From: Richard M Kreuter <kreuter@pr...> - 2007-12-13 21:24:21
There are a few unexpected failures, unhandled errors, and invalid exit
statuses on Linux/ppc and Darwin/ppc; additionally, the tests do not run
to completion on the Darwin/ppc host I test on (the topmost SBCL dies in
GC), so I only know about the test files that fail on both Linux/ppc and
Some of the failures I think might as well be marked as expected, others
I don't know enough about to say. Can anybody familiar with these
issues say something about the following?
Failure: debug.impure.lisp / (UNDEFINED-FUNCTION BUG-353)
This is marked :fails-on on x86, x86-64, alpha, mips. It fails on
Linux/ppc and Darwin/ppc, too. I think this should be expected on ppc.
(Does it not fail on sparc?)
Unhandled error dynamic-extent.impure.lisp
There are two tests in this file that fail on both Darwin/ppc and
(assert-no-consing (dxclosure 42))
This test fails only on Darwin/ppc:
(assert-no-consing (cons-on-stack 42))
I'm not at all familiar with dx issues. Are these to be expected?
Failure: hash.impure.lisp / (HASH-TABLE WEAKNESS REMOVAL)
This fails on Linux/ppc, Darwin/ppc. I think this is just a buggy test:
there's a comment saying that the test may not be bulletproof on gencgc,
and ppc uses gencgc now. Wouldn't it better to call this an expected
error on builds with gencgc than to have it warn and report success?
Invalid exit status: room.test.sh
This fails on Linux/ppc, evidently by running out of heap space. I've
run this a half dozen times, and it fails in exactly the same way each
time, but I don't understand what's going on. A trace file is here: