$ENV{foo}=undef #12337
Comments
From @cpansproutperldelta for 5.17.3 has this under Incompatible Changes: =head2 C<$ENV{foo}=undef> deletes value from environ, like C<delete $ENV{foo}> This facilitates use of C<local()> with C<%ENV> entries. In previous versions I don’t recall seeing this discussed. This change makes Perl less consistent with itself. For undef to be converted to the empty string is normal. For $hash{key} = undef to delete the hash element is abnormal. And I don’t see what it provides, except inconsistency. We already have delete local. I think this should be reverted. Flags: Site configuration information for perl 5.17.3: Configured by sprout at Mon Jul 30 23:46:13 PDT 2012. Summary of my perl5 (revision 5 version 17 subversion 3) configuration: Locally applied patches: @INC for perl 5.17.3: Environment for perl 5.17.3: |
From @doyOn Sun, Aug 19, 2012 at 12:20:21PM -0700, Father Chrysostomos wrote:
+1 -doy |
|
The RT System itself - Status changed from 'new' to 'open' |
From Eirik-Berg.Hanssen@allverden.noOn Mon, Aug 20, 2012 at 11:19 AM, Jesse Luehrs <doy@tozt.net> wrote:
Briefly discussed, July 25th. Questioned: Answered: But yeah, the answer seems a bit too brief. As you note, C<< delete As it is part of a bigger change, I wouldn't ask for it all to be Eirik |
From @cpansproutOn Mon Aug 20 03:16:41 2012, Eirik-Berg.Hanssen@allverden.no wrote:
I do recall that now, but I did not understand it at the time. I
Anything’s doable. :-) It’s probably simple, too. -- Father Chrysostomos |
From [Unknown Contact. See original ticket]On Mon Aug 20 03:16:41 2012, Eirik-Berg.Hanssen@allverden.no wrote:
I do recall that now, but I did not understand it at the time. I
Anything’s doable. :-) It’s probably simple, too. -- Father Chrysostomos |
From @rjbs* Father Chrysostomos via RT <perlbug-comment@perl.org> [2012-08-20T11:26:08]
I also thought so. I wonder whether Chip was not aware that "delete local" had been made to work; Chip, can you weigh in on this? -- |
From @cpansproutOn Fri Aug 24 18:20:10 2012, perl.p5p@rjbs.manxome.org wrote:
It turns out that perldelta entry that inspired this ticket is wrong. $ ./perl -Ilib -le '$ENV{PATH} = undef; print exists $ENV{PATH}; local Nothing more to do here, as far as I’m concerned. -- Father Chrysostomos |
From [Unknown Contact. See original ticket]On Fri Aug 24 18:20:10 2012, perl.p5p@rjbs.manxome.org wrote:
It turns out that perldelta entry that inspired this ticket is wrong. $ ./perl -Ilib -le '$ENV{PATH} = undef; print exists $ENV{PATH}; local Nothing more to do here, as far as I’m concerned. -- Father Chrysostomos |
|
@cpansprout - Status changed from 'open' to 'resolved' |
From @xsawyerxOn 10/13/2012 04:03 PM, Father Chrysostomos via RT wrote:
|
From @cpansproutOn Mon Oct 15 15:04:28 2012, xsawyerx@gmail.com wrote:
No. The VMS issue only have to do with clobbering %ENV as a whole, as -- Father Chrysostomos |
From [Unknown Contact. See original ticket]On Mon Oct 15 15:04:28 2012, xsawyerx@gmail.com wrote:
No. The VMS issue only have to do with clobbering %ENV as a whole, as -- Father Chrysostomos |
From @chipdudeOn Fri, Aug 24, 2012 at 09:19:40PM -0400, Ricardo Signes wrote:
Indeed I wasn't aware. The below code works (no die) on 5.14: # Test that $ENV{X} = undef does *not* need a special case So special case of C<$ENV{X}=undef> performing unsetenv is a mistake and |
|
@cpansprout - Status changed from 'resolved' to 'open' |
From @rjbsOn Tue Oct 16 09:44:42 2012, rev.chip@gmail.com wrote:
This has yet to be resolved. -- |
From @kentfredricOn Fri Feb 08 07:59:37 2013, rjbs wrote:
I gave a shot at patching this and it seemed like the code required was Included is tests for the undef =~ "" behaviour, and tests pass. I read C<man setenv> and it doesn't seem to specify anywhere that |
From @kentfredric0001-mg.c-revert-ENV-x-undef-behaviour-to-be-empty-string.patchFrom 0c265b21c8a971f03c7c90a8c325621f586a96d0 Mon Sep 17 00:00:00 2001
From: Kent Fredric <kentfredric@gmail.com>
Date: Sun, 17 Feb 2013 06:07:59 +1300
Subject: [PATCH] mg.c : revert ENV{x} = undef behaviour to be empty string,
not key deletion
pod/perldelta.pod: document reversion of ENV{foo} = undef behaviour in delta
t/op/magic.t: add a test for ENV{foo} = undef
---
mg.c | 2 +-
pod/perldelta.pod | 7 +++++++
t/op/magic.t | 6 +++++-
3 files changed, 13 insertions(+), 2 deletions(-)
diff --git a/mg.c b/mg.c
index e4711e7..6811727 100644
--- a/mg.c
+++ b/mg.c
@@ -1101,7 +1101,7 @@ Perl_magic_setenv(pTHX_ SV *sv, MAGIC *mg)
dVAR;
STRLEN len = 0, klen;
const char * const key = MgPV_const(mg,klen);
- const char *s = NULL;
+ const char *s = "";
PERL_ARGS_ASSERT_MAGIC_SETENV;
diff --git a/pod/perldelta.pod b/pod/perldelta.pod
index 3fc33ac..c7d6e34 100644
--- a/pod/perldelta.pod
+++ b/pod/perldelta.pod
@@ -53,6 +53,13 @@ XXX For a release on a stable branch, this section aspires to be:
If any exist, they are bugs, and we request that you submit a
report. See L</Reporting Bugs> below.
