This is for backward compatibility with existing supergraphs, bolting @core support onto the @link reading. It does not do anything to support fed v1 supergraphs.
While @core is not used in new graphs today, it is supported by the JS code, and we have a lot of tests throughout our systems that rely on using @core with fed v2.
Additionally, because the implementation of API schema generation is the same between fed v1 and fed v2, supporting @core will let us land it in the router without a fallback to JS for fed v1 supergraphs.
All new code that supports @core mentions @core literally in a comment so we can rip it out easily when we migrate over customers and our tests in the router.
Marking the comments resolved because we'll follow the JS code rather than the spec (or good sense 😄 ). this just needs a rubber-stamp approval I'd say
This is for backward compatibility with existing supergraphs, bolting
@core
support onto the@link
reading. It does not do anything to support fed v1 supergraphs.While
@core
is not used in new graphs today, it is supported by the JS code, and we have a lot of tests throughout our systems that rely on using@core
with fed v2.Additionally, because the implementation of API schema generation is the same between fed v1 and fed v2, supporting
@core
will let us land it in the router without a fallback to JS for fed v1 supergraphs.All new code that supports
@core
mentions@core
literally in a comment so we can rip it out easily when we migrate over customers and our tests in the router.