> > I checked the latest version of the Binary protocol spec athttp:// > code.sixapart.com/svn/memcached/branches/binary/server/doc/pro... > > it mentions "Data Type (reserved for future use)". Well. How about we > > define it now? > > It still feels like a bad idea, and it's not stored anywhere. > > It's in there because there was space and someone really, really > wanted it. The idea is that flags will be available for user > applications, but clients will be able to have their own set of flags > that client users don't get access to. Just sounds rather messy to > me. I'd want to expose them all because I *still* think users should > be able to define their own encoding formats for their own > applications.
I did try to start pulling together the information (pre-binary protocol) here on the memcached wiki http://www.socialtext.net/memcached/index.cgi?cross_platform_serialisation_support. Unfortunately I ran out of time (I needed this for a work thing) and had to go with a limited implementation locally to me, but I still think there is value in the approach. - Cj.
