<> = 2008 = == December == === Thurs, Dec 18 === participants: Allan, Daniel, Mike, Noel, (others?) agenda: events, 2.6.2, 3.0, plan schedule for holidays, events: (Allan's summary of 'working group's ) * with Mike Creech and Brian; will check in code * need feedback and examples on batching, since readers don't actually fire many events in 2.6's implementation * mantis bug no. 1941 is an example (http://cbio.mskcc.org/cytoscape/bugs/view.php?id=1941) -- but that can be considered an effect of bad programming style. * bulk methods (for example, .addNodes() to complement .addNode()) might solve issues without event batching 2.6.2: * Mike found one bug in edge handles in layout, not yet fixed * in general, looks ok, so release is on track for early january schedule: * next conf. call on january 8th * target for 2.6.2 release: jan. 9th 3.0: * Mike will hack on application, removing GraphView and substituing CyNetworkView, to find out what parts use presentation-level parts of GraphView and what parts only viewmodel-part. * Daniel will keep working on the refactored-viewmodel branch (svn+ssh://grenache.ucsd.edu/cellar/common/svn/cytoscape3/branches/abeld-gsoc/dev/refactored-viewmodel), implementing demo presentation and experiment with using annotations for VisualProperties === Thurs, Dec 11 === conf. call notes for dec. 11 participants: Mike, Daniel, Scooter, Sarah, Allan, Piet (and maybe others) agenda: 2.6.2, ISMB, ISMB: attendance of Allan, Scooter, Piet depends on paper submission next meeting more info 2.6.2: * java webhelp on mac: only help doesn't work, error dialog shown as work-around * multiple networks in session, groups are global and it ends up in every xgmml file, and when loading, edges appear that were not there. * an ugly workaround is available (and checked in), which is good enough for 2.6.2 * Alex volunteered for making the release. * code freeeze tomorrow (dec. 12) * plugins tagged 2.6.1 won't automatically work with 2.6.2: will have to check compatibility and mark it for each. === Thurs, Dec 4 === conf call notes for dec. 4 participants: Sarah, Alex, Daniel, Alan, Kei, Noel, Gary, Brian agenda: 2.6.2, events, ISMB 2009, 2.6.2: * target date for release: dec. 19. codefreze: dec. 11 * one 'must fix in 2.6.2' bug: webstart problem on mac -- If proper fix not possible, will disable help menu as a workaround. * no other bugs deemed critical enough to delay 2.6.2 (if you disagree, speak up on mailling list) event handling: * Brian, Allan, Mike Creech looked at event-handling frameworks (eclipse, whiteboard, etc.) * see http://www.cytoscape.org/cgi-bin/moin.cgi/Cytoscape_3.0/EventHandling ISMB 2009: * this year in Sweden (june 27-july 2, Stockholm), should cytoscape have presence there? * Due date for paper: jan. 7. * where will cytoscape 3.0 be in january or june? == November == === Thurs, Nov 20 === participants: Mike, Sarah, Noel, Daniel, Gary, Scooter, Alan, Piet, Trey, Mathieu agenda: 2.6.2, webstart bug, mini-retreat, (vizmap api) 2.6.2: * some bugs fixed, so release warranted; outstanding bugs to fix: whoever the bug is assigned to decides whether fix needed for 2.6.2 * target date for 2.6.2. beta: 1. dec. with 2.6.2 release aimed at Christmas No conference call next week (Thanksgiving), Mike won't be able to attend week after that (but there will be a conf. call then). vizmap api: * MappingCalculator, VisualStyle, like in model2 branch (http://chianti.ucsd.edu/svn/csplugins/trunk/ucsd/mes/api/src/main/java/org/cytoscape/view/model2/vizmap) * many-attributes-to-one-VisualProperty mapping: only MappingCalculator has to change, change straight-forward 3.0: * moving forward: merge new model (in core3 branch) with pluggable-renderers; but not yet: * Mike will merge new-model branch to 3.0 trunk, do some work on events on that branch, and merge pluggable-renderers later * Scooter will start provenance discussion on wiki to summarize usecases (and only usecases/requirements) * event batching: as not critical right now, (can implement viewmodel without it) === Thurs, Nov 13 === conf call notes nov. 13 participants: Scooter, Sarah, Daniel, AllanK, Noel, (and some others) agenda: mini-retreat, "undefined column CyDataTable issue" mini-retreat: * official starting time: 8:45 in the morning * teleconferencing: will setup skype call. (If you want to connect via skype, send skype loginname / account info to cyto-staff list!) 9:00 San Francisco local time will mean 18:00 in Europe. * getting from airport to Queen Anne: use SuperShuttle or, if more people, cab might be better. * some ideas for discussion are listed on http://www.cytoscape.org/cgi-bin/moin.cgi/Cytoscape_3.0/MiniRetreatTwo/DanielsQuestions -- in any case, make sure to check updates to http://www.cytoscape.org/cgi-bin/moin.cgi/Cytoscape_3.0/MiniRetreatTwo (or, in general, http://www.cytoscape.org/cgi-bin/moin.cgi/RecentChanges) "undefined column CyDataTable issue": * when setting value for an undefined column: silently create or exception * problem arises because row and column are different in 3.0 core3; and generic types are used in 3.0 * other properties that need to be set when creating column: 'unique' flag, * use java annotations to be able to declare what columns a given code uses. This would complement, not replace the .addColumn() method. * when making a database-backed CyDataTable, forcing columns to be declared first would make it more efficient and nicer. * consensus: CyDataTable will use database semantics, and columns have to be created beforehand. * more complex (higher order operations) operations on CyDataTable: joins, etc. were not discussed yet. Talk about that during Saturday at mini-retreat? === Thurs, Nov 6 === 2008.nov.6 conference call notes participants: Alex, Maital, Piet, Allan, Gary, Scooter, Daniel, Noel, Mike, Trey, Brian agenda: mini-retreat, mini-retreat: * book rooms! -- Queen Anne suggested for group discounts, reference UCSF RBVI (?) * video-conferencing: