[NOT READY] Add initial support for FileGDB format

[springmeyer]

• Currently only supports single layer FileGDB
• Proper support is dependent on solving Improve handling of multilayer formats #72

[coveralls]

Coverage decreased (-1.06%) to 89.34% when pulling 30b297d on moar into b1135dd on master.

[coveralls]

Coverage decreased (-1.06%) to 89.34% when pulling 30b297d on moar into b1135dd on master.

[rclark]

@springmeyer I would say hold on this while we take on #74, or, if you want to, you could follow the patterns that are emerging in that PR to write an FGDB module into the nothing-to-see-here branch.

@GretaCB and I have chatted about this and think that to address #72 we're likely going to need to do some significant restructuring of the output from omnivore.digest(). The current plan is to move on the modularization first, aligning with the current metadata output format, and then take on what'll likely be a very breaking change to better support multiple layers. Sound good?

[GretaCB]

@rclark this is the ticket regarding tilejson spec I mentioned yesterday when we were chatting layers. Not sure if either will impact each other if we change the metadata output of mapnik-omnivore.

If we change the output, I see this affecting:

• mapbox-studio
• tilelive-omnivore

[rclark]

I think that tilejson is one thing, omnivore output is another. That said its nice to align where possible. I think it would be a good exercise (in the context of making multi-layer adjustments) to more thoroughly document the format of omnivore's output. There are examples in the readme, but we can/should be more concrete and explicit.

[GretaCB]

@rclark agreed. mapnik-omnivore needs more documentation and explicit reasoning for how we go forward.

The vector_layers property in tilejson spec is relevant to mapnik-omnivore layers since that is what the Style UI in mapbox studio uses to read layer info (via tilelive). But tilejson seems more of a consumer of mapnik-omnivore metadata, and agreed they should be considered separate from one another.

[springmeyer]

Yes, lets pause on this. Can return to the idea once other work is in place - refs #72 (comment)