Regression in Data::Dumper output between 5.18.4 and 5.28. #17176

p5pRT · 2019-10-07T12:05:37Z

Migrated from rt.perl.org#134483 (status was 'open')

Searchable as RT134483$

p5pRT · 2019-10-07T12:05:37Z

From @demerphq

I have noticed that since 5.18 the output of Data::Dumper has changed
at least once in an undocumented way. This is very annoying. I
thought there was a way to disable the XS code, but I dont see it
anymore. Regardless, optimizations have been made to DD that are clear
regressions. It would be nice if no optimizations we applied unless
they did not change the output.

Notice that even though we selected Useqq() so there should be no
single quoted constructs, in 5.28 we still get single quoted values.

$ perl -MData::Dumper -le'print Data::Dumper->new([[ 0xFFFFFFFF,
0xFFFFFFFFF ]])->Useqq(1)->Dump(); print $^V'
$VAR1 = [
"4294967295",
"68719476735"
];

v5.18.4

$ perl -MData::Dumper -le'print Data::Dumper->new([[ 0xFFFFFFFF,
0xFFFFFFFFF ]])->Useqq(1)->Dump(); print $^V'
$VAR1 = [
4294967295,
'68719476735'
];

v5.28.1

$ perl -v

This is perl 5, version 28, subversion 1 (v5.28.1) built for x86_64-linux

Perl may be copied only under the terms of either the Artistic License or the
GNU General Public License, which may be found in the Perl 5 source kit.

Complete documentation for Perl, including FAQ lists, should be found on
this system using "man perl" or "perldoc perl". If you have access to the
Internet, point your browser at http://www.perl.org/, the Perl Home Page.

--
perl -Mre=debug -e "/just|another|perl|hacker/"

p5pRT · 2019-10-07T14:59:39Z

From @jkeenan

On Mon, 07 Oct 2019 12:05:37 GMT, demerphq wrote:

I have noticed that since 5.18 the output of Data::Dumper has changed
at least once in an undocumented way. This is very annoying. I
thought there was a way to disable the XS code, but I dont see it
anymore. Regardless, optimizations have been made to DD that are clear
regressions. It would be nice if no optimizations we applied unless
they did not change the output.

Notice that even though we selected Useqq() so there should be no
single quoted constructs, in 5.28 we still get single quoted values.

$ perl -MData::Dumper -le'print Data::Dumper->new([[ 0xFFFFFFFF,
0xFFFFFFFFF ]])->Useqq(1)->Dump(); print $^V'
$VAR1 = [
"4294967295",
"68719476735"
];

v5.18.4

$ perl -MData::Dumper -le'print Data::Dumper->new([[ 0xFFFFFFFF,
0xFFFFFFFFF ]])->Useqq(1)->Dump(); print $^V'
$VAR1 = [
4294967295,
'68719476735'
];

v5.28.1

There was a change, but it was a conscious change. See:
https://rt-archive.perl.org/perl5/Ticket/Display.html?id=74798

Using the attached program, I was able to bisect the change using this invocation:

#####
perl Porting/bisect.pl --start=v5.18.0 --end=v5.20.0 -- ./perl -Ilib ~/tmp/134483-data-dumper-output.pl
#####

Output:

#####
9baac1a is the first bad commit
commit 9baac1a
Author: Slaven Rezic <srezic@iconmobile.com>
Date: Wed Jul 10 14:18:18 2013 +1000

Data::Dumper: useqq implementation for xs

Tests are mainly unchanged, just a "cheat" and a couple of TODOs were removed.
#####

Thank you very much.

--
James E Keenan (jkeenan@cpan.org)

p5pRT · 2019-10-07T14:59:39Z

From @jkeenan

134483-data-dumper.pl

p5pRT · 2019-10-07T14:59:39Z

The RT System itself - Status changed from 'new' to 'open'

p5pRT · 2019-10-07T16:13:28Z

From @demerphq

Ok, well I'll assume this is an oversight, but the implementation in that
patch is clearly wrong, if useqq is selected there should not be any single
quoted strings. This broke code that depends on useqq being respected.

Yves

On Mon, 7 Oct 2019, 16:59 James E Keenan via RT, <perlbug-followup@perl.org>
wrote:

On Mon, 07 Oct 2019 12:05:37 GMT, demerphq wrote:

I have noticed that since 5.18 the output of Data::Dumper has changed
at least once in an undocumented way. This is very annoying. I
thought there was a way to disable the XS code, but I dont see it
anymore. Regardless, optimizations have been made to DD that are clear
regressions. It would be nice if no optimizations we applied unless
they did not change the output.

Notice that even though we selected Useqq() so there should be no
single quoted constructs, in 5.28 we still get single quoted values.

$ perl -MData::Dumper -le'print Data::Dumper->new([[ 0xFFFFFFFF,
0xFFFFFFFFF ]])->Useqq(1)->Dump(); print $^V'
$VAR1 = [
"4294967295",
"68719476735"
];

v5.18.4

$ perl -MData::Dumper -le'print Data::Dumper->new([[ 0xFFFFFFFF,
0xFFFFFFFFF ]])->Useqq(1)->Dump(); print $^V'
$VAR1 = [
4294967295,
'68719476735'
];

v5.28.1

There was a change, but it was a conscious change. See:
https://rt-archive.perl.org/perl5/Ticket/Display.html?id=74798

Using the attached program, I was able to bisect the change using this
invocation:

#####
perl Porting/bisect.pl --start=v5.18.0 --end=v5.20.0 -- ./perl -Ilib
~/tmp/134483-data-dumper-output.pl
#####

Output:

#####
9baac1a is the first bad commit
commit 9baac1a
Author: Slaven Rezic <srezic@iconmobile.com>
Date: Wed Jul 10 14:18:18 2013 +1000

Data::Dumper: useqq implementation for xs

Tests are mainly unchanged, just a "cheat" and a couple of TODOs were
removed.
#####

Thank you very much.

