Date
May 16, 2014
Attendees
Unknown User (abecker) - elsewhere but provided notes
Mario Juric - on travel
Goals
- Blockers, Just Did, and Do Next
Discussion Items
Jim Bosch
- Yusra has taken over the shapeletPSF Fitting stuff from Jim.
- Perry has mostly finished his comparison work so that's now in Jim's court to review and finish the last bits of it.
- Next week is the last week of their sprint. They are in pretty good shape to finish it off including everything they'd planned. Jim feels that it's worked pretty well; everyone has bought into the system; and they'll set up the next one when they're done with the first one.
Jacek
- Nno blockers
- Qserv meeting notes + 2014_05 sprint on DB Team board is the best place to look to see what we are up to
- Things related to the broader DM: logging is expected to be all finished & packaged mid-next week before/during next Wednesday meeting.
- They will probably start looking into migrating QSERV after that.
- Now is a good time to raise any concerns that you might have with the prototype. It won't be frozen after that but it's better now than later.
- K-T: So what you're saying is - it won't be a prototype after next week.
- Jacek: From the QSERV perspective, yes.
- K-T: The main point is that other people within DM should take a look at the interface - make sure it's OK. There are some things in terms of performance that he's working on that we hope to improve a lot, for example, in the case where the logging is below threshold. Otherwise...if you don't like anything, now is the time to say.
K-T
- Things are moving ahead pretty with the Nightly Computing. Steve has stuff up and running; moving things from a simulated camera to a simulated Distributor and he'll be getting it out all the way to the end of computing soon. It's looking pretty good. They need to analyze it to understand its behavior some more.
- K-T's work is moving forward. He continues to work on the Butler Design document and is making headway there. But he's not ready to make a release there yet.
- At yesterday's Middleware meeting, Donald Petravick mentioned he has access to an experimental machine with a very large RAM disk; he wondered if anyone in DM has a need for such a configuration. K-T talked to Jacek about that and Jacek felt - not now. K-T mentioned some other domains that might want to test this configuration: MOPS or global photometric calibration. He doesn't think that most of the other work being done, such as multifit, would benefit from that hardware. K-T thinks the size is in the TB. Simon will ask around for others who might find it useful.
Simon
- Simon iterated with Dominique to see how on how to deal with defect in CFHT since it appears that the original version was a little bit of a hack just to get things going.
- Simon also helped someone installing catalogs with installing on Ubuntu 12.04. He asked him to send his problem to DM_Stack users because Simon cannot replicate it using an EC2 instance but the user hasn't done so.
- Simon: wonders if the project is thinking about setting aside a little money to run EC2 instances with several different images automatically. It seems like a easy way to go. We could host our own VM's but EC2 has a bunch of AMIs already so we don't have to build them.
- K-T: Yes, We're about to start a discussion about what our developer cluster and our integration cluster are going to look like. We have a pot of money to spend on developer hardware. That would be one possible application of it.
- Simon: Yes, it's not expensive. It costs 4 cents an hour and it takes about an hour to build at most.
Dick
- Finished the tutorial on Processing Stripe 82 Images.
- He's been working on a tutorial for PhoSim processing which is nearly complete apart from integrating the output tables into a multi-color catalog.
- He got really great reviews from Yusra on the Stripe82 Tutorial and from Simon on the task of creating astrometry_net index files; both of which were very helpful and helped him improve the whole documentation suite. He mentioned that they ran into an inelegancy in the stack of having a fairly dated version of the astrometry_net package.
- Simon confirms that Dustin is working on the astrometry_net update. As of Tuesday he is no longer blocked by Jim. Simon will ping Dustin Monday if he hasn't heard anything more.
- He got really great reviews from Yusra on the Stripe82 Tutorial and from Simon on the task of creating astrometry_net index files; both of which were very helpful and helped him improve the whole documentation suite. He mentioned that they ran into an inelegancy in the stack of having a fairly dated version of the astrometry_net package.
- There are a number of tasks which would be useful to do prior to the August meeting; Dick has been entering them into Jira. They include, for instance, working the DM Developer Guide, migrating Trac/Twiki tutorials over to Confluence for git, eups, etc and brought up-to-date.
- Dick also wants to start working on a high level description of cameraGeometry just to introduce the concept and what it involves and, roughly speaking, how it's represented in software. This is not intended to go into the interface for how one uses the interface to use cameraGeom which is likely something which one would see in Doxygen. He'll ping Simon no that task just to get things right.
- Dick has seen the traffic on preparing cameraCFHT. There are a number of steps in that process; he'd like to capture that process while it's still fresh in everyone's minds.
- You can talk about it in two levels: 1) conceptually, what are steps that have to happen if I have camera X that I want to represent in software, and 2) what are the details needed if I'm the developer tasked with writing the software. The latter would be a Tutorial for the DM developer.
AndyB
- No blockers
- Andy is working on his first sprint for S14, looking at the distribution of stars in the phoSim galaxy model to understand the rate of extreme-color stars that will lead to DCR dipoles.
- In the course of writing S14 code Andy found 2 bugs in the W14 code that impact the results of the report (both are in the same piece of code, written in a rush near the end of the development cycle.). He fixed one and reported on it yesterday. He is working on resolving the second one now, and will release another update when it gets fixed. Both changes are used in the S14 code, so its not entirely backtracking.