Re: perl-5.005 fails to build on powerpc
On Mon, Jan 24, 2000 at 10:38:44AM -0500, Daniel Jacobowitz wrote:
> On Mon, Jan 24, 2000 at 07:57:26AM -0700, Matt Porter wrote:
> > On Mon, Jan 24, 2000 at 12:10:07AM -0800, Alex Romosan wrote:
> > > Matt Porter <email@example.com> writes:
> > >
> > > > Is anybody else having any luck? ndbm fails in two tests and I don't
> > > > see any talk of this problem. Full test log is in bug #55932.
> > > >
> > >
> > > i had no problems building perl 5.005.03-5.2. i can make the debs
> > > available if somebody wants them.
> > Uh oh, I'm more worried now that my fully upgraded potato box (excluding
> > the broken console-data) can't build 5.005.03-5.x.
> > Are you a developer? Can you make an NMU to master of these?
> Hmm, the build failed for the build daemon also:
> lib/ipc_sysv.......FAILED at test 0
> That might just be the 2.3.10 kernel I'm still running; it has large
> issues but is the only one currently able to boot my machine. By hand
> the test says:
> Test returned status 0 (wstat 11, 0xb)
> And there's always this beautiful sight:
> drow:~buildd/build/perl-5.005-5.005.03/t% ./perl lib/ipc_sysv.t
> ok 1
> ok 2
> zsh: segmentation fault ./perl lib/ipc_sysv.t
I'd suspect a kernel problem with 2.3.10, unless you're missing SysV IPC
support. My 2.2.9 PReP kernel only fails on the anydbm.t and ndbm.t when
it tries an ndbm store. I get an errno 22, invalid argument. This is at
line 101 in anydbm.t and line 106 in ndbm.t. That line tries to make a
call with a "null" argument.
Exluding ndbm support causes the anydbm.t test to fail elsewhere.
This is Linux Country. On a quiet night, you can hear Windows reboot.