[zeromq-dev] [otish] "Why ZeroMQ"

[Pieter Hintjens]

On Tue, Jul 27, 2010 at 1:32 PM, Andrew Hume <andrew at research.att.com> wrote:

> there is long established use of
> - naming with respect to the computational entity
> - for things to be plugged together: input and output

What seems to be emerging is that the socket type names make sense
only within the context of the pattern.

I.e. for a pipeline pattern, IN and OUT.  For the request-response and
peer patterns, there is no discrete input or output.  For pubsub,
there is, and it would be consistent to call the publisher socket OUT
and the subscriber socket IN.

The problem we identified some time ago was that the patterns (the key
to understanding 0MQ at this level) are not explicitly named.

So we have to make that mental leap to identify and learn the
patterns, and it's hard for many users to do that.  I don't believe
there are masses of 0MQ developers who silently "get it" and don't go
through the pain.

Did anyone here actually find the socket type names unsurprising and
easy to understand?

-Pieter