Jules Bean wrote:
Magnus Therning wrote:
Also from experience, I get a good feeling about software that compiles without warnings. It suggests the author cares and is indicative of some level of quality.
In contrast, I find almost all the GHC warnings to be useless, and therefore turn them off. I don't find they have a significant correlation with code quality.
YMMV :)
I strongly disagree with this. There is a huge difference between f (x:xs) = ... and f (x:xs) = ... f [] = error "f: we expect a nonempty list" The reason for this is that in the second case you express to somebody who reads your code (including yourself) that this omission was intentional. The same holds for other warnings (although I sometimes am annoyed by the shadowing warnings, I agree :). My default is to start developing, then adding -Wall -Werror and make it compile again. Regards, -- Jochem Berndsen | jochem@functor.nl GPG: 0xE6FABFAB