RE: Response Code 2501

["Gould, James" <JGould@verisign.com>]

Scott,

I don't see value in the 2501 response, since as you indicated the server is
not supposed to send an unsolicited response.  Even if the server sends the
response before closing the connection, the client will most likely not be
in a state to read it.  From a protocol perspective, simply closing the
connection would be more straight forward and would not introduce
unsolicited packets.  

Jim 

> ----------
> From: 	Hollenbeck, Scott
> Sent: 	Monday, August 12, 2002 10:52 AM
> To: 	'ietf-provreg@cafax.se'
> Subject: 	Response Code 2501
> 
> While working through the new state diagram to be added to the EPP core
> document, I had to ponder idle timeouts and they're addressed.  Right now
> there's an error code defined that allows a server to notify a client of a
> timeout situation:
> 
> 2501 "Timeout; server ending session"
> 
> Is this error code really needed, though?  Servers aren't supposed to send
> a
> response to a client without having first received a command, so if a
> client
> dies or creates a session that's been alive for "a long time" the server
> shouldn't be sending this as an unsolicited response.  It seems to make
> more
> sense in this case for the server to just close the connection, and if the
> client tries to write something it'll find the connection closed.
> 
> Thoughts?
> 
> FWIW this and the TCP header thing are the last two things I need to
> address
> before being able to release the updated documents.
> 
> -Scott-
> 
>