"Craig A. Berry" <[EMAIL PROTECTED]> writes:
>
> The root of most of the test failures seems to be an exceeded quota.
...
> I believe the recent piping mods use mailboxes with rather large
> buffers, so we might be seeing more of this particular error in the
> future.
1. The piping code uses several mailboxes per subprocess.
2. The default is to use large mailboxes, because that's what we used in
the past, at least partially to avoid "subprocess hung because MBX filled up"
problems.
3. The test suite uses subprocesses with wild abandon.
Put 1, 2 & 3 together and your i/o quotas are in for some major abuse.
But try setting the logical:
$ DEFINE PERL_MBX_SIZE 256
before running the tests and you'll likely avoid this particular source
of problems.
> At 7:12 PM -0400 9/3/00, John Peacock wrote:
>>This is a build failure report for perl from [EMAIL PROTECTED],
>>generated with the help of perlbug 1.32 running under perl v5.7.0.
> [snip]
>>I do know that we have a somewhat old compiler ()DEC C V5.6-003 on
>>OpenVMS
>>Alpha V7.1), but Perl itself seems to compile and work OK. That line in
>>VMS.C is:
>>
>> _ckvmssts(sys$crembx(0,chan,mbxbufsiz,mbxbufsiz,0,0,0));
>>
>>but I am a gross novice in VMS, so I am immediately out of my depth. I
>>can only offer my system for testing purposes to try and figure out what
>>is wrong. If it helps, I did try "SET PROC/PRIV=ALL" before running the
>>tests (I should have all rights) and it had no effect.
>>u=20.29 s=0 cu=0 cs=0 files=239 tests=10272
>>
>>Fatal VMS error (status=10772) at PERL_SRC:[PERL-5_7_0]VMS.C;1, line
>>1007
>>at [.io]openpid.t line 48.
>>[.io]openpid............FAILED on test 2
--
Drexel University \V --Chuck Lane
======]---------->--------*------------<-------[===========
(215) 895-1545 _/ \ Particle Physics
FAX: (215) 895-5934 /\ /~~~~~~~~~~~ [EMAIL PROTECTED]
