On Tue, Jul 12, 2011 at 08:59:36AM +0100, Simon Marlow wrote:
> On 08/07/2011 14:24, Ian Lynagh wrote:
> >
> > Random is now only needed to build dph
>
> So are we not shipping random any more?
Right. It turned out that haskell98 already had its own copy in Random,
so there is no need to ship it.
With the recently increased level of activity on the random package,
having it decoupled from GHC's release cycle will also help people get
the benefits of new work sooner.
> I see you added reqlib('random') to the tests that depend on random,
> but that's bad because it means these tests are not being run by our
> nightly builds (which typically don't have dph).
Well, validate will still run them. But in general, any reqlib's are
unfortunate, yes.
It might be worth putting a small PRNG in the individual tests (which
would also have the benefit of isolating the test results from changes
in the random package).
Thanks
Ian
_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc
