Re:Simon Peyton Jones – Haskell Foundation Podcast
by renox on Wednesday October 19, 2022 @04:53PM
(#33267794)
> Bottom has to be a member of every type because functions aren't guaranteed to terminate,
I don't understand: if the function you called doesn't terminate then you're blocked so I don't see why this should impact the types..
Re:Simon Peyton Jones – Haskell Foundation Podcast
by foldr on Thursday October 20, 2022 @06:29AM
(#33272740)
More info here: https://wiki.haskell.org/Bottom As bottom is the value of a non-terminating computation and a computation of any type may fail to terminate, bottom must inhabit all types. The same goes for 'error'.
That admittedly doesn't entail that Haskell ought to have special functions like 'error' that invite the programmer to explicitly introduce bottom values as a mechanism of signaling errors. However, in the real world, executing any function may raise an exceptional condition at any time (e.g. an out of memory error), regardless of programming language.