[plt-scheme] Disabling contract checking

From: Williams, M. Douglas (M.DOUGLAS.WILLIAMS at saic.com)
Date: Mon Apr 10 14:29:15 EDT 2006

I concur that it would be nice to be able to turn it off.

I haven't checked recently, but when I first added contracts to the Science
Collection I took an order of magnitude hit on performance with tight loops
calling into a contract.  [This happened to be basically all of my
examples.]  It was discussed on the mailing lists at the time.

I didn't use contracts for the differential equation solvers because they
essentially always work that way.  However, I did leave all of the others in
because I generally like the safer code.

I haven't checked on the performance between code with / without contracts
recently.  Has there been any substantial change to improve performance?

Doug

> -----Original Message-----
> From: plt-scheme-bounces at list.cs.brown.edu [mailto:plt-scheme-
> bounces at list.cs.brown.edu] On Behalf Of Lauri Alanko
> Sent: Monday, April 10, 2006 12:07 PM
> To: plt-scheme at list.cs.brown.edu
> Subject: [plt-scheme] Disabling contract checking
> 
> Contracts, though stupendously expressive, are still essentially just
> "assert" on steroids. However, the reason why assertions are so
> ubiquitously popular is that they can be disabled in production code,
> removing all overhead. But am I blind, or is there really no such option
> for contract.ss? Why?
> 
> I know that the theory is that contracts are placed only on inter-module
> boundaries, and that inter-module boundaries are not crossed too often,
> but neither of these is always true, so the unability to disable
> contracts discourages one from placing precise contracts on
> performance-critical functions. This seems like a shame.
> 
> What, in any case, would be the proper way to implement a compile-time
> option such as this? A transformation-time parameter value? Or perhaps a
> custom classloader that provides an alternate version of a
> macro-defining module?
> 
> 
> Lauri
> _________________________________________________
>   For list-related administrative tasks:
>   http://list.cs.brown.edu/mailman/listinfo/plt-scheme


Posted on the users mailing list.