Coordination Standup Notes 2014-06-13

Date

Jun 13, 2014

Dominique is able to run his full CFHT dataset thru the pipeline giving astrometric solutions on all the chipswhich is a huge improvement over where thigs were about a week ago. So they got that up and running.
Simon has come across a DESC-related blocker in that on the system that he's installing the build system system-level python is getting picked up somehow and causing seg-faults when he runs the codes so he's going to be checking with Mario on this.
- Jacek: mentions that his group has been fighting with the same problem on Qserv. He doesn't know how related this is but they do use the eups package so it could be related. They have a ticket open and one of his group made some changes to fix that. Jacek is providing Andy and Simon with the Ticket number.

Andy has been working on modeling how to estimate the amplitudes of refraction and the differential refraction just given bulk source colors since that's all we're going to have for most of the LSST sources.
- He was not able to get a really good model using the linear fits so he's going somewhat off the grid and using machine learning code to see if he can get some good nonlinear regressions put together. He's found that we could potentially do a good enough job by using decision trees to regress DCR or refraction amplitude on source colors.
- Since this is not officially being checked into the DMS stack and it's under contrib, he decided that he can use scikit-learn for right now ( http://scikit-learn.org/stable/).

No blockers, nothing major to report
The group keeps having daily meeting hangouts. It seems to work nicely and he thinks they will continue.
- It's at 1pm PST (not the 4pm reported in the last Coord Notes).
They are also continuing having their weekly 4 hour code development meetings.
They have an interesting new involvement occurring. They had one person from CERN who actually has interest from CMS to use Qserv there. That is still in very early discussions so it's for information only at this point.
Dick: comments that he is working on a Confluence page related to use of git and gitolite and it talks about the appropriate process for checking in code, etc. Jacek has a recent wiki page on that topic. What is its location?
- Jacek: https://dev.lsstcorp.org/trac/wiki/db/Qserv/gitMerging

Dick had a conversation with Jim Bosch about
- photometric measurement; and
- finding a learning curve on what we're doing and what we're going to do.

He expects to continue to converse off and on as he puts the material together.

Dick has been paying attention to the experience that Dominique has been having refining what goes on with obsCFHT.
- Dick would like to work with Simon and Dominique to put together a Confluence page on what one would do if you wanted to add a new camera to the system. He realized that Dominique was dealing with only a part of that since he already had a camera defined and he just needed to add distortion net, etc.
- He also wants to include additional information on how to read defect files, etc.
- So Dick would like to work with them sometime but not terribly soon.
He's about to finish migrating the trac page on using git. He has it just about ready except he plans to take the scripts he's written to support the tutorial and put them into a Tutorials package to be installed in git. This will allow him to exercise all the instructions using git and gitolite. Once he finishes that and merges what Jacek sends him regarding git, this product will be ready for review.

Robyn installed a doxygen build into buildbot; the build is triggered on every git change. See: http://lsst-web.ncsa.illinois.edu/doxygen/x_masterDoxyDoc/ for the results.
- Relating to Robert's email requesting all developers to document their 'Tasks", Dick asked if Tickets were going to get assigned to complete that work.
- Jim responded that they should be put into Tickets since all work is supposed to be managed through tickets. He wondered if we need to ask for more guidance on this since he is worried if we're told everyone is supposed to document their own 'Tasks', not all will get done. There are many packages which are not owned by any one person. If Robert and Mario want to get this done, someone needs to be in charge of making issues and assigning them to people.
- Dick: do we have an inventory of all 'Tasks'?
- Jim: That was in Robert's email.
There's a request to update the demo2012 benchmark files for v8.0.0. Robyn made a Ticket and will push it through..

Nothing new for him. No blockers.
Perry has been trying to survey the stack to find all the places which might break when they switch to meas_base from meas_algorithms.

Robert Lupton - assigning the tickets to get all 'Tasks' doxygenated or providing the <Developer:task> correspondence to his delegate.