Open GoogleCodeExporter opened 9 years ago
I would like to propose a change to the id policy, to allow people the option
of having release URIs
http://purl.obolibrary.org/obo/ONT/releases/YYYY-MM-DD/ONT.owl
This is very easy to manage. See the instructions I wrote here:
www.obofoundry.org/wiki/index.php/Making_a_Google_Code_project_for_an_ontology
The release manager performs an svn copy to a /releases folder
There can be a generic purl partial redirect for each ontology
http://purl.obolibrary.org/obo/ONT/releases/ --> $websvnroot/releases/
I cannot commit to registering every release artifact for every release using
the OCLC interface, and I cannot ask others to do this either. We must make
virtue the path of least resistance.
Jie for now you can go to the svn folder for the above
Original comment by cmung...@gmail.com
on 17 Jan 2013 at 10:34
Could the release manager be configured to create the PURL at the same time it
does the SVN copy?
If that's not possible then the above would work.
Original comment by mcour...@gmail.com
on 18 Jan 2013 at 4:56
The "release manager" is currently either a human or a periodically invoked
cron or jenkins job, depending on the ontology.
For human-triggered releases, the human could create a partial redirect for the
datestamp to the svn release copy, but this is busy work, everyone has enough
on their plate.
In theory someone could add an extension to Oort to talk to the OCLC API but we
have no plans for this.
Personally I don't like clogging up the OCLC database with an entry for every
release, it's hard enough to search and figure out what's going on as it is,
our dependence on OCLC makes me nervous enough already.
Original comment by cmung...@gmail.com
on 19 Jan 2013 at 2:34
Depending 1/2 on one and 1/2 on another is the worst of all possibilities.
Adding API access to OCLC or a PURL server isn't a large task. Can it be added
to the task list? I'd like us to have a single solution. The requirement for
that solution is that it is possible to delegate maintenance to the ontology
developers (that's part of the architecture). Alternatives are to run our own
purl server, or to try out Callimachus which I'm told will have partial url
redirects in the next version.
What language does the implementation need to be in order to work with Jenkins?
I Callimachus works out then perhaps we can deploy it on Amazon free level of
service and take Berkeley out of the loop.
I'll start a separate issue proposing that we deploy to AWS
Original comment by alanruttenberg@gmail.com
on 21 Jan 2013 at 6:53
Original issue reported on code.google.com by
mcour...@gmail.com
on 17 Jan 2013 at 9:18