Continuous queries syntax change

[richthegeek]

I suggest a change to how continuous queries are created, and possibly managed.

The changes are as follows:

1. CQs are identified by the CONTINUOUSLY param, instead of by the INTO param
2. The INTO query now simply redirects output to a second series (similar to MongoDB's $out
3. The CONTINUOUSLY param can work on DELETE queries as well, and the mode of identifying CQ deletes is consistent.
4. The CONTINUOUSLY param can be followed by AS :identifer to solve issue Named Continuous Queries #525 in a manner consistent with existing language.

Whilst this would be a breaking change, I believe it's a clearer syntax and provides multiple benefits.

An example of a query utilising these keywords
SELECT sum(value) FROM orders GROUP BY time(1d) INTO daily_order_totals CONTINUOUSLY AS daily_order_totals

[pauldix]

I think the delete queries would be better served through #479 (other scheduling options for continuous queries). Those same scheduling options could be applied to deletes. I guess the continuously keyword would be one of the scheduling options.

Not sure about this one. Wonder what other people think about it.

[jvshahid]

I like the idea. Don't think delete queries belong here. Initially to make this backward compatible we can provide a config option for comparability mode. Also, query identifier should be optional with a default of cq<unique number>, where `unique number is an auto incrementing number consistent across all nodes, using raft or other means.

[jvshahid]

Oops, referenced the wrong issue

[richthegeek]

Closed it by accident a second time?