DESCRIPTION

Wait longer on tap data received.

Change-Id: I5279eec0f642688013c9a8d98ca4a7922cae9d42


COMMENTS

author: Jeffrey Roth
date: 2012-02-24 18:36:32.434000000

Patch Set 1: Looks good to me, but someone else must approve



-------------------------------------
author: Kenny Maynard
date: 2012-02-24 18:47:31.572000000

Patch Set 1: (1 inline comment)

Can you provide an explanation for why you want to wait longer? Blocking for 10 seconds seems like a long time to me and if a user really wants to do this there is another getNextMessage() function that allows you to specify the wait time.

Line:90, src/main/java/net/spy/memcached/TapClient.java -> If we push this the comment has to be changed here.

-------------------------------------
author: Kenny Maynard
date: 2012-03-05 18:22:18.070000000

Patch Set 1: I would prefer that you didn't submit this

-1 until there is an explanation for making this timeout greater.

-------------------------------------
author: Andreas Short
date: 2012-03-05 18:35:31.053000000

Patch Set 1:

Fundamentally, I think it's silly to receive a null once a second.  The rationale for making it longer is that heuristically, we've seen nulls from this followed by data.   It'll reduce them by having a longer timeout.

It seems like having no timeout or a very long one makes much more sense here.

Why do you think 1 second is good?

-------------------------------------
author: Kenny Maynard
date: 2012-03-05 18:47:39.700000000

Patch Set 1: Looks good to me, approved

Long default timeouts means long thread blocking. When you write the code your supposed to ignore the NULL's anyways. Also, it's not that I think 1 second is better than 10 seconds, it's just that I don't see an argument for why one is better than the other.

With that said I think a timeout of 0 is bad. We shouldn't block the thread indefinitely.

-------------------------------------