although UCSF has videoconferencing system most likely won't have anyone on other end (or does anyone have access to videoconferencing who isn't attending?), so most likely skype will be used. * agenda for mini-retreat: first viewmodel, then rendering later * will have conference call next week as usual NCRR grant application possibility: * Trey suggested submitting an application (a pre-proposal) for a "National Center for Research Resources" grant. Metanodes: * Scooter made newer version of metanodes plugin which aggregates attributes when metanode is created -- go test it! 2.6.2: * maybe release before christmas, since Scooter's metanodes plugins uses some fixes and some other bugs fixed that should be released. * many unassigned bugs in bugtracker -- those should be triaged and possibly fixed 3.0: * Mike did new model layer merge (networks load and are rendered): cytoscape3/branches/mikes_new_model_branch * Cytoscape.java a big global variable, which needs to be fixed: Cytoscape.getCurrentNetwork() and similar stuff * maven repository set up with svn repository: bundles there will be totally independent, so to work on one checking out everything else won't be needed. developer awareness tool: * Jeremy Handcock will be reminded to send mail to cytoscape-staff list. == October == {{{ === Thurs, Oct 23 === participants: Daniel, Sarah, Noel, Mike, Alex, Scooter, Gary (I think I missed some people -- Daniel) agenda: Matlab, mini-retreat, metanodes mini-retreat: * additional hotel recommendations added to wiki (http://cytoscape.org/cgi-bin/moin.cgi/Cytoscape_3.0/MiniRetreatTwo) (Queen Anne most recommended, but bit out-of-way thus maybe organize shuttles if everyone stays there) * mini-retreat agenda: added some items to wiki-page * Daniel will refresh writeup about pluggable renderers and link it from mini-retreat page Matlab: * make Matlab script engine? * large body of Matlab code that people would want to use * possibilities: * javabuilder http://www.mathworks.com/products/javabuilder/ (although that would not be a fully solution) * connecting octave to cytoscape instead (with octave being somewhat compatible with matlab) might work, but since octave is not used much by the target audience (who have the existing Matlab codebase) most likely octave is too incompatible with Matlab. links about octave-matlab (in)compatibility: http://www.gnu.org/software/octave/FAQ.html#MATLAB-compatibility http://users.powernet.co.uk/kienzle/octave/matcompat/ http://users.powernet.co.uk/kienzle/octave/matcompat/HTML/summary.html * maybe use Matlab distributed computing server (If the API of that makes it easier to interface to Matlab) Metanodes: * Scooter tried writing testcases from current usecases issues: (list of questions in post to cy-staff) * what should a plugin see? (when some metanodes are collapsed, some are not): what the user sees, or all metanodes? * 'current network' no longer exsist, since if it contains metanodes, it is not unambigous: expandedness of metanode is a view-related property Software development paper invite from Greg Wilson (Gary) Deadline is April 2009. Still interested? - delay until next week === Thurs, Oct 16 === participants: Benno, Mattew, Mike, Kei, Peng, Piet, Sarah, Noel, Alex, Allan Notes: Sarah == New core developers == * Piet has been nominated as and been made a core developer * Benno has a new core developer as well Mattew (spelled that wrong) Please see mini-retreat 2 page on wiki. Extreme echo on the call line, if anyone really needs to discuss please send an email. }}} {{{ === Thursday, October 9, 2008 === 2008. Oct. 9. Conference call notes participants: Daniel, Brian, Scooter, Sarah, Noel, Mike, Alex, AllanK, Gary, Piet(had to leave earlier) agenda: confirm mini-retreat scheduling, 3.0: subnetworks, book ==== mini-retreat ==== * final decision: nov. 14-15 (Fri-Sat) at UCSF * Alex will create [:Cytoscape 3.0/MiniRetreatTwo] page on wiki, with hotel information * Arrival on Thursday evening, so full Friday can be scheduled * possibility for teleconferencing via skype, etc: didn't really work last time * UCSF has video-conferencing room (uses standard polycon, so a video-conferencing room on other end is needed; will post requirements to cyto-staff) * want to focus on viewmodel and view ==== calls about OSGi book ==== * Manning (competitor of O'Reilly), AllanK, Mike and Kei talked with them about a book on Cytoscape (and general network visualization) * arguments for and against writing book didn't really change since discussion in Toronto * Maybe book now, between 2.6 and 3.0 is too early * consensus: too early now, tell Manning that it should be revisited after 3.0 ==== 3.0 ==== * Mike has been writing implementation of model layer, subnetworks, etc. * need unittests, to tests API, too (i.e. for "typical usage") * Scooter will write unittests from some of the usecases * CyNetwork interface is considered finished, Scooter will add javadocs * model layer is (apart from subnetworks) pretty much finished. (events, CyNetwork, etc.) * events: list of viewmodel (and view) related events is not defined yet, but for model-related events is done. * CyDataTable: having different backends was discussed, but the mechanism for getting them was not yet considered. It depends on where CyDataTable objects came from (injected by spring, etc.) * big question: is CyDataTable API complete enough or will it be problematic that other backends will be limited by it? (like having an SQL-backed implementation without join-capability in CyDataTable) * one possibility is to push the issue to plugins: the plugin that brings the SQL-backed implementation could expose a richer API. This means that those plugins that use that richer API will depend on that plugin * or add richer metadata to CyDataTable to make it possible to query capabilities Greg Wilson invited us to contribute to a special issue of "Computing in Science and Engineering" next year on the