[Haskell-cafe] A Question about IO

[Brandon Allbery]

On Sat, Feb 28, 2015 at 6:40 PM, Zongzhe Yuan <psyzy3 at nottingham.ac.uk>
wrote:

> Sometimes IO do something and return something, i wonder if the return
> type, for example is IO Int means it will return an int, could i purely
> fetch the int?


Conceptually, an IO something is a *program* that can produce, not under
your control, a something. You create this program and return it as the
value of (main :: IO a); then the runtime executes it.

(Actually, what happens in GHC is different. There are ways to get at the
value --- but you have to dig into GHC internals and use them to lie to
GHC, and it will punish you for it by (among other things) treating it as a
pure value and memoizing, sharing (or not sharing when you expect it to),
etc. as it sees fit. In short: don't. This is not some thing to force you
to jump through hoops for no good reason, with an easy escape hatch to get
the behavior you'd get from a random Perl/Java/whatever program; if you go
around its back, things *will* break unless you understand how GHC works.)

There is nothing in Monad that guarantees that you can go from Monad m => m
a -> a. Specific Monad-s (lists, Maybe, etc.) may provide functions to do
so, or expose their data constructors so that you can pattern match them;
other Monad-s (IO, ST, STM) do not, and you cannot get at "the value
inside" --- indeed, there may not be a value inside. (This isn't even
specific to Monad; you can't pattern match a Data.Map.Map, either, and it's
not a Monad. It does provide toList, though, as well as its own access
methods.)

-- 
brandon s allbery kf8nh                               sine nomine associates
allbery.b at gmail.com                                  ballbery at sinenomine.net
unix, openafs, kerberos, infrastructure, xmonad        http://sinenomine.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/haskell-cafe/attachments/20150228/9fa83598/attachment.html>