PCEnv developers meeting 20080509

Location: Auckland

Present: Alan Garny (Phone from UK), Andrew Miller, Justin Marsh, Randall Britten


Summary by Randall Britten (and a bit of editing from Alan Garny on the bits related to him)

General

AG: tested ten Tusscher 2006 model in three different programs.  PCEnv (version 0.3.1 under Windows) differs from the other two (COR and AGOS).  Mostly the same AP, but slight differences during repolarisation.  AG will double check the model against the trunk version of PCEnv under Linux, as well as the latest snapshots (both under Windows and Linux). If the differences still exist then he will check units at component connections, and add a tracker item if the problem is genuine.

RB: Regarding tracker item 286 (MathML rendering), suggested that STIX beta fonts be installed along with PCEnv as part of PCEnv installer if STIX license allows.  Reasons: slightly better, and not all Windows systems will have the MS Cambria Math fonts.

AG: There might be difficulties on installing to system fonts locations on Windows.

AG: Kyoto's approach to DAE system decomposition.  Japanese/English language barrier making this harder to grasp.  Paper that has been circulated is still a rough draft.  AG will try get a simple example showing in detail how approach works, as well as trying to get source code.

AM: Appears from paper that Kyoto algorithm relies on heuristic approximation to graph theory problems.

AM: Will proceed with a recursive search that will time-out after user configurable time, or allow user to cancel.  Recursive search will work from both sides of the problem, alternating between trying to identify the smallest system within the larger system, and then trying to identify how to reduce the larger system.

Discussion of readiness for release of PCEnv 0.4

Major drivers for releasing ASAP:

  • Tracker item 380 (which is now fixed) is a bug with the current release which prevents some models being run.
  • MathML rendering is available.
  • Code generation is available (c and Fortran).


Tracker item 382 will be used to track blockers.

Currently:

104: widgets misaligned with tree cells.

361: Cannot launch PCEnv by opening CellML file.  Caused by recent rework by Mozilla of security restrictions on XUL application scripts.

286: Some minor remaining issues on MathML rendering

AG: Will look over tracker to see if he feels there are any others.


YAPE/PCEnv/COR planning

RB: Overall approach: do this in 3 stages

  1. Add features one at a time, without regard for overall layout
  2. Refine overall layout, largely guided by layout presented by Alan at recent CellML workshop.  Layout hard-coded at this stage.
  3. Add feature to allow layout to be dynamically controlled by user (like Eclipse and other similar IDE's)

All agreed this was a sensible approach.

AG: would like to incorporate a more robust and feature rich XML text editor by using Scintilla, the same editor component that is used by Notepad++. COR language will come later, since refinements to its grammar need to be discussed.

AG: will enter as tracker items any features from presentation at recent CellML workshop.  Many tracker items that are features that COR has that we would like PCEnv to have were already added last year.

RB: Suggested AG work on Matalb code generation since this would benefit a lot of users and AG has already implemented this before in COR.

AM: Asked whether COR's Matlab generation was one computation per time step, or dealt in a more Matlab consistent style with expressions that could be done on entire vectors.

(After meeting, RB tested this, and found that COR Matlab is one computation per time step.)

JM: We must ensure that code style/architecture does not later preclude converting UI to allow for dynamic layout.

AM: Current style relies on XBL, and thus will not have this problem.

The issue of the name was discussed.

RB: Started to prefer Physiome CellML Environment (PCEnv).  Does not like YAPE.

AM: We should get www.pcenv.org in the mean time.

RB: Will try revive the discussion for renaming, since it was frozen in order to discuss the PCEnv goals at the CellML workshop, but due to short time, progress was minimal.  A tracker item will be used for the name discussion, and the tracker item highlighted by mailing cellml-discussion.