On Tue, Jan 24, 2012 at 7:00 PM, Martin Lucina <[email protected]> wrote:
> I don't think Martin is ignoring anyone intentionally, just trying to > get the best result with the least possible resources. I don't think you're saying the rest of us have the goal of getting worse results, or spending more time than we need to. Introducing new features without breaking old ones is not wasteful, but making large branches no--one uses is. You're quite right to remind us that identities are complex and agreed to be removed (and already partially gone) but that's a different issue. We have a clear cost/benefit that can be discussed and decided. Same with devices, swap, etc. The cost/benefit for changes like labels aren't known upfront, which isn't helped by a "here's a bunch of new code, take it or leave it" style. Thus, if you have any hope of getting such changes accepted one MUST be compatible and give people time to weigh the economics. This is how every sane project does it (e.g. how Linux system calls migrate towards POSIX). It is more economical than attempting to make incompatible change on the basis of "trust me, it's worth it", and then having to abandon that. -Pieter _______________________________________________ zeromq-dev mailing list [email protected] http://lists.zeromq.org/mailman/listinfo/zeromq-dev
