- From: James M Snell <jasnell@gmail.com>
- Date: Wed, 6 Nov 2013 08:18:59 -0800
- To: Julian Reschke <julian.reschke@gmx.de>
- Cc: Mark Nottingham <mnot@mnot.net>, IETF Apps Discuss <apps-discuss@ietf.org>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
On Wed, Nov 6, 2013 at 7:36 AM, Julian Reschke <julian.reschke@gmx.de> wrote: > On 2013-11-06 16:27, James M Snell wrote: >> >> Whether and how a server surfaces links is undefined by this spec. There >> are cases where exposing the link would be wholly unnecessary. There are >> also cases where the server could significantly alter or augment the >> established link with additional detail. My goal with this spec is only >> to define the method characteristics. > > > I have no problem with that, but in the end something needs to come out that > is usable. > > I agree that the server may want to alter/augment the links. Maybe the > response body then should reflect what was set? > Thus far I haven't had any practical need for this... so, again, I would go back to the question about viable, non-theoretical use cases. > >> Regarding the main question about whether or not to include target >> attributes in the uniqueness constraint, the question becomes: which >> target attributes should be considered? > > > All of them? > "All of them" becomes impractical for the reasons I've already given, particularly without clearly defined, viable, non-theoretical use cases. That said, here's a compromise: Let's expand the uniqueness constraint to include anchor, hreflang and type. Doing so ought to cover the vast overwhelming majority of the possibly viable use cases. I can also say that the server MUST consider the remaining target attributes to be significant in that, *if* the server chooses to surface those links in some representable manner, the most recently received values for those MUST be included. Is that better? (btw, getting back to your earlier question about the anchor attribute, I had honestly completely forgotten about the anchor target attribute! I've never actually seen it used in any practical sense so it had completely slipped my mind... to answer that particular question, yes, the server MUST consider the anchor attribute as being significant. I'll add a few more examples) - James > Best regards, Julian >
Received on Wednesday, 6 November 2013 16:19:46 UTC