News for hackers. Stuff that matters.
Re:A Go “clone” of the great and famous Requests library by nutate on Saturday June 27, 2015 @09:13AM (#9790033)
I definitely support this sentiment w/r/t the single threaded mindset of node and their async all the things approach. The only question with Go is if we guarantee sync libs, then where is the good async example code? I agree libraries should be synchronous, but does it really hurt to not have some best practices "how to do it async" in the docs? or should it be... in the code?

Kind of a meta question, but an interesting one.

Re:A Go “clone” of the great and famous Requests library by tptacek on Saturday June 27, 2015 @02:50PM (#9791146)
The "good asynchronous" code in Golang is straight-line code that would be called "synchronous" in Node.

Part of the problem is that even as an attempt at a "non-synchronous" interface, this seems clumsy. It demand-creates new channels for requests. That doesn't seem right.

Re:A Go “clone” of the great and famous Requests library by levigross on Saturday June 27, 2015 @10:53PM (#9792277)
Whats wrong with "demand-creating" channels? The Go stdlib does it within their network library https://golang.org/src/net/singleflight.go#L67
Re:A Go “clone” of the great and famous Requests library by tptacek on Sunday June 28, 2015 @09:49AM (#9793439)
Because you can't select on an arbitrary collection of channels.

It might make sense for a library to return a new channel for something you do once, or once per goroutine.

It doesn't make sense to use returned channels as a basis for async dispatch.