PMR2/Workflows meeting 4/5/2012

Present: David "Andre" Nickerson, Dougal Cowan, Randall Britten, Hugh Sorby, Martyn Nash

Discussion on MAP client

  • Hugh: Working on code/design for MAP workspace, writing code for the workspace manager.
  • Andre: suggested Eclipse.
  • Hugh: wants plugin system to be simpler.
  • Randall: is this reviewed in a document?
  • Hugh: documentation mainly about Python vs other technology choices etc. Peter and Thor have the documents.
    • Action item Hugh: send document to Randall.
  • Randall: Why did Alan Garny choose a different plugin architecture?
  • Hugh: want to abstract away VCS issues, but still provide low level access when required.
  • Randall: what is the status of large binary data?
  • Andre: what counts as "large"?
  • Randall: there is a way to work around the Mercurial 10MB limit.
  • Randall: suggested that if VCS does not store binary data/blobs, then it should store a hash/digest (MD5, SHA etc.).
  • Marty: What about version control for the processing tools?
  • Andre: process described in metadata annotations, these are version controlled.
  • Hugh: Dougal wrote up three test cases, currently implementing Massoud's case - segment rabbit legs through to OpenSim simulation.
  • Randall: explained one of Peter Hunter's Ricordo ideas.
  • Andre: server side processing brings in server limitations.
  • Dougal: BeSTGRID data fabric and computation grid could provide resources.
  • Andre: grid computing requires complex infrastructure that we don't want to have to build ourselves.
  • Dougal/Andre: either save data produced or describe process so it can be reproduced (metadata).
  • Marty: preferably save data to make it easier to adjust or work from a particular step of a multi-step process.
  • Andre: not all intermediate data is serialized.

General PMR2 discussion

  • Hugh: Poul needs musculo-skeletal models up on the repository, will speak to Dougal about this - bones and muscles of arms and legs.
  • There was a discussion of date and ownership of data. What happened to AnatML bones from skeleton using Farro arm? AnatML only had bones.