Issue 368 in redis: Timeout set to 0 does not disable timeout, makes it more angry and wrong.

Status: New
Owner: ----
Labels: Type-Defect Priority-Medium

New issue 368 by zed.shaw: Timeout set to 0 does not disable timeout, makes
 
it more angry and wrong.
http://code.google.com/p/redis/issues/detail?id=368

What steps will reproduce the problem?
1. Set redis.conf timeout 0, which should disable timeouts.
2. Connect and hang out for a period of time.
3. Watch redis disable idle clients, despite timeout 0 saying it should  
never do this.

What is the expected output? What do you see instead?

Redis does not close "idle" connections.

What version of the product are you using? On what operating system?

2.0.0. So far reproduced on Debian and ArchLinux.

Please provide any additional information below.

Ultimately there's no logical reason why a server that is intended to  
function like a database server should be closing connections at all.  The 

default should be that it tries very very hard to keep clients connected,  
and then people who want it the other way set an option.  So far I've tried
 
4 different redis libraries in 3 different languages and all of them have  
to go through fairly horrible *****1s to be able to recognize that the socket
 
is closed and retry the connection.  Otherwise they just blow up huge and  
then your application is forced to constantly check for exceptions on every
 
potential request.

Since it's unlikely that this will change, at least make the timeout 0 work
 
and potentially warn people as to why there is a timeout in the first place
 
so that we don't shoot ourselves in the foot trying to abuse some unclear  
design flaw.