Id should not be mandatory

[hobofan]

While most objects can be Identified by an Id, some objects (e.g. on-the-fly statistics) can't be.

That's why the schemas should not have id as a required parameter:
https://github.com/storeness/api-blueprint/blob/master/schemas/single.json#L17
https://github.com/storeness/api-blueprint/blob/master/schemas/list.json#L30

[MichaelHirn]

I removed the schemas completely as they turned out to be very out-of-date with the use of attributes. Schemas get created now right from the Attributes and the problem with required parameters such as id disappear. Check out 856a618

Schemas would have one advantage, that they secure a standard on a higher order. But in practice this turned out to to be too restrictive. Again the answer is hercule with Attributes. Hercule is missing at the moment the feature to transclude transclusion variables. With that we could build higher abstractions of request/response Attributes/Schemas, with which we could guarantee e.g. that a list response has a type of "list.*". But as far as I experienced it, this advantages can't justify the time that would go into providing this with hercule