io.load_tiles defaults to old tile list $DESIMODEL/data/footprint/desi-tiles.fits

[ashleyjross]

Can we make it point to the main survey tile list instead (/global/cfs/cdirs/desi/survey/ops/surveyops/trunk/ops/tiles-main.ecsv) or something equivalent? (This would include all of the backup tiles, but we want to be inclusive here, I think.)

This has caused problems with people making mocks and assuming

tiles = desimodel.io.load_tiles()
point = foot.is_point_in_desi(tiles, ra, dec)

would cut to anything that could possibly be observed in Y1. (The actual tiles are different around the edges.)

We will of course work around (explicitly using correct tile lists), but it would be very good to not have this issue happen again for anyone in the future.

@schlafly (let me know anyone else to tag)

[schlafly]

I'm fine with something like that as the default if present; feel free to submit a PR.

[geordie666]

@ashleyjross: You can assign me to this task, if you'd like. I'm a little busy today but could get to it later in the week.

[ashleyjross]

@geordie666 that would be great, thanks!

[araichoor]

just a comment, in case, about tiles-main.ecsv:

• one shall not forget to cut on IN_DESI==True! that can surely be included in the updated routine;
• one could add a programs argument to the routine;
• keep in mind that the list of tiles in tiles-main.ecsv can slightly change from time to time, e.g.:
  • when we retire one tile (setting IN_DESI=False for it) + replace it with a new tile at the same location, but with a different TILEID (e.g. https://desi.lbl.gov/trac/changeset/1533/data); we did that few times in the past, I don t know if that will happen again.
  • when we nudged the tile centers (https://desi.lbl.gov/trac/changeset/2566/data); same maybe that won t happen again.

of course, such operations are never done on validated tiles; but the list of "potential tiles to be observed in the future" could in theory slightly change with time.

[ashleyjross]

I agree about all of these caveats. The most important, thing, though, is that if used in default mode it won't somehow yield a smaller footprint than what has already been observed (like it does now.)

[geordie666]

Addressed in #168.