dustin at spy.net
Sat Mar 1 23:42:36 UTC 2008
On Mar 1, 2008, at 12:36, Aaron Stone wrote:
> If we get a mostly-working protocol out the door, and find some
> problems, but forget to clearly specify rules for new protocol
> that's a killer. We have to block on making sure we get the protocol
> version stuff right the first time.
The only thing that's not working today is stats. Stats is kind of
free-form, and nobody seems to want to freeze it.
What you have is a protocol implementation that meets the design
requirements of the binary protocol as we discussed it in the first
meeting. It's pretty much been there a while.
When it comes up in discussion, people want to extend it or change.
Semantics change, packet formats change, etc... I'd think it'd be
better to ship *something* (at least something that meets the original
goals) and let people actually try to use it than to just have this
ever-in-design-phase protocol that fails to make everyone happy (hence
the perl6 analogy).
Besides stats, is there anything you could do in the text protocol
that you can't do in the binary protocol? If there's not, the goals
have been met and we're done.
More information about the memcached