ABI CellML Meeting minutes 2009-07-15
Present: Catherine Lloyd, Mike Cooling, Jonna Terkildsen, Dougal Cowan, Poul Nielsen, Randall Britten, Tommy Yu, Amir Shirkhani, James Lawson, David "Andre" Nickerson, Justin Marsh
Apologies: Peter Hunter
3) Exposures - what are they and how should they be used?
Apologies: Peter Hunter
This week's agenda:
0) Time delays- Catherine received an email this morning from Dom Layfield about time delays. Catherine spoke to Andrew, who says there is a draft spec of time delays.
- Randall said that there is a newer tracker item than the one Don referenced; tracker item 1543. The current description is not in a very user friendly form.
- Andrew suggested we encourage Dom to participate in the discussion about time delays.
- Andre said that we need someone to implement the ideas we have so far into the API so we can test them; we don't have someone with the time to do this at present.
- A breakaway was called.
- Catherine asked if we can freeze the draft.
- James says yes. Alan has made a lot of comments and corrections to the 1.0 draft, and the fixes have been made. It is ready to freeze.
3) Exposures - what are they and how should they be used?
- James explained the current PMR1 style usage of exposures, and the problems he sees with this; too many exposures that are hard to manage, and currently multiple exposures of single models in the listing.
- James explained that at the moment, having a single exposure for each model in the listing is the only way to avoid a big mess. This is a presentational issue.
- James is currently making a new exposure for every commit. This breaks the "software release" analogy, and is very time-consuming. James would prefer a workspace focus over an exposure focus.
- Mike and others think the listing issue is a user interface problem; this can be dealt with by designing or allowing appropriate filters.
- There was a discussion around workspaces, exposures and the main listing/s. Workspace centric listings are seen as problematic because workspaces can contain multiple models, for example, or may just be scratch-pad or resource type containers.
- Catherine says we need documentation! A glossary and guides.
- Jonna is confused by the repository listing, due to current naming inconsistencies.
- Catherine suggests we need a best practice as soon as possible.
- Andrew wonders if we need a curated list, e.g. of published models.
- James suggests that people should only expose a model once they are sure it is ready for release; this is more akin to the "software release" analogy.
- There was a discussion about multiple exposures.
- It was decided to call a breakaway meeting to continue the exposure discussion.
- Tommy has implemented Peter's request, and released PMR2 0.1.1.
- James has made changes requested by Peter.
- James is continuing to do unit checking, but is not yet making exposures.
- James has been working with Jonna on using the repository, and is taking notes for fixes and features.
- Catherine is working concurrently on several large models.
- Catherine reported that the group has decided on a Monday deadline for putting up the new site.
- Tommy said that ITS should have a VM cloned by tomorrow, then it will be time to migrate.
- James reported that final quality control is underway.
- There was a discussion about UK/US spelling. It was decided that we won't worry about it, unless there is mixed spelling on a single page.
- Justin has been doing bugfixes, fixing things highlighted by James.
- Jonna reported a bug whereby opening from the windows explorer context menu (open with) does not work. This is a known issue to do with Administrator rights and spaces in paths on Windows.
- Andrew has got the build system on Windows mostly working, bar a couple of issues. (?) It is working on Linux, and broken on Mac, with a known cause.
- The build system is almost there!
- Justin has looked at Jonna's tracker items.
- There was an extensive discussion about opening 1.1 models in OpenCell, to do with XML base.
- Andrew reported that Sarala is now OK with the Java API, and Morgan has not yet got back to him about it.
- Randall reported that Sarala has put together Java examples with DocBook documentation - there is no Doxygen hyperlinking possible for the Java. Andrew said that integrating the Doxygen linking would be possible, but a big job that would involve creating our own custom extensions to Doxygen.
- Amir has been adding notes to current Python documentation.
- James has left the API section of the website to Andrew and Justin.
- James will send out 1.0 of the metadata spec for a final look and the freeze it on Monday.
- James says that we are currently using metadata for things it was not designed for (such as the PMR2 exposures listing); this needs to be fixed.
- Jonna wants to know if the initial value fix she is currently using is going to be permanent, or a temporary fix; she wants to share the model that uses it.
- Randall said that this brings us to the next step of the process; are we going to make this change permanent, and if so, how?
- Andrew suggested that we need to look more deeply at the initial conditions problem, and perhaps try to tackle it using constraints.
- Randall suggested that we know that we need initial values - what is the next step?
- Andrew said that implementing constraints would be the next step, but that we need to standardize the initial condition hinting system. We could possibly publish a technical note on the current implementation for Jonna to refer to.
- Poul said that we might need a technical notes section on the new website for things like this.
- Andre asked if we have tried other DAE solvers yet - we have not.
- Randall has tried other solvers by hand-coding.