OpenAerialMap/Meeting Mar 19, 2015

From OpenStreetMap Wiki
Jump to: navigation, search

Mar 19 19:01:47 cristiano: Hello everyone - OAM meeting starting in couple of minutes
Mar 19 19:02:15 cristiano: Blake: can you do take care of the log please?
Mar 19 19:02:58 cristiano: Here's a direct link to the agenda for today:
Mar 19 19:03:56 dodobas: hi
Mar 19 19:04:27 lossyrob: hi, Rob here.
Mar 19 19:04:51 cristiano: Hi lossyrob and dodobas
Mar 19 19:05:44 cristiano: We'll wait a couple of minutes to confirm that Blake can do the log and to allow people to go over the agenda
Mar 19 19:06:15 cristiano: lossyrob: feel free to add anything on the first item if you want to follow up on the discussion post BoF
Mar 19 19:06:25 lossyrob: sure
Mar 19 19:08:42 FrankW: I'm here for a few minutes (btw, thanks for the great summary of the BOF lossyrob)
Mar 19 19:09:08 lossyrob: FrankW: np!
Mar 19 19:10:06 cristiano: OK, it looks like the logs are taken care of, so we can start
Mar 19 19:10:38 cristiano: FrankW: is there anything you want to touch on before you have to leave?
Mar 19 19:10:46 cristiano: we can jump through the agenda
Mar 19 19:11:54 FrankW: cristiano: nope, just monitoring.
Mar 19 19:13:02 cristiano: OK. We will start working on the next contract pretty soon - the processing/tiling node, last item on the agenda - so we will need your inputs on that for sure :-)
Mar 19 19:13:48 cristiano: OK, lossyrob thanks so much for organizing the BoF and the great summary
Mar 19 19:14:26 lossyrob: surely. there was a lot of interested parties sitting around the table, and a pretty great discussion
Mar 19 19:15:04 cristiano: Do you want to expand on the architecture discussion other than what you reported?
Mar 19 19:15:26 cristiano: otherwise we can touch on the branding and licensing
Mar 19 19:16:02 lossyrob: If there's any questions about what I outlined I can try to elaborate
Mar 19 19:18:10 cristiano: I feel the architecture discussion may get too broad here for an IRC discussion, but I think that we should include some kind of licensing selection method
Mar 19 19:18:35 cristiano: especially for the HOT usecase, the is going to be need to sort/filter by license
Mar 19 19:18:55 cristiano: so, we should account for that "switch" in the design
Mar 19 19:19:17 lossyrob: I totally see a need for that. And that's why I think the OAM catalog can pull from this network of same-licensed open data, as well as other sources which are differently licensed
Mar 19 19:20:08 cristiano: and that may not be the default, but instead an advanced option in the catalog interface to find those dataset that don't have a completely open license
Mar 19 19:20:09 lossyrob: which is why I see a separation of concerns here...OAM is trying to index a broad set of imagery for access through the OAM technology. Others might want to access a purely open network in some other fashion.
Mar 19 19:21:19 lossyrob: so really what I proposed is a way to create a discoverable network of imagery that is hosted on disparate sources, that some other technology could draw from.
Mar 19 19:22:07 lossyrob: and those networks could be segregated by licensing, so that issue sort of goes away...if there is a network of only open data, that's fine, OAM catalog can pull from it. But it can alls pull from/index other networks (that can be based on the same technology)
Mar 19 19:22:27 lossyrob: and that allows providers to participate in the networks that have the appropriate licensing
Mar 19 19:23:03 lossyrob: and if someone wants to contribute to a network of same-licensed open imagery, then they can...and that directly benefits the OAM index, because the OAM catalog is indexing that open data set.
