A rewrite of range to suit the ephemeral nature of cloud architecture. This is not a replacement for the old range, since this understands only a subset of features and also have added few features which are very valuable in cloud environment.
- a new querying language
- dynamic store
- reverse lookup are faster
- options for fast querying (ie, return the first result)
- written in go
%range1== expand range1%range1:KEYS== show all the KEYS in range1%range1:FOO== show the value for key FOO in range1%RANGE== toplevel (here RANGE is a keyword)%%RANGE== second level (can go any level down with more%till you hit leaf node)%%range1== second level w.r.trange1(you can go any level down with%)*hostname== get cluster where this hostname is present*value;KEY== get the cluster where KEY=value*value;KEY:HINT== get the cluster where KEY=value, HINT is to scope within a toplevel
%range1 , %range2== union (space is optional)%range1 ,- %range2== set difference%range1 ,& %range2== intersection%range1,(%range1 ,& %range2)== set operations with grouping using brackets
%*value== cluster operation on reverse lookup%*value;KEY== cluster operation on reverse lookup with KEY where KEY=value%*value;KEY:HINT== cluster operation on reverse lookup where KEY=value, HINT is to scope within a toplevel%(%range1:FOO,(%range1:BAR ,& %range2:MOO))== set operations with grouping using brackets%(*value1;KEY1:HINT1 ,& *value2;KEY2:HINT2)== cluster lookup the result of a set operation done on reverse lookups
I would suggest you to read expr.peg to understand all the possbile query combinations. The AST evaluator evaluates from Right to Left.
If you are planning to use etcdstore as the store for the range then we need to setup etcd cluster.
WIP
Probably many, please open issues. Patches are most welcome.