Spatial Data on the Web IG F2F - Day 2/2
06 June 2018
jtandy: Since Amersfoort, do you have a status update?
hugoledoux: Yes.
hugoledoux: To speak of CityJSON, we need to speak about CityGML first.
… CityGML is both a data model and a GML encoding, and both are called CityGML
… Two implementations of the data model: CityGML, and 3DCityDB in a relational database (never been standardized)
… CityJSON could be seen as a third implementation of the CityGML data model
… In the last few months, we continued developing the specification and worked on tools
… 0.6 version released. All modules have been mapped, some modules don't support everything, but that's mostly a choice!
… We can fully read/write CityJSON and convert to/from CityGML.
… We've been able to test most of the assertions
… That is the first (and important!) milestone
… Is it CityGML compliant? Yes, mostly, but there are a few "if".
… citygml4j is the conversion tool
… We also released azul on Mac to read CityJSON natively.
… We also developed cjio (Python command-line interface)
… Still beta but we can do a lot of operations on files.
… Everything merged into one coherent package to process CityJSON files.
… "pip install cjio" to install the tool.
… The validation with the schema is included
… I started to analyze file sizes between CityGML and CityJSON.
… E.g. One part of Montreal is 56MB is CityGML, 7.8MB in CityJSON. So almost a factor 10.
… Now removing spaces and tabs in CityGML removes 25% of the space
… [going through things left to tackle, see slides]
brinkwoman: Jeremy and I had a talk with CityGML folks to explain what the IG is doing and mention CityJSON
… There were worried about the governance of CityJSON. They are developing CityGML 3.0 right now and have always had plans to develop a JSON serialization.
… They haven't started yet.
jtandy: Steve Smith agreed with the points that you've made, Hugo.
brinkwoman: They were not saying that CityJSON should be an exact implementation of the CityGML data model.
… They want to make sure that the design is correct though.
jtandy: They want to make sure that all of the different encoding of the CityGML data model have a formalized mapping
… In the JSON encoding that we've seen, there's a certain amount or artistry to make things easy for developers.
… You talked about picking up all the models. CityJSON is compliant with CityGML...
hugoledoux: Mostly.
jtandy: I'm hearing that you're trying to have CityJSON compatible with the CityGML information model.
hugoledoux: In 2.0, there was no difference between the data model and the GML. Right now, I have a mapping for all geometry, semantics, etc. Raster files in GML, I've never seen a file that use that in CityGML.
… We can convert a CityJSON to CityGML and back to CityJSON without losing information.
… v3 is a distinct beast. They are adding so much more complexity in the model that I don't think I can keep a mapping.
… I don't think it's possible to map the abstract classes, the new hierachy. It's also difficult for me to assess because they release the model every few months.
… The process hasn't been very transparent.
… Difficult to assess whether it will be possible to assess mapping to CityJSON.
ClemensPortele: In terms of communication, we need to be careful about "compliance". Two aspects:
… Not everything in CityGML is implemented in CityJSON. It's a profile, but what is there is compliant.
hugoledoux: Something I proposed yesterday for a way forward for v3 is indeed to create a profile of most used features of v3.0. Happy to be fully compliant for that profile.
eparsons: With that work on v3, do they recognize that there's only going to be an XML encoding in the future?
brinkwoman: No, they also want a JSON encoding.
eparsons: But adding more complexity is going to make that much harder, surely.
brinkwoman: We didn't get to that yesterday.
hugoledoux: It's been a difficult discussion so far. The plan right now seems to be to work on the data model, and from there to work on the encodings.
… For me, these two things need to go hand in hand.
jtandy: Yesterday with Carston and Steve, they said that where the Ian Coady happens, they're not precious about where it happens, CityGML SWG or elsewhere in OGC or W3C.
… We didn't talk about profiles because I hadn't realized it was an important point to make.
… Let's assume that there are profiles of CityGML.
… They talked about the v3 model being reasonable stable for the past 5 months. I felt a desire to be more open.
… In terms of sequencing of the work, we talked about using a testbed approach, where we can take the CityJSON approach and the work that Steve has been doing, compare them, and challenge the model based on what makes things easy/hard.
josephab_: That's the point that I wanted to make.
