On Sat, 2011-05-21 at 12:03 +0200, Niklas Larsson wrote: > This comes from: > commit 58532eb46041aec8d4cbb48b054cb5b001edb43c > Author: Spencer Janssen <[email protected]> > Date: Thu Apr 14 01:11:05 2011 -0500 > > Add capability sets to the event system. Contains code from Duncan > Coutts. > > It's a bit too posixy, it uses getppid() which isn't in the Windows > C-runtime lib. I just ifdeffed in 0 there instead to get it working. It is > pretty cumbersome to find out what the parent process is on Windows, but it > can be done with the toolhelp api in case there is a real need for it.
I somehow have the misfortune of submitting my broken code just before meeting up with SPJ in person, so he is able to embarrass me in public. I think this happened at ICFP two years running! :-) (Public embarrassment obviously being the right and proper punishment for breaking the build). So I reverted that version of the patch and I split up the event into two, one for the process pid and one for the parent pid. We now don't generate the latter on windows, which is ok since it's mainly useful for users of forkProcess. I notice that my new patch also broke the build on windows, but not in any fundamental way. I've pushed another fix and hopefully it'll be working come Monday morning. Duncan _______________________________________________ Cvs-ghc mailing list [email protected] http://www.haskell.org/mailman/listinfo/cvs-ghc
