Hmmm... While bfg worked for the ".fits" files in the "tests" directory, it doesn't seem to work for ".fits.gz" files nor for the ".ipynb" files. Although bfg says that it deleted the files they are still there. Gonna have to try something else...

So apparently this worked just fine:
git filter-branch --force --index-filter 'git rm --cached -r --ignore-unmatch python/pfs/drp/stella/*.ipynb' --prune-empty --tag-name-filter cat – --all
git filter-branch --force --index-filter 'git rm --cached -r --ignore-unmatch tests/data/calExp/2016-01-12/v000000*/PFFAr2.fits.gz' --prune-empty --tag-name-filter cat – --all
git reflog expire --expire=now --all
git gc --prune=now
git gc --aggressive --prune=now
git push origin master --force

All fits files and not needed ipython notebooks have been removed from the repo

All ... and not needed ipython notebooks have been removed from the repo

Which notebooks are needed?

createDetGeom.ipynb to create the detector geometry fits files

and

createRefSpec.ipynb to create the reference spectrum

I don't think any executable scripts should be notebooks, please export them as straight python, and make sure that they can be run from the command line (by adding argparse argument parsing etc.).

If you have educational notebooks, please move them into a new product.

"createDetGeom" has been moved to a task to be executed from bin.src/createDetGeom.py (command line task didn't work because there are no data to be processed and no butler to be associated with. Error: "WARNING: Not running the task because there is no data to process; you may preview data using "--show data"")

"createRefSpec" has been moved to a command line task

See drp_stella tickets/INFRA-32

Comments on the code at https://github.com/Subaru-PFS/drp_stella/pull/1 – I think a bit more thought could make the scripts a clearer, but I don't think there are any obvious regressions here, so we should probably file another issue for refactoring.

Confirmed that drp_stella is now pruned to a sensible size.

Will address provenance of detector geometry in PIPE2D-37. Other comments have been addressed.