And I mention it in the blog post, and I mention it in the README, but I feel I should reinforce this:
This. Is. Alpha. Please don't use it in your mission-critical code.
I'd like to run a lot more tests on it. I'd like other people to run tests on it and let me know what they think. This is my first time writing anything this complex, and I want any and all feedback I can get on it.
Thanks!
The basic gist of it is similar to a phone tree. You route a message subscribing to a topic, basing the message ID on the topic name. As the message routes through the network, each Node stores the information about who is subscribed. When an event happens, that information is passed back through to the subscribed nodes.
I may have to tweak it a little, as I did Pastry, but I'm pretty sure the same basic concept will work for my needs.
Spotted a typo on the line "OnNodeExit", "Applcation".
And I guess my point is, I've tested this. I can't seem to find any problems with it. But I do not have unit tests for the larger routing algorithm, only the routing within a Node. In my tests, it always worked out, but I would put my confidence in this at about 70%.
In essence, I was ready to release this software to the Go community, have them kick it around a bit, see what came of it. I was not expecting a HN frontpage. So now I'm slightly panicked, and hoping to God that people don't think I'm an idiot for the rest of forever.
If you're going to have any component that's a single point of failure, redis is a pretty good bet for it.
I just really didn't want to have a single point of failure. Really badly. Pastry has no single point of failure.
[edit] Realised this thread isn't actually all about me. Oops! Sorry for answering a question not directed at me.