Meeting Minutes 16th October 2006
Meeting Minutes 2nd October 2006
Present: Michael , Matt Halstead, Gareth de Walters, Poul Nielsen, Andrew Miller
Andrew: All the technical issues on OS X from last week have been resolved, except relating to the compiler. otools still doesn't compile, and the suggestion from Apple's mail list has been that code modifications are needed to make it compile. Apple has sent a mail which seems to say that distribution of their gcc and ld binaries 'is not permitted'. Poul has asked Peter if we can get a new OS X box for this. Andrew noted that the box we have now is PPC, and so binaries from it will on work on this architecture, while the new binaries will only work on Intel architecture Macs (the newer ones). Poul believes that it is better to generate Intel binaries. Andrew will keep builds / tests on the PPC OSX going now to prevent regressions, but will put further work on getting Mac PCEnv going on hold until the Intel Mac is available.
More progress has been made with MingW32 builds. The CellML API now compiles on a cross-compiled Linux hosted, MingW32 targeted environment, and the binaries behave the same as those on Linux. The PCEnv UI is still not working correctly, however.
Poul noted that there had been some correspondence with Andre via Campfire with a proposed workflow for CellML put up. Poul thinks that we need to get workflows like this up quickly, as we are receiving lots of models, e.g. from Penny Noble's group. Matt thinks that we still need to decide if we want to go with Andre's ontology, and that we should pull parts out of it first and get those working.
Matt noted that we need to get the CellML site migrated to Zope 3. Michael noted that we could get this simply by following Plone, and using Five.
We need to get PCNG, and the Plone Software Centre migrated for this to work.
Gareth noted that we currently have four sites at the institute using Plone: physiome-project.org, cellml.org (on Zope instance), and CMISS and the internal CMS on separate instances. All running Plone 2.1.2 (with security patches) on Zope 2.7.8, with Python 2.3.5.
Matt asked how we are backing up the repository. At the moment, the ZopeFS is being rsync'd hot from the live site. This could result in data loss if the site grows while it is being rsynced. Michael suggests that Reposo could be used to safely perform a hot backup.
Michael asked about the testing environment we use. Currently, people develop and test on their own local Zopes, and Gareth tests configurations on a separate Zope instance on his local machine, and/or on bioeng2. Matt is keen for there to be a common, well maintained testing environment. Michael thinks that we should be able to easily create clean testing environments, so we don't have to worry about the risk that they get trashed.
Michael noted that we may be able to use Python Eggs (zip files with metadata + code) to distribute our products / libraries. He also suggested that Gareth could use ZExplorer to traverse Zope objects, and plan for migrations.
Michael will be here for 6 months.
Matt noted that we need to get the types of RDF data from CellML to also go into CMISS models. Poul was concerned that there was no easy way to find CMISS models associated with e.g. a certain organ. Poul thinks this should be done by letting physiome.org display trees from ontological data, and then see the associated CMISS and CellML models. Matt also noted that Peter is keen to see the old physiome ontology product available again and working with the new data. Matt is concerned that this will take too much of his time away from developing the ontology.
Gareth will look at site-tools, and see how we can use them at the institute.
Andrew: All the technical issues on OS X from last week have been resolved, except relating to the compiler. otools still doesn't compile, and the suggestion from Apple's mail list has been that code modifications are needed to make it compile. Apple has sent a mail which seems to say that distribution of their gcc and ld binaries 'is not permitted'. Poul has asked Peter if we can get a new OS X box for this. Andrew noted that the box we have now is PPC, and so binaries from it will on work on this architecture, while the new binaries will only work on Intel architecture Macs (the newer ones). Poul believes that it is better to generate Intel binaries. Andrew will keep builds / tests on the PPC OSX going now to prevent regressions, but will put further work on getting Mac PCEnv going on hold until the Intel Mac is available.
More progress has been made with MingW32 builds. The CellML API now compiles on a cross-compiled Linux hosted, MingW32 targeted environment, and the binaries behave the same as those on Linux. The PCEnv UI is still not working correctly, however.
Poul noted that there had been some correspondence with Andre via Campfire with a proposed workflow for CellML put up. Poul thinks that we need to get workflows like this up quickly, as we are receiving lots of models, e.g. from Penny Noble's group. Matt thinks that we still need to decide if we want to go with Andre's ontology, and that we should pull parts out of it first and get those working.
Matt noted that we need to get the CellML site migrated to Zope 3. Michael noted that we could get this simply by following Plone, and using Five.
We need to get PCNG, and the Plone Software Centre migrated for this to work.
Gareth noted that we currently have four sites at the institute using Plone: physiome-project.org, cellml.org (on Zope instance), and CMISS and the internal CMS on separate instances. All running Plone 2.1.2 (with security patches) on Zope 2.7.8, with Python 2.3.5.
Matt asked how we are backing up the repository. At the moment, the ZopeFS is being rsync'd hot from the live site. This could result in data loss if the site grows while it is being rsynced. Michael suggests that Reposo could be used to safely perform a hot backup.
Michael asked about the testing environment we use. Currently, people develop and test on their own local Zopes, and Gareth tests configurations on a separate Zope instance on his local machine, and/or on bioeng2. Matt is keen for there to be a common, well maintained testing environment. Michael thinks that we should be able to easily create clean testing environments, so we don't have to worry about the risk that they get trashed.
Michael noted that we may be able to use Python Eggs (zip files with metadata + code) to distribute our products / libraries. He also suggested that Gareth could use ZExplorer to traverse Zope objects, and plan for migrations.
Michael will be here for 6 months.
Matt noted that we need to get the types of RDF data from CellML to also go into CMISS models. Poul was concerned that there was no easy way to find CMISS models associated with e.g. a certain organ. Poul thinks this should be done by letting physiome.org display trees from ontological data, and then see the associated CMISS and CellML models. Matt also noted that Peter is keen to see the old physiome ontology product available again and working with the new data. Matt is concerned that this will take too much of his time away from developing the ontology.
Gareth will look at site-tools, and see how we can use them at the institute.