2012 Software Review Planning Meeting: Mar. 9, 2012
- 1 Connecting
- 2 Agenda
- 3 Minutes
- 4 Action Items
- Since Curtis won't be available, Matt can dial directly to Eugene's office at: (757) 269-6959
- Previous Meeting (Mar. 2, 2012)
- Status of Review Organization (JLab organizers)
- Cross-Hall Collaboration
- Meeting with "all 4" halls on 3/8/2012
- Histogram Monitoring
- Single Event Display
- Physics division effort to identify and promote cross-hall collaboration
- Meeting with "all 4" halls on 3/8/2012
Dear all, We plan a followup software review preparation meeting tentatively on Wednesday March 14th, from 3:30 to 5:00 pm (EST). The hope is to then go over the "less-than-one-summarizing-page" to stare at on the sub-meetings that many of you have been doing. We likely will also discuss a bit more preparations and dry runs for the final software review. For now, the meeting is tentatively plannned for A110, but this may change. Best regards, Rolf
- Organization of Hall-D preparations for Review
- Introductory talk (preliminary slides)
- Software Manpower within collaboration (no change) (Mark)
- 2012 Offline Computing Schedule(no change) (David)
- Software Milestones/Challenges
Attendees: David L., Mark I., Eugene C., Matt S.
Status of Review Organization
No new information
- Rolf has tentatively planned a meeting for Wednesday next week to discuss the status of the cross-hall collaboration on software.
- Matt asked if a phone connection would be available. Dave said he would check on it.
- There was a meeting yesterday to discuss the topics of Histogram Monitoring and Single Event Displays. The latter topic has had work done to implement a Hall-D viewer in the Hall-B bCNU framework dating back more than 1 year.
- Ole Hanson described the Hall A system of histogram monitoring which has been in use successfully by them for some time. It is ROOT based and relies on quick processing of the data shortly after acquisition (not during). The QWeak experiment in Hall-C has adopted this and future Hall-C experiments are likely to adopt it as well.
- For hall-D, we have ROOTSpy.
- Hall-B has used the so called "mini-hbook" (no relation to the cernlib HBOOK") for some years successfully. It builds histograms in the single board computers themselves (ROC's) and sends them through the data stream occasionally. A separate process is attached to the ET system and extracts the histograms as they come by and displays them.
- We discussed incorporating the ability to accept the histograms as defined in mini-hbook into ROOTSpy. Sergey was going to send David some source code and then they would talk offline.
Organization of Hall-D preparations for Review
We filled Matt in on Curtis' suggestion to have a 10 minute introductory talk at the beginning of the Hall-D presentations at the June Software review. Curtis sketched up a few slides to give a flavor of what he was thinking. We looked at them briefly. There was general consensus that this would be a good idea. There were a few questions on some of the slides but the discussion was deferred until Curtis could be there to answer them.
Software Manpower within collaboration
No changes to the estimates of available manpower.
Offline Computing Schedule
Nothing has been done on this recently. Matt asked about the exercise we'd planned where folks who have written large pieces of code we're claiming to be XX% done have given their estimates for how long it actually took them. This could be used to calculate a percentage complete based on the total estimated time for that activity. Dave agreed to take this up again by contacting authors of some of the larger pieces of code.
We discussed the issue of milestones and data challenges. The general consensus was that any effort to do stress-tests or data challenges at this point (i.e. before the June review) might be more of a distraction than benefit. We should, however, think of a list of milestones/data challenges that could be done prior to the engineering run to help ensure preparedness. Mark pointed out that the reviewers will likely be interested in knowing details of this plan as this was emphasized in the review report from last May. Matt noted some of this has been started for simulation already with UConn's efforts to produce large samples of simulated events on the GRID.
We briefly discussed the issue of calibration and alignment. It was felt that the reviewers may be interested in knowing some details. At least, expected data rates/volumes. Some detector systems have already thought seriously about this will others are just beginning. Efforts have already begun to make this an issue that is tracked for all detector systems.
- Eugene noted that it would be in our interests (even outside of the review) to find out what other big experiments are doing for offline analysis. He has already talked to a contact at LHCb and shared some of what he learned from him. End user impressions might be of particular interest.
- Eugene asked if any other software workshops similar to the one on Parallel Computing held last year are being planned. No one knew of any. We discussed options ranging from a Hall-D specific workshop to simply inviting guest speakers to give Friday seminars. No definite action items came of the discussion other than we will consider the options some more and perhaps discuss this again.
- Check if there will be a phone connection to Rolf's meeting on Wednesday next week (David)
- Gather human estimates on time spent on large software pieces to date and compare to percentage done estimates in activity list. (David)
- Make draft plan for data challenges/milestones prior to 2015 engineering run. (David, Mark)
- Look at details of other recent, big experiments and get end user "lessons learned" (All)