[Sbcl-devel] Re: Failing tests in debug.impure.lisp on NetBSD/x86

[Richard M K. <kr...@pr...>]

Nikodemus Siivola <nik...@ra...> writes:
> On Thu, 2 Jun 2005, Richard M Kreuter wrote:
>
>> A couple tests that fail in debug.impure.lisp on NetBSD/x86 with
>> 0.9.1 and yesterday's anoncvs are conditionalized out for x86
>> linux.  I can't tell if these ought to be sensitive to the
>> operating system.  Is there any reason why these tests should get
>> run on NetBSD/x86?
>
> They are indeed sensitive to the operating system. Probable cause:
> x86-call-context. Unless there are objections I'll merge this and
> add a note about NetBSD to BUGS.

Okay, I see the description of this on Linux in BUGS.

> Does NetBSD currently pass other tests?

All but one: non-encapsulating trace at the end of debug.impure.lisp.
Non-encapsulating traces (the non-default trace mode) induce a
segmentation fault that isn't handled correctly.  I'm still trying to
figure out what's going on there.

Are these three known bugs sufficient to prevent posting a 0.9.1 build
for NetBSD on the file release page?  I can supply one, if not.

--
Richard