[nanomsg] Re: Generalizing pubsub distribution

  • From: Isaac Peterson <ridler77@xxxxxxxxx>
  • To: nanomsg@xxxxxxxxxxxxx
  • Date: Sun, 13 Dec 2015 19:34:48 -0800

I am also interested in this topic. For my normal deployed runtime,
dropping messages is appropriate for my application. However, during
regression testing on VMs, deterministic behavior is essential. It's very
difficult to write system tests against behavior that is message driven
when message dropping varies between runs.

Being able to set a flag and run in a deterministic way would greatly
simplify my system testing. Currently I have to throttle the input such
that nothing gets dropped, but that makes the test run longer
unnecessarily, based on the slowest host timing.

Cheers!
Isaac
On Dec 13, 2015 6:02 PM, "Carlos Pita" <carlosjosepita@xxxxxxxxx> wrote:


On Dec 13, 2015 5:31 PM, "Robbin Carlson" <robbin.carlson@xxxxxxxxx> wrote:


I wonder if a message bus is the appropriate complete solution for this
application?

Do you mean the bus sp in nanomsg? I think it's lossy for the same reason
pubsub is. Moreover, everyone gets everything. So I don't think it's the
right pattern for my problem. But maybe I'm not following you.

Cheers
-
Carlos

Other related posts: