AnswerDotAI / AnswerDotAI/nbdev
`nbdev_export` exports notebooks in arbitrary order
- Vorherrschende Sprache
- Jupyter Notebook
- Sterne
- 5.3k
- Forks
- 513
- Ø Merge
- 2 T. 30 Min.
- Gemergte PRs (30 T.)
- 8
Beschreibung
In my project I can't get the build-pipeline to work again. It fails with the error, that my notebooks and the library are not in sync.
I already tried:
- upgrading local nbdev (is 2.3.7.)
- running nbdev_export locally, pushing the changes (there were none)
- deleting the library dir locally, pushing that. Then run nbdev_export locally to re-generate the lib folder, pushing that
- cloning the repository into a new local folder and running nbdev_export there. Again, there are no changes when running git status
- Deleting the github-runner-cache
- Performed the exact steps done by https://github.com/actions/checkout on my local machine. Ran `nbdev_export`. Still No changes shown by git status.
To no avail. It seems that whatever version the gitlab runner pulls is different than what I see in the repo.
Here is the pipeline in question:
https://github.com/omlnaut/pymoq/actions/runs/3356523622/jobs/5561653458
Any help is appreciated :)
Update:
I figured out how to ssh into the github runner while it's executing the workflow. The issue is not that the repository is checked out incorrectly. Instead, it seems that `nbdev_export` doesn't handle the case that I have one notebook (04_...) with a `default_export` and another notebook (06_....) that has specific cells exporting to the same target. On my local machine this works as intended, but when I run `nbdev_export` via ssh on the runner, only the code from the 04_... notebook gets copied to the target file.
Beitragsleitfaden
Rechercherichtung
Beginne mit der Überprüfung des verlinkten GitHub Actions-Workflows und reproduziere `nbdev_export` mit dem Notebook 04_..., unter Verwendung von `default_export`, sowie mit dem Notebook 06_..., wobei die Zellen in dasselbe Ziel exportiert werden. Vergleiche die lokalen Ergebnisse mit denen des Runners; abgeschlossen ist die Aufgabe, wenn beide Notebooks konsistent verarbeitet werden und die generierte Bibliothek mit den Notebooks übereinstimmt.
Vom Indexierungsmodell aus dem Issue-Text verfasst.
Bewertung
- Tech-Stack
- jupyter-notebook, python
- Bereich
- build-system, tooling
- Issue-Typ
- Bug
- Schwierigkeit
- 4/5
- Geschätzter Aufwand
- 3-5 Tage
- Aktivitätsstatus
- Veraltet
- Klarheit
- Größtenteils klar
- Anfängerfreundlichkeit
- 35/100