Tom MacWright

Sure, interesting idea. Looking around I see: - http://www.public.iastate.edu/~hofmann/data/titanic.html - `Class, Age, Sex, Survived`, string values, 2201 lines - https://ww2.amstat.org/publications/jse/v3n3/datasets.dawson.html - `Class, Age, Sex, Survived`, integer values, 2201 lines -...

Instead of being the Decider on this one, I decided to write a notebook, [rdatasets](https://beta.observablehq.com/@tmcw/rdatasets), that provides access to the titanic datasets, as well as many others, and some documentation...

Started look at this a bit - at least for generating a tsvector column in a schema definition, doing something very similar to the citext example of an extended column...

Wow, that is the first I'm learning of the generated specifier 😆. So thanks for the tip, but yeah - not sure how to support that so far!

👍 Adding that Postgres also supports `WITH` with SELECT, INSERT, UPDATE, or DELETE queries.

This could be annoying, but it also has the potential to be a nice and simple community feature: I think good design can avoid making it intrusive, and @fuddl's sketch...

So ideally this includes mustache for templating, and htmlsanitize. @danzel do you have time to work on this?

@ddelemeny if you have any ideas about improving the TileJSON spec, [they are welcome in its open repository](https://github.com/mapbox/tilejson-spec/issues).

sure, would be interested in your solution, a smart solution that doesn't just do random in bbox->expensive point-in-polygon would be interesting

Stopgap solution that I'm okay is the bbox functionality I just pushed to master.