ABI Meeting minutes 2009-02-04

Last week's action items:

• Andrew took on this job, which has been done.  There was a low voter turnout (41%), and the winning candidate was OpenCell.
• This has been deferred to next week when Peter returns.

• Done.

This week's agenda:

• Catherine wants to know what the priorities are for curation, as she is the only curator while James is working on the new website.
• What is the priority; new models, a particular class of model, fixing models in the current repository?
• Andrew suggested soliciting opinions from experts in their fields, not existing CellML users, to find out what kinds of models are most desired or important.
• Poul suggested the website might have a submission system, and that curators could select models from the submitted list. Pre-publication deadlines could be made a priority so that a publication might be made with the CellML model in the repository linked.  There could be a page on the website where people could submit requests.
  
• Catherine prefers direct contact with the model authors, to obtain their assistance and code.
• Andrew suggested that externally requested models be prioritized over self-selected models.
• James suggested prioritizing new models - these usually can be curated right away due to the newer systems in place for creating models.
• Poul suggested that Catherine and James put together a workflow for a model request submission page.

• Jeelean has been making session files.
• James has now made a better template for creating session files.
• Catherine has done a simple cardiac model.
• James has made a few example sessions and has been teaching session creation.

• Catherine and James have been testing PCEnv on Mac and Linux respectively, and have found bugs which Justin is addressing.

• Jonna has found that Andrew has changed something in the DAE solver that break what was previously able to be solved.
• Randall explained the change; there was previously a bug that did not pass the error if a good initial guess was not found - this meant that PCEnv would take more guesses until one was found that allowed the DAE to be solved.  The error is not longer skipped and so these cases no longer solve.
• Jonna asked if the scaling of tolerance values is ever going to be implemented.
• Randall stated that we have been using the levmar solver out of the box - we may need to scale the residual in some way.
• Andrew has been writing code to increase the number of random guesses if a good starting point is not found in the initial 99 random guesses.
• Randall suggested that we may need to set up some automated testing of the DAE solver.
• Poul asked how the automated initial guesses are chosen; could these be included in a session file?
• Randall stated his support for using the initial_value attribute.  However, the initial_value problem must be deferred to the next release.  Can we solve this problem some other way for the 0.6 release?
• Andrew asked if the release is being held back by anything else; Justin reported that bugs were found during testing and they need to be fixed.
• Randall said that the options are to either accept the bugs as part of the experimental nature of the DAE solver and release anyway, or to push back the release to try and fix the problems using various approaches.
• Andrew said that adding his code to try more random guesses if the first 99 do not work could be added in order that the 0.6 release can solve the same DAEs that 0.5 is able to solve.
• Jonna reported that one of her models is failing, outputting a NAN (?) having a negative value where a negative value should not be possible.  It was agreed at this point that these issues should be discussed later.
• Justin said that the 0.6 release will have to be pushed back, or 0.6 released with the bugs found during testing.
• Randall suggested holding back the release and attempting a fix of the DAE issues.
• Adam is wondering what sort of interface should be implemented for the services he has been working on.

• Catherine reported that James and Dougal are about to begin migrating content over from the old site to the development site - the dev site has the information architecture laid out as folders ready to accept content.
• Catherine asked how the specifications are going to be put up; in W3C format, or some other format?
  

• This discussion was a continuation from the website agenda item, discussing how to display the specs on the new website.

• Andrew reported that we only have draft specs at present.
• Poul suggested that it is important that we have a method of showing what is currently being considered for the specs.
• Andrew outlined a variety of methods for dynamically displaying drafts, and ways for people to author drafts.
• There was a discussion about exactly what in the way of draft specs should be put up on the site, and how these will be chosen.
• Poul is keen to get normative and informative versions of the specs up on the new website.

• Tommy wrote the code for the transformation of temp documentation within CellML files to HTML for use in PMR2 Exposure in Python.  This will now allow model exposure to have documentation generated like the current PMR.
• James asked whether we could use HTML instead of temp documentation and have an editor, perhaps the Plone editor to allow the editing of documentation for the models.
• Tommy notes this is on the list of things to be done, but will see if time allows to include this before the workshop.
• Tommy asked what we want before the workshop:
• Metadata editor for PMR2?
• Mercurial access via PMR2?
• HTML interface?
• Randall said that the bulk of the work of a database/repository is in the retrieval of information - we should concentrate on this working for the workshop, even if there is no website based Mercurial access.

• Catherine has had feedback from Nicolas Rodriguez that yes they are aware of the converter and he has given us (Andrew) some valid feedback

• Catherine will organize a Skype meeting between Nicolas, Andrew, Sarala and Alice.