On Tue, Jan 7, 2014 at 3:46 AM, Anne van Kesteren <annevk@annevk.nl> wrote: > On Mon, Jan 6, 2014 at 6:54 PM, Alex Russell <slightlyoff@google.com> > wrote: > > I admit the bit where they hand a URL in is a bit of a fudge. I noted > > somewhere (can't' recall where) that they should have a generic (string? > > JSON?) payload and let apps do as they will (modulo size limits for the > > payload). But with that problem of generality solved, the overall thing > > becomes transport agnostic, and I should point out, with respect to the > > underlying message delivery system, started that way. > > > > This is a non-issue WRT their current draft, IMO. > > I agree with Tim that punting on what protocol is going to be > implemented across actors would be unacceptable. Whether we define > them in the same draft or via reference is a technicality not worth > worrying about, but there has to be a baselines that is guaranteed to > work. > This feels fanciful, at best. There are existing underlying transports. They are (or can be) specified elsewhere. They are also interchangeable under this API. We must not prevent this API from accommodating current reality (thereby robbing it of purpose), nor can we expect everyone to sign up to a new transport for no reason other than that we asked them to. Lets get concrete: is it acceptable to the folks clamoring for a specified transport if one is specified elsewhere?
Received on Tuesday, 7 January 2014 13:09:35 UTC