Spatial Data on the Web WG Coverages Sub Group Teleconference

04 May 2016

See also: IRC log


ScottSimmons, billroberts, Maik, Duo, phila, Kerry
jonblower, eparsons, Jeremy


<billroberts> https://www.w3.org/2015/spatial/wiki/Meetings:Coverage-Telecon20160504

<Kerry__> scribenick: kerry

<billroberts> https://www.w3.org/2016/04/20-sdwcov-minutes

<Kerry__> proposed: approve minutes https://www.w3.org/2016/04/20-sdwcov-minutes

<billroberts> +1

<Maik> +1

<ScottSimmons> +1

<sam> +1

RESOLUTION: accept minutes https://www.w3.org/2016/04/20-sdwcov-minutes

<billroberts> https://www.w3.org/2015/spatial/wiki/Patent_Call

<Kerry__> patent call

<billroberts> https://www.w3.org/2015/spatial/wiki/Coverage_Solution_Criteria

brief recap of previous meeting

<Kerry__> bill: billroberts discussed criteria and also updates wiki page

<Kerry__> billroberts: also had presentation from ANU team

<Kerry__> ... suggest w have update from ANU and Maik at Reading today, then plan wehat to do next

<Kerry__> ... invite sam or Duo to speak

<Kerry__> Duo: not much to say -- getting protoype working but not quite ready to show off

<Kerry__> ...we have some data in a live application but not quite uasable or interactive yet -- will take another week or 2

<Kerry__> billroberts: anything stood out as challenging or othereise of interest to us?

<Kerry__> Duo: just getting things to run and work together

<Kerry__> .... how to create sparql queries that are efficient enough for coverage data -- we have identified this as significant

<Kerry__> billroberts: spaql on large data or with complicated joins is more than an art than a science and can be implementation dependent

<Kerry__> ...some trial and error for how to best write the query

<Kerry__> which sparql?

<Kerry__> billroberts: duo: jena and fuseki

<Kerry__> billroberts: jena list is very helpful

<Kerry__> ... please let me know when you have something to show and we will schedule it

<Kerry__> billroberts: how is coveragejson going?

<Kerry__> maik:spoke at egu, got good feedback

<Kerry__> .. completely different to what exists now -- at first confused but then positive

<Maik> http://json-ld.org/playground/#startTab=tab-compacted&json-ld=https%3A%2F%2Fgist.githubusercontent.com%2Fneothemachine%2F6c7a337c82575efc08bbf43b234c9b01%2Fraw%2F6981d315eef329439696d3f802642b86b0ce56ba%2Fgrid-example.covjson&context=%7B%7D

<Kerry__> maik: been looking at what exactly json-ld means for coveragejson

<Kerry__> ... this is a coveragejson doc with json-ld context

<Kerry__> ...some perople were asking about this

<Kerry__> s/perope/people/

<Kerry__> ...also how to express quesantity types, fractions, etc, pls see our issues page and cookbook we are writing

<Maik> https://github.com/Reading-eScience-Centre/coveragejson/issues/63

<Kerry__> billroberts: rdf datacube has dealt with these questions too and might help here

<Kerry__> Maik: here is what we are looking at for rgb bands.

<Maik> https://github.com/neothemachine/xndarray

<Kerry__> ...am writing a javascript library for arrays called (missed it)

<Kerry__> ... more lightweight than others for mutidimensional data, expecially for coveragejson multidimensional range objects

<billroberts> (library called xndarray)

<Kerry__> ...will have coordinates some time

<Kerry__> ...that's it...

<Kerry__> billroberts: collaboration with beijing re CEO-ld -- have you heard anything?

<Kerry__> Maik: they have implemented coveragejson from scratch with geotiff and landsat

<Kerry__> ... see how to efficiently have an api for big amounts of statellite images

<Kerry__> billroberts: uses a geotiff to ccoveragejson converter

<Kerry__> billroberts: how do we take our scoping work and use it?

<Kerry__> .... we should aim towards producing a spec

<Kerry__> ...as well we also need a primer to esxplain what we are trying to achieve

<Kerry__> s...can use these solution criteria

<Kerry__> ...but need to go further on exploringdetails on what that would be

<Kerry__> ...thinking to continue both tracks as at present and at some point we compare these solutions to criteria and make a proposed solution

<Kerry__> ... also need to test alongside existing technology such as wcs

<Kerry__> ... to demonstrate we have something to offer

<Zakim> phila, you wanted to answer Duo's question

<Kerry__> phila: an implementation for a spec needs 2 independent implementations of every feature of the sepc, so it's a high bar!