theme "Software Engineering and Scientific Computing". Some general interest from Mike, Sarah, Scooter and Allan. So Gary will get the relevant deadlines from Greg and if compatible with people's schedules, say yes. }}} {{{ === Thursday, October 2, 2008 === participants: Daniel, Noel, Mike, Sarah, Piet, AllanK (sorry, I came late), Gary agenda: OSGi book, 3.0, mini-retreat ==== 3.0 ==== * performance tests work for Mike, not for Sarah (runs out of memory when running all tests, but tests individually work) * will split performance tests into seperate project, so that they won't run at every build * will set cruisecontrol to run after every checkin (and at least nightly) so that performance-tests will be run after every checkin; and to run performance tests after API bundle changes ==== mini-retreat ==== * nov. in Seattle would be bad for AllanK, * possibly move mini-retreat to San Francisco, or have travel funds from cytoscape consortium ==== OSGi book ==== * as mentioned on cyto-staff * skype conference tomorrow at 10, ask for invitation from Kei ==== Gary's group working on a Cytoscape-lite ==== * Ajax, web-based * required for GeneMania project }}} == September == {{{ === Thursday, September 25, 2008 === present: Daniel, Mike, Scooter, Noel, Sarah, Allan ==== Cytoscape 2.6.1 ==== * Only reported issue is OpenJDK issue * Need to change the toolkit from MToolkit to XToolkit in the installer: * {{{ sed "s/MToolkit/XToolkit/" Cytoscape_2_6_1_unix.sh > Cytoscape.sh }}} * Perhaps we should start testing with OpenJDK * should note on website the list of officially supported (tested) jdk vendors, versions * Also have property saving bug reported * Plans for 2.6.2? * No specific plans * Will evaluate as needed ==== Cytoscape 3.0 ==== * Noel raised an issue about API symmetry between the network writer and reader: CyWriter takes File, not OutputStream -- should it take OutputStream as well? * Scooter: having subnetworks, without metanodes might be needed, usecase, for example: two network with different edges but same nodes (this wouldn't be the same as pre-3.0 design, because using single RootGraph is not mandated) -> decision: implement current API plans, and see whether it is featurefull enough * Mike continues to work on 3.0 API * Sarah has added some unit tests * Focus should currently be on basic performance tests * In the future, might be useful to exercise multi-threading * Mike wanted to get people's opinion on checkstyle * General consensus: useful, but only if it doesn't get in the way of checkins, which means using a pretty minimal rule-set * Noel suggested maybe having 2 different rule-sets, one for general building and one for generating a report * early prototype might be possible by early 2009 ==== mini-retreat dates ==== * Target: November in Seattle * Allan is waiting to hear back from some people at ISB to piggy-back some meetings * Tentative dates: November 14th & 15th * On a related note: Barb would like to attend a mini-retreat to talk about usability issues * The feeling is that the San Francisco retreat (the one after Seattle) is the right target for this discussion }}} {{{ Thursday, September 18, 2008 present: ? Notes: DanielAbel (but feel free to edit -- I hope I dont mis-represent anyones opinions) == cy 2.6.1 == - considered finished, javadoc-generation problem fixed (shouldn't cause problems next release) == mini-retreat == - some discussion on CyDataTable etc. (i.e. mini-retreat results) conclusion: api not yet described well enough, no consensus yet on what naming to use. (like RootGraph or ParentGraph ) API should be easy to to understand, although metanodes part might be inherently complicated with no way to make it simple - further development: two alternatives were discussed: -- merge api plan in current cy3.0 (more of a refactor) pro: less chance of re-inventing old bugs and of total rewrite; how api changes affect plugin developers would be more understood. cons: lots of code to fix, would destabilize code when lower layers are refactored; would have to re-fix the same code several times (first when model is fixed, then when viewmodel, when view, etc.) -- create seperate branch, implement model layer there, then pull code over from cy3.0 (this would be closer to rewrite) pro: it would be easier to enforce the cleanness of the new branch cons: too much incentive to rewrite too much from scratch -- opinions differed on severity on pro/cons for each (like how much re-fixing first alternative would mean, how much bug regression would be in second). Someone commented that either way it would end up to be a circular process, with code moving both ways. - organization of project layout: -- currently 3.0 branch: one maven project with many modules (easy to compile everything, but a bit less modular) -- Mike will break it apart into separate projects - HyperEdges / HyperGraph -- current consensus: hypergraph semantics should be stored as Attributes (like "this is a connectornode, so don't count it as real node") -- no competitor "hypergraph system" identified (although many wanted to have one to steal ideas from) - mini-retreat considered success, next mini-retreat plan: mid-october, early november (before US thanksgiving) maybe in Seattle (to coincide with some other meeting?) - 'private' maven repository for cytoscape: will be one in future, but not yet in short term (although a public one might be used instead of a cytoscape-specific one) }}} == July == {{{ Thursday, July 31, 2008 AlexP, AlexA, Sarah, Brian, Noel, Gary, Mike, Kei, Peng, Trey, Allan Notes: Alex == Post Retreat Questions == - Keep an eye out for retreat wiki notes and materials: feel free to add or edit - Next year: University of Michigan by David States == 2.6.1 == - Scooter's last commits? - Mike will update release bundle instructions - Anyone volunteering to handle release while Mike is OOT == 3.0 == - Kei is working on clean build in Eclipse - Everyone should test