… Hugo is essentially validating the serialization through tools. It would be a shame if that feedback was not taken into account for the information model.
hugoledoux: Certainly for the reasons that you mention. The point of view of the developer has always been difficult in SWG. CityJSON can help them for sure, and we're happy to provide feedback.
… For the process, the problem is more a cultural thing: the GitHub repo does not allow to raise issues for instance.
ClemensPortele: Going back to the profile point. Haven't looked to the v3.0 spec, but it probably defines conformance classes. Depending on how monolithic the conformance classes are, it may be difficult to define profiles.
… Something Hugo should look into, otherwise the statement may be that CityJSON is not compliant.
hugoledoux: Right, there's e.g. the notion of Space which is now part of the core.
… Creating a profile would mean that we keep the core.
brinkwoman: Hugo, do you know when the conceptual model will be done?
ClemensPortele: September, they want to present the spec for review for spec published end of the year.
brinkwoman: OK, then the Testbed approach may not work since it won't start until next year.
… On the cultural aspect, that resonates with similar experiences of people creating models on their side, and finding it difficult to get feedback once they publish what they believe is a perfect model.
PeterR: Rendering-wise, CityJSON may be easier to process than CityGML. But deeply nested objects are sometimes a problem.
jtandy: Summary: implementation is a good feedback loop to inform the underlying model.
… The issues around the space stuff that is part of the core is a good example.
hugoledoux: Right, they appeared 3 months ago.
jtandy: Is your view that they add unwarranted complexity? Or is it that they're new and it's not clear yet?
hugoledoux: Both. My colleagues and I don't understand them the same way. Feedback I got from others is that they don't need it.
… The concept of spaces is not meaningful for developers and difficult to assess.
hugoledoux: In my opinion, they make things more complex than necessary
<hugoledoux> the latest CityGML v3 are there (zipped... and 3mo old)
josephab_: I would say, get on to the "hackathon", "testbed" thing ASAP, to show what works in implementations.
… You'll be able to quickly validate whether that's useful or not.
hugoledoux: I'm afraid that the vote would be based only on UML models.
… I haven't seen the timeline yet. For me, it should happen when people have had the chance to study the implications of the new model.
… I'm not sure that there is time to move to v3. They also said that it's not their job to create a converter from v2 to v3
ClemensPortele: Debatting whether the concept of CityJSON based on v2 makes sense and whether it should be promoted for standardization. And then there is alignment with v3, which is still not mature enough.
… We cannot do much to move CityJSON to v3. That v3 is never going to be a standard before one year.
… I think we can say that we're concerned and that profiles should be allowed.
… We shouldn't wait for v3 to come along. We should work with what we already have, meaning v2.
… Let the SWG do its job otherwise.
<brinkwoman> +1
<hugoledoux> I like that a lot
jtandy: We'll have another conversation with Carston and Steve here based on this discussion.
… Having CityJSON standardized based on v2 seems a good recommendation.
ClemensPortele: v2 has one conformance class and allows profiling, I think. If that's correct, CityJSON would be a compliant implementation.
hugoledoux: Thank you all, that was very useful to get your feedback and help. It gives me back hope in the standardization process!
Rob_Smith: From Away Team Software.
… Use case update: I added a crisis response one. The details are in the explanatory spec.
… Getting reliable geolocation information from the population in a crisis context. For instance, the volcano in Guatemala right now.
… Second update is for the area survey: drone surveying an area. This was rather unexpected for me.
… WebVMT is quite an efficient way to store geolocation info for video. Allows for rapid geoindexing.
… Searching VMT files (20Kb/min) is faster than processing HD vidoes (30Mb/min)
… Technical update: The Web player has now been integrated with YouTube. You can synchronize a video with a map. Uses the iframe API provided by Google.
… I've also created a page that allows you to load and play a file
… I've also created a mobile app to record video with WebVMT. Planning to create a short film with it.
… The app is in Android, compatible with Gingerbread and later.
… Finally, engagement update: the draft has been updated. Parameters moved out of WebVMT file and into a track attribute.
… On the community outreach, I have a growing list of contacts. Business, governments and other areas.
… Business: Dashcams, drones. Government: rescue services, aviation, police (which uses body-worn cameras).
