This other issue requires further discussions. In any case not blocking this change-request of ETag.

Let's continue on https://discourse.openehr.org/t/return-minimal-204-no-content-is-confusing/1056/17 and on SPECITS-50

Prefer: return=uid - that would work for me, thoughI feel uneasy that ‘minimal’ is left completely undefined.

, what we can do is to introduce a new Prefer: return=uid request header or something similar, which will instruct the server in returning only the unique identifier in the body (content type text/plain), beside the standard headers. Or we could specify an extra response header openEHR-uid to be sent always upon POST/PUT/DELETE.

Much as I hate the double-quotes, I have to accept that the Gods of IETF say no. I do think it is important that we come up with agreed minimal representations for calls where at least the ehrID, compositionID or templateID are returned. Honestly the 204s are really confusing for loads of their-party devs, not just hackers like me, and digging into the Headers to find the ehrID of the new EHR you just created is a pain, and hard to explain.