[EMAIL PROTECTED] (Jonathan Naylor) writes:
>Hello Julian
>On Thu, 21 Jan 1999, Julian [ISO-8859-1] Muqoz[ISO-8859-1] Dommnguez wrote:
>>
>> I have been looking the ax25 source code, and didn't find nothing wrong.
>Thats pleasing.
>> I think that this was done at the beginning, because timers were not
>> implemented, but now there is no reason for doing poll in the last frame of the
>> ax25 tx window; most of the implementation do not, and if that's causing
>> problems the best is remove it (imho).
>The reasoning for attaching a poll to the last I frame in a window was
>something me and John G8BPQ worked out (his code does it BTW). The idea
>behind it was to improve performance, normally the receiving station would
>wait T2 from last hearing a frame from the remote end, before replying
>with an RR. The logic we used goes a little further, if we have sent a
>full window and are therefore not going to send any more to him (except
>retries), why not remove T2 from the equation by asking for an immediate
>reply from the remote station.
>I don't actually see how it could break anything to be honest.
It does really. I had lots of trouble with tcp/ip connections to the LAP
here using the DAMA protocol. When the delays are getting larger the
snowball effect occurs. After disabling the 'poll after' behaviour the
connections are much more stable and I don't see snowball effects any more.
>The original textual description of AX.25 was very hazy on that point and
>so we felt able to add a little :-) It is such issues that make me a
>strong believer in SDL diagrams. If anyone wants to read a bad textual
>description of a protocol, try the LAPB/X.25 specification from the CCITT.
>I managed to get a university dissertation out of that particular mess.
Please continu this discussion, there must be something changed in the
kernel sources.
73,
Gerhard.
--
Gerhard Ahuis JO32EO It's good to be independent
[EMAIL PROTECTED] 145,550 MHz Linux the Ultimate HAM OS
Unsolicited advertisements subject to $1000 consulting fee.
