core.async example

[stream-iori]

I write a example about core.async in vert.x, i think this would be useful for someone.
hope you could checking my PR, if something doesn't correct.

[tobias]

Thanks for this! I cleaned it up a bit, and changed it to ping/pong 100 messages instead of 1: 02eaea3

The only things I changed in the logic from your original code was changing the alts! calls to <!, since we're only working with one channel in each case, and using (go-loop ...) instead of (go (loop ...)). go-loop is a convenience macro that puts a loop inside of a go.

Since it now sends 100 messages, the println's would interrupt each other, making the output unreadable. So I added a log function that uses a channel to serialize those printlns.

I've been copying the examples to a PR on https://github.com/vert-x/vertx-examples since we're about to be included as an official module. I'll copy this to that PR as well.

[stream-iori]

Ha!

Thanks for your correction. :)

Toby Crawley writes:

Thanks for this! I cleaned it up a bit, and changed it to ping/pong 100 messages instead of 1: 02eaea3

The only things I changed in the logic from your original code was changing the alts! calls to <!, since we're only working with one channel in each case, and using (go-loop ...) instead of (go (loop ...)). go-loop is a convenience macro that puts a loop inside of a go.

Since it now sends 100 messages, the println's would interrupt each other, making the output unreadable. So I added a log function that uses a channel to serialize those printlns.

I've been copying the examples to a PR on https://github.com/vert-x/vertx-examples since we're about to be included as an official module. I'll copy this to that PR as well.

---

Reply to this email directly or view it on GitHub:
#105 (comment)

[dmarjenburgh]

A related question. core.async runs go blocks on a fixed size thread-pool. Since vert.x makes sure only one thread runs through each verticle, is it really a good idea to send messages over the eventbus inside of go-blocks? These would likely be send by a different thread than the verticle thread.

I suppose using put! and take! (with the default on-caller? arg to true) will ensure things get done on the verticle thread, but there is still a thread-pool inside of core.async in the background. Do you have any thoughts on this?

[stream-iori]

These would likely be send by a different thread than the verticle thread.
I don't clear about this, i think send and reply always been invoked by Verticle Thread in Eventloop context.