… Research. Leaflet already supports some video functionality but simply as an overlay.
… I'm planning a web site for public engagement. Not a replacement for GitHub activity. Part of the outreach process.
… Sending people to technical spec details might not be the best way to attract new people.
… There isn't anywhere to display these at W3C.
… I'm also mindful of Jeremy's point that this should not be part of the company's web site, so I'm proposing to use a dedicated domain.
brinkwoman: I was wondering about police use cases. Maybe defense organizations could be interested as well. Lots of them represented at OGC, typically.
Rob_Smith: Possibly. I've had no luck in that area so far, perhaps because I was looking for funding.
MichaelGordon: I can introduce you to folks from DSTL, involved in standards. I suspect they might be interested in that topic.
brinkwoman: [mentions Belgium]
MichaelGordon: I'll chat with these folks and will put you in touch if there's interest.
<MichaelGordon> +21
jtandy: Rob is proposing setting a Web site to gather feedback. I think that's a good idea.
… I think it's separate from standard development.
<brinkwoman> We could ask for a slot to demonstrate WebVMT to the defence & intelligence DWG at the next OGC meeting
jtandy: It sounds that you've worked on identifying your end users.
… Another area is that we need to identify people that would actually help the development of the standard itself.
Rob_Smith: That leads nicely in the topic I wanted to discuss next.
… What is the best strategy to get other people involved?
… Ultimately, the answer is money. There's a lot of collaboration in the space industry, where I come from, because it costs too much to do things on your own.
… But having been turned down there, question is: who sponsors standards at OGC/W3C?
… There has to be some incentives there that it will provide some useful benefits.
… I was wondering about setting up a consortium or alliance.
… Members could pay a subscription.
… I'm wondering whether there is an acceptable approach.
… Are there better suggestions?
PeterR: It's a long road, Community Groups give you some visibility. Video and geo go naturally together. In theory, that's a good way to progress to a standard if you get enough people to contribute there.
… Idea is that you propose a Community Group at W3C. 5 supports are enough to get it created.
… That's a good way to socialize your ideas.
… It doesn't bring any funding though.
… It's a lightweight way to start standardization process.
MichaelGordon: I believe that there's enough of standardisation bodies out there. I wouldn't say that people sponsor standards, they get involved in standards they are interested in.
… For an OGC direction, it's probably worth trying to approach the main principal members that could support the work.
… Goal is to have enough initial support to be able to create the next logical thing.
… Look at large OGC members who are interested enough to invest time in it.
… I would suggest that you look at the OGC membership list and have a look at the strategy members for instance.
MichaelGordon: From there, see if they are interested in having a discussion at the next TC meeting, in Stuttgart.
jtandy: Another piece you might be interested in is Testbed activities. A large company can sponsor research mediated by OGC, and SMEs can bid to participate.
Rob_Smith: I've looked at testbeds previously, it seems it requires you to become an OGC member and to have sponsors first.
[discussion about OGC testbed rules]
<PeterR> Web Incubator Community Group provides a forum for discussing ideas with browser vendors and other interested parties
ClemensPortele: There is a GitHub repository for ideas. Sponsors look at it.
<PeterR> You might have to join the community group before posting
MichaelGordon: I cannot, on top of my head, think of anything closely related to WebVMT in the current ideas, but putting that in the repository seems a very good way to make progress.
eparsons: I was engaged by the idea of building a Web site. It demonstrates to people what you're solving.
… Good use cases on a web site that is accessible to marketing people (as opposed to technical people).
… That's really useful.
… Second comment is that you need to think what are the best SDOs for this work. My gut feeling is that W3C is better suited, because of its history with video.
… No mobile people involved in capturing video for instance.
… You'll catch these more through W3C than through OGC.
Rob_Smith: Capturing, and indexing seem like good use cases for the Web
eparsons: Right, machine learning is being used to determine location of videos. Improving that would help.
PeterR: I put my comment into the chat window. A link to the Web Platform Incubator Community Group.
… It's a good forum to get.
tidoust: [mentions CG, IG, +1 to specific domain, need to identify implementors and what it means to implement WebVMT, and prototype using Web technologies to "look more web-friendly"]
Rob_Smith: Back to strategy funnel, can I request moving to Incubation?
