On 9 December 2010 15:58, chung <chvng at mail.ntua.gr> wrote:

> On Thu, 2010-12-09 at 11:03 +0200, Nina Jeliazkova wrote:
>
> > Hi,
> >
> > On 9 December 2010 10:48, Andreas Maunz <andreas at maunz.de> wrote:
> >
> > > Hi Pantelis,
> > >
> > > chung wrote on 12/08/2010 06:06 PM:
> > >
> > >     I have some question, since we currently work on accounting and
> we've
> > >> managed to handle local limitations on the per user number of models
> and
> > >> running/queued tasks allowing for 2000 models, 1000 BibTeX entries and
> 2
> > >> asynchronous tasks (these bounds were chosen arbitrarily). With the
> > >> current set up, it is feasible (and easy) to provide to users lists of
> > >> resources that they have created, exploiting our local database and
> the
> > >> SSO identity service. The question is how this should be done? I was
> > >> thinking of a boolean URL parameter so that the request:
> > >>
> > >> POST on /resource?onlyMine=true
> > >> Authorization:{token}
> > >>
> > >> will return to the client the list of resources for which he is the
> > >> creator. But what would be of real interest, is to provide to the user
> a
> > >> list of URIs for all resources to which he has (any) access.
> > >>
> > >
> > > I was considering such a "wildcard authorization query" a while ago for
> the
> > > implementation in OpenSSO.
> > > But it turned out that an automatically generated list, to which a
> given
> > > user has any access, was not feasible.
> > > Each query must be bound to a specific pair of URI and ACTION.
> > >
> > >
>
>
> BTW, I think we don't really need the /{userid}/ part in the URL. The
> service knows who you are by your token. So we can keep it simpler and
> also replace {bookmarkid} by {category}. This boils down to a list of
> URI templates:
> GET /bookmark : all bookmark categories
> GET /bookmark/myModels: User's favorite models
> GET /bookmark/myAlgorithms
> GET /bookmark/myTasks
> These are manually updated by the user applying a POST
>
> What do you think?
>
>

Yep, my first version was without {userid} ... but it is much easier to
protect the bookmark service itself if userid is there, otherwise there is
no reasonable way to protect it via current AA.

Then I would prefer to have search parameter for models, algorithms, etc,
but this is not critical. The annotea ontology has a property to assign
"Topic" of the bookmark, so the type of the bookmark could naturally be
there.

To answer the question in the previous emai - yes,  something like "My
favourites" or http://www.delicious.com/ , etc.

Regards,
Nina



> Yes, that's why we are considering additional service for "bookmarking"
> URLs
> > of interest
> >
> > /bookmark/{userid}/{bookmarkid}
> >
> > It may include not only "own" resources but let's say somebody else model
> > URI or dataset which are interesting to keep for some reason. There are
> > relevant ontologies to represent bookmarks:
> >
> > Bookmark : http://www.w3.org/2002/01/bookmark#
> > Annotea: http://www.w3.org/2000/10/annotation-ns#
> >
> > If there is a consensus to include it into the API , would be great,
> > otherwise we'll be using it internally for toxpredict and read across
> > applications.
> >
> > Regards,
> > Nina
> >
> >
> > > Regards
> > > Andreas
> > >
> > > --
> > > http://www.maunz.de
> > >
> > > There are only 10 types of people in the world: those who understand
> binary
> > > and those who don't.
> > >
> > > _______________________________________________
> > > Development mailing list
> > > Development at opentox.org
> > > http://www.opentox.org/mailman/listinfo/development
> > >
> > _______________________________________________
> > Development mailing list
> > Development at opentox.org
> > http://www.opentox.org/mailman/listinfo/development
> >
>
>
> _______________________________________________
> Development mailing list
> Development at opentox.org
> http://www.opentox.org/mailman/listinfo/development
>