However, this seems to result in rather large Link headers, and it seems a strange precedent to stuff JSON in headers.
What is the design rationale behind this?
Aren't we better off just linking to a Link Hints document (which could be cached and, with HTTP/2, be pushed)?
Following Appendix A of https://tools.ietf.org/html/draft-nottingham-link-hint-02, the Link Hint JSON structure can be serialized in a
Link
header.However, this seems to result in rather large
Link
headers, and it seems a strange precedent to stuff JSON in headers.What is the design rationale behind this? Aren't we better off just linking to a Link Hints document (which could be cached and, with HTTP/2, be pushed)?