<Kerry__> ... needs to be as good as html or css...

<Kerry__> ...if ANU does na REading does and they both implement everything, but we really would like a third too

<Kerry__> .... and a common test suite is needed too, and exmaples

<Kerry__> ... if we are going for something as solid as that then we need all this

<Kerry__> billroberts: then we also need to be canvassing for people outside this meeting for implementations too

<Kerry__> phila: If we say 'who wnats to do cool map overlays" etc we can get a huge amount of interest if we do this right

<Kerry__> billroberts: amybe SWIRRL too

<Kerry__> phila: SWIRRL as a commercial player is a plus -- working with the universities here

<Kerry__> billroberts: we could do this if it works for our use cases

<Kerry__> ....evidence of commercial application is extra power it seems

<Kerry__> phila: yes. shows it has value

<Kerry__> billroberts: is a spec that works towards a rec plus a primer as a technical note ok?

<Kerry__> phila: that is one way, but not the only way.

<Kerry__> phila: matbe unlike ssn, the test suite may be important itself as supplmentary material

<Kerry__> ... also whatever OGC needs

<Kerry__> billroberts: so... how do we get to that point/

<Kerry__> billroberts: will cjheck over which BPs need we need to take into account

<Kerry__> ..we may need to move requiremnt to formal testable form

<phila> Activity Streams

<Kerry__> kerry: suggest looking at WCS too for an idea

<Kerry__> Duo: testing framework -- who would develop this suite?

<phila> Build the suite as you go, then rationalise into a doc

<Kerry__> ... we'd like to strt building ad testing against this asap

<Kerry__> billroberts: agrees... we will need volunteers

<Kerry__> ScottSimmons: we can use ogc teamengine for test development as it will need one for spec anyway

<Kerry__> ...ogc does note require multiple implementations but the test suite is very important

<Kerry__> ...this staurday there might be a decision that will affect the standard tier wrt reference implementations

<Kerry__> billroberts: "abstract test suite"

<Kerry__> ScottSimmons: set out in words that exaplins what a physical test suite would look like

<Kerry__> [scott looks for an abstract test suite to share]

<ScottSimmons> look at Annex A in http://docs.opengeospatial.org/is/14-100r2/14-100r2.html

<Kerry__> billroberts: so we need our requirements formalised, a test suite, and in the meantime example implementations to carry on. then we can evaluate implementations against the requirements

<Kerry__> ... implementations will also help us to understand appropriateness for requirements

<Kerry__> phila: requirements are always important becuase they help you develop the test suite -- as an anchor-- but alsway leave out or cover up stuff

<Kerry__> ...suc has huge assumptions

<Kerry__> ...we have a good UCR

<Kerry__> ..does not say "the Web has to exist' as it is assumed

<Kerry__> ...but just doing what the requirements says can be silly

<Kerry__> we have one verty advanced and one rapidly developing group, plus commercial interst, plus chinese group so we have time to work on this

<Kerry__> ..if we can develop some of that test data too and get in place by (northern) summer break we'd be cooking with gas

<Kerry__> ...i would be happy to say then that we are making progress, but if we get to september still messing around, im not so sure..

<Kerry__> billroberts: all good.

<Kerry__> billroberts: will take first attempt at requirements

<phila> Kerry__: Don't forget that the WG's UCR already exists :-) https://www.w3.org/TR/sdw-ucr/

<Kerry__> billroberts: any other questions?

<phila> s/sdw-ucr/TR\/sdw-ucr/

<Kerry__> Maik: how concrete or abstract would they be?

<Kerry__> ... itsounds like they have to be abstract

<Kerry__> billroberts: needs to be concrete while expressing what, not how

<Kerry__> ...requirements doc, not design doc

<Kerry__> ...not excessively prejudging the soluti0ons

<Kerry__> ...will try writing something

<Kerry__> billroberts: close meeting

<billroberts> thanks kerry!

<billroberts> thanks everyone

Summary of Action Items

Summary of Resolutions

  1. accept minutes https://www.w3.org/2016/04/20-sdwcov-minutes
[End of minutes]