"Budel" <[EMAIL PROTECTED]> writes:
> Or as in:
> 6.13.2 Time Slot Rules
> 1. Successive Beginning of Slot frames (BOS) must be separated by at least
> 25 ms.
> 2. Devices responding to BOS frames must commence transmission of the first
> BOF (X�FF�) of the 11
> BOFs required in NDM within 10 ms and complete the response within 70 ms.
OK, so it must complete the response within 70 ms, that means we must add a
little since the other side uses some time to receive the data. IrDA frames
are however not fully in flight, and only a fraction of a bit is actually
in the air. That means that when the last bit is sent, it is actually
received at the same time at the other side. So that leaves us with the
delivery latency which is just a few ms at most.
> Does not look ambiguous to me.
Not to me either, but there must be a reason why so many stack have
problems with keeping this time limit.
> Assuming the measurements of irdadump are correct, there seems to be a
That is a possibility, but I really don't think so. Irdadump measures the
time difference between the last two received frames, so there should
really be no problem with it. I take the timestamp in userspace, but I
think it might be possible to poll the kernel for the timestamp to get even
more accurate measurments.
> timing problem in those stacks. Maybe the developer of the stack of those
> machines can offer some explanation?
I have tried to ask, but nobody seems to answer that question! Maybe I
should try and ask one of the gurus directly! Well I'm actually meeting one
of them next week. I'm going for a second visit to HP labs to talk about
IrDA and other interesting stuff.
-- Dag
--
/ Dag Brattli | The Linux-IrDA Project /
// University of Tromsoe, Norway | Infrared communication for Linux //
/// http://www.cs.uit.no/~dagb | http://www.cs.uit.no/linux-irda/ ///
