Use "runtime schema" instead of "full schema"

[martinbonnin]

Iterating on the naming and definition. Feedbacks welcome.

[benjie]

I don’t think this accurately conveys the intent. Today many people have a source SDL with directives and extends that are combined with resolvers and what not to build their runtime schema. This runtime schema is introspected and printed for use by clients/tools/etc; that, to me, is the “runtime schema description“ (typically in SDL).

When you have the runtime schema printed including the introspection schema and its entry points plus builtin types and directives, I would call that the “complete runtime schema description” (and would require that it use the schema keyword even if the root types had their default names).

[martinbonnin]

@benjie I agree with you but also "complete runtime schema description” is a mouth full. I don't expect people will memorize that. (Also schema description is a bit ambiguous as it could be the description of the schema definition)

Generally I think there's some ambiguity what a "schema" is:

• Is this a runtime artifact, including resolvers
• Is this a build artifact: only the descriptions of the types (and capabilities?), without resolvers

Interestingly, there's no definition in the spec index for schema so this is probably the first thing to agree on.

Today many people have a source SDL with directives and extends that are combined with resolvers and what not to build their runtime schema

What difference are you making between a "runtime schema" and a "service"?