Zyre

Thanks for the intro and your efforts. I'm still to digest all that you wrote.
One off-the-cuff observation: I wonder if you have investigated 'lessons learned' from MPI and, more practically, the various implementations. I do recall trying to follow a debate about one aspect of the asynchronous/reliability requirement in a part of the spec, but it was a bit beyond me at the time…
I'm often surprised to discover how few web/enterprise people are aware of it. It does deal with a different use case, and different architecture - I mention MPI because I think there you might find more 'talented engineers' (which I am not) who have tackled and thought about messaging 'issues'.
HTH
HH

There are a couple of major resources on webhooks.

Here is a recent talk: http://vimeo.com/4537957
Here is the original epic slide deck: http://www.slideshare.net/progrium/web-hooks-and-the-programmable-world-of-tomorrow-presentation

It's a fairly grassroots effort that me and others are evangelizing, but I must say we are doing a good job. There is a lot of support in a lot of important places and it's increasing every week! I think webhooks are VERY important to consider with your work. At the very least it's a great way to "hook" support of RestMS into our web application ecosystem. But I think it's important to consider this simple concept in your architectures.

Where are you located? Perhaps we should try and talk sometime.

Also, it's not about a "spec" … I'm all about standards, particularly open ones, but I also like iterative and organic growth. Webhooks are general enough to have many specs involved. The most promising right now is PubSubHubbub, which uses webhooks for the age old pubsub problem. I believe this overlaps with some of what you're trying to accomplish.

I think your work is important and valuable, but let's not try to approach this monolithically (not sure that you are). There are many existing standards and ideas that can work together to create a much more flexible open toolkit to solve these kinds of problems.

Ok, well I'm in San Francisco … so unless I travel out there we probably won't get a chance to talk in person. I'll try to stay on top of RestMS as best I can.

@Jeff, maybe one of these days I'll be in SF. Used to go there quite often.

There is a lot of work going on in this area, and webhooks and PubSubHubbub are definitely interesting. These are about delivery, which is just one of the aspects of messaging. I'd say there are five main aspects: publishing to a server, routing inside the server, queuing inside the server, delivering to the recipient, and federating servers together. There are many more aspects when one comes to the details but these are the top level slice.

RestMS tries to provide answers for all of these but it's most definitely a work in progress and it is deliberately unfinished wrt delivery. We've already started looking at how that should really be done, and there are some options: webhooks, BWTP, and other asynchronous HTTP-like protocols.

If you look at the RestMS architecture and process then you'll see that it's easy for anyone to propose and submit a spec that implements, e.g. webhooks for delivery within RestMS.

In this respect RestMS is not a protocol but a family of protocols based on an open process.