Abe Timmerman wrote:
Here's the output of mmk test_harness:

Failed Test                       Stat Wstat Total Fail  Failed  List of Failed
-------------------------------------------------------------------------------
[-.ext.Compress.Zlib.t]04def.t      44  1024  1769    2   0.11%  42-43

I have not investigated this yet.

[-.ext.List.Util.t]p_tainted.t                  ??   ??       %  ??

This is failing because the module is being invoked in VMS filename syntax so __FILE__ is returning a VMS format pathname. The test is assuming that __FILE__ is returning a UNIX format pathname.

[-.lib.Archive.Tar.t]04_resolved_   44  1024     9    1  11.11%  9

I have not investigated this yet.

[-.lib.ExtUtils.t]FIRST_MAKEFILE.   44  1024     7    2  28.57%  4-5

On OpenVMS, the file "jakefile" should have a trailing "." unless the OpenVMS CRTL is put in UNIX filename mode. Attached is a patch first_makefile_t.gdiff to fix the compare.

[-.lib.ExtUtils.t]Manifest.t        44  1024    49    2   4.08%  26 29
[-.lib.ExtUtils.t]PL_FILES.t        44  1024     9    5  55.56%  3-7

I think that these are failing because of the issue addressed by the patch mm_vms_pm.gdiff, but it has been a while since I investigated this.

[-.lib.File]Compare.t                           13    1   7.69%  12
[-.lib.Test.Simple.t]is_deeply_fa               73    1   1.37%  73

I have to re-investigate these, I seem to remember that this is a problem in the test code that is not VMS specific.

[-.lib]vmsish.t                                 25    4  16.00%  22-25

I have not looked at this yet.

io/crlf_through.t                              942  404  42.89%  4-5 13-14 17-
> io/through.t 942 210 22.29% 4-5 13-14 17-

Still have not looked at the "through" tests. There was a recent change to VMS.C that improved the number of subtests that passed.

op/stat.t                                       86    3   3.49%  81-83

I do not recall looking at this one.

pod/pod2usage2.t                    44  1024    15   12  80.00%  2-3 5-13 15

I looked this up but forgot what I found, but it does not look like VMS will be passing this test anytime soon.

 (10 subtests UNEXPECTEDLY SUCCEEDED), 90 tests and 284 subtests skipped.

What does "UNEXPECTEDLY SUCCEEDED" mean?

-John
[EMAIL PROTECTED]
Personal Opinion Only
--- /rsync_root/perl/lib/ExtUtils/t/FIRST_MAKEFILE.t    Sat May 21 05:45:55 2005
+++ lib/ExtUtils/t/FIRST_MAKEFILE.t     Sun Sep 18 21:57:37 2005
@@ -36,5 +36,5 @@
 
 ok( -e 'jakefile', 'FIRST_MAKEFILE honored' );
 
-ok( grep(/^Writing jakefile for Big::Dummy/, @mpl_out) == 1,
+ok( grep(/^Writing jakefile(?:\.)? for Big::Dummy/, @mpl_out) == 1,
                                        'Makefile.PL output looks right' );
--- /rsync_root/perl/lib/ExtUtils/MM_VMS.pm     Tue Aug  2 07:09:07 2005
+++ lib/ExtUtils/MM_VMS.pm      Sun Sep 18 21:58:46 2005
@@ -447,10 +447,20 @@
     $self->{NOOP}               = 'Continue';
     $self->{NOECHO}             ||= '@ ';
 
-    $self->{MAKEFILE}           ||= 'Descrip.MMS';
+    $self->{MAKEFILE}          ||= $self->{FIRST_MAKEFILE} || 'Descrip.MMS';
     $self->{FIRST_MAKEFILE}     ||= $self->{MAKEFILE};
     $self->{MAKE_APERL_FILE}    ||= 'Makeaperl.MMS';
-    $self->{MAKEFILE_OLD}       ||= '$(FIRST_MAKEFILE)_old';
+    $self->{MAKEFILE_OLD}       ||= 
$self->eliminate_macros('$(FIRST_MAKEFILE)_old');
+#
+#   if an extension is not specified, then MMS/MMK assumes an
+#   an extension of .MMS.  If there really is no extension,
+#   then a trailing "." needs to be appended to specifify a
+#   a null extension.
+#
+    $self->{MAKEFILE} .= '.' unless $self->{MAKEFILE} =~ m/\./;
+    $self->{FIRST_MAKEFILE} .= '.' unless $self->{FIRST_MAKEFILE} =~ m/\./;
+    $self->{MAKE_APERL_FILE} .= '.' unless $self->{MAKE_APERL_FILE} =~ m/\./;
+    $self->{MAKEFILE_OLD} .= '.' unless $self->{MAKEFILE_OLD} =~ m/\./;
 
     $self->{MACROSTART}         ||= '/Macro=(';
     $self->{MACROEND}           ||= ')';

Reply via email to