On 8/30/2011 8:31 AM, bearophile wrote:
Time ago I have suggested a "trusted purity" to allow the implementation of pure function with memoization, but this observational purity seems better.
"observational purity" seems like another word for "logical const". This has been debated here many times.
