Coordination Standup Notes 2014-07-08

Date Jul 8, 2014

Middleware notes are available at Middleware Meeting Notes 2014-07-08
Steve and Greg are making progress. The sprint was closed and the unfinished items were carried over to the new sprint.
They are working on finishing up some of the last things and making sure that everything is visualizable and presentable if Mario ever has a presentation day. Perhaps he'll have one at the AllHands meeting.
Greg has been gaining experience creating dockable containers for condor and including eups stacks. That may be of use for Simon.
Talked with Condor and they would like to understand our needs a little bit better so K-T is working on a document for them
K-T was able to give the butler some attention last week and a bit more on Monday but it's still not at the point that he can present anything yet.

Perry has merged the first part of the transition to meas_base - the part that allows you to use meas_base instead of meas_algorithms. He's now working on some issues with slot usage in afwTable to make the full transition a little bit smoother.
Jim is about to return to working on similar stuff, adding table aliases, which has been long promised but never actually done.
Robyn commented that there was a problem that Perry needs to address which is occurring in the Demo2012 (used as an end-to-end test). The stack builds with everything that Perry has integrated. However Demo2012, which is not incorporated into the stack but is generated during a post-stack build step, is no longer building.
- Jim: That does a column by column comparison of...
- Robyn: yes it does. However the column by column comparison happens after the compilation which is failing. There is a object - maybe called something like tableVector - that is not being found. It's a new field.
- Jim: there a thread on this?
- Robyn: I sent the first sign of it to Perry over the weekend; Robert also noticed it in a build he'd triggered and we exchanged email. I'll send you the emails.
K-T wanted to apologize for unfairly suggesting that meas_base might be a cause in DominiqueB's stack processing slowdown. It appears to be entirely convolved in afw and they don't know why.
- Jim: meas_base should not be enabled by people anywhere yet.

Andy has been working towards finishing his DCR modeling sprints. In particular, he's been looking at how well we need to know source colors in order to be able to model the refraction in DCR.
This sprint will be finishing up in the next couple of days. Andy wonders if he should transition to Robert's Call for Task documentation ahead of continuing his work on the DCR so he can get the documentation done before the All Hands meeting.
- K-T suggested Mario or Robert would be the ones to ask. K-T offered his opinion that if interrupting the thread of DCR tasks is necessary to get the documentation done before the AllHands, that is a good idea. AllHands is the prime deadline right now.
- Andy will transition away from the DCR tasks in order to get the documentation finished in time.

Robyn said the buildbot how-to documentation is in reasonable if not polished condition. She's sending out the subtasks for review.
K-T: Is it staying on lsst-buildx or is it moving to lsst-build eventually.
- Robyn: it'll stay for the meantime on lsst-buildx – primarily because the move involves coordinating with other people and I want to get this released.
K-T: Were you aware of the issue with lsstsw2 home directory?
- Robyn: What issue?
- K-T: It's getting huge.
- Robyn: It was shadowing the system stack builds. But it should not be active now at all.
- K-T: It's on the order of 100's of GB (Editor: actually 103Gb) with many copies of afw, afwdata, and sim stuff. The LSST-admins noticed and are concerned.
- Robyn: I'll clear it.
- Post meeting note. The directory consuming the majority space was the shadow stack. Since a DM stack accretes versions of packages as their source is modified or their build product is rebuilt with different versioned dependencies, the space required grows. This is an issue which needs to be addressed as we enter construction. We may want to ream the stack of non-release versions after each Release is cut.