb will be enabled once the issue with dmri_covariance.ipynb is fixed.

[jhlegarreta]

@oesteban Although my local environment may be slightly different wrt the one on GHA, I cannot make sense of the failure in dmri_covariance.ipynb:
https://github.com/jhlegarreta/nifreeze/actions/runs/14815869226/job/41596188587#step:11:75

Suggestions are welcome.

[oesteban]

Looks like the most plausible explanation is the data file not being downloaded properly: https://stackoverflow.com/questions/60191681/cannot-load-file-containing-pickled-data-python-npy-i-o

Perhaps adding allow_pickle=True to the np.load() function call could also give us more info.

[jhlegarreta]

Perhaps adding allow_pickle=True to the np.load() function call could also give us more info.

That is the first thing I tried, and Python reported back saying that data was not pickled and that it could not open them:
https://stackoverflow.com/a/67480110

It has happened through multiple runs, so unless no data is effectively being downloaded (e.g. I missed some step), the data download step does not seem to be the issue.

[oesteban]

I bet the problem is:

nifreeze/.github/workflows/notebooks.yml

Line 47 in f4fd8be

curl -L -o "${TEST_DATA_HOME}/hcpdata.npz" https://gin.g-node.org/nipreps-data/tests-nifreeze/src/master/hcpdata.npz

Looks like the URL for the git interface on the g-node website, so it's not the link to the actual data. I've tried https://gin.g-node.org/nipreps-data/tests-nifreeze/raw/master/hcpdata.npz.

On my computer (and cutting the download at 35MB after almost 10 minutes:

(base) oesteban@hos65851:~/tmp/nifreeze-notebooks$ curl https://gin.g-node.org/nipreps-data/tests-nifreeze/raw/master/hcpdata.npz -o rawlink.npz
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 35.5M    0 35.5M    0     0  71713      0 --:--:--  0:08:39 --:--:-- 31624^C
(base) oesteban@hos65851:~/tmp/nifreeze-notebooks$ curl https://gin.g-node.org/nipreps-data/tests-nifreeze/src/master/hcpdata.npz -o currentlink.npz
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 15513    0 15513    0     0   2670      0 --:--:--  0:00:05 --:--:--  3859
(base) oesteban@hos65851:~/tmp/nifreeze-notebooks$ ls -lah
total 36M
drwxrwxr-x  2 oesteban oesteban 4.0K May  7 15:46 .
drwxrwxr-x 35 oesteban oesteban 4.0K May  7 15:36 ..
-rw-rw-r--  1 oesteban oesteban  16K May  7 15:46 currentlink.npz
-rw-rw-r--  1 oesteban oesteban  36M May  7 15:45 rawlink.npz

The latest nightly run of the action seems to confirm this (total size ~15kB, so definitely not getting the full dataset):

   % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed

  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0
  0     0    0     0    0     0      0      0 --:--:--  0:00:01 --:--:--     0
  0     0    0     0    0     0      0      0 --:--:--  0:00:02 --:--:--     0
  0     0    0     0    0     0      0      0 --:--:--  0:00:03 --:--:--     0
  0     0    0     0    0     0      0      0 --:--:--  0:00:04 --:--:--     0
100  8192    0  8192    0     0   1533      0 --:--:--  0:00:05 --:--:--  1533
100 15513    0 15513    0     0   2847      0 --:--:--  0:00:05 --:--:--  4020

With the download speed gin allows, it looks like we do need to upload the file to OSF.

[jhlegarreta]

👍 Switched to OSF: PR #115. Notebook testing is passing now:
https://github.com/jhlegarreta/nifreeze/actions/runs/14895700573/job/41837681690

The only missing piece is to enable bold_realignment.ipynb (x-ref #95 (comment)).

[jhlegarreta]

The only missing piece is to enable bold_realignment.ipynb (x-ref #95 (comment)).

PR #117.