Re: [Wish] Use entity-tags for solving out of order PATCH requests

[Sergio Garcia Murillo <[e-mail]>]

Hi Lorenzo

On Wed, Mar 2, 2022 at 6:24 PM Lorenzo Miniero <[e-mail]> wrote:

>
> I think the changes look fine, I just have a couple of comments:
>
>         * What error should be returned in case a PATCH to trigger an
>           ICE restart does NOT contain the "*" If-Match value? The
>           draft says it's a MUST, but doesn't elaborate on the
>           consequences if that doesn't happen. Intuitively a 412 could
>           be returned as well, but the PATCH for ICE restart could
>           actually contains the last valid ETag, and so in that case a
>           412 sounds "wrong".
>

Not sure if ETags are mandated on the client side for non-GET requests, so
while we mandate the If-Match="*" on the whip spec, i think whip
implementations should just accept request without it for ice restarts and
do not return an error.


>         * It's also not clear to me if ETags apply to other requests
>           addressed to a WHIP resource. Thinking of the DELETE to tear
>           down the PeerConnection, for instance, would that need the
>           right ETag to be provided as well? Or would ETags be limited
>           exclusively to PATCH? If so, it might be better to clarify it
>           in the doc to limit the scope.
>

I have the same doubt as before, I don't think it is mandatory from the
HTTP specs to send the etags on the DELETE request, so we should just state
that it is only required for PATCH request.

Best regards
Sergio