On Sat, Aug 29, 2009 at 07:39:20AM +0100, Nicholas Clark wrote:
> There's a bit of context, and then it goes BOOM!
>
> On Fri, Aug 28, 2009 at 08:20:34PM +0100, Nicholas Clark wrote:
> > I don't see why it won't work, so I've committed it as
> > 1d8d82d3d8b73345e6276774
I can replicate the exact failures for anything annotated "runperl" with this
patch on OS X:
--- a/t/test.pl
+++ b/t/test.pl
@@ -518,6 +518,8 @@ sub runperl {
my $runperl = &_create_runperl;
my $result;
+return if $Perl =~ m!^/!;
+
my $tainted = ${^TAINT};
my %args = @_;
exists $args{switches} && grep m/^-T$/, @{$args{switches}} and $tainted = $
It's to do with paths that contain a directory 't' being made absolute on
Win32. But I don't think that it's *all* cases of tests that spawn child
perl programs.
> On Sat, Aug 29, 2009 at 04:21:00AM +0100, Steve Hay wrote:
> > Automated smoke report for 5.11.0 patch
> > a2c4b2d6db383103b96e4733616b17d0a3bb0bd9 blead-1969-ga2c4b2d
>
> > Failures: (common-args) -DCCTYPE=MSVC90
> > -DINST_TOP=$(INST_DRV)\Smoke\doesntexist
> > [default] -Duselargefiles
> > [default] -DDEBUGGING -Duselargefiles
> > [default] -Duselargefiles -Dusemymalloc
> > [default] -DDEBUGGING -Duselargefiles -Dusemymalloc
> > ../ext/B/t/concise.t........................................FAILED
> > 2-5, 148-157
uses runperl in t/test.pl
> > ../ext/B/t/lint.t...........................................FAILED
> > 1-29
uses runperl in t/test.pl
> > ../ext/B/t/o.t..............................................FAILED
> > 2-5, 7-9
uses runperl in t/test.pl
> > ../ext/B/t/optree_check.t...................................FAILED
> > 11-13
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/B/t/optree_concise.t.................................FAILED
> > 12-14, 16
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/B/t/optree_constants.t...............................FAILED
> > 27-28
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/B/t/optree_misc.t....................................FAILED
> > 2
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/B/t/optree_samples.t.................................FAILED
> > 11, 20
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/B/t/optree_sort.t....................................FAILED
> > 3, 5, 7, 9
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/B/t/optree_specials.t................................FAILED
> > 2-8
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/B/t/optree_varinit.t.................................FAILED
> > 6-8, 13-15, 19-22
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/B/t/showlex.t........................................FAILED
> > 2-3, 9-10
uses runperl in t/test.pl (via ext/B/t/OptreeCheck.pm)
> > ../ext/Devel-DProf/t/DProf.t................................FAILED
> > Bad plan. You planned 20 tests but ran 0.
Gets the perl to run from which_perl in t/test.pl
But I can't see where things are going wrong.
> > ../ext/Devel-PPPort/t/ppphtest.t............................FAILED
> > 2-10, 12-15, 18-23, 27, 29-30, 32-40, 42-45
> > 47-50, 55, 57-61, 63-64, 66-68, 70-71, 73-75
> > 80-81, 83-85, 87, 89-94, 96, 98, 100-113
> > 115-129, 131, 134, 136-138, 142-144, 147
> > 149-158, 160-162, 164-182, 184-189, 191-209
> > 211-214, 216-223, 225-226, 228-230, 232-234
Seems to use a very similar logic to which_perl, but I can't actually work
out where it is going wrong
> > ../ext/IO-Compress/t/010examples-bzip2.t....................FAILED
> > 2-4, 6-8, 10-12, 14-15, 17-19
$Perl = qq["$Perl"] if $^O eq 'MSWin32' ;
> > ../ext/IO-Compress/t/010examples-zlib.t.....................FAILED
> > 2-4, 6-8, 10-12, 14-15, 17-19
$Perl = qq["$Perl"] if $^O eq 'MSWin32' ;
> > ../ext/IO-Compress/t/cz-05examples.t........................FAILED
> > 2-4, 6-8, 10-12, 14-15, 17-19, 21-22, 24-26
$Perl = qq["$Perl"] if $^O eq 'MSWin32' ;
> > ../ext/Test-Harness/t/perl5lib.t............................FAILED
> > 1
sub has_crazy_patch {
my $sentinel = 'blirpzoffle';
local $ENV{PERL5LIB} = $sentinel;
my $command = join ' ',
map {qq{"$_"}} ( $^X, '-e', 'print join q(:), @INC' );
my $path = `$command`;
my @got = ( $path =~ /($sentinel)/g );
return @got > 1;
}
but that's to determine whether to skip, so I can't figure out why it
takes out the entire test if it fails to execute the contents of the ``
Is `` fatal on Win32 sometimes?
> > ../ext/Test-Harness/t/process.t.............................FAILED
> > 1-27
TAP::Parser::Iterator::Process uses IPC::Open3::open3, which deep down is
doing this:
$pid = eval { system 1, @_ }; # 1 == P_NOWAIT
I can't work out why that is failing.
> > ../ext/re/t/lexical_debug.t.................................FAILED
> > 1, 3, 5, 7-10
uses runperl in t/test.pl
> > ../ext/re/t/regop.t.........................................FAILED
> > 1, 3-19
> > Bad plan. You planned 54 tests but ran 19.
uses runperl in t/test.pl
> I think that the connection is that all the failures spawn a new perl
> But I'm not sure why they don't like spawning ../../t/perl when ../../perl
> worked just fine. Something interpolating, where "\." is \ . but "\t" is
> a tab character? But what has that sort of interpolation rule?
>
> Doesn't everything in Perl that interpolates "\t" to a tab also interpolate
> "\." to a dot?
and runperl calls this code:
my $runperl = which_perl();
if ($runperl =~ m/\s/) {
$runperl = qq{"$runperl"};
}
which, I assume, means that we're passing something like this to the Win32
command line:
"C:\Path with Space\perl\t\perl" -le "print qq{Hello world}"
Does it treat \t inside "" as something to interpolate to a tab character, but
leave "unknown" escape sequences as 2 characters?
If so, is there any way in Win32 to have the path above safely escaped?
Nicholas Clark
