maintperl@22000 status on VMS
Craig A. Berry
I see the following failures (with bug tracking info included where
available):
lib/ExtUtils/t/Constant..............FAILED at test 27
http://rt.cpan.org/NoAuth/Bug.html?id=4687
lib/ExtUtils/t/INST_PREFIX...........FAILED at test 41
no ticket #, but see
http://www.nntp.perl.org/group/perl.vmsperl/12215
lib/Test/Harness/t/callback..........FAILED at test 17
lib/Test/Harness/t/inc_taint.........FAILED at test 1
lib/Test/Harness/t/strap-analyze.....FAILED at test 93
lib/Test/Harness/t/test-harness......FAILED at test 164
http://rt.cpan.org/NoAuth/Bug.html?id=4745
lib/vmsish...........................FAILED at test 25
Failed 7 test scripts out of 781, 99.10% okay.
This is a genuine bug in using stat() while the "vmsish 'time'"
pragma is in effect.
Other serious issues include:
MakeMaker recompiles and reliinks all the extensions when making the
test target (for reasons different from those that caused similar
behavior recently on other platforms). See
http://rt.cpan.org/NoAuth/Bug.html?id=4675
Extension building dies on older VMS systems because of portability
issues with chmod on a directory. See:
http://rt.cpan.org/NoAuth/Bug.html?id=4676
On VAX, the basic test num.t fails because test #47 assumes a greater
floating point range than is available by default. See
http://www.nntp.perl.org/group/perl.perl5.porters/86742
The MakeMaker issues are probably the most serious ones and are all
new since 5.8.2. Patches for most of them exist in some form, either
in the rt queue or in vmsperl discussions. I'm a bit confused about
what I should do given the code freeze deadline, holiday schedules,
and Nick's understandable reluctance to integrate new things from
CPAN at the last minute. I guess I'll take a stab at throwing
together a consolidated VMS-only patch today and see what happens.
--
________________________________________
Craig A. Berry
mailto:craig...@mac.com
"... getting out of a sonnet is much more
difficult than getting in."
Brad Leithauser
Nicholas Clark
> With maintperl@22000 using Compaq C V6.5-001 on OpenVMS Alpha V7.3-1,
> I see the following failures (with bug tracking info included where
> available):
Thanks for testing. I was worried that some of the recent changes may have
completely broken the build on VMS
Was that a threaded or a non-threaded build? If you didn't do both, could you
try the other?
> The MakeMaker issues are probably the most serious ones and are all
> new since 5.8.2. Patches for most of them exist in some form, either
> in the rt queue or in vmsperl discussions. I'm a bit confused about
> what I should do given the code freeze deadline, holiday schedules,
> and Nick's understandable reluctance to integrate new things from
> CPAN at the last minute. I guess I'll take a stab at throwing
> together a consolidated VMS-only patch today and see what happens.
But I'm quite happy to rollback MakeMaker at the last minute, to 6.17
(the version that shipped with 5.8.2)
I'm happier to do this than to slip 2 weeks for a new MM, given that
there's no guarantee that it will not exhibit some other problem that
in turn prevents its inclusion.
Nicholas Clark
Craig A. Berry
>But I'm quite happy to rollback MakeMaker at the last minute, to 6.17
>(the version that shipped with 5.8.2)
Will that break the sdbm build? I lost track of the exact sequence
of fixes and counterfixes between sdbm and MakeMaker.
Rafael Garcia-Suarez
> At 7:58 PM +0000 12/30/03, Nicholas Clark wrote:
> >But I'm quite happy to rollback MakeMaker at the last minute, to 6.17
> >(the version that shipped with 5.8.2)
>
> Will that break the sdbm build? I lost track of the exact sequence
> of fixes and counterfixes between sdbm and MakeMaker.
No ; the sdbm build broke afterwards.
Craig A. Berry
>> With maintperl@22000 using Compaq C V6.5-001 on OpenVMS Alpha V7.3-1,
>> I see the following failures (with bug tracking info included where
>> available):
>
>Thanks for testing. I was worried that some of the recent changes may have
>completely broken the build on VMS
>Was that a threaded or a non-threaded build? If you didn't do both, could you
>try the other?
That was threaded:
$ search config.sh archname
archname='VMS_AXP-ithread'
I'll try a non-threaded, but they have been doing the same recently
(about a week ago, don't remember the patchlevel).
Rafael Garcia-Suarez
> lib/Test/Harness/t/callback..........FAILED at test 17
> lib/Test/Harness/t/inc_taint.........FAILED at test 1
> lib/Test/Harness/t/strap-analyze.....FAILED at test 93
> lib/Test/Harness/t/test-harness......FAILED at test 164
>
> http://rt.cpan.org/NoAuth/Bug.html?id=4745
I applied that patch as #22016 to blead. It's also in Andy's copy
so the next Test::Harness release will have it.