Using error instead of fail results in an ErrorCall instead of
IOError to be emitted, which is subsequently not caught by wrapIOError
and thus escapes the MonadError machinery.
This means the user would need to have two layers of error handling in place,
which is probably not the intented behaviour.
acowley
commented
9 years ago
Using
errorinstead offailresults in anErrorCallinstead ofIOErrorto be emitted, which is subsequently not caught bywrapIOErrorand thus escapes theMonadErrormachinery.This means the user would need to have two layers of error handling in place, which is probably not the intented behaviour.