Miriad Meeting 28 October 2010

Jump to: navigation, search

On October 28, 2010, preceding the Blitz65 fest, a number of MIRIAD developers plan to get together in Berkeley to discuss present and future of MIRIAD.


Place and Time

Campbell Hall, October 28th, 10am-1pm and 2pm-4pm PDT, room 635.

In addition to being on the phone, remote developers might want to look into some screen collaborative tool. VNC comes to mind, but the MMM group has been using YuuGuu. OpenMeetings might be another alternative?


  • Steve Croft
  • Douglas Friedel (telephone)
  • Colby Gutierrez-Kraybill (telephone)
  • Gordon Hurford
  • David MacMahon
  • Jim McTiernan
  • Peter Teuben
  • Peter Williams
  • Mel Wright
  • Jun-Hui Zhao (telephone)

Potential Agenda Items

please add your favorite items and if you want/can present something on this

  • recent developments ( wcslib, fft, using gpu parallel processing... )
    • wcslib - peter T
    • fft / fftw
  • build methods
    • old style - peter T
    • new style - Peter W
  • ATCA coordination
  • SMA status
  • wrappers and user interfaces ( pythonization, user tasks and scripts....)
    • we have a LOT of python interfaces
  • memory management on large datasets - Steve
    • abstracting memory access to better handle memory management? - Colby
  • Frequency-dependent polarization leakage calibration - Peter W
  • CASA and MIRIAD interfaces - Peter T
  • pipelines (e.g. drpacs, cf. recent mmm disucssion] - Peter T
  • Strategic plan for MIRIAD (AKA "MIRIAD in the Age of CASA and Giant Datasets")
  • Organizing the MIRIAD community - Peter W
    • Centralizing websites, mailing lists, installation instructions, etc.
      • Specific platform support
    • Quantifying the userbase
    • Ties in to the myriad Python interfaces, ATCA coordination, strategic plan, etc
  • Old sleeping projects
    • continuous integration testing
      • hudson [1] -- Colby
      • tinderbox [2] (we use that for CARMA)
    • regression tests with and without real telescope data
    • scripts to run many-channel mapping & deconvolution in parallel - Peter T
    • BIMA compatibility (e.g. listobs, gmake-routines)
  • Miscellaneous items we rather not talk about?
    • autotools vs. cmake (CASA is now using cmake), any other popular options?
    • provide update system that does not require full repository checkouts, instead providing an updating tool?
    • abandoning CVS? use SVN/trac? or git? or....
    • requiring Fortran-90 style or F95?


This agenda will be modified as we go through the day:

  • Morning session (10-1pm)
    • 10.00 welcome (Mel?) - resolve technical issues communication with each other
    • 10.15 introduction (Peter) - set agenda for the day
    • 10.30 etc etc etc
  • Afternoon session (2-4pm)
    • 2.00
Personal tools