On March 4, 2014 at 8:01:26 AM, Martin Sustrik (sustrik@xxxxxxxxxx) wrote: Maybe I have been over-cautious with the version numbering. If people generally feel OK with moving to 1.0, it can be done on the next release. There is value in asserting stability. I’d be most concerned if we thought this stuff was really alpha. One of the folks I was working with on nanomsg expressed concern over the 0.2 number… and yet clearly this is more well baked than other “1.0” items on the ‘net. So calling this 1.0 will *probably* lead to wider adoption. The other thing is *interface* stability, where interfaces are not just API/ABI, but also *protocol* level details. The protocol asserts version 0 (despite the RFC indicating version 1). Having another independent implementation of the protocols (not just via foreign function interface) will help us be more “sure” here, and I’m working towards that end. It may be a bit longer, yet, but I hope to have a “pure Go” version of the SP protocols available…. these wil support all the same transports that the C version does. (inproc will be done using go channels.) The other thing I’m thinking of is the possibility of offering a websocket transport. This would offer several useful benefits — HTTPS and support for traversing firewalls and proxies. - Garrett