In the current jbeta, we have "improved error reporting", and it looks like https://code.jsoftware.com/wiki/Vocabulary/ErrorMessages has been updated to include some information about this mechanism.
However, this is error reporting on primitives and it could probably be extended to support error reporting on library code. But I'm not sure if that's more of a documentation issue or an implementation issue. ---------------- My motivation for going here (and, thus, the subject line of this email message) was that I was trying to figure out how to compute euler's constant. https://en.wikipedia.org/wiki/Euler%27s_constant The wikipedia page mentioned the digamma function, so I tried: require'math/calculus' ! deriv_jcalculus_ 1 Ideally this would have given me a verb, but instead I got a domain error which was somewhat opaque given the context. Studying the page on J errors shows me that the second line of the error report means that this was a synthetic domain error. Conceptually speaking, though, this should probably be a nonce error (the math exists to find the derivative of the gamma function, but it has not been implemented here). So that gives me three problems: (1) I still need to figure out how to compute euler's constant, and (2) I should probably change the error reporting in the math/calculus implementation (which, by the way, matches the previous d. implementation -- so it's somewhat understandable why it's the way it is), and (3) It seems like the second line of an error report from 13!:8 should provide a line from higher up in the stack, since 13!:8 by itself isn't particularly meaningful. (Perhaps this should be the topmost explicit line which does not contain the invoking 13!:8 unless there is no such line?) For now, I'm just leaving this note here... FYI, -- Raul ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
