patch applied (ghc): mark stop event handle as invalid once closed, making shutdowns more graceful.

[Sigbjorn Finne]

I guess an apology is in order from yours truly for having my
hard-drive going bad before getting around to pushing this
back onto HEAD & leaving it to Ian to clean up (thanks!)
Won't happen again...I hope :)

I also hope that Simon's offer isn't totally serious..

--sigbjorn

Simon Marlow wrote:
> Sven Panne wrote:
>> Am Mittwoch, 15. November 2006 14:30 schrieb Ian Lynagh:
>>
>>> [...]
>>> It has been in the 6.6 branch since 16 Oct, but has only just entered
>>> the HEAD. [...]
>>
>>
>> The direction of merging is a little bit chaotic in recent times 
>> IMHO. In good old CVS days we had the rule that changes from the HEAD 
>> should be pushed to the branches where appropriate, not the other way 
>> round. It would be good if we used this rule again IHMO, as it makes 
>> tracking things easier (by humans, at least).
>
> I don't feel terribly strongly about this, but the reason I've 
> committed fixes to 6.6 sometimes is that (a) its sometimes more 
> convenient, if I have an up to date 6.6 tree to test on, and (b) it's 
> easy to see which patches are in 6.6 and not the HEAD, but we can't do 
> that in the other direction.  So we'll never forget to merge a patch 
> from 6.6 into HEAD, but we will almost certainly forget to merge in 
> the other direction.
>
> So I'll leave this up to Ian as the caretaker of the 6.6 branch.  I'm 
> happy to stop pushing new patches to 6.6 if that's what folk would 
> prefer.
>
> Cheers,
>     Simon
> _______________________________________________
> Cvs-ghc mailing list
> Cvs-ghc at haskell.org
> http://www.haskell.org/mailman/listinfo/cvs-ghc