Lot of us who use range have instances running in Amazon AWS where the node list is not static due the auto-scaling property of AMZ. IMHO we can solve this in following ways

1. manually edit content in YAML to keep it in sync
2. generate the YAML file automatically though a separate system
3. give an option to udf to generate the content

(1) is not an option because it won't scale. Problem with (2) is, if we use version control to version the changes in YAML file, we might run into race conditions where the final YAML files cannot be merged.

(3) seems to be better because users can write their on functions to generate content on fly without the need of editing YAML file. Down side for this is, a slow user defined function will make the end point drag. (udf could be a REST call to make interaction easy).

key1:
  - udf(fn(someargs))

where udf is a keyword to let the system know we are looking for at an user defined function, fn could be the function name with someargs as the argument passed to the function (or fn could be a REST endpoint using someargs as its GET param)

I am willing to write a patch if you are interested. Please feel to close the issue if you think it is not in the scope :-)

From the clusters data in this repo

When server starts.

WARN  [2016-01-20 13:28:17.552] Could not parse query: %!!(MISSING)s(bool=false)

Queries

$ erg %dc2-mysql:KEYS
ALL,CLUSTER,DOC,DOWN,EXTRA,PORT,TYPE,UP
$ erg %dc2-mysql:UP
Error:  Could not parse query: %!s(bool=false)