Add RFC for incremental delivery

[robrichard]

No description provided.

[sungam3r]

I co not see this in the example above.

[robrichard]

The content-type: multipart/mixed; boundary="-" header should be in the HTTP response headers. The example is the HTTP response body. Each part of the response body contains it's own set of headers, but the header defining the boundary needs to be on the HTTP response, not in the body.

[sungam3r]

it's worth mentioning

[jaydenseric]

Pretty interesting; I'll have to think about how this fits in with the GraphQL multipart request spec 🤔

@robrichard have you considered how they could be unified? E.g. someone using @defer and @stream directives alongside file uploads.

[robrichard]

@jaydenseric I don't think there will be any conflict with the multipart request spec, since that spec only defines the request format, and this spec only describes the response format. Should be able to use @defer and @stream directives alongside file uploads without any issues.

[IvanGoncharov]

I think it ready to be merged.
On last GraphQL WG we agreed to add support @stream/@defer across the stack.

[benjie]

Must the boundary be "-" specifically? Could other boundaries be used?

[robrichard]

I think we should support arbitrary boundaries. fetch-multipart-graphql currently only supports "-" but I will update that and this document to allow any boundary.

[IvanGoncharov]

@benjie @robrichard Sorry, I didn't notice this comment before merging.
On the last "GraphQL over HTTP" WG we decided on merging and just forgot to actually do it.
Hope it didn't disrupt your workflow and @benjie suggestion can be addressed as separate PR.

[benjie]

I've filed an issue here: #135 (@IvanGoncharov if you could assign it to @robrichard that would be helpful).

[acao]

@jaydenseric to clarify, i think what @robrichard is saying is that non-file fields can use @defer and @stream with multipart alongside your multipart file spec. in slack today @robrichard confirmed internally they have a client that is successfully able to do as such.

@jaydenseric speaking of which, would love to help advance file spec, and maybe we can revisit how @defer and @stream impact your spec if at all, and revise them if/as needed? I'm going to open an issue in your spec repo to discuss this further!