[Haskell-cafe] Re: Timeouts that don't cause data growth.

[Simon Marlow]

On 23/03/10 17:40, David Leimbach wrote:
> Trying to understand why the code here:
> http://moonpatio.com/fastcgi/hpaste.fcgi/view?id=8823#a8823  exhausts
> memory.
>
> I need to have timeouts in a program I'm writing that will run an
> interactive polling session of some remote resources, and know when to
> give up and handle that error. Unfortunately this code dies pretty
> quickly and produces an -hc  graph like the one attached.
>
> It seems that System.Timeout can't be used for this.  I should note that
> if the code is changed to use an inifnite timeout (-1) that this problem
> doesn't occur. Is this a bug in System.Timeout, or is there something I
> should be doing to keep the data size down?

The leak is caused by the Data.Unique library, and coincidentally it was 
fixed recently.  6.12.2 will have the fix.

Cheers,
	Simon