 Implicit params in Haskell
started 1/28/2004; 6:07:23 AM - last post 1/29/2004; 10:30:36 AM
|
|
andrew cooke - Implicit params in Haskell 
1/28/2004; 6:07:23 AM (reads: 235, responses: 8)
|
|
|
|
Ehud Lamm - Re: Implicit params in Haskell
1/28/2004; 6:23:46 AM (reads: 224, responses: 2)
|
|
|
You are an editor! use the home page!
|
|
andrew cooke - Re: Implicit params in Haskell
1/28/2004; 7:29:41 AM (reads: 211, responses: 1)
|
|
|
v busy - didn't have time to check if it's sensible.
|
|
Ehud Lamm - Re: Implicit params in Haskell
1/28/2004; 7:44:41 AM (reads: 209, responses: 0)
|
|
|
me2...
|
|
Christopher Hendrie - Re: Implicit params in Haskell
1/28/2004; 8:43:54 AM (reads: 210, responses: 0)
|
|
|
It looks interesting, but I find the implicit parameter stuff sufficiently different that I don't trust my intuition as to correctness. I hope we can see some formal treatments of various implicit parameter mechanisms soon -- I'd be a lot more comfortable with it if I could see the key concepts in as simple a setting as possible. Has anybody written a paper on, say, lambda-calculus or PCF or whatever + implicit parameters?
|
|
Dave Herman - Re: Implicit params in Haskell
1/28/2004; 6:46:58 PM (reads: 179, responses: 0)
|
|
|
|
|
|
|
Actually, Ben proposes the converse of implicit parameters, i.e.
implicit return values.
His argument goes like
* implicit parameters subsume the Reader monad.
* implicit return values subsume the Writer monad.
* together they subsume lots of monads... IO, ST, etc.
My view:
Implicit parameters already have a flaw, that is some transformations
that are valid in haskell become incorrect. (Which ones exactly, I can't
tell form memory :)
My intuition is that implicit return values breaks
referential transparency. That's because they allow to to have state
threaded through computation in a completely implicit way.
|
|
andrew cooke - Re: Implicit params in Haskell
1/29/2004; 4:24:23 AM (reads: 153, responses: 0)
|
|
|
My intuition is that implicit return values breaks referential transparency.
the google link a couple of posts back said [...] implicit parameters [...] provide[s] dynamically scoped variables [...] so (unless i'm confused about what the terms mean - which is possible, since i couldn't define referential transparency clearly if you paid me) they certainly do break referential transparency. i always thought dynamic scoping was supposed to be the work of the devil, so it will be interesting to see what the "motivating arguments" are in that paper.
|
|
Andris Birkmanis - Re: Implicit params in Haskell
1/29/2004; 10:30:36 AM (reads: 125, responses: 0)
|
|
|
I've read the cited paper in a slightly different edition, where the difference between problems of implicit parameters in call-by-need and call-by-value languages is more pronounced. Or did I just misread that...
|
|
|
|