Meeting Minutes 07 November 2007

Present: Sarala Dissanayake, Vignesh Kumar, James Lawson, Catherine Lloyd, Justin Marsh, Poul Nielsen, Andrew Miller, Jonna Terkildsen, Tommy Yu

Apologies: Randall Britten


Agenda

Update on CellML 1.1.1 and 1.2

  • Reiterated that the CellML 1.1.1 specification (if eventually proposed) is to signal deprecation of reaction elements

  • CellML 1.2 is to introduce further fixes (if any) and other additional updates.

    • New features such as typing and matrices, more flexibility, not limiting to scalar types

    • Also enable structure types, such as lists and sets.

  • Poul still need to finish checking up the document about steps to remove reaction elements

  • Poul indicated that the specifics to all these additions are outlined in his talk during the CellML workshop earlier this year.

    • Action item: Poul will gather those material about CellML 1.1.1/1.2

    • Action item: Sarala will gather the charts and diagrams relating to that.

  • Andrew noted that the normative and informative parts of the updated specification will need to be written.

    • The Zwiki in Plone is not good because history is lost when database is packed

    • Andrew suggested that the work in progress version of this document be kept in Subversion.

    • Possibly Git, as a distributed source control manager could aid in this

      • Other benefits and potential issues with Git was discussed here.

      • Action item: Andrew will set a public Git repository for the development of this document as a trial.

  • Poul noted that the development of CellML is really driven by the community, so there should be a process to involve them into the specification drafting process, that CellML should accommodate their use cases.

    • For instance in the current version, the content MathML cannot represent state machines.

    • James and Catherine also came across a model that would need to be represented using matrices, which cannot be created using CellML currently.

  • Action item: Poul will send email about starting the drafting of the next CellML specification

    • Still need to resolve the format of the document (such as DocBook, reStructuredText, or other formats)

Putting up a CellML 1.1 model into the repository

  • Jonna needs to put up a CellML 1.1 file into the repository, but the current repository cannot support any CellML 1.1 models (or anything else other than CellML 1.0 files)

  • James suggests creating a folder in her own members page on cellml.org and put models inside there for the mean time, and replace the page to direct users to the permanent location of the model within the repository when it finally supports CellML 1.1.

  • Tommy noted that the repository to support this is in the works, which brings to the next topic

Update on PMR for CellML 1.1

  • Tommy noted that he is working on integrating Git with Plone, which will serve as a possible foundation of the next CellML repository if this test trial is successful. A very limited working version has been created, but it still got a long way to go.

    • The reason to use Git is because of its distributed nature.

    • Since CellML is really similar to (or even is) software code it really should be tracked as software code rather than as a document.

    • Each model will have its own repository

      • Poul questions why, it was explained that Git doesn't allow checkout of subset of files within the repository yet, and so this forces a repository to contain a single model only, with a "super repository" containing these sub-repositories as submodules.

      • Along with Git's distributed nature, this will allow model authors to have control of their own repositories for their models until they are ready to publish it into the repository.

  • Current plan is to expose an interface to Git via Plone.

  • Andrew noted that Windows users may have trouble using Git, and so the web upload interface will be needed.

    • Mercurial could be a good alternative to Git since it's written in Python, which means it's cross-platform.

      • Although there are reports of performance issues.

      • It may also eventually support Git's data structures so it could be an alternative client.

  • There are still a number of issues to be solved, but the plan is to let this test go on and see its results.

Updates on PCEnv 0.3

  • There is still the clone from output issue that needs to be resolved in PCEnv, which Andrew will work on today.

  • Still need to write functional tests and documentation

    • James and Justin are on that

    • Justin wrote basic documentation for the Javascript API.

James on tasks for Vignesh

  • Vignesh is the new summer student. He will be helping James out with curating CellML models. James will help him get started.