and provide feedback - Follow up on Daniel's work - Mike will start it and then distribute tasks - First CyWest Meeting in San Diego (beach house!) Sept 12-13 }}} == June == {{{ Thursday, June 19, 2008 Sarah, Maital, Noel, Brian, Scooter, Mike, Kei, Peng, Allan, Gary Notes: Alex == 2.6.1 == - Scooter is going to squeeze in a couple key features for tunables and then give Mike the green light when ready to package and ship. == 3.0 == - Everyone should have reviewed the view portion of the API: http://chianti.ucsd.edu/svn/csplugins/trunk/ucsd/mes/api/src/main/java/org/cytoscape/view/ - Terminology: ViewModel and Presentation - X,Y,Z,T in VisualProperty may not be ideal for continuous updating, e.g., in animated layout; may lead to performance hit. Keep for now. - Should the VisualProperty enum implement an interface to "keep the door open" for plugin developers, e.g., to add their own Node_Shape? -- Does this control belong in the ViewModel or Presentation? ViewModel, if you expect to store your Node_Shape. -- The advantage of using an enum is the ability to simply get at all possible values. -- Alternatively, we could use an osgi registry to support a new Node_Shape. -- Or, add User_Defined in the enum. This allows it to be serialized and keeps it simple. This mechanism could use Attributes to support multiple shapes. -- Of course, if the plugin is not loaded when the file is reopened, then Node_Shape would default back to "circle" or something. -- Ideally, we can "leave the door open" with having to "have it open", meaning keeping for 99% of the cases. This could potentially be done by extending enum. -- Within the "1%" we'll want to make sure we are taking care of power users/developers who need custom shapes e.g., mims group? -- Perhaps we need to think top-down here, to balance the design and consider the Presentation concerns/needs. -- Important to distinguish _custom shapes_ (size and color controlled by vizMapper) and _custom graphics_ (a *substitute* node that has a fixed shape, size, color, but behaves fully as a node, rather than as a sticker on top of a node). There are use cases that blur these, however, and thus suggest having a single system that can have overrides per parameter. E.g., a custom graphic like a photo that you still want to have vizMapper controlling size; or a donut node with vizMapper-controled node in the center and a custom color on the outer rim. -- Fill color and opacity -or- Paint? Breakdown -or- lump? - Next week we will continue the discussion of ViewModel and Attributes == Retreat == - Detailed schedule is up on wiki. Please sign-up to lead sessions with Gary. - http://cytoscape.org/cgi-bin/moin.cgi/CytoscapeRetreat2008 }}} == May == {{{ Thursday June 12. 2008 Allan, Peng, Kei, Mike, Sarah, Noel, Brian, Kristina, Trey, Gary, Scooter Notes: Sarah == Retreat Update == - Some issues with sign up/paying via credit card - Big dinner Thursday evening - Need to know who is interested in a dinner for the cyto staff group on Sat. night? - Let Gary know about specific requests or presenting on application showcase day. - Funding, Trey and Gary working it out == 2.6.1 == - Bugs have been assigned, everyone needs to triage and get back to Mike ASAP - Mike will put together release and get it out to group to test == Mike 3.0 API == - Everyone needs to look through the API emailed - Some concern about the distinction between global/local attributes - Need to make sure we can still use them through things like VizMapper etc - There is some concern about the memory management in regards to the attributes - Open issues: - Are we happy with Attributes? - Are we punting groups, splitting the concept, etc? - Ignore groups in the model for now. - Keep the use case of metanodes/named-selection for later and if we need it we can iterate back over the model API later - Start thinking view model, read the 3.0 discussion. Iterate over based on interface discussed for graph model/attributes }}} {{{ Thursday May 22, 2008 Notes: Alex Alex, Brian, Sarah, Mike, Peng, Kei, David, Scooter, Allan == Maven Builds == * Brian raised some issues with bundling during build * Mike will continue to assess build/package options beyond Felix; we can work around current issues in the meantime == Model Issues == http://www.cytoscape.org/cgi-bin/moin.cgi/Cytoscape_3.0/ModelDiscussions === Event Model === * Here, "events" refer to events that travel up the hierarchy, e.g., node added to graph; events communicated between layers * Our modules form a directed acyclic graph and event listening should "listen up the graph" - for example, the View module should listen for events in CyNetwork (CyNetwork is higher up in the hierarchy) - VizMapper changes CyNetwork; CyNetwork fires event; View detects event and controls layout, etc. - This is "event driven" and more consistent with MVC * Synchronous or Asynchronous? Currently, we have synchronous, but asynchronous would clearly be advantageous for clean operation and external application integration * Two approaches: (1) an Event Type Model, (2) an Inheritance Model (see advantages and disadvantages on wiki page) * Mike is currently leaning towards Inheritance; consensus agreement. * Single broad listener or multiple type-specific listener? Both? Or, a "specific" listeners can be more general? * User interface events will be handled within layers and then rely on the event hierarchy to pass between layers * OSGi advocates a "whiteboard model": register yourself as a listening service; handler checks registry * Network Loading use case: "add node" events should be ignored until done - listen for "start load", ignore "add node" for *that* network, until "end load"... this would require processing each "add node" to determine whether it can be ignored - alternatively, wait to register for "add node" events until *after* loading is complete * We should investigate additional patterns, since we are not likely to be the only ones