Feedback on draft-gregorio-atompub-multipart-00

[Aristotle Pagaltzis <pagaltzis@xxxxxx>]

Hi,

I have mixed feelings over this one. In principle, I like the
idea, and the spec itself is well written.

However, it seems to completely ignore RFC 2387 beyond
superficialities, to the extent that supporting this spec
precludes an Atompub implementation from supporting any other
`multipart/related`-based functionality, because it specifically
requires the server to disregard the relationship of parts as
defined by RFC 2387 along with all facilities available to
express different relationships.

The spec should instead be written such as to advise clients how
to construct proper two-part `multipart/related` entities in
accordance to the spirit of RFC 2387, and not any pass comment on
how a server might interpret `multipart/related` entities other
than those which match this particular use.

That leaves the door open for using `multipart/related` in the
atomic creation of multiple interlinked Atom Entries, f.ex.

Other nits include ignorance of the cid: URI scheme and the use
of boolean markup, but they’re not my main complaints.

Regards,
-- 
Aristotle Pagaltzis // <http://plasmasturm.org/>