nng is quickly approaching what I consider “production ready” state.
Ideally I’d have an epoll() implementation for Linux written (we have
kqueue for BSD thanks to Liam Staskawicz!) and perhaps a few more documents
written. (I’m going to be moving the TLS functions into a separate header
like I did for HTTP as well.)
At this point, I’d like to open up the testing to everyone. Please don’t
put this in production yet (at least not “critical” production) because I
can’t promise its bug free, *but* if you have test cases that simulate
production, or can use it in production in ways that won’t be tragic if it
fails, please do so!
Also, this is now a good time for folks who’ve been contemplating writing
language bindings to begin the process. Again, I may shuffle some things
around, but by and large I feel good about “nng”. I also unreservedly say
that I think it is likely *more* robust than nanomsg has ever been. (At
this point most of my reservations against production use relate more to my
ability to finish stabilizing the API than any concerns about fragility in
nng itself.)
I will probably cut a “release” soon (pre-release really), but for folks
who can build and test from github, please do!
Also, if anyone finds any significant missing functionality that would
impair their ability to adopt nng, especially in cases where they are
already using nanomsg, I *really* want to know that ASAP, so that I can
prioritize closing any such gaps asap.
Thanks!
- Garrett
