Teleconference 12 May 2009, 10:00-11:00 h
Urban Baby LarKC
Attendants: Daniele, Emanuele (CEFRIEL); Luka (CycEur); Barry, Matthias (UIBK); Vassil (Ontotext); Axel, Alexey, Georgina (HLRS)
Agenda:
- Review of Urban Baby LarKC status
- Review of open points (Emanuele)
- AOB
Minutes
Current Status: Emanuele merged test case 1 and test case 2 into a single one. Now the ?GeoIdentifier inserts into the Collection<?InformationSet> both the graph of the main streets of Milano and the two small graphs loaded using ?AllegroGraph ?GeoExtention.
AP: TO DISCUSS IN MILAN! Several SELECTs selecting data. The merge of those results should be passed to a REASONER. HOW TO DO THAT? TRANSFORM plug-in? DECIDE plug-in? other plug-in? Discuss in Milan about a general approach to do so in LarKC.
- Emanuele@WP5 I discovered that the platform caches data, happy with it, but how can I control such behavior?
- Vassil: it should be a decision of the DECIDE plug-in to keep the data or to remove it. See Javadoc API to see the methods to do so.
- Data is persistently stored. Therefore, we may get different answers depending on the data that is stored every time. In order to avoid this, we should be able to remove data from the cache.
- Passing the data between plugins by reference, would avoid that data is stored persistently in the platform.
AP: Vassil will send an example on how to delete data
AP: TO DISCUSS IN MILAN! DATA CACHING? Who should care about it in LarKC? Should the platform assist on this? Should every DECIDE developer care about it independently?
- Emanuele@Daniele, a got a null pointer exception when (in my best guess) the graph is unconnected. Could you check it?
AP: Emanuele will check with Daniele alone
* Emanuele@WP5 which is the correct way to rise exceptions? Does the platform start listening again for other queries after a caught exception?
- Should it be the DECIDE or should it be the PLATFORM???
AP: TO DISCUSS IN MILAN and get an agreement!!!
AP: Luka to check current status and propose a general solution.
* Emanuele@WP5/Mick can we put together a merger of Collection<?InformationSet>? It would help in having multiple identifiers.
- Related to the first discussed point
- Emanuele@WP5 where can I put a config file for the UBL plugins? I mean a place which is independent from the place where the platform in installed and that can be access using a local path.
- A plug–in which need a configuration info, it is currently coded in the plug-in.
Should the platform handle the config file of the plug-in or should the plug-in themselves do so? => It affects DESIGN of the platform!!
- The plug-in config file is general for the concrete plug-in, not dependant on the concrete execution or query and not pipeline related. Should this then be stored in the platform? Should the platform handle these settings? Should the plug-in do so? Who needs to access this file, plug-in or platform?
AP: TO BE DECIDED IN MILAN: best practice to put it in a place that can be accessed through a local path.
- Proposal from Luka to create a folder in the platform to store these files. Plug-in specific setting files? (Emanuele example of Tomcat, webinfo). WSDL file to point to this file of the plug –in so that the platform can find it.
- Emanuele@WP5 can we fix the problem of the library of UBL plugins? Where can we put them in order to make sure that they got loaded correctly by the platform?
- Some developers are committing files and changing paths, dependencies,…
AP: To be solved in MILAN!!: Library versions conflicts, where to store the libraries, …
- Emanuele@Vassil: any news about the UBL selecter?
- In terms of performance, is the correct way to execute a SELECT implementing the construct query and pass the result to a REASONER? Vassil: it depends on the REASON plug-in.
AP: Vassil to send example of execution of a construct query over the data layer api. Emanuele will try to implement it.
Emanuele@Barry&Kono: any news about the Ontobroker implementation of the shortpath reasoner?
- It must be ready before review
- Matthias: plug-in itself must be ready next week. With the pipeline, we have to see how we are progressing.