Interesting. I just told GHC "let 5 = 3" and got no complaint.

lazy pattern matching semantics, see: ghci -XBangPatterns Prelude> let { 0 = 1 } in 0 0 Prelude> let { ! 0 = 1 } in 0 *** Exception: <interactive>:3:7-13: Non-exhaustive patterns in pattern binding

2014-06-17 12:51 GMT+02:00 Roman Cheplyaka :

* Alexey Muranov [2014-06-17 03:25:10-0700]

...

Is this the expected behavior that

1 = 0

does not raise any error? What does this mean?

1 is a valid (nullary) pattern, so yes. This pattern binding obviously fails.

Prelude> let x@1 = 0 Prelude> x *** Exception: <interactive>:2:5-11: Irrefutable pattern failed for pattern x@1

If it can make things clearer, you can do something similar with any constructor: > let Just x = Just 5 in x > let Just x = Nothing in x *** Exception: <interactive>:19:5-20: Irrefutable pattern failed for pattern Data.Maybe.Just x

On Tuesday, June 17, 2014 1:08:45 PM UTC+2, Brent Yorgey wrote: The general syntax of assignments is

pattern = expression

One most often uses a pattern consisting of a single variable, but any pattern will do. (For example, try writing [x,y,z] = [1,2,3] as a top-level declaration and see what happens!) 1 is a pattern which matches the number 1. 0 is obviously an expression. Of course, the pattern 1 does not match the expression 0. However, binding is lazy: for example, writing [x,y] = [1,2,3] does not in and of itself cause an error; it will only cause an error if you try to use x or y. However, since the pattern 1 contains no variables, there is no way to ever force the pattern-matching to actually take place.

So 1 = 0 just sits there, sort of like an ugly, angry dog who wants to bite people except that it is locked in a cage with no door. Occasionally people walking by look at it pityingly, but mostly no one pays it any attention.

Thanks for the detailed explanation, i think i've understood. Alexey.