[Haskell-cafe] More on performance

[Albert Y. C. Lai]

Jon Harrop wrote:
> IRL the specification often dictates the complexity. If your code fails to 
> satisfy the spec then it is wrong. Are you saying that Haskell code can never 
> satisfy any such specification?

In addition to RL, it it should and it can in theory too:

http://www.cs.toronto.edu/~hehner/aPToP/  in particular chapter 4 
sections 4.2 and 4.3.

I also admire that Musser and Saini give asymptotic costs as an 
indispensible part of the STL specification. Everyone should at least do 
that much.