Closed guynir42 closed 2 months ago
This also changes the MD5sum so it is not going to be trivial to do.
Perhaps what is best is just to overwrite the file on the archive. Then we'd need to update all other Cutouts that the sum has changed...
I'm closing this as I am coming around to thinking this is not allowed by the logic of how we manage data right now.
Example: we have a list of Cutouts objects, and we saved it to disk and to archive, but then decided we want to delete one (or a few) objects from the list.
Then to delete the local file is easy, just remove the dataset (either HDF5 group or the FITS HDU) and if the file doesn't have any more datasets, delete the file.
BUT, if there's also a file on the archive then we are in trouble because you need to send a request to delete a part of a file, and that is currently not supported.