search

XENON1T / pax

The XENON1T raw data processor [deprecated]
BSD 3-Clause "New" or "Revised" License
16 stars 15 forks source link

Remove Deprecated Data Files #604

Closed pdeperio closed 6 years ago

pdeperio commented 7 years ago

Can we remove old input data files that should never be used again? (They should be stored in the version control history anyway, right?) For example, to avoid issues such as https://github.com/XENON1T/hax/pull/140, or past confusions when running reconstruction with wrong maps.

tunnell commented 7 years ago

Of course, delete away!

pdeperio commented 7 years ago

@weiyuehuan cleaned up the NN files in #612 (thanks!)

Can the others responsible please clean up their maps? Please make sure that by cleaning those files we can still do SR0/SR1 analysis/simulations.

s1_xyz: @l-althueser s2_xy_patterns: @pelssers s1_aft: @darrylmasson s1_xyz: @berget2 s2_xy: @adambrown1

JelleAalbers commented 7 years ago

If you have old data processed with these maps, keeping them around allows you to re-correct. They'll be in the repo somewhere, but it will be harder for most people to do some study comparing pax versions. I don't know if that's worth some confusion potential like https://github.com/XENON1T/hax/pull/140 -- though weren't these map names set by the corrections db now?

pdeperio commented 7 years ago

@JelleAalbers, I was also worried about an (undocumented) issue with a previous reprocessing that took an obsolete NN map filename from an outdated RunsDB, and was only discovered after someone analyzed the position distribution. (The other maps are indeed taken care of in hax now, but as https://github.com/XENON1T/hax/pull/140 shows, we can still make this kind of mistake there.)

I agree it's a bit more difficult to do direct comparisons without all the maps in one package, but I hope it's enough to have them all in the existing conda environments.