wiki:DM/SAT/Meetings
Last modified 3 years ago Last modified on 05/27/11 11:14:13

SAT Meeting Agendas and Minutes

The DM System Architecture Team currently meets weekly. The meetings are not closed, but they are intended as working meetings of the team, to which others will be specifically invited as relevant topics arise on the team's agenda.

Future agenda items

The following items have been raised as needing attention from the SAT. Please propose additional items either as mail to the SAT members (at least to the System Architect) or to lsst-data, with the Subject: line clearly indicating that it's an SAT agenda request (include "[SAT]", at least). Inserting Trac comments here is also a way to add to the list.

PT2 planning through PT3 execution

  1. State of eups and the build system (added 2010.3.19) and its handling of dependencies -- serious review on 2010.7.9 (11/5 RHL: desperately need to release latest version; need to handle diamond dependencies, have proposal +)
  2. Unification of persistence patterns (added 2010.4.2, review after PT1 development stabilizes) (11/5 gpdf and KTL: defer until after PT1.2)
    • Metadata persistence, SDQA persistence, Source-redesign reflection/persistence mechanism, Boost persistence, etc.
  3. Review of provenance (at orchestration layer or lower? assigned ids or dataset hashes?) (added 2010.05.25) (11/5 KTL: post PT1.2)
  4. Dependencies of afw, etc., on database schema; repackaging? (added 2010.4.2) (11/5: Near-term needs addressed already; major redesign of Source needed post PT1.2)
  5. Coding standards for the use of global functions - review of feedback from Apps and MW groups (JK: please conclude by end of DC3b) (11/5: if time permits)
  6. Stage/pipeline policy refactoring to separate science control parameters from plumbing/interface details. (11/5: better separated in PT1.1 pipelines already; further review as part of staging review, partly in PT1.2?)
  7. Support of pipelines (like parts of DayMOPS) with algorithms that require more space than physical memory: should we, and if so how? (11/5: post-PT1.2)
  8. Handling time-varying camera conditions (in ImSim and in real life) (added 2010.4.30, related to "conditions database") (11/5: Butler handles some of this, post-PT1.2)
  9. Compiler-dependent #ifdefs (added 2010.4.30), and ...
  10. Language and Unix standards, compiler feature control macros, etc. (added 2010.5.7)
  11. Review of what actually ended up in the Butler for PT1, lessons learned towards PT2 (added 2010.5.20) (11/5: should review for PT1.2, review Paul Price changes)
  12. Refinement of the orchestration: 1) how should failed processings be reported; 2) how should the orchestration link from SFM to SourceAssoc behave? (1, 2 added 2010.5.20 from suggestions by Robyn)
  13. MOPS algorithms (added from Tim, 2010.6.4)
  14. Revisit HEALPix decision (added by Russell, Tim, 2010.6.23) (11/5: tangent planes for PT1.2)
  15. Review of DC3b-PT2 database schema & lessons-learned from PT1 & consequences of Source redesign
  16. Review of existing SDQA classes, alternatives
  17. pipette design review

LSST reference design and planning (for PDR) and/or DC4 planning

  1. Persistence of (fat) footprints in Sources (added as result of DC3b meeting 2010.4.6)
  2. Chunk size of data from ImSim and camera: should it be amplifier, CCD, etc.? (added from RHL, 2010.4.16)
  3. General issue of balance between centralized coding standards and developer independence
  4. Overall review of pipeline framework
  5. Use of NaNs?, persistence of NaNs? (added as a result of DC3b stand-up meeting, 2010.4.26)
  6. Conditions database issues (particularly conditions that evolve in time)
  7. Need for a database of policies (and other control information) that vary across the sky (e.g., providing different parameters for the PSF measurement algorithm in the galactic plane) or perhaps by time of year (added 2010.4.30)
  8. Large-memory architectures (e.g., for MOPS; added from Mike Freemon 2010.6.4)
  9. Multi-threading / thread-safety issues (added from Robert 2010.5.28)

Record of meetings

Meetings are (as of March 2010) scheduled for 11:00-12:30 PT on Fridays on the DM teleconference line.

Retired agenda items

The following items have been addressed and are retired, at least for now (PT1). They might be revived for future cycles of DC3b or DC4.

  • High-level interface for rendezvous of image data with ancillary data (e.g., flats)
  • API for logical-to-physical mapping for data and dataset identifiers
  • File organization for DC3b (added 2010.3.19)
  • Integration plan for DC3b-PT1; match of orchestration capabilities to PT1 requirements
  • PSF redesign (added by TA 2010.4.2) - perhaps week of 2010.4.26?