- From: Andrew Wilson <atwilson@google.com>
- Date: Mon, 28 Jan 2013 09:56:15 +0100
- To: public-web-notification <public-web-notification@w3.org>
- Message-ID: <CAArhhitxB8KaKpzu+EX1tfbfBFS-=_mGs9tybyxLb7kKV7_yTw@mail.gmail.com>
So, it was pointed out to me that there's no longer any show() API, and that my experience with the original webkitNotifications API has poisoned my brain :) Seems the intent of the specification is to never throw an exception for permissions errors, but to always report them through the error callback - I'll follow up with the WebKit folks. On Mon, Jan 28, 2013 at 9:45 AM, Andrew Wilson <atwilson@google.com> wrote: > Hi all, > > As implementations of the notifications spec have landed in WebKit and > elsewhere, I wanted to clarify when exactly the permissions check should be > done. > > My reading of the specification ( > http://www.w3.org/TR/notifications/#showing-a-notification) is that the > permissions check should always (and only) be done when invoking show(). I > think that WebKit may instead do a permissions check in the constructor, > which I suspect is incorrect and may lead to some browser compatibility > issues. Additionally, since the application can call show() at an > arbitrarily later time from invoking the constructor, only checking > permissions in the constructor is incorrect since the user may have changed > his permissions between the constructor and the call to show(). > > Am I correct in my reading of the specification, and do we perhaps want to > change the spec to do a permission check in both the constructor *and* > show()? > > -atw >
Received on Monday, 28 January 2013 08:56:45 UTC