computed => isolate

= pull applyComputationsFromSDL into Subschema class

* Subschema class was initially useful when performing delegation outside of wrapSchema context. The class constructor performs some of the preprocessing logic done by wrapSchema. It can be also used to do preprocessing based on the SDL.

= short splitFieldsFromSubschema further => isolateFields

= run isolateFields automatically when using stitchSchemas.

also remove remnant of named endpoint code

Initial suggestions

…ls into gm-dynamic-merge-fields

…h fieldSelectionSet isolation

On the other hand, will ignore the price and weight given by its own query in favor of the Product service, so may give different results when queried directly and when queried via Gateway.

So isolation definitely needs to go behind a flag as opt-in. Name of flag might be `fieldSelectionSetsViaGateway` or something like that to describe the different behavior rather than the internal implementation. Anything sound catchier?

no need to modify subschemaConfig.schema when isolating schemas, this should all be done by transforms

…ls into gm-dynamic-merge-fields

…ls into gm-dynamic-merge-fields

computed => isolate

= pull applyComputationsFromSDL into Subschema class

* Subschema class was initially useful when performing delegation outside of wrapSchema context. The class constructor performs some of the preprocessing logic done by wrapSchema. It can be also used to do preprocessing based on the SDL.

= short splitFieldsFromSubschema further => isolateFields

= run isolateFields automatically when using stitchSchemas.

also remove remnant of named endpoint code

…h fieldSelectionSet isolation

On the other hand, will ignore the price and weight given by its own query in favor of the Product service, so may give different results when queried directly and when queried via Gateway.

So isolation definitely needs to go behind a flag as opt-in. Name of flag might be `fieldSelectionSetsViaGateway` or something like that to describe the different behavior rather than the internal implementation. Anything sound catchier?

no need to modify subschemaConfig.schema when isolating schemas, this should all be done by transforms

…ls into gm-dynamic-merge-fields

…ls into gm-dynamic-merge-fields

default for federate argument for directive \@requires can be set by each schema

finish changing computed => federated and static => non-federated