[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Chicken-hackers] [PATCH] ensure scrutiny doesn't walk nodes more th
Mario Domenech Goulart
Re: [Chicken-hackers] [PATCH] ensure scrutiny doesn't walk nodes more than once
Wed, 30 May 2012 23:05:28 -0400
Gnus/5.13 (Gnus v5.13) Emacs/23.1.91 (gnu/linux)
On Wed, 30 May 2012 22:15:22 +0200 (CEST) Felix <address@hidden> wrote:
> The attached patch fixes a problem with the scrutinizer which causes
> nodes to be walked more than once. A case detected by Christian and
> Mario turned out that this behaviour (which is known - see bug #751)
> can cause code to make incorrect assumptions about argument types and
> so drop necessary type checks.
> The tests all seem to pass (with the applied patch and after a
> complete self-compile with the patched compiler). Still, this is
> a non-trivial change and may break stuff.
> This patch fixes bug #751 (and #855 which is a duplicate of the
Thanks for your patch.
I've built chicken with it (with bootstraping) and run make check.
Everything seems to work. The test case for #855 works fine too.
I've run salmonella with that chicken on the full set of eggs for
Linux. The diff against the salmonella log run on 2012-05-30's morning
is here: http://parenteses.org/mario/misc/751-patch-diff/
Salmonella for the chicken with the patch for #751 was run with the
--keep-repo option, so it didn't set the local egg repo empty after
installing each egg. That's why there are some extra differences
against the log for 2012-05-30's morning (which doesn't use
The only relevant differences are the test results for zmq and
svn-client. Coincidently, both are mysteries. Some svn-client tests
started failing out of the blue on 2012-05-18 and keep failing. If I
run salmonella on it using the same chicken as the one which fails for
the daily salmonella, it works.
There's a specific zmq test that have been consistently failing on
call-cc.org. However it works with your patch. (OTOH, if I remeber
correctly, Moritz said he could not reproduce the failure.)
So, as far as I can tell, your patch doesn't break anything. If nobody
has objections, I'd like to apply it.