Attendance:  

  Ed Maros,  John Whelan,  Warren,  Joe,  Erik,  Philip, Erik,  Kent,
  Masha, Stuart, Jolien, Duncan,  Alan,  Patrick.


Action:

  Jolien to send announcement about upcoming release date and
  emphasizing the need for testing.
   

Announcements

  1.  Remember that lal and lalwrapper are under /lal now.  So update
  your scripts to pick up the dso's from there.

  2.  Loose jobs in mpiAPI in 0.1.97 

  3.  Kent would like to be able to tag the release on Monday before
  the GriPhyN meeting.   But if that does not happen,  then would be a
  week. 


Status of LDAS pre-release:

  Have made a lot of progress fixing bugs.  Down to 2/3 bugs that are
  really a concern.   Put onto test on Monday.  When new lal and
  lalwrapper,  they put on a new version because of bug in manager
  API.  Put new version from dev to Hanford and MIT.   MIT was more
  difficult,  but they have it up now.  Have not pushed jobs through
  the system at MIT.  All previous pre-releases have 10-40% failure
  rate.  Last night,  finally down to 0.2% failure rate.   Version
  0.1.100 will be pushed to Hanford, MIT, but do not want to put it
  onto test until there is an announcement.  

  There remains problems:  cannot properly handle proc frames.  Have
  done 40-50% of testing.  No documentation verification of this
  release.  But going well otherwise.

  Greg:  some passwords were eliminated.  The passwords dso_XXXX
  should still be there.  They are gradually eliminating other shared
  user/passwords.  

  Greg generated an SFT file yesterday,  but could not use FrDump
  because it is not up to date with the spec.   

  Next push of LDAS at MIT.  Doing a bit of clean-up.  It takes a
  while to make these changes to the system.  It should stay stable
  for the weekend -- including later today and friday.  But Kent can
  change his mind.  

  Livingston has not yet been upgraded yet.  OS upgrades were being
  done yesterday.  They will push a version of LDAS to LLO next 

  Had not properly cleaned up job directories from 0.0.23 which was
  causing problems.    We had that pain going to 0.0.47.   


Status of LDAS : 

  0.1.97 from sw on Monday.  The upgrade was failure painless.
  Everything seems to be working.  What about the eventMon API?  There
  are efficiency issues.  Unfortunately there is FILO effect in tcl.  

  If they have only one dataserver,  they would be getting connection
  refused.  There is a unix limit of 5 on each server.  Each job has
  its own dataserver.  


DSO Authors:

  Stochastic:  The only outstanding problem right now is the database
  issue; summ_value.  They resolved some issues with normalizations.
  Kent and people at LIGO were confused.  John to send a follow up
  e-mail about this and what is going wrong.

  Inspiral:  Things have been going well.  Just going to be using it.
  E7 frames that were corrupted;  Scott replaced the frames.   No
  indication of why they got corrupted.

  Power:  many changes to move stuff into lal from lalwrapper.  A
  standalone code will be available in lalapps soon.  

  SFT:  Greg generated SFT's yesterday.  The problem with proc frames
  will be fixed with highest priority today.   That will be his
  priority.  


Plans for post this release,  but for next release:

  1.  Put into tcl channel:  big job.  Every API needs to be updated.

  2.  Diskcache API:  will re-write the cache code into C++.  He has a
  draft of the specification for persistent.  

  3.  Fix performance problem in eventmonAPI

  4.  datacondAPI:  thread problem.  Less than once a week,  so hard
  to modify.   Request to implement complex math,  cloning,  better
  meta-data API. 

  5.  port to gcc-3


Next Meeting:

  Thursday,  25 April 2002 at 10:30 PDT