Add spectral-cube reprojection tutorial

[keflavich]

see review on radio-astro-tools/tutorials#17

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[adrn]

Sorry for the build failures - I just fixed those in #505 - do you mind rebasing on main?

[jonathansick]

The specific technical build issue here is with the kernel that the notebook is saved with. I've created an issue because we should definitely handle this gracefully on our end. astropy/nbcollection#16

In the meantime, It looks like the notebooks we are running use a generic python3 kernel:

  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },

If you can switch it in the notebook while we fixe the tooling, that'd be 👍

[kelle]

Hmmm, no other tutorials have an Index like this. I'd like to discuss this.

Reply via ReviewNB

[kelle]

We discussed this and decided to handle the Index as part of the template so every tutorial will have it automatically.

[keflavich]

we also highlighted this as a nice low-hanging fruit issue to tackle in e.g. astropy/learn-astropy#26

[jonathansick]

In GitHub Actions, we got a network timeout downloading a file:

filename_1 = download_file("https://almascience.nrao.edu/dataPortal/member.uid___A001_X1465_X3a33.BrickMaser_sci.spw71.cube.I.manual.image.pbcor.fits",

I'll re-run the GitHub Actions job now to see if it clears (if it persists we might want to look into a retry loop? or loading the files from somewhere else?)

[keflavich]

unfortunately, that error is expected. NRAO's ALMA server has been timing out on me all week.

We can try the ESO and Japan mirrors, but a retry loop is the most reliable. It's super clunky, though.

[jonathansick]

Good news it looks like the re-run of the GHA workflow worked!

We can try the ESO and Japan mirrors, but a retry loop is the most reliable. It's super clunky, though.

On one hand, we could think about adding a retry-loop (with exponential back-offs and all that) to astropy.utils.data.download_file or a wrapper of it. Until we start caching built tutorials, so we don't need to rely on every tutorial being runnable all the time, either a retry or just caching to a more solid static file server is our best bet.

Since it's working right now, I think we can keep going, but it's something we might need to come back to soon.

[eblur]

This looks great! I had a few minor comments -- mainly to do with adding some exposition around the science context for performing these tasks. Adding that context will greatly aid in the search discovery potential for this tutorial.

[kelle]

I'm confused. It looks like the build is failing. @jonathansick , are you saying it can be merged anyway?

[jonathansick]

It was working, but not now. In this case, it looks like the server really is wonky and we should find a work-around for the download issue before merging.

[kelle]

Maybe we should serve the FITS file from the astropy data server instead of downloading from the ALMA server.

[keflavich]

I'd be happy with that; I just picked data that were publicly available b/c I wasn't sure how much we could afford to put on the astropy server

[eteq]

Line #1.    filename_1 = download_file("https://almascience.nrao.edu/dataPortal/member.uid___A001_X1465_X3a33.BrickMaser_sci.spw71.cube.I.manual.image.pbcor.fits",

Maybe this should use astroquery.alma instead? Might that be more reliable than the direct download?

Reply via ReviewNB

[keflavich]

no, it will not. The archive is the bottleneck, not the access path to the archive.

[jonathansick]

@keflavich I can't download the datasets at all:

• https://almascience.nrao.edu/dataPortal/member.uid___A001_X1465_X3a33.BrickMaser_sci.spw71.cube.I.manual.image.pbcor.fits
• https://almascience.nrao.edu/dataPortal/member.uid___A001_X87d_X141.a_sma1_sci.spw27.cube.I.pbcor.fits

(500 server errors)

Are there different URLs we can try? Once we download the datasets I'll add them to the Astropy data server for more stable access.

[jonathansick]

Oh nevermind; the eso.org archive works. Maybe we can change the links to avoid using the archive server?