[rabbitmq-discuss] Roadmap Questions

[Matthias Radestock]

Darien, Keith,

Darien Kindlund wrote:
> After doing some functionality comparisons between v1.5.x and v1.6.x
> branches, I have some basic questions about the upcoming release
> schedules.

There are quite a lot of changes going from v1.5.x to v1.6.0. We don't
normally let that many changes accumulate between releases, but our aim
was to get a stable version into major OS distributions, such as Ubuntu,
Debian and Fedora (as of yesterday!), which is quite a protracted
process. Now that's all done we are completely focused on getting v1.6.0
out.

> Specifically, I'm looking for a stable/tested (in other words,
> released) version of RabbitMQ that supports the following features:
> 
> 1) Basic QoS (allowing consumers to set the prefetch window to any
> number, preferably 1)
> 2) Basic prioritization (allowing producers to set a priority of 1
> through 10 [according to the spec?]) which RabbitMQ attempts to honor
> on a best effort basis
> 3) Clean compatibility with OpenAMQ's C library (aka WireAPI?) -- or
> clean compatibility with *some* type of C-based library that is
> compatible with a recent version of RabbitMQ (maybe:
> http://hg.rabbitmq.com/rabbitmq-c/ ?)
> 4) [Ideally], headers exchange support -- although I can make due
> without this feature in the near term

#1 and #4 will be in v1.6.0.

#2 is fairly high up our todo list, but won't happen until after the new
persister has been released since it touches areas of the code which
change substantially with the new persister code.

Note that there are a number of ways to mimic priority-like behaviour in
the current version, and even more with basic.qos in place.

#3 hasn't been asked for much, so it's currently a low priority task.
Having said that, any feedback on the experimental rabbitmq-c client is
of course welcome. Also, we are aiming for compatibility with OpenAMQ
when RabbitMQ implements AMQP 0-9-1.

> I realize the developers may be tackling other higher-priority items
> such as a high performance persister implementation, which may
> effectively block any v1.6.x release from happening anytime soon.

The new persister won't appear until after v1.6.x.

> Can someone let me know if any of these features will be backported to
> a future v1.5.x release?

I do not anticipate any more v1.5.x releases, and if there are any they
will be for low-risk/high-impact bug fixes only.

> If so, when would be an approximate timeframe? If not, is there any
> status update as to when a v1.6.x would be released? If there is no
> v1.6.x release timeframe, can you list out what are the major
> blockers are (aside from the persister problem)? I was also wondering
> if there were any external dependencies, such as a new AMQP spec
> version getting "ratified", before a v1.6.x gets released?

There aren't any major blockers left for v1.6.0. There is still a bit of
qa left to do, and a few minor bug fixes to make, so I'd expect to see
v1.6.0 to be released by mid June. The new persister should be ready in
early July.


Regards,

Matthias.