Meeting Minutes 23 May 2007
Meeting Minutes 23 May 2007
Present: James Lawson, Catherine Lloyd, Jonna Terklidsen, Poul Nielsen, Tommy Yu, Randall Britten, Dave Brooks, Andrew Miller
Andrew has started the process of making CCGS use the newly separated components. As part of the work on his PhD (which is creating a program for dealing with mixed black-box / white-box models) he is creating a program that is in some ways similar to CCGS, and as part of this process, has identified that MaLaES needs to be able to generate expressions where the overall expression is in different units.
He also has fixed some minor PCEnv issues that were blocking other people's work. In particular, he fixed some issues with error detection that were making it hard for Catherine to find errors, and also fixed an issue with loading models from the website, which resulted from updating Mozilla about a month ago.
Andrew noted that bugs in snapshot releases of PCEnv were being treated as 'urgent' issues by Peter, despite the fact that snapshots are not supposed to be formal releases of any kind. The underlying problem is that the last release was so old that everyone ends up using the snapshot to get the latest features. To fix the problem we need to get people who aren't prepared to take the risk associated with bleeding edge code off the snapshots and onto stable releases.
There was a lengthy discussion about what process should be adopted in terms of making the next release and in making regular releases. The outcomes were that we should make a release fairly soon, that we should involve the community in running functional tests, but that we don't need to branch, make release candidates, and wait for a period of time to make releases. Instead, we can just get interested users such as James and Jonna to run the standard functional tests, and any other ad hoc tests they want to do on the spot, and then make a release immediately if the tests pass. Any bugs which get missed by this process get fixed in the next week. There was also discussion about how often releases should be made. Given that Andrew is only working 10 hours, it was decided that every 2 months might be one idea, if Andrew spends 2 weeks out of the 2 months on it, although this might depend on the level of changes which are helpful to the general external user.
There was also discussion about how the new functional testing regime would be managed. Andrew suggested that Litmus might be a good project to use (Litmus is the Mozilla Foundations functional test management product) since it is geared towards community run functional tests. For now, it will just be run on Andrew's developer system (bioeng44), but eventually, it would be good to get it moved onto a publicly visible webserver.
There is a need for an expansion of the models available for use in these functional tests. The decision was not to put a lot of effort into this at one point, but let it evolve over time.
James and Catherine have been working on fixing up models with no reaction elements. Many of them actually have full equations in the CellML model, and Catherine is recoding many of them from scratch, because the mathematics in the reactions was not a good enough representation of the paper or the biology to make it feasible to automatically convert the reactions into non-reaction based equations.
James has completed 4 models, and Catherine has completed 8. Catherine has also produced a new endocrine model, as an initial run-through for Peter's endocrine project.
Andrew suggested that to submit the new models into the repository, she uses the docbook in the existing models as a template, and then just uploads the images separately to the Plone site. He noted that source figures (e.g. the Xfig sources) would also be worth putting up, because they are a valuable resource for generating things like SVG.
There is still a problem with papers which have multiple models. Matt's originally proposed naming system allowed for models to be designated A, B, C, to represent this, but the repository design used at present doesn't allow for an easy way to support this. Tommy is looking into it.
Tommy has been working on the metadata editor. He has also been in correspondence with Mike about getting site tools up.
Andrew asked what the process should be for requesting that Mike look at problems on the site. Tommy will send Andrew Mike's e-mail. Andrew suggested that perhaps Mike could attend CellML meetings. Poul thinks it would be good if he could attend CellML meetings, but at least is keen for him to meet with Tommy, and perhaps Randall, at which point issues at the CellML meeting could be passed on to him.
Andrew has started the process of making CCGS use the newly separated components. As part of the work on his PhD (which is creating a program for dealing with mixed black-box / white-box models) he is creating a program that is in some ways similar to CCGS, and as part of this process, has identified that MaLaES needs to be able to generate expressions where the overall expression is in different units.
He also has fixed some minor PCEnv issues that were blocking other people's work. In particular, he fixed some issues with error detection that were making it hard for Catherine to find errors, and also fixed an issue with loading models from the website, which resulted from updating Mozilla about a month ago.
Andrew noted that bugs in snapshot releases of PCEnv were being treated as 'urgent' issues by Peter, despite the fact that snapshots are not supposed to be formal releases of any kind. The underlying problem is that the last release was so old that everyone ends up using the snapshot to get the latest features. To fix the problem we need to get people who aren't prepared to take the risk associated with bleeding edge code off the snapshots and onto stable releases.
There was a lengthy discussion about what process should be adopted in terms of making the next release and in making regular releases. The outcomes were that we should make a release fairly soon, that we should involve the community in running functional tests, but that we don't need to branch, make release candidates, and wait for a period of time to make releases. Instead, we can just get interested users such as James and Jonna to run the standard functional tests, and any other ad hoc tests they want to do on the spot, and then make a release immediately if the tests pass. Any bugs which get missed by this process get fixed in the next week. There was also discussion about how often releases should be made. Given that Andrew is only working 10 hours, it was decided that every 2 months might be one idea, if Andrew spends 2 weeks out of the 2 months on it, although this might depend on the level of changes which are helpful to the general external user.
There was also discussion about how the new functional testing regime would be managed. Andrew suggested that Litmus might be a good project to use (Litmus is the Mozilla Foundations functional test management product) since it is geared towards community run functional tests. For now, it will just be run on Andrew's developer system (bioeng44), but eventually, it would be good to get it moved onto a publicly visible webserver.
There is a need for an expansion of the models available for use in these functional tests. The decision was not to put a lot of effort into this at one point, but let it evolve over time.
James and Catherine have been working on fixing up models with no reaction elements. Many of them actually have full equations in the CellML model, and Catherine is recoding many of them from scratch, because the mathematics in the reactions was not a good enough representation of the paper or the biology to make it feasible to automatically convert the reactions into non-reaction based equations.
James has completed 4 models, and Catherine has completed 8. Catherine has also produced a new endocrine model, as an initial run-through for Peter's endocrine project.
Andrew suggested that to submit the new models into the repository, she uses the docbook in the existing models as a template, and then just uploads the images separately to the Plone site. He noted that source figures (e.g. the Xfig sources) would also be worth putting up, because they are a valuable resource for generating things like SVG.
There is still a problem with papers which have multiple models. Matt's originally proposed naming system allowed for models to be designated A, B, C, to represent this, but the repository design used at present doesn't allow for an easy way to support this. Tommy is looking into it.
Tommy has been working on the metadata editor. He has also been in correspondence with Mike about getting site tools up.
Andrew asked what the process should be for requesting that Mike look at problems on the site. Tommy will send Andrew Mike's e-mail. Andrew suggested that perhaps Mike could attend CellML meetings. Poul thinks it would be good if he could attend CellML meetings, but at least is keen for him to meet with Tommy, and perhaps Randall, at which point issues at the CellML meeting could be passed on to him.