Paul Price added a comment - 03/Feb/16 12:22 PM I'm cleaning up the clutter, and hope you will find it improved in the next version.

Paul Price added a comment - 03/Feb/16 1:54 PM Russell Owen , now that the commits are rearranged and there's more documentation than code, would you mind having another look please? I hope you'll find that what you considered complicated before is now much simpler (and it turns out that the complication was there prior to this ticket). price@price-laptop:~/LSST/pipe/tasks (tickets/DM-4878=) $ git sub commit 2f3f40535da66ec17ab668f764f7602e4c969624 Author: Paul Price <price@astro.princeton.edu> Date: Thu Dec 18 22:41:51 2014 -0500   add task to propagate flags from CCDs to coadds This is used in processCoadd and the detection and measurement parts of the multiband processing scheme to flag stars on the coadd as having been used for PSF determination. Based on code written by Hironao Miyatake.   python/lsst/pipe/tasks/multiBand.py | 5 + python/lsst/pipe/tasks/processCoadd.py | 6 + python/lsst/pipe/tasks/propagateVisitFlags.py | 197 ++++++++++++++++++++++++++ 3 files changed, 208 insertions(+)   commit 8f42c159f2057679cb7505dc7bcc61ff293075b6 Author: Paul Price <price@astro.princeton.edu> Date: Wed Feb 3 15:49:38 2016 -0500   AssembleCoaddTask: make assembleMetadata clearer Reorganised the metadata reading loop into a list comprehension for clarity (it's now closer to appropriate the comment). Removed the 'bbox' variable in favor of specifying the bounding box explicitly, so it's clear that it's only reading a single pixel. Added an assert to guard against vector length mismatches   python/lsst/pipe/tasks/assembleCoadd.py | 18 ++++++++++-------- 1 file changed, 10 insertions(+), 8 deletions(-)   commit 1b5babd7693a08ef71f28200e2fbd8f5ea7e20d2 Author: Paul Price <price@astro.princeton.edu> Date: Wed Feb 3 15:50:19 2016 -0500   make coadd input catalogs contiguous We know how many inputs (both CCDs and visits) we have for the warp ("tempExp") and for the coadd, so reserve space in the catalogs for that many so they can be contiguous. This prevents errors when propagating flags to the coadd due to non-contiguous catalogs.   python/lsst/pipe/tasks/assembleCoadd.py | 3 +++ python/lsst/pipe/tasks/coaddInputRecorder.py | 9 ++++++--- python/lsst/pipe/tasks/makeCoaddTempExp.py | 2 +- 3 files changed, 10 insertions(+), 4 deletions(-)   commit 4139813fcb8a090d2f3d4000f54c66ae7e4948c9 Author: Paul Price <price@astro.princeton.edu> Date: Fri Jul 24 16:19:46 2015 -0400   PropagateVisitFlagsTask: guard against non-contiguous ccdInputs It is safer to read the ccdInputs value by value instead of by column, since the latter way fails if the Catalog is non-contiguous. Since we don't expect that this catalog is large, this should not have too bad an effect on the performance in comparison with the other operations.   python/lsst/pipe/tasks/propagateVisitFlags.py | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-)   commit 7a88d2f0a431723d1f6c4de548592258429097ff Author: Paul Price <price@astro.princeton.edu> Date: Wed Feb 3 15:11:51 2016 -0500   CoaddTempExpInputRecorder: add docstring for __init__   python/lsst/pipe/tasks/coaddInputRecorder.py | 7 +++++++ 1 file changed, 7 insertions(+)

Russell Owen added a comment - 03/Feb/16 2:10 PM I like your cleanup. This looks good to merge.

Paul Price added a comment - 03/Feb/16 7:02 PM Thanks Russell! I found a test failure when running Jenkins before merging, which I quashed by disabling flag propagation (there's no 'src' dataset in the test, nor flags). Now it passes. Merged to master.

John Swinbank added a comment - 08/Feb/16 6:50 AM Added to release notes: https://confluence.lsstcorp.org/display/DM/Data+Release+Production+WIP+W16+release+notes#DataReleaseProductionWIPW16releasenotes-Propagateflagsfromindividualvisitstocoadds