PROPOSAL: Move WebVMT to Incubation in Strategy funnel
<brinkwoman> +1
<eparsons> +1
<tidoust> +1
<ClemensPortele> +1
<MichaelGordon> +1
<Rob_Smith> +1
<PeterR> +1
<jtandy> +1
<josephab_> +1
Resolved: Move WebVMT to Incubation in Strategy funnel
Action: tidoust to enact the resolution on WebVMT in the Strategy funnel
jtandy: From discussions with Linda yesterday, our feeling is that it would be useful to put SSN Extensions into the funnel
tidoust: It would be in Incubation, I believe
MichaelGordon: I think it should be on that, to capture all things that we're dealing with.
[Discussion on what incubation means and on what should appear in the funnel: SSN extensions. SSN v2]
tidoust: I believe it makes sense to put "SSN v2". That's the standardization question we need to answer. Not doing "real" standardization work is a perfectly valid answer.
jtandy: I propose to add SSN v2 and explain the work on the SSN extensions.
PROPOSAL: add SSN v2 to strategy funnel at Incubation phase
<MichaelGordon> +1
<jtandy> +1
<brinkwoman> +1
<ClemensPortele> +1
<Rob_Smith> +1
<tidoust> +1
<PeterR> +1
<josephab_> +1
Resolved: add SSN v2 to strategy funnel at Incubation phase
jtandy: We should let Armin and Simon know as well.
marqh: I've had a couple of conversations about CoverageJSON.
marqh: There are a couple of pieces in the Spatial Data on the Web IG repo
… Question on ongoing maintenance. Some parts that are not fully aligned with ISO standard (ISO 19123?)
… Should we track that in the strategy funnel?
brinkwoman: If there is work that anyone is willing to do on it, then it should be added to the funnel, yes
jtandy: There appears to be a will to move it from something that has a more prominent status.
billroberts: I was one of the editors of the note that we did in the SDW WG, but the brains were Jon Blower.
… We've been busy on other stuff. From a technical point, it has a lot of promises but I don't know who could take the lead.
… That's not a priority for me so I'm not raising my hand.
… Chris Little wanted to use it in some context.
marqh: I would like to put my hand up. There are some strong reasons from a company perspective. Right now, there's a chicken and egg issue for me to get involved.
… But I'm happy to stand up in principle.
… [Talking about implementations for OGC and ISO]
… With the caveat that I may pull the plug at any moment, I'm happy to stand up.
jtandy: I'm content with that. We're talking about investigating here.
… In order to get the ball rolling on that. Can you draft a bit of text and create an issue in the Strategy funnel?
marqh: Yes, I'll draft something and we can take it for there.
<PeterR> +1
<MichaelGordon> +1
<brinkwoman> strategy funnel is here: https://github.com/w3c/strategy/projects/2
PROPOSAL: Put CoverageJSON at Investigation phase in the Strategy funnel
<tidoust> +1
<jtandy> +1
<ClemensPortele> +1
<PeterR> +1
<Rob_Smith> +1
<brinkwoman> +1
<eparsons> +1
<josephab_> +1
<MichaelGordon> +1
Resolved: Put CoverageJSON at Investigation phase in the Strategy funnel
jtandy: Anything that people want to put in the funnel?
[none heard]
billroberts: not much happening but will start soon
billroberts lost momentum because of clashes of schedules, now using online method, needs someone to push it along
billroberts idea of having concentrated date for sdwig focus is good plan
billroberts: had a chat with Ian Coady
billroberts: UN stuff. will stay in touch with ... to ensure other initiatives stay coordinated
billroberts: reg contributors will start to contribute over next month or two
billroberts: UNGGIM update on Monday. thanks)
jtandy: ONS uk wants to get involved in this activity
michaelgordon: last year we sponsored; ONS was asked to get involved;
jtandy: is there any help needed to get ONS more engaged?
billroberts: Michael and me to keep contacts warm. all sound enthusiastic, but need to keep prodding to get more activity
MichaelGordon: i'll give denise a nudge. she's on/off unggim stuff;
jtandy: is there a stats DWG? yes in OGC, by Ian
billroberts: i've been talking to ian about that.