facing these issues. == Next Week == * Gary's attributes use cases }}} {{{ Thursday May 15, 2008 Notes: Sarah Mike, Kei, Peng, Scooter, Allan, Sarah, Brian, Maital == 3.0 Summary == * Mike put together a diagram of steps required to get to 3.0, not yet posted. Graphical view of project plan currently on the 3.0 page. * Note it will be an iterative process, not a straight line. == Build Stuff == * Issue with assembly not packaging libs the way we need? * pax tool that creates osgi packages * Brian & Mike will work out how to clean up the build == Attributes == * Mike has proposed two possible solutions * New use case: a global pool of unmapped attributes * Look back at Gary’s SUID proposal which would handle both local and global attributes. Provides a database abstraction for attributes and their mapping to networks. * Need to address various use cases at the application layer and plugin writer layer * Need the idea of foreign keys in order to establish a basic relationship that is accessible to users and plugin writers * Make sure we handle user mapping (users won’t see the SUID’s) where nodes may have duplicate “names” (which are not necessarily guaranteed to be unique), this may be a tool by tool basis? * Gary has captured use cases and will add these. * Everyone ADD COMMENTS! }}} {{{ Thursday May 8, 2008 Gary, Allan, Scooter, Brian, Noel, Alex, Peng, Kei, Mike Notes: Alex == Cytoscape: The Book == * Wiki page established: http://www.cytoscape.org/cgi-bin/moin.cgi/Cytoscape_Book * Lots of information there * Review the outline and modify to get a solid plan together == Open Helix == * CEO emailed Mike asking if it was legal for them to use Cytoscape in their training material * Apparently they have been getting requests for Cytoscape (and GenMAPP) training! * Folks pay them to put together free tutorials == The Merge == * Done! * Remember, if you have a bug fix, you have to make it in both 2.6 and 3.0 * Maven: we still have a few things to sort out (e.g., Felix) and we want to keep it simple for the plugin developer == 3.0 Development == * Updated wiki page on model discussion: http://www.cytoscape.org/cgi-bin/moin.cgi/Cytoscape_3.0/ModelDiscussions * Review new content regarding network model and model API proposals * There are remaining open issues that need to be addressed (preferably, one at a time): - fundamental issue: node, edge and network relationship/coupling - need for speed/performance, need for simplicity for plugin developers - alternative proposal illustrates the type of simplicity a plugin writer needs, not necessarily a specific impl - degree of coupling at various "layers" of the hierarchy is at issue - some simple examples like "traversing" or "getting node name" need to be very easy - how should attributes be modeled and accessed - separate core and plugin APIs? - the concern for impl and javadoc view are separable - the javadoc necessarily follows the impl, but we should all have the same javadoc goal in mind before settling the impl plan - need for overview vision of modules to settle higher-level issues and contextualize impl work, including * attribute * groups * hyperedges }}} == April == {{{ Thursday April 24, 2008 Allan, Gary, Alex, Mike, Peng, Scooter, Sarah Notes: Allan == Presentations, Book chapter == * invitation for a Cytoscape book chapter, picking up on Nature Protocols work - Humana Press, book on systems biology of yeast - analysis of gene expression with Cytoscape -- can mention SGD as source of yeast - leverage from Nature Protocols package - draft due March 2009 - Allan and Scooter may co-author? * invitation for tutorials - with BigCat group at Maastricht , all travel expenses paid, June 23-27th, actual talk is June 26th - European Molecular Biology Organization: 26-30th September, 2 hour lecture, assume travel expenses paid but doesn't say in email, Mike will look into this. == GSoC 2008 == * between now and May 21st is community bonding period, get people prepared * mentors should let Mike know when to add students to Cytostaff * should they code to 2.6 or 3.0? - probably 2.6 for people building plugins, possibly 3.0 for people working more closely in areas related to 3.0 == Retreat Update == * have local organizing committee * still raising money * do we need a registration deadline? - currently booked room is for 100, which should be sufficient, but may want larger room for showcase * Wednesday (possibly Tuesday) for hackathon, Thursday application showcase, Friday tutorials / decision making, Saturday decision making * possible conflicting sessions with ISMB saturday morning tutorials, BioPathways meeting == 3.0 == * merging: we need to fork off 3.0 version for future development - merge of 2.6 with 3.0 is organizational, not re-coding. - should we merge before we first re-start the discussions? - merge will be week of April 28th - May 2nd - Mike will make assignments and send email - look at Mike's email 1-4 for instructions for merging. == 2.6.1 Release == * what bugs need to be fixed? * critical bugs should be fixed, but non-critical features should be added to 3.0 * get 2.6.1 out in the next couple of weeks * Scooter and Sarah working on a logging prototype for 2.6.1, this can be post-merge * bug fixes are post-merge }}} {{{ Thursday April 17, 2008 Maital, Sarah, Alex, Peng, Kei, Allan, Kristina Notes: Alex == 2.6 Release == * It's out the door! * Image export on the mac still has problems == GSoC 2008 == * We have the final 9! * Offer travel scholarship to retreat to Pekka, contact Tero * Start working with students on the 21st == Retreat Update == * http://www.cytoscape.org/cgi-bin/moin.cgi/CytoscapeRetreat2008 * Any feedback on dates, etc? * Organizing committee is forming: send feedback * Online registration will be setup soon * Let us know if/when conflict with ISMB events arise, there is still flexibility in the schedule == 3.0 Issues == * Expect email on how to merge * Image of project plan now on