[PATCH t/op/stat.t] Declaring -t STDIN w/pipe TODO

Something funny is going on with the -t test in t/op/stat.t (#44).
When run manually, it passes.  When run with mmk test or perl harness,
it fails.  This leads me to believe that test.com and Test::Harness
are both interfering with STDIN on VMS.

    perl t/op/stat.t                # pass
    perl t/op/stat.t | cat          # pass
    perl -e 1 | perl t/op/stat.t    # fail

test.com is doing:
    $ MCR Sys$Disk:[]Perl. "-I[-.lib]" $switch $test 2>&1|

and Test::Harness does:
    MCR $1$dkb300:[schwern.src.bleadperl.][000000]perl.exe;1  op/stat.t|

I thought maybe it was MCR's fault, but I stripped that out (I don't
think its necessary anymore) and still the test failed.

So I can only conclude that something inside VMS's pipe implementation
is interfering with STDIN.  I don't see any movement on fixing this,
so this declares it TODO.

--- t/op/stat.t	2001/12/07 22:30:22	1.1
+++ t/op/stat.t	2001/12/07 22:33:08
@@ -284,7 +284,11 @@
         close(TTY);
     }
     ok(! -t TTY,    '!-t on closed TTY filehandle');
-    ok(-t,          '-t on STDIN');
+
+    {
+        local $TODO = 'STDIN not a tty when output is to pipe' if $Is_VMS;
+        ok(-t,          '-t on STDIN');
+    }
 }
 
 my $Null = File::Spec->devnull;



-- 

Michael G. Schwern   <schwern@pobox.com>    http://www.pobox.com/~schwern/
Perl Quality Assurance	    <perl-qa@perl.org>	       Kwalitee Is Job One
stretch your colon out,
put some effort into it,
and shit through that paste.
	-- japhy

• [PATCH t/op/stat.t] Declaring -t STDIN w/pipe TODO by Michael G Schwern
• Re: [PATCH t/op/stat.t] Declaring -t STDIN w/pipe TODO by Jarkko Hietaniemi