> The specific problems in the example could be solved by changing how the data is represented. Finding the "right" representation for a given set of data is a

> The specific problems in the example could be solved by changing how the data is represented. Finding the "right" representation for a given set of data is an interesting problem, but most (all) of the time the representation is specified by someone/something else. In the past I've written a [preprocessor][1] that adds some power to the representation while avoiding general purpose computation. For example, (buckets (let ([(regional region (name policy) ...) ((['name name] ['region region] ['lifecycle_policy policy]) ...)]) (regional us-west (alpha-hourly (delete_after_seconds 345600)) (alpha-daily (delete_after_seconds 2592000)) (alpha-monthly (delete_after_seconds 31536000)) (bravo-hourly (delete_after_seconds 345600)) (bravo-daily (delete_after_seconds 259200)) (bravo-monthly (delete_after_seconds 31536000))))) Macros, basically. Arithmetic would help there, but that might be too much. [1]: https://github.com/dgoffredo/llama https://github.com/dgoffredo/llama