Re: RFC3080 MSG/RPY Window

> argghh... I overlooked this. Sure, this clarifies greatly. However, I
> still do not get if there is a windowing mechanism for msgno. I assume
> even if there is none, a client should be clever enough not to send
> 10,000 messages without any reply - but this is not explicitely
> outruled. My apologies if I am again missing the obvious, but I will
> make sure I understand things crystal-clear as in my experience only
> this helps to avoid (at least some) later security and interop issues.

never use "windowing" or "seqno" in the same sentence with "msgno".

the first two terms are used by beep in order to avoid starvation. the third
term is used by the application so it can correlate requests and responses.

there is nothing that explicitly prevents one side from sending lots of requests
and not waiting for replies. the same is true for applications that use tcp
directly. this assumes, of course,that there is infinite buffering in the layers
below and/or zero-latency end-to-end.

if you look at the way that syslog uses the ANS messages, that's exactly it's
strategy. however, there is this little issue of getting end-to-end
acknowledgements back to the application-layer...

/mtr