wiki * Summary of graph/attribute model email thread is still needed - This needs to be reconstructed/organized on the wiki - Gary will start a page for this and folks can recontribute their points where appropriate - Look for it here: http://www.cytoscape.org/cgi-bin/moin.cgi/Cytoscape_3.0#head-bfe4e806702979251d6c21d9e5e82fe51cf08730 }}} {{{ Thursday April 10, 2008 Sarah, Mike, Kei, Peng, Brian, Noel, Gary, Kristina, Alex Notes: Sarah == 2.6 Release == * Trunk of current Cytoscape is going to be the 2.6 "branch" * Few people have responded with some small bugs * Code has been checked into the trunk that will need to be tested if the release is rebuilt now * Will release now, 2.6.1 will come out shortly == 3.0 == * Look at plan on the 3.0 page * Note any fixes made in the 2.x branch will need to be fixed in the 3.0 branch, there will be no merges after the post-2.6 release merge * Encourage everyone to check out the 3.0 branch and get comfortable with maven * Break out the merge by package and everyone takes a piece. Mike will send an email about how to do the diff/merge and assign pieces. * Sarah will link image of project plan on wiki (and keep it updated) * Everyone else will take a look at the plan and give feedback or just be aware == Google Summer of Code == * This week look at projects and request # of students * Need to be sure we have enough mentors for all the projects/students * Currently 7 mentors have signed up, need to know how many are definite by tomorrow (Friday) * Could use these projects as test cases rather than projects that will certainly be used * Need to think carefully about the projects proposed in case they are not useful for moving towards 3.0 * Also need to talk to various groups that are have already done some of this (see NetworkAnalyzer) * Consider involving students who are interested outside the google program as well }}} {{{Thursday April 3, 2008 Sarah, Mike, Kei, Peng, Maital, Gary, Alex Notes: Alex == 2.6 Release == * Any issues? * Rework help docs in future == 2.6.1 Release == * Changes should go to 2.6.x branch * For the time being, don't bother with trunk since there is no 2.7 planned (may change...) == 3.0 Planning Overview == * Separate repository due to major directory changes and build process * Deprecations cleaned out, initial attempt at modularization, moved to Maven (i/o Ant) - Initial organization to help see the potential modules and the major issues * Strategy: start with current code and build in new architecture - treat each module independently: impl, test, then continue - the ultimate test for each module is that Cytoscape still works, in addition to functional and architectural goals * The refactoring strategy will be outlined on the 3.0 wiki page, including schedule, to encourage as much input as possible while also ensuring that things continue to move along == Google Summer of Code == * Mentors can now assign themselves to applications they *might* want to mentor (go ahead and sign up for more than 1 at this point) * This action will begin to rank the applications (mentored > non-mentored). * By April 8th, I want to see all applications ranked like this, then we can begin a round of scoring * In the end, we will assign multiple mentors per student * Mentors can work with multiple students, but can only be the primary mentor on a single project. }}} == March == {{{ Thursday March 27, 2008 Sarah, Scooter, Alex, Maital, Mike, Ben, Kei, Peng, Trey Notes: Ben == 2.6 Release == * A few outstanding bugs (see mantis): - inability to import vizmap.props - image export (pdf export from freehep is broken) - node size width * Scooter to make some user manual changes on Monday. * Put release files/candidate together early next week for internal testing * "Gold" release build files by next Thurs == 2.6.1 Release == * Scooter to cleanup UI and "hammer on" group node plugin with the goal of it becoming a core plugin for the release }}} == February == {{{ Thursday February 28, 2008 Mike, Trey, Scooter, Sarah, Alex, Gary, Allan, Noel, Maital, Alex Notes: Alex == Google Summer of Code 2008 == * Ideas Page looks good: http://conklinwolf.ucsf.edu/genmappwiki/Google_Summer_of_Code_2008 * Keep adding ideas * Sign up as mentors * Will apply on Mon or Tue * Main focus now should be on attracting students to apply for our projects == Bug Day Wrap-up == * Lots of bugs: good and bad * $5 per person per bug; no cap * Will talk to Louis about execution == Error Dialog == * Scooter: managing session errors illuminated utility of a general ErrorDialog service * Error list you can append multiple errors * Task dialog currently only handles errors singly (an not very well) * Could improve on what already exists * But, error dialog is more user friendly * Can multiple tasks call addError to this dialog? * 2.6? Concerns about quick and dirty solution for long-term issues. On the other hand, slate will be cleaned for 3.0 anyways == Case Sensitivity == * Allan: identifier matching are case-sensitive * Defer to 3.0 to find stable solution to broad problem == CytoRetreat 2008 == * Toronto: July 15-19: http://www.cytoscape.org/cgi-bin/moin.cgi/CytoRetreat2008Planning * Give feedback on schedule * Will open registration soon * Need to determine fee for symposium ($50 or $75, students and presenters free) == Everyone should be fixing bugs == * If a bug is a suggestion for a change in behaviour, you have the right to reject on principle. Just add note to bug, then close. * Cancellation of network and attribute download: leaves unwanted data behind * Will implement if relatively painless. Real fix will come in 3.0 }}} {{{ Thursday February 21, 2008 Trey, Mike, Scooter, Sarah, Alex, Gary, Allan, Noel Notes: Sarah == Category Names for Plugins == * Mike at Agilent has the VistaClara plugin that doesn’t fit in any current category * Should we consider adding categories as requested or wait until we see more than one plugin that fits and then