Mar 19 19:23:16 cristiano: yes, totally agree on that. And if we can integrate that with some kind of authtentication through the catalog, then users will have a seamless experience
Mar 19 19:23:43 lossyrob: if digital globe wants to not publish on the open network, that's fine to...they publish to some other network that is accessable only through an authentication token through OAM, but that imagery is also indexed by the OAM catalog
Mar 19 19:23:58 lossyrob: right
Mar 19 19:24:02 cristiano: exactly
Mar 19 19:24:08 lossyrob: so to the user of OAM it's totallly opaque
Mar 19 19:24:27 lossyrob: but in the backend, its segregated enough where licensing because the division of the networks
Mar 19 19:24:45 lossyrob: and this means that other applications can also benefit directly from the open network, without having to build in licensing tools
Mar 19 19:25:02 cristiano: the catalog should also be proactive to make sure those sources are all reliable, with continuos monitoring and testing
Mar 19 19:25:29 cristiano: that way the imagery flow into OAM could be as seamless as possible
Mar 19 19:25:57 lossyrob: tooling could be built to validate the imagery, although I'm a bit unclear what the monitoring and testing is
Mar 19 19:26:16 lossyrob: but if the network is specified as simply as possible, tooling can be built to make those sort of checks\check for updates
Mar 19 19:26:24 wildintellect: sounds like nagios style ping to ensure it's available
Mar 19 19:26:26 lossyrob: it does put the onus on the provider to provide up to date metadata
Mar 19 19:26:28 cristiano: making sure the source are reliable and can serve imagery TMSs when called
Mar 19 19:26:40 lossyrob: they don't have TMS ready imagery though
Mar 19 19:26:50 cristiano: and work between nodes that can accepts loads when need to load balance stuff
Mar 19 19:26:55 wildintellect: that was the change in architecture
Mar 19 19:26:57 lossyrob: this is for the raw RGB imagery that is not yet tiled for a TMS
Mar 19 19:27:17 lossyrob: right maybe I didn't make that point clear
Mar 19 19:27:32 lossyrob: in the discussion, there was talk about providing imagery vs providing tiles
Mar 19 19:27:50 lossyrob: tiles can be made from imagery, but requires processing: reprojecting to Web Mercator, tiling, indexing
Mar 19 19:28:18 lossyrob: and then, if I provide a set of imagery, I won't necessarily be willing to also host the set of tiles from that imagery. so where are those stored?
Mar 19 19:28:35 lossyrob: that's something that OAM processing node can do, make tiles out of sets of imagery
Mar 19 19:28:50 lossyrob: (not the whole network, because that would be waaaay too many tiles to bake and store)
Mar 19 19:29:22 lossyrob: but that again is a externalized technical challenge that is placed on the tooling around the imagery network
Mar 19 19:29:25 cristiano: right, so there would be OAM nodes who volunteer that capacity
Mar 19 19:29:28 lossyrob: network(s)
Mar 19 19:29:43 lossyrob: that sounds like a good idea.
Mar 19 19:29:56 lossyrob: so nodes would volunteer CPU to bake the tiles and storage to host the tiles
Mar 19 19:30:06 cristiano: i.e. and HOT OAM node would offer tiling/processing for what it can support for its mission
Mar 19 19:30:22 cristiano: *an HOT OAM
Mar 19 19:30:49 lossyrob: there's a way you can serve imagery "on a map" e.g. show the footprint and allow the download of imagery (a la Libra) which I think would be a first step
Mar 19 19:31:21 lossyrob: then there's actually baking the tiles and having some sort of TMS service that serves out the imagery as a TMS layer, and perhaps just baked in the freshest image onto the tile
Mar 19 19:31:25 cristiano: yea, footprints and overviews that then the user can request to tile
Mar 19 19:31:34 lossyrob: I'm not sure what's in scope for the period of work in OAM catalog
Mar 19 19:31:38 cristiano: that should not be the default though
Mar 19 19:32:34 cristiano: the OAM catalog work will also include the API development, so these functions are to be considered in
Mar 19 19:33:26 lossyrob: ok. so it seems like the baking/serving of tiles is part of the catalog UI? I'm unclear on the architecture there, but I think it's orthogonal to the architecture I was outlining
Mar 19 19:33:40 cristiano: anyway, I think I like the idea of the two steps when tiling can't be done a priori
Mar 19 19:33:43 lossyrob: i.e. tooling built on top of the networking idea
Mar 19 19:34:09 cristiano: well, the catalog will send and receive instructions for the data sources
Mar 19 19:34:38 cristiano: so, it will know if a source is not tiles and it'll only serve an overview or footprint
Mar 19 19:34:39 lossyrob: what instructions would it send to the data sources?
