Resolved: minutes confirmed
<trackbot> action-193: Rob Atkinson to Move jmeter test suite to within w3c systems -- due 2018-09-05 -- OPEN
roba: still working on this
ncar: is it planned to do QSA in this test suite, too?
roba: Yes, QSA as fallback
… also planning to check headers etc
ncar: How to check that all resources are handled according to the spec?
roba: you have to nominate one test resource. You can e. g. use void to find those.
ncar: If there is wide adoption, we cannot rely that resources are RDF
… so we need as simple a test case as possible
… without knowledge of data cubes or anything
roba: the tests will be against one known resource
ncar: will use it to test our APIs
roba: delayed due to OGC paper work, not much time for implementation
so we leave action-193 open.
ncar: we spent last week looking at which issues should be lifted
… ending with issue #748
… looking at #785 https://github.com/w3c/dxwg/issues/785
roba: this seems like cartesian products
ncar: we still hae all other dimensions, too
roba: suggest "links should be at the level that the implementation supports"
ncar: this is about a link header saying which profiles are available
… if you want to know which media types are available there are other methods
… looking at example 5 in 7.1
… not current according to the spec
… all links are resource links with combinations of media types and profiles
roba: there should be nothing magic about it (cartesian product)
… can type be a list of types?
… we could answer "yes, it's correct, we just didn't provide an example"
ncar: can we mark it for 3PWD?
roba: if I suggest 10 different profiles and media types (100 possible links):is that unreasonable?
… should we limit the number of options?
… are the semantics an exhaustive list or a suggested list?
ncar: should be exhaustive list
roba: so what do you do if you have further dimensions (spatial resolution etc.)?
ncar: Kam is right in that the URI doesn't change if we really do content negotiation
… so if we just give "link" Profile-URI we should be fine (without type)
<longish discussion about processing of profile vs media type>
<conclusion seems to be that we should focus on profiles>
roba: profile should take presedence if present
ncar: we simply don't say anything about media types
roba: we should change the example so that there is one with an
… explicit type and one that works dynamically
… If a resource only supports one particular media type, what should the server say
<ncar> I've used Link headers for pagination and so does GitHub (Issues list)
roba: back to agenda
… we should address #785 in 3PWD
ncar: the issue is not how link headers work but what we think should be in it
<ncar> LarsG: we should ask Mark Nottingham or dret as they wrote the Link header spec
<ncar> LarsG: we try and solve this within a week and, if not, as them for a special teleconf
roba: OK, so we'll discuss it in the issue
Action: LarsG to write proposal in #785 regarding how to handle multiple dimensions in link headers
<trackbot> Created ACTION-326 - Write proposal in #785 regarding how to handle multiple dimensions in link headers [on Lars G. Svensson - due 2019-05-16].
ncar: we should confirm that some issues are _not_ for 3PWD but for 4PWD
<ncar> Milestone Conneg 3PWD: https://github.com/w3c/dxwg/milestone/20
roba: it's a problem to implement conneg when you don't have access to the actual server implementation but only working with redirects
… i. e. #603
ncar: #504 could be closed as resolved
ncar: #589 should be in the milestone
… it's about using specific keywords, we don't prescribe that
roba. what about #575
ncar: still significant work. will give it a go in this milestone
roba: points 16 and 17 are addressed already
ncar: we should review #575 in the next sprint
ncar: 14 issues are tagged for 3PWD, is that enough?
ncar: will address as many as possible