2012/12/15 Ian Lynagh <i...@well-typed.com>: > > Hi Niklas, > > On Sat, Dec 15, 2012 at 04:02:24PM +0100, Niklas Larsson wrote: >> >> It seems like the build on Windows HEAD is trying to build the PowerPC >> assembly in rts/AdjustorAsm.S. Somehow the powerpc_HOST_ARCH must be >> defined, I can't figure out how though. > > Does it happen for you with a non-parallel (i.e. without using > "make -j") build? If so, can you reproduce it by running the last > command that "make" runs by hand? If you can, then hopefully it will be > easy to diagnose, e.g. by minimising the code being read. >
I built with "-j 4" on both my desktop and laptop. I'll retry without "-j", it's strangely deterministic for a race condition though. >> It seems to work fine on commit d7d25db3e58f3edfb950ebfa466c1e1e06d175d6 >> from a week ago. > > Hmm, no patch since then jumps out at me as a likely cause. If anyone's > able to narrow down which patch cause it then that may also be useful. > I guess I may have tried that one without "-j". >> I have one patch though. When I was messing around, I taught sync-all how >> to checkout all the repos based on a date. It is somewhat tested, use with >> caution. > > Thanks! I'll take a look. > I should add a bit of documentation to the sync-all help (especially that it can't travel forward in time. You have to checkout the head of the branch again to go there, maybe it could take a branch as an argument too, now it follows the current branch). And it shouldn't really try to checkout when there is no prior commit. But it doesn't do any harm. Niklas > > Thanks > Ian > _______________________________________________ Cvs-ghc mailing list Cvs-ghc@haskell.org http://www.haskell.org/mailman/listinfo/cvs-ghc