Feedback on Error messages in ghc.

[Ketil Malde]

Peter A Jonsson <pj at ludd.ltu.se> writes:

> I read the summary of the survey and noticed you wanted feedback on
> where error messages could be improved.

I have some more comments, but let me say that I'm mostly happy with
the error messages as they are.  One thing that occasionally happens
is that my (explicit) type signature is wrong, and GHC seems to take
it as fiat, and give all kinds of rather misleading error messages.
(No example, but I can try to create one if it's necessary).

Another thing I really would like, is runtime errors with line
numbers in them.  Sometimes I get an error from "head", "read", "!" or
similar during testing, and it would be really nice to know exactly
which invocation that failed.  In particular since I don't get a stack
backtrace.  (I can achieve this reasonably well with CPP macros, but
it is a chore.) 

-k
-- 
If I haven't seen further, it is by standing in the footprints of giants