Re: Deepseq proposed patch

Yes, gmapQ is very slow. This might not matter for forcing mostly unevaluated data, but for evaluated data it would really hurt. On Fri, Jun 11, 2010 at 8:20 AM, Stephen Tetley wrote:

On 10 June 2010 22:15, Julian Blake Kongslie wrote: [SNIP]

...

The point of my email was more that I'm not sure it makes sense to have a "NFData" class, where "Data" would appear to suffice and be easier for programmers to work with.

Don't Data.Data and traversals through gmapQ etc have a reputation for being particularly slow? (most newer generics libraries benchmark against it to show their swiftness).

That seems one good reason against this broad instance.

Best wishes

Stephen _______________________________________________ Libraries mailing list Libraries@haskell.org http://www.haskell.org/mailman/listinfo/libraries