Adding map API

[ajakubo1]

Any suggestions on map API?

Google maps seem to be a sensible choice, as those are free for non-profit applications:

https://developers.google.com/maps/pricing-and-plans/#details

Do you know of any other sensible map alternative?

[zen]

OpenStreetMap - but I'm not sure if they have API.

I'm in general OK with Google, but it would be great to make it pluggable.

[ajakubo1]

You mean in case we want to swap it?

Yeah, no problem, I'll try to abstract it :).

[zen]

Yes. Google changed their different policies many times, I would not rely on this for 100%

[ajakubo1]

Hmm, this one looks quite interesting as well: http://leafletjs.com/

And they did a stable release 1.0.0 a couple of months ago.

[zen]

Yeah, that looks pretty cool and it's really lightweight. I would give it a try

[lechup]

It's open source so I think we should go leaflet + OSM.

PS: I hope we do not need to host OSM map service? ;)

[lechup]

Oh, and I assigned @ajakubo1 to this issue. Any objections? :)

[ajakubo1]

Nope, none at all. I'll read up and create a prototype :).

[lechup]

@ajakubo1 do we want to host tile service connected to Leaflet?

I've checked pricing for Mapbox: https://www.mapbox.com/pricing/

And it looks much more expensive than Google's Map API (checking free app/web): https://developers.google.com/maps/pricing-and-plans/#details

We have 30*25000 monthly limit on Google, 50000 monthly limit on mapbox ...

Maybe we should stick to Google Maps?

[ajakubo1]

Hi, @lechup - for now I'm using OpenMaps from what I know. I have some problems with the build configuration though (not because of OpenMaps, integration with webpack+react is little bit weird to me) :/. I should have something usable by the end of the week.

[lechup]

Ok, cool. Do You mean OSM? Like here? https://switch2osm.org/using-tiles/getting-started-with-leaflet/

What limits does it have? Is it usable? I mean it is not sluggish and loads quite fast?

[ajakubo1]

@lechup yeah, that's the one.

I can see that they suggest using your own data server:

https://operations.osmfoundation.org/policies/api/

And is it sluggish? Well, to tell you the truth I don't feel it's sluggish. The whole thing takes a second to load :).

I think I've handled most problems I had. Things left to do:

• [ ] Sort-out component struture and add code to display pins properly
• [ ] Add a fetch library + a polyfill
• [x] create Redux actions/reducers + appropriate libraries which make life easier (Like Immutable)
• [x] take care of test problems (build fails because of that)

• [ ] Add tests :)

  • [x] Adding tests for DataHolder and SmogMap components

I'll do some of those today I think :).

[lechup]

@ajakubo1 ok, so for first version we just use their tiles service right? Then we could write to API holders whether we should get our solution or not.

I could set up OSM tile service, but still we would need to proxy it through our backend and require token (and timestamp?) to actually do not let to abuse it... http://stackoverflow.com/a/31640941/479931

I do not have any idea what kind of HW do we need for our usage but still it is doable - but increases running costs a bit...

[ajakubo1]

@lechup Yeah, I think we can use theirs for now, but I would have the OSM tile service in the back of my head. To tell you the truth that's something which @zen should consider more (as he would deploy it all, I think ;)).

Anyway, if you have a couple of seconds you can checkout this branch: https://github.com/EnviroMonitor/EnviroMonitorFrontend/tree/adding_map

I have a feeling this pull request might be quite big, but I want to add everything which we might need in futther development.

Anyway, if I add the fetch library, and input some test data, I'll open the pull request for it.

[zen]

Folks, I can set up tile server if that's needed. Just let me know if we run into any perf problems with current setup. Still, that's fine for our global instance, but what about users deploying their own backend/frontend? It can be a showstopper for them :(

[ajakubo1]

For internal usage (I mean one person wants to set it up and has like 2 sensors or something), it should be fine using the global OSM. But - if our project would be used for country-wide-monitoring or city-wide-monitoring, then I guess we would have to suggest setting up their own OSM tile service for performance.

I'm not sure if we should block the page somehow for bigger traffic? Nah, that would be weird, it's open-source anyway :). So anyone can hack it in their own branch.

BTW - I'll check what is the current page size with map fully loaded.

[lechup]

@zen @ajakubo1 Shouldn't we go into way we are hosting all instances? I mean we could support local Smog Alarm Groups:

• allow other pages integration through iframe - map tiles + sensors + stats (token auth)
• allow integration through API (token auth), so user could present his data on Google Map or whatever tile service?

I think it will be much more user friendly than "set up Your own soft infrastructure". What do You think guys?

[zen]

I'm all for hosting it for others, and maybe you are right that if someone chooses to host it himself, it's out of our scope. So, I would assume that we host central instance and allow users to set up their own frontend to use our data. They need to post some sensor data though, to get access to resources.

[ajakubo1]

Yeah, I'm OK with setting up our own server and allowing users to use it, if they are contributing to the overall data scope. That's quite user-friendly, and everyone is getting something out of it :).

Btw. I've just checked and the whole thing weights about 1.1 MB. And half of that is our scripts (well, mostly vendor scripts really). On my modem it takes about 2s to load right now (last time I've checked I've had better connection).

[lechup]

@ajakubo1 1.1 MB - I think we should go with some kind of loader for slower connections. React apps just need some more data at the beginning but later it just works.

[lechup]

Done in #15