Re: [Webpush] WGLC for draft-ietf-webpush-protocol-05

[JR Conlin <[e-mail]>]

I noted my opinion in the follow up. Basically, while the spec goes into
great detail about other aspects of the protocol, sets are left mostly as
"they may exist", and "it's up to the PS and UA".  I will also admit that I
was initially confused because I hadn't noticed it was only between UA and
PS, but a second read caught that.

I am fine as is, because I don't think that outside groups will be impacted
by my concern, nor do I believe that they would fail to implement a
solution if they misunderstood it.
On May 31, 2016 4:25 PM, "Brian Raymor" <[e-mail]> wrote:

>
> On May 31, 2016, at 12:01 PM, JR Conlin <[e-mail]> wrote:
>
> Thanks for your feedback.
>
> > There are some aspects of the published spec that are optional and seem
> a bit "hand wavey"* (e.g. "subscription sets")
>
> How do you feel this could be improved? What details are missing from your
> perspective?
>
> > I believe that the recommendation for expired or removed push
> subscription endpoints was to return 410, not 404 (see section 7.3). I note
> that 7.3.1 suggests 410 for expired subscription sets, so it might help if
> things were consistent.
>
> Good catch.
> 7.3.1 should be 404 and not 410.
> 410 (Gone) is returned when a push service decides to cease delivery
> attempts before the TTL expires.
> 404 (Not Found) is returned when a push services expires a subscription or
> set.
>
>
>
>
>
>
>