So these function orders have different behaviour in GHCI:
safetailpatter xs = tail xs
safetailpatter [] = []
safetailpatter [] = []
safetailpatter xs = tail xs
The former generates the following warning and the following error when passing in []
ex3.hs:66:1: warning: [-Woverlapping-patterns]
Pattern match is redundant
In an equation for ‘safetailpatter’: safetailpatter [] = ...
*** Exception: Prelude.tail: empty list
As such does the order of definition matter and why? I don't see why the former overlaps while the latter doesn't as the same definitions are given.
Patterns are matched in order. But, what's going on here is that you are not actually parttern matching on safetailpatter xs.
In regular english safetailpatter xs = tail xs means: safetailpatter on any variable is tail of that variable.
What you want to match is: safetailpatter (x:xs) = tail (x:xs), which stands for: safetailpatter when applied on a list with at least one element, is tail of such a list
Knowing this, In the code
safetailpatter xs = tail xs
safetailpatter [] = []
Will check in order, and because the first equation matches any list input, you get a runtime error on []. Whereas
safetailpatter (x:xs) = tail xs
safetailpatter [] = []
matches in order, and since [] doesn't match the first equation, It'll run the second, with no runtime error
As @chepner says, this is call irrefutable pattern. Meaning that pattern matching is happening but there is no chance to fail. Same as if you match against _