[Haskell-cafe] There is no null; Maybe/Option types

[Evan Laforge]

> Most of the (non-IO) runtime errors I get using Haskell software is due to head or fromJust, it's actually quite annoying.

Just singling this one out because I've heard it before.  I've never
gotten a runtime error from these.  It seems quite easy to just not
use these functions.

It's already basically been said, but to add more personal experience:

For the original blog post, I think the title is pretty easily
falsifiable just from practical experience: haskell's Maybe *did* save
me from NPEs and that's why I don't see them.  The problem is not that
a key lookup may fail in a map or whatever and now you have to
propagate that Nothing, the problem is that they pop out of just about
any code at just about any time and are not covered by the tests
because you can't test all code in all ways, so production apps wind
up crashing from them.  That's the thing I see all the time in java,
and don't see in haskell.  And I don't think it's just redefining the
problem either, I only rarely get "unexpected Lefts" once the app has
gotten past the tests.

So that's the practical result I actually care about.