Meeting Minutes 9th October 2006

Meeting Minutes 2nd October 2006
Present: Poul Nielsen, Shane Blackett, Andrew Miller

Poul noted that there had been a proposal for a review paper on CellML. This has been organised by Alan Garny, and is planned for the Philosophical Transactions of the Royal Society.

Poul also noted that it looks like Tommy will be back here before the end of November.

Poul noted that Matt and Mike Dunstan will be here on Wednesday.

The CellML tools position still needs to be given to the leveller. Poul also still needs to get back to the appplicant for the repository curation position.

Poul is still waiting for Peter to discuss how to reply to the e-mail Shane got sent regarding FieldML.

Andrew has been having some issues getting the OSX build to work. The current system configuration means that it is not possible to right-click on anything, which makes the UI unusable at present. There is also an issue with getting a gcc (and in particular, a linker) to ship with PCEnv on OSX. Binutils doesn't provide a Mach-O linker, and the otools linker from Apple doesn't seem to build on the latest Xcode. Andrew will need to look into whether or not we can assume that there will be a gcc on the system, and if not, what we can do to get a gcc / linker to ship with PCEnv.

Andrew has also encountered trouble getting the Windows build ready. Was previously using STLport on Windows (because the MingW libstdc++ didn't support wide characters, and only came as a static library, which meant that each DLL had separate copies of template instantiations). Unfortunately, STLport has issues with template instantiation too, because it uses static members on templates with get implicitly instantiated in each DLL (so the static members in the DLLs are not instantiated correctly). To get around this, it will be necessary to go back to libstdc++, which works on Linux. It turned out that MingW would build correctly on Cygwin, but Andrew has managed to get a Win32 targeted Linux hosted cross-compiler built, with support for wide characters in libstdc++ (it required a patch to the libstdc++ configure.ac, which was found in the gcc bugtracker). Andrew is making progress (gradually) with the libstdc++ approach.

Shane has been working on getting the GPU to process textures, so parameters like the contrast can be changed on-the-fly. He has got it working on nVidia cards, and also on MESA. He doesn't have access to other graphics cards to test on, since Carey and Andre (who had ATI cards) have left.