[zeromq-dev] Moving to 0MQ/3.0, yes or no?

[Pieter Hintjens]

On Sun, Apr 10, 2011 at 7:22 PM, Martin Sustrik <sustrik at 250bpm.com> wrote:

> My point now is that the discussion around 3.0 proved that no substantial
> fixes to 0MQ design (say removing identities) can be done even if major
> version is bumped.

If you want 0MQ to be a success, then IMO you need to find some middle
ground between "it's a minor version upgrade and therefore the API
cannot change one iota", and "it's a major version upgrade and
therefore I'm free to rip the heart out of 90% of real applications".

You're hostage to your users (and this will get worse as 0MQ gets more
successful), and you are working in an open source model. Pain is not
profitable here. My advice is to make small, incremental, and valuable
changes, with upfront discussion, and not be obsessed by the release
policies, which have not helped here. It's been painful to watch the
flipping between total caution, and total disregard, both extremes
missing the goal by quite a margin. Focus on giving people real value,
on solving current problems, and on creating broad discussions, and
it'll all work.

Also, it is a bad idea, IMO, to describe current 0MQ semantics as
"broken". You are basically telling people, "I was incompetent last
year, but you can trust me next year". Not great marketing.

-Pieter