| | 1 | = 18 September 2014 meeting = |
| | 2 | |
| | 3 | == Attendees == |
| | 4 | |
| | 5 | Greg, Eric, Conrad, TomG |
| | 6 | |
| | 7 | == Agenda == |
| | 8 | |
| | 9 | Presentation of Hydra architecture and migration path |
| | 10 | |
| | 11 | == Action Items == |
| | 12 | |
| | 13 | Conrad: Toolshed and Tools API |
| | 14 | |
| | 15 | Greg: Session and Application object API |
| | 16 | |
| | 17 | Eric: Continue wx ui implementation. |
| | 18 | |
| | 19 | TomG: Continue ambient occlusion implementation |
| | 20 | |
| | 21 | == Results == |
| | 22 | |
| | 23 | [http://www.cgl.ucsf.edu/chimera/data/hydra-sep2014/hydraarch.html Hydra architecture and migration path slides] |
| | 24 | |
| | 25 | TomG explained Hydra Session object that holds all application global state. |
| | 26 | |
| | 27 | Discussed the advantages of avoiding circular dependencies between Chimera 2 modules. |
| | 28 | Previously Scooter advocated this saying Cytoscape modules have a dependency graph that is a tree |
| | 29 | and communicating to a module that is not a dependent is only done by quequeing events. |
| | 30 | Eric and TomG did not understand the value of this. Requires further discussion, maybe with Scooter. |
| | 31 | |
| | 32 | Conrad suggested current Hydra Session be divided into an Application class and a Session class where |
| | 33 | the latter stores global state and the former initializes that state at start-up. This will allow code for |
| | 34 | different application start-up modes (e.g. nogui vs gui) to be cleanly separated. |
| | 35 | |
| | 36 | Discussed migration path from Hydra to a Chimera 2 release. Conrad suggests that APIs for Session, |
| | 37 | Application, ToolShed and Tools to establish the framework is an immediate need, even if such APIs |
| | 38 | are only for in-lab developers after an initial release. |
| | 39 | |