Mar 19 19:34:59 cristiano: sorry - the OAM tiling/processing nodes
Mar 19 19:35:02 lossyrob: hmmm yeah that could be a description in the metadata for the provider
Mar 19 19:35:08 lossyrob: gotcha
Mar 19 19:35:51 cristiano: anyway, that maybe leads us to the catalog proposals
Mar 19 19:36:20 cristiano: but first just wanted to touch on the branding discussion
Mar 19 19:37:14 cristiano: and OAM vs other names… OAM came up as an analogy with OSM initially, but I think we should be open to any name that really conveys what we are building
Mar 19 19:38:04 lossyrob: I think OAM is a great name for the catalog
Mar 19 19:38:35 lossyrob: my suggestion is not to drop OAM at all. but to differentiate it from a network of open imagery data that OAM can pull from, but also that other tooling can pull from
Mar 19 19:38:45 lossyrob: that I think is of broader interest to communities
Mar 19 19:39:06 lossyrob: where the same-licenseness is desired, and tooling can actually be built around plugging providers into that network
Mar 19 19:39:30 cristiano: OK. I was thinking OAM may not be appropriate or very intuitive for example for the portable version itself
Mar 19 19:40:32 lossyrob: you mean the catalog UI/imagery processing that would happen in the field?
Mar 19 19:40:46 cristiano: and we should def make sure we use software licensing that allows for making those different tailored versions where needed
Mar 19 19:41:00 cristiano: right, that would be more like a toolkit
Mar 19 19:41:18 cristiano: a field client for OAM that plugs into the main OAM when connected
Mar 19 19:42:46 cristiano: OK, anyway, I listed the 5 proposals we received. For two of them I haven't received approval for public sharing yet, the other three should work for anyone to look at
Mar 19 19:43:11 cristiano: Feel free to read through and provide any feedback
Mar 19 19:44:17 cristiano: We will start reviewing them today after finalizing criteria
Mar 19 19:46:47 lossyrob: there were two proposals that are DDF based, correct?
Mar 19 19:47:31 cristiano: Yes. Connexta and Cohesive Integrations
Mar 19 19:48:28 cristiano: There are also have some weights in there. They should sum up to 1 and then we multiple each criteria score.
Mar 19 19:48:38 lossyrob: I wish I knew more about that project
Mar 19 19:50:20 cristiano: me too, but I've been reading and digging through quite a bit to make sure we can make a good evaluation
Mar 19 19:50:39 lossyrob: I think a criteria that I would like to see listed is a plan / perceived ability to engage with the community of interested parties (which seems to be quite a few in number) in developing the solution. Which I think is quite a challenge, given the time frame.
Mar 19 19:51:51 cristiano: Agree, that is probably part of sustainability, but I'll propose it as a separate one
Mar 19 19:52:40 cristiano: anyway, I will let you know if the other two proposals will be public too, so you have a good view of all the options
Mar 19 19:54:08 cristiano: I need to leave in ten minutes, so let's go through the other points quickly unless you have other questions/comments on the catalog proposals
Mar 19 19:55:26 cristiano: We will try to have the selection by Monday, and start a contract soon after that
Mar 19 19:57:00 cristiano: We are hoping to get a good working prototype by the end of April, so that we can do a 2 day in person sprint in Washington DC in early May
Mar 19 19:57:34 cristiano: The HOT summit is APR30-MAY2, so we are planning an OAM sprint right after that
Mar 19 19:58:01 cristiano:
Mar 19 19:58:52 cristiano: It would be nice to have developers from interested parties join in for the sprint
Mar 19 19:59:58 cristiano: I will send a message to the list, but if anyone here thinks he/she may be available on those dates, please let me know
Mar 19 20:00:38 lossyrob: I'd certainly like to attend, I'll have to check that the date works.
Mar 19 20:01:23 cristiano: awesome. And the last point on the agenda is partly related to that. We would like to start recruiting someone for the development of the processing/tiling node
Mar 19 20:02:27 lossyrob: is that going to be through another tech challenge?
Mar 19 20:02:28 cristiano: so, I'm going to share the first draft of requirements, but I will obviously would like your input based on recent discussions on architecture.
Mar 19 20:02:49 cristiano: not sure, it may be faster and more efficient through a direct contract
Mar 19 20:03:13 cristiano: and we are open to suggestions and advice
Mar 19 20:04:18 lossyrob: I'd be happy to look over the requirement and see if I have any input or ideas. Will you share that as a google doc?
Mar 19 20:04:46 cristiano: yes, I will send it out later today or tomorrow
Mar 19 20:04:52 lossyrob: ok great.
Mar 19 20:05:05 cristiano: OK. I gotta go now. Let's continue over on mailing list. Thanks lossyrob and others for attending today
Mar 19 20:05:14 lossyrob: thanks cristiano.
Mar 19 20:05:24 cristiano: have a good rest of the day. See ya!
Mar 19 20:05:39 lossyrob: cheers