Sarbajits API ingest performance issues. What does this mean for using the API vs native Jena tools to ingest data. This doesn’t make sense to Ralph either.
1b. On wiki agenda now.
Brian - Opera Nora project - 250M triples,a ll come in via API. 200K pubs, lots of orgs. Takes about 2 days total to load all through the API. Beefy server.
Michel - Usage of the API for the amount of of data doesn’t seem right. Http isn’t made to xfer a gig of data.
Don - issue was to have VIVO be triple store agnostic and TDB isn’t ACID compliant.
Andrew - Jena tools does load data.
Michel - Sparql endpoint itself isn’t efficient. Could create a webservice without using a sparql endpoint.
Benjamin - Difficulty of jena tools and harvester tool is there’s no way to queue inference tools for specific objects. So when you load RDF via API VIVO chooses those triples to inference and index on. So if you bypass the API then everything needs to be inferenced and indexed.
Michel - so webservices can use ftp for example to up. Large datasets are a problem.
Andrew - Can we re-index, re-inference things in just one named graph.
Benjamin - currently the index has an API where you can re-index specified triples, we could use a re-inference feature like that too.
Brian - would like a re-inference by named graph.
Action - Andrew - tickets needed for reinference by a list and/or named graph.
Brian - can respond to Sarbajits email. What’s needed for documentation. Tomcat params - max post and max-swallow-size needs to be documented. Otherwise can get a 401.
Don - Doesn’t TIB and Mike Conlon use the API a lot. Might want to link back to older Sarbajit slacks on his API vs SDB vs TDB performance issues
1a. Updates? "Site admins" unable to add People (email)
Andrew will follow up.
2. v1.12 - Andrew looking for agreement on schedule. WIll consult with leadership group.
A. 1918 --- Andrew wants the firsttime files tested.
Don can test this on a stock VIVO instance.
Brian - need to also verify that GUI changes will also remain.
The GUI should take
Compares firsttime against stored copy of last time firsttime files was loaded against the state in the triplestore. So if you change the triple in the gui, it will not load the entry.
Don - so if you change the triple in the GUI, and then you want to persist the change in the files, you will have to blow away those triples in.
Don - Could changes in the GUI also be written to the files.
A. ii - 1934 -- Simple change to test and review.
C. iv. Do we want the war install in 1.12. Self contained .war which populates VIVO home on first startup. Work has been done.
Don - is this in addition to allowing users to have VIVO install on build.
Andrew - this is instead of. So the old deploy into the tomcat dir won’t work anymore.
Don - this seems like a “breaking change” to some degree. It should be noted as such.
Benjamin - it does make his deploys easier.
Andrew - 3rd tier builds still work.
Ticket 1443 - oct 17 link to github project - vivo custom webapp is andrews example of how to do a 3rd tier.
William - so many benefits from this. Can build and then run in a container.
Andrew - argument that the existing install bit makes the installs easier, but now can be more modular.
Don - nice to have it modular because we can have build images and then more hardened runtime images.
Benjamin - would replace existing pom.xml and also not set the installer directory directive.
B. 1734 remove on-time perms in ORCID language strings.
Matthias or Dominik will move it through
1652. Ontology Change - Administrative Unit is a new sub-class of a foaf:Org
Ralph thinks this is still in place for 1.12 - Ralph will check with Ont group.
Andrew - Agenda 2 - 1.12 release
Targeting lot’s of work and collaboration to release into the new year.
Bruce - estimate on amount of 1.12 work.
Andrew - March??
Bruce - suggestion for a marketing survey for VIVO in a box. So Bruce can still work on the survey as 1.12 is done.
Next meeting
Don wants to explore using the i18n files to store local institutional RDF label and description changes.
Indicating note-taker