create a category? * Resolution: leave categories as they are, in 3.0 consider doing “tags” instead == Plugins Page == * Bug causing plugins to show up in lists multiple times (per each version) * Google for cytoscape plugins takes us to the old plugins page. Alex will add link to the old pages back to the new plugin page. * Eventually move the plugins database to the cytoscape.org server == Bug Day Friday == * Thanks Alex! * Add info to the header for bug day, link to the beta release, bug tracker and the wiki page for testing. * Still working out how much and how we give out gift certificates. * Sign up on Skype to let people contact you for questions == 2.6 Status == * Bug fixing * Check any bugs you’ve reported, if something needs feedback please try to provide it == Report == * Trey had discussion with several NIH managers * Recommended NCBC (centers) grants, submission date isn’t until Sept 2009 * Talked to Dan Galahan (NCI) and Genie Larken may be willing to permit a special grant application * Plenty of groups are interested so should be able to get soemthing == 3.0? == * Wait until next week or until 2.6 is taken care of for further discussion }}} {{{ Thursday, February 14, 2008 Mike, Gary, Scooter, Allan, Noel, Gary, Sarah, Alex, Maital, Kristina, Brian Notes: Ben = Cytoscape 2.6 / Testing Day = * Must go out before testing day. + It's ok to get it out there with bugs as long as known bugs are announced + Concern about plugins not working with release, so we say beta is targeted toward plugin developers + Beta release 1 on 2/15. Beta release on 2/22 * Prize: Top 50 bug finders get $5 Amazon gift card - leftovers to be decided * Bugs entered in mantis * Use MeBeam and basic chatroom to facilitate communication during testing day = Bugs = * Based on feedback, Mike has been moving bugs to postponed project & reassigning them * Hidden Cytoscape directory bug + Cytoscape dir gets corrupt if plugin has problem loading and cannot be started + Cytoscape needs to be able to startup regardless of what is in .cytoscape, check robustness of plugin manager + During testing day, make bogus track_plugins.xml + Possible problem with corrupt vizmapper.props file * Scooter made update to Task Manager to make it cancelable up to a certain point (before data structures start to get created) = Cytoscape 3.0 = * 3.0 discussions to be postponed until after 2.6 release and testing day }}} {{{ Thursday, February 7, 2008 Mike, Noel, Gary, Sarah, Alex, Maital, Kei, Allan Notes: Gary Stanford Medical School (Allan) – Agilent collaboration using LitSearch. Needs ~1800 attributes, which is slow. Wondering if anyone else has dealt with this many attributes before? It doesn’t seem so. Open questions: Could Gaggle manage this and just import the attributes you want? There is also the old data matrix code, but that . This is a use case for 3.0. Cytoscape 2.5.2 – no one has any outstanding bugs. Mike will publish this today. Cytoscape 2.6 – do we need a contest for fixing the most bugs? People don’t think they need motivation, but rather help with time management. We need to find a way to motivate people outside of the core team to do testing. Alex suggests scheduling specific time to get together virtually to fix bugs e.g. open up an IRC chat. People feel we don’t need a prize. Testing could involve undergrads, but they may not be able to test well – they need some training on Cytoscape first. We should do this early to give us enough time for bug fixing. Also, would be nice if all participants could get a prize, not just one person. Decision: Alex will organize a virtual testing session, likely on Fri.Feb.22. Status: we are feature complete, just need to test and fix bugs Cytoscape 3.0 – Mike updated the 3.0 site. Please add your use cases. Mike has been working on his own branch to clean up code and remove deprecation warnings, mostly related to an external library. Did things like remove references to the rootgraph. Mike will keep experimenting on the branch, simplify visual styles. Noticed that current GUI is very spread out, so would be interesting to see how it could be moved into one code area. Merge model layer with CyNetwork, then extract it into an OSGI bundle. Planning the next stages of 3.0. Mike will create a plan for the model layer on the wiki to focus discussions. Regroup around the model discussion and finish that up Attributes Identifiers Deadlines: schedule a call to discuss each proposal to ensure people all review the proposal by a given date. Open issues that are dependent on other layers can be identified and returned to later. }}} == January == {{{ Thursday Jan. 31, 2008 Mike, Sarah, Allan, Gary, Noel Notes: Scooter = ISMB Tutorial = * Tutorial was not accepted * We should definitely propose a Demo ($400 for a 20 minute session) * Will also propose a BOF session covering "Network and Pathway Visualization and Analysis" = 2.5.2 Release = * Release is available for testing * Will hold announcement pending backport of Vizmap class cast exception bug from 2.6 * Goal is to release on Monday = 2.6 = * In bug fixing mode * Goal for 2.6 is to make it a very solid release * Beta release (to allow plugin developers to port) + Sarah wants to do some fixes to the plugin manager first + Allan has some work to do on the workflow manager and themes integration + Goal is a beta in 2 weeks + Mike will do some bug triaging in the interim = 3.0 = * Still waiting for use cases * What do we want our use cases to cover? + Want to put functionality in context + Want to have rationale for new functionality or things that are currently hard + Want to record reasons for design decisions + Put use cases up for areas of controversy or difficulty * Update Cytoscape 3.0 page on wiki + Specifically update new features list, 3.0 goals, links to RFCs + Include task/status table * Mike wants to look at Attribute interface + Look at Samad's proposal + Sarah will send along Hector's