[JTS] Support for indexes

I like the spec very much. What I am currently missing is the possibility to add information about indexes. This would be especially important for me, when it comes to unique-indexes for more than one column. I could imagine the following:

"indexes": [{
  "fields": ["email"]
},
{
  "fields": ["name", "birthday"],
  "type": "unique"
}
]

where “type” is optional and might be at least “unique”. Maybe it could be extended to “type” like “bitmap”, “btree”, “partition”, … but well, ok, than we it might be too database-vendor-specific.

[akariv]

I agree that indexes are used for optimisation in a certain use case, and do not describe the data.

On the other hand, uniqueness does describe the data - and should be part of the constraints IMO.

I would possibly go with adding a new constraint (uniqueWith?) which would be defined on one field and would include names of other fields in the schema which, when combined, provide unique n-tuples of values.

[rufuspollock]

@pwalsh agree on WONTFIX though wonder if this would be worthy of a pattern to document the OKI approach to how this gets specified. In fact, I think a general pattern for adding sql specific info to JTS would be useful - though def a separate issue!