Re: so_rerror

[christos%zoulas.com@localhost (Christos Zoulas)]

On Nov 6, 12:25am, roy%marples.name@localhost (Roy Marples) wrote:
-- Subject: Re: so_rerror

| But that's they key point isn't it? Two participating hosts.
| 
| My argument is that if I create a client/server setup using UDP on one 
| host only, I would expect the receiver to log an error if it could not 
| receive something the sender claimed it sent because at this point it's 
| all self contained.
| 
| Protocols such as TCP and UDP are designed for on the wire, but in this 
| instance there is no wire.

Yes and the prevailing opinion from others is that the "medium" should
not affect the protocol behavior. I.e. UDP connections between two hosts
and the same host should behave the same way. As I said before I am not
that much a purist about it and I am ok to report more errors when they
are available. The issue is backwards compatibility, and changing the
existing behavior should be done carefully and deliberately.

| > Therefore, to my mind it makes no sense to insist that a particular
| > type of packet loss event (receive buffer overflow) *must* always be
| > reported, as there are other packet loss events which will never be
| > reported to the receiver.
| 
| With more than one host I agree entirely, but for a single host I disagree.
| 
| > I'm with Christos: if you want some sort of reliability and still
| > insists on using UDP, you need a protocol layer above UDP, with
| > sequencing, retransmissions, performance enhancements with large
| > windows, flow control, fairness, etc., so ... perhaps the insistence
| > on UDP isn't well-thought-out and instead TCP begins to make more
| > sense as a choice?
| 
| This is the thing I don't understand though.
| I'm not trying to make anything more reliable.
| I'm advocating that we should report things that are unreliable - we got 
| something but could not deliver it.
| How the unreliable is solved I leave as an exercise to the reader, but 
| we should at least empower them with knowledge that it happened so they 
| can try to fix it if they so want.

I agree as I wrote above, the issue is to do it in a non-intrusive way or
make sure that nothing breaks... And we discover more and more things
breaking.

christos