Possibly inconsistent handling of FITS data. about desimodel

[weaverba137]

This is left over from a failed attempt to convert FITS data into a "pure" numpy array. The code still works so resolving this discrepancy is low priority.

[weaverba137]

@sbailey this is a pretty simple fix, but I need to know, what is the significance of fits.getdata(fiberposfile, upper=True) in load_fiberpos() and why is upper=True not applied in load_tiles()?

[sbailey]

We were trying to standardize upon data from FITS files looking like columns are all upper case (so retro...). The FITS standard is case insensitive and astropy.io.fits automagically handles both upper and lower for the returned object, but if the data are ever cast to a astropy Table or a numpy array, it has to make a choice.

load_tiles() reads footprint/desi-tiles.fits which was written with IDL and already had upper case names. load_fiberpos() reads focalplane/fiberpos.fits which was written in python with lower case column names. upper=True was an attempt to move that to upper case (which doesn't completely work due to my favorite topic of astropy.io.fits bugs). We could just re-write fiberpos.fits with upper-case column names and be done with it.

fiberpos.fits needs to be updated with the latest focal plane geometry anyway. Let's use UPPERCASE columns when we do that.

[weaverba137]

Did fiberpos.fits ever get updated with UPPERCASE columns?

[weaverba137]

As noted in #30, the use of fits.getdata() may be causing problems, and it is probably time to get rid of it. So, as previously asked, dit the fiberpos.fits file ever get updated to have uppercase columns?