joelr1:
This is using stage2. Does it look any better?
OVERALL SUMMARY for test run started at Mon Jul 10 15:11:22 BST 2006 952 total tests, which gave rise to 4583 test cases, of which 11 caused framework failures 1099 were skipped
3185 expected passes 24 expected failures 0 unexpected passes 217 unexpected failures
Not too bad (mips64 is around the same), but not the same as the linux head: OVERALL SUMMARY for test run started at Sun Jul 9 22:35:16 BST 2006 1401 total tests, which gave rise to 7573 test cases, of which 0 caused framework failures 1357 were skipped 6056 expected passes 50 expected failures 0 unexpected passes 110 unexpected failures So the next step would be to diff the two results, and individually run the tests that failed. Quite possibly many are failing for non-scary reasons (like missing libs, or missing arch-specific output cases) http://www.haskell.org/pipermail/cvs-all/2006-July/048188.html -- Don