#1813: t/op/infnan.t: failures on Darwin/PPC
---------------------+------------------------------------------------------
Reporter: jkeenan | Owner: plobsing
Type: bug | Status: new
Priority: normal | Milestone:
Component: core | Version: 2.8.0
Severity: medium | Keywords:
Lang: | Patch:
Platform: darwin |
---------------------+------------------------------------------------------
Changes (by plobsing):
* owner: dukeleto => plobsing
* status: assigned => new
Comment:
Replying to [comment:5 jkeenan]:
> Replying to [comment:3 plobsing]:
> > Status Update:
> > * the responsible changes were reverted on trunk and moved to
branches/infnan
> > * the affected test files are t/op/inf_nan.t, t/dynoplibs/trans-
infnan.t, and t/op/string.t
> > * this problem is only manifesting on darwin/ppc/g++. gcc does not
exhibit this problem
>
> Unfortunately, I cannot confirm that the source of these test failures
is the use of g++. I have gotten the same pattern of test failures on
Darwin/PPC with three different configurations.
>
> First, my 'standard' approach, i.e., the same one I've used on this box
since late 2006:
> {{{
> CC="/usr/bin/gcc"
> CX="/usr/bin/g++"
> /usr/local/bin/perl Configure.pl --cc="$CC" \
> --cxx="$CX" --link="$CX" \
> --ld="$CX" \
> --configure_trace \
> $@
> }}}
> This uses g++ for `--link` and `--ld`. This led to the errors reported
originally in this TT as well as TT #1814 and TT #1815.
>
> Second, stripping away all the command-line options:
> {{{
> perl Configure.pl
> }}}
> This results in the same failures. This is not really surprising
because, in the absence of command-line options, ''Configure.pl'' will
make its selections for `--link` and `--ld` on the basis of the `%Config`
from the Perl used to invoke ''Configure.pl''. And, on this box, once I
follow all the symbolic links, you end up with g++ for `--link` and `--ld`
just as I got in the first case.
>
> Third: I set `--link` and `--ld` on the command line to gcc, thereby
overriding (I think) what ''Configure.pl'' would normally default to.
> {{{
> perl Configure.pl --cc=/usr/bin/gcc --cxx=/usr/bin/g++ \
> > --link=/usr/bin/gcc --ld=/usr/bin/gcc --configure_trace
> }}}
> In other words, I purge g++ from the configuration and build as much as
I know how to. In ''lib/Parrot/Config/Generated.pm'', this leads to the
following values:
> {{{
> 'ld' => '/usr/bin/gcc',
> ...
> 'link' => '/usr/bin/gcc',
> }}}
> But I still get:
> {{{
> t/op/inf_nan.t
> (Wstat: 0 Tests: 37 Failed: 10)
> Failed tests: 1, 10-12, 16-17, 25-28
> t/op/string.t
> (Wstat: 0 Tests: 291 Failed: 4)
> Failed tests: 280-283
> t/dynoplibs/trans-infnan.t
> (Wstat: 0 Tests: 70 Failed: 10)
> Failed tests: 1, 7-8, 18-19, 58, 61, 64, 67-68
> }}}
> ... as reported in
[http://smolder.parrot.org/app/projects/report_details/554 Smolder 554].
>
> So it's not, or not simply, g++. I wish I had better news to report.
>
> Thank you very much.
>
> kid51
Can you retest on origin/infnan2 branch? It makes use of math.h inf and
nan test functions where available.
--
Ticket URL: <https://trac.parrot.org/parrot/ticket/1813#comment:7>
Parrot <https://trac.parrot.org/parrot/>
Parrot Development
_______________________________________________
parrot-tickets mailing list
[email protected]
http://lists.parrot.org/mailman/listinfo/parrot-tickets
