On Tue, Oct 31, 2006 at 04:35:04PM -0800, Philip M. Gollucci wrote:
> Tim Bunce wrote:
> >I'd appreciate testing of this release candidate:
> >
> >  http://www.data-plan.com/public/DBI-1.53-RC2.tar.gz
> Hi, these are "stock" installs of perl via source not port on
> FreeBSD 7.0-current / i386.  All configured the same way.
> 
> I got ugly messages in perl versions that don't bundle Storable in CORE, 
> so I batch installed it.

Storable is marked as a prerequisite in Makefile.PL and META.yml.
I think that's sufficient, especially as only very old perl's are
affected. Thanks for the reminder though.

> Also, Test::Pod.

Only used by t/pod.t and should be skipped cleanly if not installed.

> Both of these failed in 
> various versions, but I just left them failed.
> 
> Attached are 19 outputs 5.6.2 5.8.1->5.8.8 with and without threads, and 
> bleed with and without threads. Perl -V are in the individual files.

Thanks!

Could you look into the t/40profile.t failure for 5.6.2 some more:

        t/40profile..............dubious
                Test returned status 9 (wstat 2304, 0x900)
        DIED. FAILED tests 6-8, 26, 31, 35-36, 38-39
                Failed 9/45 tests, 80.00% okay

(upgrading Test::More and/or running the test verbosely may give better
diagnostics.)

> The threaded versions failed the threads test --- ~1.48->1.51 or so 
> didn't seem to do this.

"Seem"? Could you recheck 1.52 and 1.51 for me?

> This is with libpthread.so NOT libc_r.so.

Is libpthread the normal choice on freebsd 7?
(I presume so from your Config command line.)

> I could probably in a reasonable time duplicate this on SunOS 10 and/or 
> netbsd 3.0 if you think its useful.

Given that
 - the pure-perl tests also failed
 - freebsd has a history of threading issues
 - no one else has reported these failures
 - my own tests with a threaded perl don't fail
 - http://cpantesters.perl.org/show/DBI.html#DBI-1.52 shows several
   freebsd variants (threads/64bit) passed - though most were freebsd 5.x
I think this may be a local perl/FreeBSD-7 issue rather than a DBI issue.

So, yes, it would be great if you could rerun these tests on SunOS
and/or netbsd.

Thanks Philip!

Tim.

Reply via email to