At 12:08 AM 5/17/2002 -0400, Michael G Schwern wrote:
>On Wed, May 08, 2002 at 04:30:16PM -0500, Craig A. Berry wrote:
>> This tests out fine against bleadperl on OpenVMS Alpha 7.1 and 7.2-1. There
>> were a few test failures when built with Perl 5.6.1 on VMS 7.1 but I don't
>> know if this was expected to work. I wouldn't make backporting that high a
>> priority, but if anyone is interested the full test output is below:
>
><snip>
>
>> t/command...........
>> glob failed (can't start child: invalid argument) at
>/DISK8/BERRYC/EXTUTILS-MAKEMAKER-5_93_01/blib/lib/ExtUtils/Command.pm line 53.
>> # Failed test (t/command.t at line 48)
>> # got: '0'
>> # expected: '1'
>> # Failed test (t/command.t at line 49)
>> # undef
>> # doesn't match '(?-xism:00setup_dummy.t)'
>> # Looks like you failed 2 tests of 24.
>
>Could you find out what $file and @ARGV are at line 45?
main::([.t]command.t:12): chdir 't';
DB<1> b 45
DB<2> c
main::([.t]command.t:45):and; ($ARGV[0] = $file) =~ s/.\z/\?/;
DB<2> p $file
00setup_dummy.t
DB<4> p $ARGV[0]
Use of uninitialized value in print at (eval 7)[/perl_root/lib/perl5db.pl:1521] line 2.
DB<5> p scalar(@ARGV);
0
DB<6> s
main::([.t]command.t:46): ExtUtils::Command::expand_wildcards();
DB<6> p $ARGV[0]
00setup_dummy.?
Is this because you are feeding '?' to glob as a wildcard character? On
VMS, '*' means match many characters and '%' means match one character. '?'
has no special wildcard meaning as this example illustrates:
$ perl -e "print join(' | ',glob('t/i*.%'));"
t/inst.t | t/installed.t | t/inst_prefix.t
$ perl -e "print join(' | ',glob('t/i*.?'));"
$
>> t/installed.........
>> Can't cd to (/perl_root/lib/site_perl/VMS_AXP/auto/dbd/sybase/diag/) sybase : no
>such file or directory
>> Can't cd to /perl_root/lib/site_perl/VMS_AXP/auto/dbd/sybase/bin../../.. at
>/perl_root/lib/File/Find.pm line 535.
>> # Looks like you planned 42 tests but only ran 15.
>> # Looks like your test died just after 15.
>
>This is just never going to work. Too deep.
Are we intentiolly rooting through site_perl?
>> t/mkbootstrap.......
>> # Failed test (t/mkbootstrap.t at line 87)
>> # ''
>> # doesn't match '(?-xism:Unable to open dasboot\.bs)'
>> # Looks like you failed 1 tests of 18.
>
>Possibly tripping up on the different meaning of -w on VMS. It's expecting
>to be able to write read-only files and uses -w to confirm this.
It's just that I had too many privs. It's creating a read-only file but
allowing me to create a higher-versioned file of the same name. If I
downgrade privileges it won't let me do that. However, due to a bug that's
only recently been fixed, 5.6.1 will use the account's default privileges
rather than the actual current process privileges when checking with -w,
which could in some cases throw things off for this test. I'd leave it alone.
