[rabbitmq-discuss] RFC - new exchange type idea

[Alvaro Videla]

As a follow up just see today's hacker news top article: http://news.ycombinator.com/item?id=3899080  

Yes, that the article is top on HN might not mean much… but at least the idea suggested there is popular, ie: people might want that.

-Alvaro  

--  
Sent with Sparrow (http://www.sparrowmailapp.com/?sig)


On Wednesday, April 18, 2012 at 1:42 PM, Alexis Richardson wrote:

> I'm in favour of anything that makes it easier to get started.
>  
> I am continually surprised to hear things like "rabbit has queues but
> not pubsub" as well as "rabbit is good for pubsub but not queues".
> Not to forget "rabbit does not do persistence" as well as "rabbit is
> best for persistence" and of course "rabbit uses mnesia for message
> persistence" when it doesn't. In addition all kinds of FUD gets slung
> about.
>  
> The Redis community has been great at innovating from the bottom
> upwards, to deliver simplicity. This is immediately visible when you
> look at the slew of MQ-alike layers that people have created. These
> tend to focus on the core functions of pubsub, queue, and sometimes
> 1-1 and/or RPC; while ignoring eg flow control, paging, and other
> features. Some layer a work queue system on top of Redis, like
> Celery. Resque is a good example of that.
>  
> Here are more examples:
>  
> http://thechangelog.com/post/6791886560/thoonk-py-persistent-and-fast-push-feeds-queues-and-jobs
> ("a project that takes a unified approach to feeds, jobs, and queues,
> all with a Redis backend." and looks a lot like rabbit)
>  
> https://github.com/djhworld/redis-mq ("I felt tools like the
> excellent RabbitMQ were too complex for small projects like the ones I
> tend to make.")
>  
> https://github.com/ServiceStack/ServiceStack/wiki/Messaging-and-redis
> (WCF for hipsters)
>  
> http://restmq.com/ ("There are HTTP clients for pratically all
> languages. That's all that it takes to use RestMQ within your
> application. No special protocols or strategies, just dynamically
> created queues.")
>  
> http://lumberjaph.net/perl/2010/04/14/presque-a-redis-tatsumaki-based-message-queue.html
> ("a small message queue service build on top of redis and Tatsumaki.
> It’s heavily inspired by RestMQ and resque.")
>  
> http://flask.pocoo.org/snippets/73/ ("if you just want a very basic
> queue functionality to get started")
>  
> https://github.com/josiahcarlson/rpqueue (priority queue)
>  
> All of these projects are motivated by ease of use and a simple API.
> Let's make Rabbit easier to use.
>  
> alexis
>  
>  
>  
>  
> On Wed, Apr 11, 2012 at 5:36 PM, Jason J. W. Williams
> <jasonjwwilliams at gmail.com (mailto:jasonjwwilliams at gmail.com)> wrote:
> > I think this is a great idea. It makes it easier to on onboard those who initially need a simple message queue and end up using Redis. Whatever message queue tech people start with for their first quick and dirty needs they tend to stick with. So I think this is a great idea for getting folks converted to Rabbit early.
> >  
> > Yes functionally anonymous queues function the same way, but conceptually they break the idea exchanges are where you publish messages since you supply the "empty" exchange.
> >  
> > Whether its a new exchange type or a change to the library API for anonymous queues, I like the direction Alvaro's going for folks getting their feet wet.
> >  
> > -J
> >  
> > Sent via iPhone
> >  
> > Is your email Premiere?
> >  
> > On Apr 11, 2012, at 7:13, Alvaro Videla <videlalvaro at gmail.com (mailto:videlalvaro at gmail.com)> wrote:
> >  
> > > Hi,
> > >  
> > > I've been thinking on creating a new exchange type whose goal will be to simplify the ways messages are routed to queues.
> > >  
> > > So now for example to publish a message to a queue I need to do something like:
> > >  
> > > $ch->exchange_declare('foobar', 'direct');
> > > $ch->queue_declare('foobar');
> > > $ch->queue_bind('foobar', 'foobar');
> > > $ch->basic_publish($msg, 'foobar');
> > >  
> > > I want to be able to do this:
> > >  
> > > $ch->exchange_declare('foobar', 'new-exchange-type');
> > > $ch->basic_publish($msg, 'foobar');
> > >  
> > > So that means that by declaring a "foobar" exchange with the new type I will bind a new queue to that exchange during exchange creation on the broker side.
> > >  
> > > The queue will have similar durability properties as the exchange.
> > >  
> > > The exchange won't accept binding requests, so 1 exchange of that type will only have 1 queue bound to it.
> > >  
> > > The idea is also to not perform any query while searching for a queue where to route messages, so routing speed *should* be a bit faster than a normal direct exchange.
> > >  
> > > The exchange should act basically as a direct exchange.
> > >  
> > > Then as a user of RabbitMQ that just want to have a queue server for delayed jobs this is the most basic setup I think you can have.
> > >  
> > > BTW, I know you can publish messages to the anon exchange by using the queue name as routing key, but this requires that particular knowledge of AMQP.
> > >  
> > > What do you guys think?
> > >  
> > > -Alvaro
> > > _______________________________________________
> > > rabbitmq-discuss mailing list
> > > rabbitmq-discuss at lists.rabbitmq.com (mailto:rabbitmq-discuss at lists.rabbitmq.com)
> > > https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
> > >  
> >  
> > _______________________________________________
> > rabbitmq-discuss mailing list
> > rabbitmq-discuss at lists.rabbitmq.com (mailto:rabbitmq-discuss at lists.rabbitmq.com)
> > https://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss
> >  
>  
>  
>  


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rabbitmq.com/pipermail/rabbitmq-discuss/attachments/20120427/665f4404/attachment.htm>