AP Pipeline Meeting, 2020-05-18

BlueJeans: https://bluejeans.com/426716450

attending:

• Eric Bellm
• Unknown User (cmorrison)
• Dino Bektesevic
• Unknown User (gkovacs)
• Ian Sullivan
• Krzysztof Findeisen
• Meredith Rawls
• Spencer Nelson

regrets:

• John Swinbank (vacation)

Topics for discussion:

• Gen 3 conversion (all):
  • Krzysztof Findeisen : working on getting gen3 compatible test data; not blocked

• Precursor processing (Meredith Rawls ):
  • still working on DECam calibration products–last weekly moved things around.  DM-17538.  meeting with John P after this to move ahead
• report from NCSA (Unknown User (emorganson)  + Unknown User (manninga) et al):
  • out today
• Alert Distribution (Spencer Nelson ):
  • two reviews out right now, at which point it could be distrubuted.
  • Eric suggests have Carl from ANTARES give it an external look to make sure nothing obvious is missing
  • on Scimma in the meantime
• ap_association/SDM/fakes/alerts (Unknown User (cmorrison)):
  • alert packets now created by the stack by default; saved in current wd by default
    • Eric suggests making this configurable
  • returning to fake injection; has postprocessing done but still thinking about how best to deal with injection (one, two, or three input tables?  impacts task structure)
• Special Projects (Ian Sullivan ):
  • working on tests for assemble coadds etc.  Looking at existing test suite, but it's not regularly run–they had made some heavyweight choices about how to set up the test dataset.  Ian aiming to make a very lightweight test suite.  Still in progress
• Image differencing (Unknown User (gkovacs) ):
  • still on ZOGY bugfixes–will probably split ticket. 
  • first test runs in next few days, then will write tests
  • wondering of whether to use detection on the ZOGY Scorr image
    • It's the optimal approach; Eric suggests to file a ticket and put it on the backlog for now
  • also noticed that source detection in the current code uses a simple Gaussian PSF rather than the known PSF–why? 
    • Let's follow up with RHL. 
    • Dino thought he remembered this was configurable–characeterizeTask uses Gaussian, calibrateTask uses real PSF?
• Review CI (https://chronograf-demo.lsst.codes/):
  • breakage last night?  nightlies have been broken since Saturday/Sunday, so the CI job doesn't even start (on #dmj_s-ap_verify)
  • problem is entirely upstream from us; KF says K-T reports it's being looked at
  • nothing new in Chronograph (no data since Sat. though due to above)
• Review outstanding action items
• AOB
  • Reviews
  • No meeting next week (Memorial Day)
  • Chris M: main laptop is currently in Houston being repaired; has other computer for now

Action Items