Ah, got it. On Sat, Aug 5, 2017 at 3:23 AM, Martin Thomson <martin.thomson@gmail.com> wrote: > On 5 August 2017 at 10:55, Jana Iyengar <jri@google.com> wrote: > >> Martin, >> >> I think the idea of sequential Push ID is a good one. You have to specify >> a MAX_PUSH_ID, without which the client won't know how far out to expect >> CANCEL_PUSHes can be... this also requires then a MAX_PUSH_ID_UPDATE >> message which can be used to move the MAX up. >> >> One note: >> >> >>> PUSH_PROMISE is sent by the server, and CANCEL_PUSH may be sent by the >>>>>> server. The text currently says that CANCEL_PUSH on a non-existent Push ID >>>>>> results in error, which does not work with the possibility of reordering >>>>>> between PUSH_PROMISE and CANCEL_PUSH. >>>>>> >>>>> >>>>> That's just a bug. >>>>> >>>> >> I don't think so. The spec allows it, and I can easily see it happen in >> practice. >> > > By bug, I meant that I made a mistake in the PR and forgot about the > possibility of reordering, that's all. >
Received on Saturday, 5 August 2017 10:25:03 UTC