Concerning Int vs. Integer:
Wouldn't it be better if Standard Haskell had "save" Int's, that is a version that
simply
would abort the program when encountering a overflow.
This can be implemented without too much of a penalty on the more relevant benchmarks.
And if Int32 and friends from the Hugs/GHC libs are promoted to the standard library
in some form then the report has to change on two points:
1. state that an exceptional result (for all fixed precision types?) yield an error,
instead of being undefined.
2. note that the standard library will contain an integer without overflow checking
That implementations would, for the moment, implement this new Int by just using the
current
situation only would be a minor temporary inconvenience.
On a general note, should executing a program not always yield a defined result?
That is, either yield the correct result OR produce an error message?
Sietse Achterop