jtandy: sdwig to ensure linkages between OGC and W3C.
jtandy: need to find body of workers to help move stuff on
<marqh> https://portal.opengeospatial.org/files/?artifact_id=79320
marqh: comment: draft charter for stats domain wg was explicitly discussed at last ogc tc,although no project in portal yet; doc in pending
marqh: underneath tech committee in pending documents
jtandy: any comments? bill: any more coments or requests?
billroberts: not for now, will follow up online
jtandy: ogc interest group project page on github; focus days first week of each month, communicate intensively
jtandy: bill did you see the sdwig projects on github? no... not yet.
jtandy: what we talked about yesterday. you'll find your project down at the bottom, it is actually #2 in list. please edit the description and link to any work
jtandy: identify hot topics for focus days.
jtandy: in the project you can see cards / kanban style. as we come up to focus days we can see what you want help on.
jtandy: thank you bill.
billroberts: i will fill this in
jtandy: take the moving objects item will take a small amount of time
jtandy: no input from jano
jtandy: is this important enough to do something right now?
jtandy: ok to leave it at pending, but need to get jano to indicate
rob_smith: moving objects is of interest to me, sad to see; it could be relevant to webvmt. understand the need to put it off.
jtandY: rob would you take an action to poke jano via the mailing list?
rob_smith: yes
jtandy: anymore about moving objects/.? no
15 minute break
yes, we will and send to you
jtandy: notify Michael if you know of work that is implementing best practices
MichealGordon: review of what has happened
MichaelGordon: maintenance activity. Best Practices in Practice first draft complete
MichaelGordon: used in evolution of OWS in OGC
MichaelGordon: first draft of best practices in practice provides quick links across examples in best practices
MichaelGordon: have found unnumbered examples in the bp document.
MichaelGordon: added details from...
MichaelGordon: API for GNAF
josephab_: final URI to be available in July; will have 13/14 best practices implemented.
josephab_: don't have a commercial use case for the final example
MichaelGordon: looking for more examples; have a place to put this stuff eventually pull into best practices, examples of implementation
MichaelGordon: some best practices don't have examples
MichaelGordon: no actual maintenance has been done
MichaelGordon: examples not in HTML. Can't link to best practices without ids.
<tidoust> [Right, ReSpec does not support numbering of examples out of the box, IIRC]
jtandy: as ex-editor recommend pull request, then merging those in. Change history should be updated.
josephab_: pull request, on overall thing for the group. for individual work items, Michael is in charge of that resource, we prefer a pull request rather than a direct commit.
MichaelGordon: other topic is in next 6 mo, what is intended outcome?
MichaelGordon: adoption activities
MichaelGordon: if there are gaps please advise
edparsons: write a report based on examples that we found, these are the impacts that the bp paper has had.
edparsons: to get to a quantitative measurment of what impact has had
jtandy: who would write?
edparsons: perhaps reach out to one of the trade magazines
jtandy: MichaelGordon, can you take this under advisement?
MichaelGordon: yes
ClemensPortele: have 14 best practices, then tiny examples in there. what would be more helpful,might be to take wfs3 implementation report
ClemensPortele: describe what best practices are implemented, what aren't. one markdown page per implementation. easier than tiny examples
ClemensPortele: similar to doc I wrote for INSPIRE. That's one other thing for discussion, we may be extending data on the web best practices.
ClemensPortele: 50 best practices is very hard to read. stick to our project for now.
edparsons: hot topic influence by best practices
jtandy: tricky to tell whether would have happened anyway.