--
James E Keenan (jkeenan@cpan.org)

p5pRT · 2019-10-07T16:23:08Z

From @jkeenan

On 10/7/19 12:12 PM, demerphq wrote:

Ok, well I'll assume this is an oversight, but the implementation in
that patch is clearly wrong, if useqq is selected there should not be
any single quoted strings. This broke code that depends on useqq being
respected.

Yves

Corrected implementation welcome (though we'd have to discuss the timing
of its application, as this code has been out in the wild for 6 years).

On Mon, 7 Oct 2019, 16:59 James E Keenan via RT,
<perlbug-followup@perl.org <mailto:perlbug-followup@perl.org>> wrote:

On Mon\, 07 Oct 2019 12&#8203;:05&#8203;:37 GMT\, demerphq wrote&#8203;:
 > I have noticed that since 5\.18 the output of Data&#8203;::Dumper has changed
 > at least once  in an undocumented way\. This is very annoying\. I
 > thought there was a way to disable the XS code\, but I dont see  it
 > anymore\. Regardless\, optimizations have been made to DD that are
clear
 > regressions\. It would be nice if no optimizations we applied unless
 > they did not change the output\.
 >
 > Notice that even though we selected Useqq\(\) so there should be no
 > single quoted constructs\, in 5\.28 we still get single quoted values\.
 >
 > $ perl \-MData&#8203;::Dumper \-le'print Data&#8203;::Dumper\->new\(\[\[ 0xFFFFFFFF\,
 > 0xFFFFFFFFF \]\]\)\->Useqq\(1\)\->Dump\(\); print $^V'
 > $VAR1 = \[
 >           "4294967295"\,
 >           "68719476735"
 >         \];
 >
 > v5\.18\.4
 >
 > $ perl \-MData&#8203;::Dumper \-le'print Data&#8203;::Dumper\->new\(\[\[ 0xFFFFFFFF\,
 > 0xFFFFFFFFF \]\]\)\->Useqq\(1\)\->Dump\(\); print $^V'
 > $VAR1 = \[
 >           4294967295\,
 >           '68719476735'
 >         \];
 >
 > v5\.28\.1

There was a change\, but it was a conscious change\.  See&#8203;:
https://rt-archive.perl.org/perl5/Ticket/Display.html?id=74798

Using the attached program\, I was able to bisect the change using
this invocation&#8203;:

\#\#\#\#\#
perl Porting/bisect\.pl \<http&#8203;://bisect\.pl> \-\-start=v5\.18\.0
\-\-end=v5\.20\.0 \-\- \./perl \-Ilib ~/tmp/134483\-data\-dumper\-output\.pl
\<http&#8203;://134483\-data\-dumper\-output\.pl>
\#\#\#\#\#

Output&#8203;:

\#\#\#\#\#
9baac1a3613bd641a847683d7877b3cfab3244bc is the first bad commit
commit 9baac1a3613bd641a847683d7877b3cfab3244bc
Author&#8203;: Slaven Rezic \<srezic@&#8203;iconmobile\.com
\<mailto&#8203;:srezic@&#8203;iconmobile\.com>>
Date&#8203;:   Wed Jul 10 14&#8203;:18&#8203;:18 2013 \+1000

Data&#8203;::Dumper&#8203;: useqq implementation for xs

Tests are mainly unchanged\, just a "cheat" and a couple of TODOs
were removed\.
\#\#\#\#\#

Thank you very much\.

\-\- 
James E Keenan \(jkeenan@&#8203;cpan\.org \<mailto&#8203;:jkeenan@&#8203;cpan\.org>\)

p5pRT · 2019-10-07T16:44:41Z

From @Tux

On Mon, 7 Oct 2019 12:22:43 -0400, James E Keenan <jkeenan@pobox.com>
wrote:

On 10/7/19 12:12 PM, demerphq wrote:

Ok, well I'll assume this is an oversight, but the implementation in
that patch is clearly wrong, if useqq is selected there should not
be any single quoted strings. This broke code that depends on useqq
being respected.

Yves

Corrected implementation welcome (though we'd have to discuss the
timing of its application, as this code has been out in the wild for
6 years).

And it obviously needs additional tests

--
H.Merijn Brand http://tux.nl Perl Monger http://amsterdam.pm.org/
using perl5.00307 .. 5.31 porting perl5 on HP-UX, AIX, and Linux
https://useplaintext.email https://tux.nl http://www.test-smoke.org
http://qa.perl.org http://www.goldmark.org/jeff/stupid-disclaimers/

khwilliamson · 2022-04-17T17:47:55Z

@demerphq, what to do about this ticket?

p5pRT added the Severity Low label Oct 19, 2019

jkeenan added the dist-Data-Dumper issues in the dual-life blead-first Data-Dumper distribution label Jul 5, 2021

xenu removed the Severity Low label Dec 29, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Regression in Data::Dumper output between 5.18.4 and 5.28. #17176

Regression in Data::Dumper output between 5.18.4 and 5.28. #17176

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

khwilliamson commented Apr 17, 2022

Regression in Data::Dumper output between 5.18.4 and 5.28. #17176

Regression in Data::Dumper output between 5.18.4 and 5.28. #17176

Comments

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

From @demerphq

p5pRT commented Oct 7, 2019

From @jkeenan

p5pRT commented Oct 7, 2019

From @jkeenan

p5pRT commented Oct 7, 2019

p5pRT commented Oct 7, 2019

From @demerphq

p5pRT commented Oct 7, 2019

From @jkeenan

p5pRT commented Oct 7, 2019

From @Tux

khwilliamson commented Apr 17, 2022