Meeting Minutes 12 March 2007
Meeting Minutes 12 March 2007
Present: Poul Nielsen, Tommy Yu, James Lawson, Andrew Miller, Jonna Terkildsen, Poul Nielsen, Peter Hunter
Poul proposed that the CellML meeting time be moved to 10 am on Wednesdays, because Matt can't make it on Mondays. There was no objection, so the next CellML meeting will be held on the 21st of March.
Poul also noted that the current developments page is currently not being updated, and suggested that it be updated. Andrew noted that he had been keeping several other lists up to date, including the list of features for PCEnv and the API that we established some time ago (although the work on these lists is now mostly completed). The problem with having multiple lists is that the lists can get desynchronised if there are several places to update, and this could potentially mean that too much overhead is required to keep things up to date. Andrew suggested that we look at the model used for Mozilla development, where there is a bugtracker (and bugs are much more than just bugs in the traditional sense, they include proposed features, and so on). This means that the bug tracker will do much of the work required to produce lists, while still providing a means for us to keep more detailed comments, and maintain dependencies between bugs. Peter was keen for Andrew to show him how this works on the Mozilla Bugzilla, for the sake of comparison.
Peter noted there has been some correspondence with Andre about how we can open models in PCEnv and other software programs, and deal with the case where the software is not already installed. Andrew and Tommy noted that there is no reasonable way for us to detect whether these software programs are installed that will work across all programs (for example, (PCEnv, COR, JSim) * (Linux Firefox, Linux Opera, Linux Netscape, Linux Galeon, Linux epiphany, Linux Konquerer, Mac Firefox, Mac Opera, Mac Safari, Mac Camino, Win32 IE, Win32 Firefox, Win32 Netscape, Win32 Opera) is already 42 different configurations we would have to maintain, although we could deal with some cases in blocks as COR is less portable). Everyone at the meeting was happy with maintaining the separate links for each program, but having this unconditionally pop up a new window saying that the model was about to open, and containing installation instructions for the package.
Peter was also keen for there to be work to develop a 'layout metadata' top-level file, which refers to models in the CellML repository. There was extensive discussion about how the workflow would work for this, and the consensus was that after being created in CellML (to refer, presumably, to models on the filesystem), PCEnv would allow the user to right click on the model and bring up a dialog to change the model base URI to a http URL referring to the repository. It was also agreed that we didn't need to describe any information above what the user can set up with the mouse (so the PCEnv specific layout information will basically be the percentage of the screen that each pane takes up, and which graph is associated with each panel). Poul suggested that we could make the order of graphs in the list important, so that the first graph pane would take the first graph, the second pane the second graph, and so on.
There was also discussion about how the actual layouts available could be made more user friendly. There was consensus that there should be buttons which set the sizes to predefined values (so for example, the two graph button would arrange the splitters so one graph pane was collapsed, and the other two each took 50% of the vertical screen space). It was noted that in some point in the future, we might want horizontal panels, but this isn't a high priority so was deferred.
Poul proposed that the CellML meeting time be moved to 10 am on Wednesdays, because Matt can't make it on Mondays. There was no objection, so the next CellML meeting will be held on the 21st of March.
Poul also noted that the current developments page is currently not being updated, and suggested that it be updated. Andrew noted that he had been keeping several other lists up to date, including the list of features for PCEnv and the API that we established some time ago (although the work on these lists is now mostly completed). The problem with having multiple lists is that the lists can get desynchronised if there are several places to update, and this could potentially mean that too much overhead is required to keep things up to date. Andrew suggested that we look at the model used for Mozilla development, where there is a bugtracker (and bugs are much more than just bugs in the traditional sense, they include proposed features, and so on). This means that the bug tracker will do much of the work required to produce lists, while still providing a means for us to keep more detailed comments, and maintain dependencies between bugs. Peter was keen for Andrew to show him how this works on the Mozilla Bugzilla, for the sake of comparison.
Peter noted there has been some correspondence with Andre about how we can open models in PCEnv and other software programs, and deal with the case where the software is not already installed. Andrew and Tommy noted that there is no reasonable way for us to detect whether these software programs are installed that will work across all programs (for example, (PCEnv, COR, JSim) * (Linux Firefox, Linux Opera, Linux Netscape, Linux Galeon, Linux epiphany, Linux Konquerer, Mac Firefox, Mac Opera, Mac Safari, Mac Camino, Win32 IE, Win32 Firefox, Win32 Netscape, Win32 Opera) is already 42 different configurations we would have to maintain, although we could deal with some cases in blocks as COR is less portable). Everyone at the meeting was happy with maintaining the separate links for each program, but having this unconditionally pop up a new window saying that the model was about to open, and containing installation instructions for the package.
Peter was also keen for there to be work to develop a 'layout metadata' top-level file, which refers to models in the CellML repository. There was extensive discussion about how the workflow would work for this, and the consensus was that after being created in CellML (to refer, presumably, to models on the filesystem), PCEnv would allow the user to right click on the model and bring up a dialog to change the model base URI to a http URL referring to the repository. It was also agreed that we didn't need to describe any information above what the user can set up with the mouse (so the PCEnv specific layout information will basically be the percentage of the screen that each pane takes up, and which graph is associated with each panel). Poul suggested that we could make the order of graphs in the list important, so that the first graph pane would take the first graph, the second pane the second graph, and so on.
There was also discussion about how the actual layouts available could be made more user friendly. There was consensus that there should be buttons which set the sizes to predefined values (so for example, the two graph button would arrange the splitters so one graph pane was collapsed, and the other two each took 50% of the vertical screen space). It was noted that in some point in the future, we might want horizontal panels, but this isn't a high priority so was deferred.