proposal. * Discussion points/shopping lists + HyperGraphs/HyperEdges -- are they worth it? + Break down by layers + Big issue: events + "Glue things" - Events - How layers interact with each other (e.g. Application and View) o Mike's idea: # Application is at top of heap. Application depends on everything else, but nothing else depends on the application # Two examples - JActiveModules: pretty easy to see how it might work - Cytoscape Editor: more difficult to see how it might interact # Command layer is important in this context also = Other Discussion = * Mike has been trying to remove all deprecation warnings + Definitely problematic given the current structure + Has identified areas where there are opportunities to clean up current code + Got deprecation warnings down to around 110 }}} {{{ Thursday Jan. 24, 2008 Peng, Sarah, Allan, Gary, Noel, Brian, Maital, Ben Notes: Sarah = 2.5.2 = * Allan has been using it, no problems * Maital has been using it, none to report = 2.6 = * Bug fixing! * There are tasks on the wiki that need to be updated * Sarah and Allan will work to get the Workflow Theme up on the site * Pathway Commons plugin is nearly ready, may not be fully integrated with the web service API * On track for March 1st release = 3.0 = * Use cases still need to be filled in. * Have comments on example code ready for Mike's return = First Cytoscape Talk in Spanish = * Allan has been invited to give a talk on tropical diseases on Cytoscape and the Lit Search plugin. Would like to give a dry-run for anyone who speaks spanish. * Cytoscape & List Search being used at a University of Antioch(?) for an interactome of a specific parasite = Funding for Summer Intern = * Allan may have funding for a summer intern to work with him on a plugin of some sort * Has to be someone who does not need a visa to work in the US * If anyone has a student who would like to do an internship let Allan know }}} {{{ Thursday Jan. 17, 2008 Maital, Sarah, Allan, Kristina, Scooter, Noel, Peng, Gary Notes: Sarah 2.5.2 -Sarah/Scooter have used it some, need others to bang on it a bit 2.6 March 1st Deadline VizMapper - Getting used more heavily now, but it's got some quirks. - There appear to be a lot of bugs on the VM - Needs to be very robust before we release. Everyone needs to hammer on the VM - When reading in the cys file would be nice to not apply every style as it gets loaded - Kei is working on all VM bugs that are reported on Mantis so if you find new ones report them 3.0 - Waiting on use cases for graph model - Samad has proposed a good API for attributes - Mike's API is in his csplugins directory Retreat - Booking space - Will be planning events shortly so send ideas to Gary }}} {{{ Thursday January 10, 2008 Scooter, Mike, Kei, Gary, Alex, Sarah, Trey, Allan Notes: Scooter == SBGN == * Having a workshop in Okinawa. Should there be a Cytoscape representative? Costs are an issue for most of us. Biopax will be sending Emek Demir. Gary will ask Emek to represent Cytoscape at the meeting. == 2008 Cytoscape Retreat == * Strong preference for holding the Toronto retreat immediately before ISMB. The retreat will be July 16-19 in Toronto, Canada, which overlaps with ISMB by one day. == Cytoscape 2.5.2 == * One outstanding bug (XGMML reader/writer problem). Scooter is awaiting for feedback from user. == Cytoscape 2.6 == * Complete in terms of features, now on to bug fixing. * Outstanding "issues": Vizmap UI issues, Table import UI issues, Vizmap enhancement request * Tentative release date: March 1, 2008 * Focus on really polishing and getting 2.6 to be a quality release * We are unanimous that postponing 2.6 release is the right thing and will be recommended to the board == Cytoscape 3.0 == * CyAttributes * Need both local and global attributes * What will the interface look like? * Mike is concerned that the code underlying CyAttributes has lots of boilerplate * Generics would be a good approach * Lists and Map attribute types complicate things significantly * Do we want runtime checks or compile-time checks * Arbitrary attributes vs. type-specific * Lean towards type-specific * API needs to be extraordinarily efficient * May need to avoid keeping large lists of attributes in memory * Interesting idea, but not an API issue at this point * Complex type: * Replace with the ability to recursively define attributes with the four basic types that you could combine into lists and maps * Primary direction: use generics, if possible. * What do we put into CyNode, CyEdge, vs. CyAttribute? * Everything goes into CyAttribute, including edge type * Will require work to provide better UI for merging allowing the user to distinguish between names and edge types * Need to develop use cases for Cytoscape 3.0 }}} {{{ Thursday January 2, 2008 Scooter, Gary, Alex, Ben, Sarah, Maital, Mike, Peng, Trey, Allan, Noel Notes: Sarah == Grant Proposal == * Trey is trying to put together a P41 grant * Part 1: Development of new technology * Part 2: Collaborative activities * Part 3: Service * Need info on the collaborative research being done using Cytoscape, list the one or two most important scientific collaborations you have (coauthoring). Include some text (paragraph) about the project. == 2.6 == * Bug fixing! * Mike will assign some bugs == 2.5.2 == * Mike put out a release right before xmas * Gary’s lab has tested it, most bugs look fixed * Ben has checked in a fix to graphics export formats * Mike will put together another release after this * Need people to run through the test cases == 3.0 == * Scooter is putting together a 2 layer interface proposal * Lowest layer uses ints (Mike’s initial API) * Plugin API is all object based, allows for lazy-loading of CyNode/CyEdge objects for large graphs * Plan to discuss CyAttributes next week == Large Networks == * Mike implemented a mysql backed implementation of the API he put together, works, uses less mem etc. Got stuck on CyNode/CyEdge being held in memory }}}