<brinkwoman> This is sort of an 'implementation report' from the Dutch geoportal PDOK: https://docs.google.com/spreadsheets/d/1_itsV7vgotV61eYycDYrcAi5hD0q9mSB4U58gDrMrqU/edit?usp=sharing
ClemensPortele: I waited until bp was available to submit change request.
josephab_: i chaired research program that had its work reviewed against sdwbp. Looked at gaps, we thought the user community needed this
josephab_: conformance / non-conformance criteria, and impact for each project
brinkwoman: put link in irc above to geoportal in NL report quickly gives overview.
jtandy: trying to encourage more people to be web-friendly content (slightly subjective) but often see publishing playbook these days
jtandy: if you start here, and follow this tutorial, you will come out with web-friendly result. don't suggest to randomly pick a scenario, could use what you did to make that into a playbook
jtandy: it might be challenging to write, but if we want to encourage people to follow the best practices we could give examples of how to do that
ClemensPortele: ER are not playbook style
MichaelGordon: target audience
edparsons: SDI cookbook was quite influential
edparsons: that content is obsolete but the method is what you are talking about
PeterR: here is the spreadsheet of how MapML responds to sdwbp
josephab_: could provide retrospective look
jtandy: Clemens talked about engineering report, is kind of like the case study idea.
ClemensPortele: need a template, not too many; implementation reports have tick boxes
brinkwoman: as example of cookbook, like the github guide
ClemensPortele: the github example is easier than sdwbp cookbook, because of the tool nature
brinkwoman: helps newcomers with getting started
jtandy: more comments for Michael?
jtandy: bp project needs to be updated
MichaelGordon: have a good chunk of work there
ClemensPortele: gaps is a separate project; more important to focus on outreach and communication
<brinkwoman> +1 to clemens
<Zakim> jtandy, you wanted to ask about DXWG
jtandy: dxwg might use content profile negotiation
jtandy: no issue associated, useful to create a ticket linkage to dxwg
jtandy: where are we going to publish and how to drive people to read
MichaelGordon: if we end up with different artifacts that will help people implement, might need an overarching resource to direct people
jtandy: 3 things of poss. interest: use of press releases when we have useful stuff from w3c and ogc
jtandy: occurs that fit / no-fit e.g. unggim is a place where we try to get people to publish geo data, is there an inroad to unggim to understand
group: yes
jtandy: visible part is the document thats published, the conversations that need to be had are the real workj
brinkwoman: geonovum is planning event to maximize SDI, in Amherstfoort (sp?). Invite help to contribute.
<ClemensPortele> https://www.w3.org/TR/websub/
<brinkwoman> SDI.Next, oct 31 in the Netherlands, probably Amersfoort
ClemensPortele: how w3c does implementation reports is links in document. they have link to github repo with implementations listed.
ClemensPortele: eventually could create something more shiny, turn over to a journalist eventually.
ClemensPortele: update note to add implementation reports
jtandy: From Chris one issue identified and fixed
tidoust: confirmed
jtandy: Another issue re unix time not really an issue
brinkwoman: issue 987 covered by dr-shorthair in Git, already in owl time widely used - however issue raised has merit
brinkwoman: Replace schema data type with new data type in owl
jtandy: Is feedback expected ?
ClemensPortele: No proposal is on the table to resolve the issue
jtandy: Is something broken or is this a purist problem ?
jtandy: Need dr-shorthair feedback as to the impact of changing the spec ?
ClemensPortele: Would a bugfix be compatible with current version
jtandy: Can we add this discussion to the issue
tidoust: Yes will do
jtandy: Anything we have not talked about ?
Rob_Smith: GPX forum chat on new version of GPX - GPXv2
Rob_Smith: Suggested to get in touch with tidoust
<Rob_Smith> Dan Foster
jtandy: GPX id'd as a de facto standard
tidoust: No contact yet ... what did he want to achieve ?
Rob_Smith: GPX predates wearable tech etc - needs update
Rob_Smith: Suggested w3c would be good forum to talk about it
Rob_Smith: GPX has mistakes never corrected
brinkwoman: TPAC ?
brinkwoman: Logistics - days etc
tidoust: Thurs/Fri scheduled but same as data exchange so may move to Monday/Tuesday
jtandy: Last week of October
jtandy: Lyon
tidoust: Promises Good Food !!
brinkwoman: May want to organise Wednesday meetings
PeterR: Too Late to Organise CG Meeting ?
tidoust: Maybe...
PeterR: Want to talk to browser devs
jtandy: Wednesday best day for that...
<tidoust> [FYI, there seems to be a couple of slots left for CG meetings on Friday]
jtandy: Thanks Everyone !!!!
jtandy: Always room to improve meeting - let jtandy and brinkwoman know
jtandy: Meeting Closed