Cheapo WMS, Part II

[bhousel]

Like #2409 but uses whoots-js library to calculate the bounding box with more correct math.
Also added a wms.html debug page.

cc @tmcw @lucaswoj @mourner

[lucaswoj]

The CI failure should be fixed by #2610. Can you rebase and re-push? Thanks!

[lucaswoj]

I'm 👍 on merging this pending:

• consensus on what to call the token. There was some debate about this in Cheapo WMS support #2409 (comment) cc @jfirebaugh @tmcw @mourner
• a quick code review of whoots-js. I will do this myself shortly.

[lucaswoj]

What do you think about namespacing the bbox token behind a wms TileJSON scheme, like our tms scheme? #2565

[bhousel]

What do you think about namespacing the bbox token behind a wms TileJSON scheme, like our tms scheme?

I thought about that, but worried that's misleading because we are only supporting a very specific subset of wms servers. We'd likely see people opening issues because they try to add wms://something as a source and it doesn't do what they expect.

[lucaswoj]

My understanding is that {bbox-epsg-3857} only makes sense within the context of wms.

Putting the wms-specific tokens to a wms namespace will increase the self-documenting-ness of the feature and give us maneuvering room to improve wms support in the future.

Our likelihood of getting weird wms support requests is unrelated to the existence of a wms scheme.

[lucaswoj]

cc @yhahn @tmcw @jfirebaugh: As the lead TileJSON contributors, what's your feeling on this?

[tmcw]

I love this PR!

• I don't think we should add this to TileJSON. WMS already has its own metadata stuff and allowing this to be configured WMS-style is an unnecessary can of worms. Also, WMS is not quite a tiling 'scheme', since it isn't tiles, we're just requiring it as tiles.
• {bbox-epsg-3857} seems fine to me.

I see why we might want to scope this bigger, but don't feel like we need to or should. I think what we have here is pretty great.

[lucaswoj]

Ok! Let's 🚢!

[lucaswoj]

... as soon as I finish the hotfix release for #2657

[bhousel]

Just pushed a few commits to bump to the v2 classless library, and added a test.
I'll merge later today when CircleCI goes green..

We should think about how we want to communicate this feature. Should I document it in style-spec?

[bhousel]

merged as
09c0c96 Add test for {bbox-epsg-3857} token replacement
e793682 Bump to WhooTS v2
2ebe717 Add wms.html debug page
4e9a523 Use whoots-js to calculate EPSG:3857 bounding box

[1ec5]

This is being ported to the native libraries in mapbox/mapbox-gl-native#5628.