Bug #38848 for Test-Simple: eq_array changes the magic nature of the "expected" array

Fri Aug 29 14:47:06 2008 djerius [...] cpan.org - Ticket created

Subject:

eq_array changes the magic nature of the "expected" array

When comparing an input array with magic values to a fiducial array without, eq_array transfers magic to the fiducial array. Here's an example using YAML (which creates a magic value): use YAML; use Devel::Peek; use Test::More; my $obj = [ 'snake' ]; YAML::DumpFile( 'foo', $obj ); Dump $obj; my $array = YAML::LoadFile( 'foo' ); eq_array( $array, $obj ); Dump $obj; And the result: % perl eq_array.pl SV = RV(0xe4d660) at 0xe4d650 REFCNT = 1 FLAGS = (PADMY,ROK) RV = 0xe21df0 SV = PVAV(0xe22fb0) at 0xe21df0 REFCNT = 1 FLAGS = () ARRAY = 0xe396b0 FILL = 0 MAX = 0 ARYLEN = 0x0 FLAGS = (REAL) Elt No. 0 SV = PV(0xe1eb68) at 0xe22060 REFCNT = 1 FLAGS = (POK,pPOK) PV = 0xe35560 "snake"\0 CUR = 5 LEN = 8 SV = RV(0xe4d660) at 0xe4d650 REFCNT = 1 FLAGS = (PADMY,ROK) RV = 0xe21df0 SV = PVAV(0xe22fb0) at 0xe21df0 REFCNT = 1 FLAGS = (RMG) MAGIC = 0x1032e00 MG_VIRTUAL = &PL_vtbl_arylen_p MG_TYPE = PERL_MAGIC_arylen_p(@) MG_FLAGS = 0x02 REFCOUNTED MG_OBJ = 0x1520110 SV = PVMG(0x1124160) at 0x1520110 REFCNT = 1 FLAGS = (GMG,SMG,pIOK) IV = 0 NV = 0 PV = 0 MAGIC = 0x152bd10 MG_VIRTUAL = &PL_vtbl_arylen MG_TYPE = PERL_MAGIC_arylen(#) MG_OBJ = 0xe21df0 ARRAY = 0xe396b0 FILL = 0 MAX = 0 ARYLEN = 0x1520110 FLAGS = (REAL) Elt No. 0 SV = PV(0xe1eb68) at 0xe22060 REFCNT = 1 FLAGS = (POK,pPOK) PV = 0xe35560 "snake"\0 CUR = 5 LEN = 8 I believe $obj should remain unchanged. The fallout from this is that Clone, which doesn't seem to be able to clone the YAML created hash, also doesn't work on $obj. Thanks, Diab

Fri Aug 29 14:51:00 2008 djerius [...] cpan.org - Correspondence added

My apologies; the output was mangled in the last message. One more time: SV = RV(0x7ce660) at 0x7ce650 REFCNT = 1 FLAGS = (PADMY,ROK) RV = 0x7a2df0 SV = PVAV(0x7a3fb0) at 0x7a2df0 REFCNT = 1 FLAGS = () ARRAY = 0x7ba6b0 FILL = 0 MAX = 0 ARYLEN = 0x0 FLAGS = (REAL) Elt No. 0 SV = PV(0x79fb68) at 0x7a3060 REFCNT = 1 FLAGS = (POK,pPOK) PV = 0x7b6560 "snake"\0 CUR = 5 LEN = 8 SV = RV(0x7ce660) at 0x7ce650 REFCNT = 1 FLAGS = (PADMY,ROK) RV = 0x7a2df0 SV = PVAV(0x7a3fb0) at 0x7a2df0 REFCNT = 1 FLAGS = (RMG) MAGIC = 0xe1c410 MG_VIRTUAL = &PL_vtbl_arylen_p MG_TYPE = PERL_MAGIC_arylen_p(@) MG_FLAGS = 0x02 REFCOUNTED MG_OBJ = 0xea1040 SV = PVMG(0xaa50e0) at 0xea1040 REFCNT = 1 FLAGS = (GMG,SMG,pIOK) IV = 0 NV = 0 PV = 0 MAGIC = 0xe1c440 MG_VIRTUAL = &PL_vtbl_arylen MG_TYPE = PERL_MAGIC_arylen(#) MG_OBJ = 0x7a2df0 ARRAY = 0x7ba6b0 FILL = 0 MAX = 0 ARYLEN = 0xea1040 FLAGS = (REAL) Elt No. 0 SV = PV(0x79fb68) at 0x7a3060 REFCNT = 1 FLAGS = (POK,pPOK) PV = 0x7b6560 "snake"\0 CUR = 5 LEN = 8

Fri Aug 29 14:51:13 2008 djerius [...] cpan.org - Status changed from 'new' to 'open'

Fri Aug 29 14:54:00 2008 djerius [...] cpan.org - Correspondence added

On Fri Aug 29 14:51:00 2008, DJERIUS wrote: Show quoted text

> My apologies; the output was mangled in the last message. One more time:

Aargh. Once again mangled. Please see attached file.

Download eq_array.output
application/octet-stream 1.1k

Message body not shown because it is not plain text.

