On Mon, Nov 23, 2015 at 12:15 AM, Dmitry Gutov wrote: > IMO, there's no too much need to exhaustively test `json-encode', > aside from one encode test (with any value), one decode test, and > maybe one try-to-encode-something-unusual-and-end-up-with-an-error > test. > > Testing all contents of json-readtable would seem excessive to me. > But hey, if we miscalculate, we can add more tests as we fix > regressions, later. Alright! I've attached an updated patch. -- Simen