- From: Arvind Jain <arvind@google.com>
- Date: Sat, 23 Aug 2014 07:04:27 -0700
- To: Sigbjorn Finne <sof@opera.com>, Jonas Sicking <jonas@sicking.cc>
- Cc: Ilya Grigorik <igrigorik@google.com>, David Benjamin <davidben@chromium.org>, Adam Barth <abarth@chromium.org>, public-web-perf <public-web-perf@w3.org>
+jonas I have uploaded a new draft to not add Beacon-Age if age is zero. https://w3c.github.io/web-performance/specs/Beacon/Overview.html#sec-processing-model Arvind On Tue, Aug 19, 2014 at 4:59 AM, Sigbjorn Finne <sof@opera.com> wrote: > Den 18.08.2014 20:07, skreiv Ilya Grigorik: > >> On Mon, Aug 18, 2014 at 10:59 AM, David Benjamin <davidben@chromium.org> >> wrote: >> >>> On Mon, Aug 18, 2014 at 1:51 PM, Ilya Grigorik <igrigorik@google.com> >>> wrote: >>> >>>> Yes, until there is a bug and some UA omits it by accident (or >>>> intentionally), at which point I'm back to UA detection: if X UA then no >>>> delay, if Y UA then I don't trust the timestamp... at which point, I >>>> guess >>>> I would need to move that detection into JS-land and pick the method >>>> that I >>>> trust. >>>> >>> >>> Firefox appears to already ship sendBeacon without the Beacon-Age header >>> anyway. (I imagine their implementation predates the header.) >>> >> >> Yes, but they landed a fix for that since and first shipped version had no >> delay. But your point still stands. >> >> tl;dr: I'm ok with omitting it. Just trying to think of cases where we >> might regret this decision later :) >> > > If the requirement is that an implementation is only allowed to include > "Beacon-Age: N" for N > 0, that might reduce your concerns? > > --sigbjorn
Received on Saturday, 23 August 2014 14:04:56 UTC