Fri Aug 29 15:15:37 2008 djerius [...] cpan.org - Correspondence added

It seems that the input need not have magic. A revised example using YAML::Tiny, which doesn't create magic values is attached. It looks like both of the arrays passed to eq_array are changed. The example shows that both arrays are now incorrectly handled by Clone. I'm including that just for completeness; I'm not sure if it's Clone's problem or Test::More's.

Download eq_array.output
application/octet-stream 2.4k

Message body not shown because it is not plain text.

use YAML::Tiny qw( DumpFile LoadFile ); use Test::More; use Clone qw( clone ); use Data::Dumper; use Devel::Peek; my $obj = [ 'snake' ]; DumpFile( 'foo', $obj ); my $array = LoadFile( 'foo' ); print STDERR "Before (\$array):\n"; Dump $array; print STDERR "Before (\$obj):\n"; Dump $obj; eq_array( $array, $obj ); print STDERR "\nAfter (\$array):\n"; Dump $array; print STDERR "\nAfter (\$obj):\n"; Dump $obj; print STDERR "\nCloned \$array:\n"; print STDERR Dumper clone( $array ); print STDERR "\nCloned \$obj:\n"; print STDERR Dumper clone( $obj );

Sat Sep 06 15:17:24 2008 schwern [...] pobox.com - Correspondence added

Subject:	Re: [rt.cpan.org #38848] eq_array changes the magic nature of the "expected" array
Date:	Sat, 06 Sep 2008 12:15:45 -0700
To:	bug-Test-Simple [...] rt.cpan.org
From:	Michael G Schwern <schwern [...] pobox.com>

Diab Jerius via RT wrote: Show quoted text

> Queue: Test-Simple > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=38848 > > > It seems that the input need not have magic. A revised example using > YAML::Tiny, which doesn't create magic values is attached. It looks > like both of the arrays passed to eq_array are changed. The example > shows that both arrays are now incorrectly handled by Clone. I'm > including that just for completeness; I'm not sure if it's Clone's > problem or Test::More's.

Sorry, I can't reproduce this. What version of perl and all the involved modules (Test::More, Clone, YAML::Tiny) are you using? Also, eq_array() is discouraged. Use is_deeply(). -- We do what we must because we can. For the good of all of us, Except the ones who are dead. -- Jonathan Coulton, "Still Alive"

Sat Sep 06 18:15:32 2008 mschwern [...] cpan.org - Severity Unimportant added

Sun Nov 09 04:08:37 2008 mschwern [...] cpan.org - Status changed from 'open' to 'stalled'

Fri Jun 04 16:26:45 2010 djerius [...] cpan.org - Correspondence added

Show quoted text

> Sorry, I can't reproduce this. What version of perl and all the involved > modules (Test::More, Clone, YAML::Tiny) are you using? > > Also, eq_array() is discouraged. Use is_deeply().

Yikes!! My profuse apologies for letting this slip through the cracks. I've attached a much simpler test case & output which just uses Test::More (version 0.94). Under Perl 5.8.8 (both Linux & OS X Leopard) the arrays passed to eq_array do not acquire magic through that call. Using Perl 5.10.0 (both Linux & OS X Leopard) they do.

Subject:

test2.pl

use Test::More; use Devel::Peek; select STDERR; print "Perl = $]\n"; print "Test::More = ", Test::More->VERSION, "\n\n"; my $obj = [ 'snake' ]; my $array = [ 'snake' ]; print "Before (\$array):\n"; Dump $array; print "Before (\$obj):\n"; Dump $obj; eq_array( $array, $obj ); print "\nAfter (\$array):\n"; Dump $array; print "\nAfter (\$obj):\n"; Dump $obj;

Subject:

eq_array_added_magic.out

Download eq_array_added_magic.out
application/octet-stream 2.4k

Message body not shown because it is not plain text.

Subject:

eq_array_no_added_magic.out

Download eq_array_no_added_magic.out
application/octet-stream 1.6k

Message body not shown because it is not plain text.

Fri Jun 04 16:26:46 2010 The RT System itself - Status changed from 'stalled' to 'open'

Fri Jun 04 17:46:15 2010 mschwern [...] cpan.org - Correspondence added

On Fri Jun 04 16:26:45 2010, DJERIUS wrote: Show quoted text

> Yikes!! My profuse apologies for letting this slip through the cracks. > > I've attached a much simpler test case & output which just uses > Test::More (version 0.94). Under Perl 5.8.8 (both Linux & OS X Leopard) > the arrays passed to eq_array do not acquire magic through that call. > Using Perl 5.10.0 (both Linux & OS X Leopard) they do.

Thanks for picking this one up. Looks like you found a bug in perl. I tracked it down and its $#array which triggers magic (oddly enough, $scalar = @array does not). 5.10.0 and 5.10.1 have this problem but 5.8.9, 5.12.0 and 5.12.1 don't. So it looks like it was fixed. $ perl -wle 'my @array = (1,2); use Devel::Peek; Dump \@array; my $idx = $#array; Dump \@array' I don't think there's a whole lot Test::More can do about this. It could eliminate all uses of $#array from the code, but its pretty likely something else will use it. So I'm going to mark this as rejected. Feel free to reopen if you want to.

Fri Jun 04 17:46:16 2010 mschwern [...] cpan.org - Status changed from 'open' to 'rejected'

Bug #38848 for Test-Simple: eq_array changes the magic nature of the "expected" array

Preferred bug tracker