Just discussion

Nobody has claimed this yet.

@flippmoke this worked 💯 on my first try. Pretty slick. I just thought I'd open a ticket here where we can discuss what's happening next.

- Right now, is the output one line per feature per zoom level? Or one line per **tiled**-feature per zoom level? That is, are you doing simplification and clipping yet?

- I guess the reduce functions for actually building vector tiles are still to come? Do you think that the details of how that works can be kept distinct from decisions about the architecture for storing the outputs of the map step?

- From the perspective of figuring out how the storage architecture will work, I do think that we'll probably want to generate a property-store that is separate from the clipped, simplified geometry store. I guess what would be cool is if `m2f` were to set the properties aside in another file (?) so that they aren't duplicated in each zoom/tile of the `m2z` output.

- Having a distinct property store means the reduce step will need to be able to do lookups based on feature ids. Do you have any thoughts about how to handle geojson features that don't bring their own ids? Or do bring ids that aren't numbers?

No source file, test, or entry point is named. Review the discussion around m2f, m2z, vector-tile reduction, and property storage; this issue needs a decided implementation direction before a newcomer can start.

Written by the indexing model from the issue text.