Re: [Sbcl-help] bus error on new darwin ppc binary

[Nikodemus S. <nik...@ra...>]

On Sun, Nov 2, 2008 at 1:42 PM, Giovanni Gigante <gi...@ci...> wrote:
>
>> I believe current SBCL assumes at least 10.4.
>>
>> In src/runtime/Config.ppc-darwin, try changing the line
>>
>> CFLAGS = -g -Wall -O2 -fdollars-in-identifiers -mmacosx-version-min=10.4
>>
>> to
>>
>> CFLAGS = -g -Wall -O2 -fdollars-in-identifiers -mmacosx-version-min=10.3
>>
>> and build from source using the 1.0.2.
>>
>> Let us know how it goes.
>
> I have built 1.0.22 from source using 1.0.2 and removing the
> --macosx-version-min.

Great. Brian (or someone else): any ideas how to best

 1. detect 10.3 at runtime? (So we can give a sensible error message.)

 2. detect 10.3 at build-time? (So we can omit the minimum version stuff.)

> The compilation apparently succeeded, but when I run the tests, I get this:
>
> Finished running tests.
> Status:
> Expected failure:    float.pure.lisp / (SCALE-FLOAT-OVERFLOW BUG-372)
> Expected failure:    float.pure.lisp / (ADDITION-OVERFLOW BUG-372)
> Expected failure:    debug.impure.lisp / (UNDEFINED-FUNCTION BUG-346)
> Failure:             debug.impure.lisp / (UNDEFINED-FUNCTION BUG-353)

This is non-serious: the test in question fails pretty much
everywhere, but has just been marked as expected to fail. It has
possibly been failing forever on PPC as well? (And the test is for
backtrace niceness when calling an undefined function, so it's not a
huge issue.)

> Unhandled error      dynamic-extent.impure.lisp

This is probably just a DX feature that is not implemented on PPC, but
has not been marked as expected to fail there. Can you run

  sh run-test.sh dynamic-extent.impure.lisp &> log

and then look for the error in the log, and send it to the list?

Cheers,

 -- Nikodemus