($!!) :: (NFData a) => (a -> b) ->a -> b f $!! x = x `deepseq` f x

Could be useful.

force :: (NFData a) => a -> a force x = x `deepseq` x

I'm not convinced this forces x at the point where you want it to. For instance, foldlw f a (x:xs) = foldlw f (force (f a x)) xs would not give you the foldl' you might be looking for - indeed it is identical to the space-leaking foldl, apart from the extra class constraint. Regards, Malcolm

At 5:55 PM +1000 7/1/11, Ivan Lazar Miljenovic wrote:

I propose we add a operator - tentatively labelled $!! (and which Hayoo reports isn't currently exported by a Hackage package) - as a deepseq analogue $! to the deepseq library:

($!!) :: (NFData a) => (a -> b) ->a -> b f $!! x = x `deepseq` f x

infixr 0 $!!

+1 Note that I included the ($!!) operator in my original DeepSeq module (http://haskell.org/pipermail/haskell/2001-August/001586.html). Dean

It's been two weeks since I raised this, and as such discussion time is over. For the complete discussion, please see http://www.haskell.org/pipermail/libraries/2011-July/016540.html In total, there were five respondees. On 1 July 2011 17:55, Ivan Lazar Miljenovic wrote:

I propose we add a operator - tentatively labelled $!! (and which Hayoo reports isn't currently exported by a Hackage package) - as a deepseq analogue $! to the deepseq library:

($!!) :: (NFData a) => (a -> b) ->a -> b f $!! x = x `deepseq` f x

infixr 0 $!!

All respondees agreed with this proposal; as such it passes.

As a sub-proposal, I also propose that we add a "force" function (the name of which also seems to be unused) to make it easier to test forcing of values:

force :: (NFData a) => a -> a force x = x `deepseq` x

* Two people agreed with this proposal. * Two people indicated that it may produce confusing results (one of which at least can be seen as a vote against this proposal), * The fifth respondee didn't mention it. As such, including myself in the tally, I would consider that this proposal passes, but that we should add sufficient documentation to cover Malcolm's and Sebastian's concerns (specifically that using force inside a left-fold would actually be more analogous to foldl than foldl', and that a reminder should be made that this will evaluate the value to NF rather than HNF, which is indeed the point). I will now go and create a ticket on the GHC Trac. @Simon: as the maintainer of deepseq, would you prefer I create a patch with these changes, or will you do so yourself? -- Ivan Lazar Miljenovic Ivan.Miljenovic@gmail.com IvanMiljenovic.wordpress.com