+=head2 C<$ENV{foo} = undef> no longer deletes value from environ
+
+5.17.3 Introduced a change where assiging C<undef> to an C<%ENV> key was equivalent
+to C<delete $ENV{foo}>.
+
+This release reverts that change.
+
[ List each incompatible change as a =head2 entry ]
=head1 Deprecations
diff --git a/t/op/magic.t b/t/op/magic.t
index 990de55..5421d01 100644
--- a/t/op/magic.t
+++ b/t/op/magic.t
@@ -5,7 +5,7 @@ BEGIN {
chdir 't' if -d 't';
@INC = '../lib';
require './test.pl';
- plan (tests => 178);
+ plan (tests => 179);
}
# Test that defined() returns true for magic variables created on the fly,
@@ -647,9 +647,13 @@ SKIP: {
}
$ENV{__NoNeSuCh} = 'foo';
+ $ENV{__NoNeSuCh2} = 'foo';
$0 = 'bar';
env_is(__NoNeSuCh => 'foo', 'setting $0 does not break %ENV');
+ $ENV{__NoNeSuCh2} = undef;
+ env_is(__NoNeSuCh2 => '', 'setting a key as undef does not delete it');
+
# stringify a glob
$ENV{foo} = *TODO;
env_is(foo => '*main::TODO', 'ENV store of stringified glob');
--
1.7.12.4
|
From @kentfredricJust regenerated the patch with a cleanup of the ENV variable I added for |
From @kentfredric0001-mg.c-revert-ENV-x-undef-behaviour-to-be-empty-string.patchFrom 0c265b21c8a971f03c7c90a8c325621f586a96d0 Mon Sep 17 00:00:00 2001
From: Kent Fredric <kentfredric@gmail.com>
Date: Sun, 17 Feb 2013 06:07:59 +1300
Subject: [PATCH] mg.c : revert ENV{x} = undef behaviour to be empty string,
not key deletion
pod/perldelta.pod: document reversion of ENV{foo} = undef behaviour in delta
t/op/magic.t: add a test for ENV{foo} = undef
---
mg.c | 2 +-
pod/perldelta.pod | 7 +++++++
t/op/magic.t | 6 +++++-
3 files changed, 13 insertions(+), 2 deletions(-)
diff --git a/mg.c b/mg.c
index e4711e7..6811727 100644
--- a/mg.c
+++ b/mg.c
@@ -1101,7 +1101,7 @@ Perl_magic_setenv(pTHX_ SV *sv, MAGIC *mg)
dVAR;
STRLEN len = 0, klen;
const char * const key = MgPV_const(mg,klen);
- const char *s = NULL;
+ const char *s = "";
PERL_ARGS_ASSERT_MAGIC_SETENV;
diff --git a/pod/perldelta.pod b/pod/perldelta.pod
index 3fc33ac..c7d6e34 100644
--- a/pod/perldelta.pod
+++ b/pod/perldelta.pod
@@ -53,6 +53,13 @@ XXX For a release on a stable branch, this section aspires to be:
If any exist, they are bugs, and we request that you submit a
report. See L</Reporting Bugs> below.
+=head2 C<$ENV{foo} = undef> no longer deletes value from environ
+
+5.17.3 Introduced a change where assiging C<undef> to an C<%ENV> key was equivalent
+to C<delete $ENV{foo}>.
+
+This release reverts that change.
+
[ List each incompatible change as a =head2 entry ]
=head1 Deprecations
diff --git a/t/op/magic.t b/t/op/magic.t
index 990de55..5421d01 100644
--- a/t/op/magic.t
+++ b/t/op/magic.t
@@ -5,7 +5,7 @@ BEGIN {
chdir 't' if -d 't';
@INC = '../lib';
require './test.pl';
- plan (tests => 178);
+ plan (tests => 179);
}
# Test that defined() returns true for magic variables created on the fly,
@@ -647,9 +647,13 @@ SKIP: {
}
$ENV{__NoNeSuCh} = 'foo';
+ $ENV{__NoNeSuCh2} = 'foo';
$0 = 'bar';
env_is(__NoNeSuCh => 'foo', 'setting $0 does not break %ENV');
+ $ENV{__NoNeSuCh2} = undef;
+ env_is(__NoNeSuCh2 => '', 'setting a key as undef does not delete it');
+
# stringify a glob
$ENV{foo} = *TODO;
env_is(foo => '*main::TODO', 'ENV store of stringified glob');
--
1.7.12.4
|
From [Unknown Contact. See original ticket]Just regenerated the patch with a cleanup of the ENV variable I added for |
From @cpansproutOn Sat Feb 16 10:46:39 2013, kentfredric wrote:
Thank you. Applied as 888a67f. -- Father Chrysostomos |
From [Unknown Contact. See original ticket]On Sat Feb 16 10:46:39 2013, kentfredric wrote:
Thank you. Applied as 888a67f. -- Father Chrysostomos |
|
@cpansprout - Status changed from 'open' to 'resolved' |
|
@rjbs - Status changed from 'resolved' to 'open' |
From @rjbsThis was applied, but only after not being smoke-me'd first. It breaks Win32, t/op/magic.t -- |
From @tonycozOn Wed Feb 20 18:56:12 2013, rjbs wrote:
Steve fixed it, see: http://www.nntp.perl.org/group/perl.daily-build.reports/2013/02/msg137334.html Re-closing. Tony |
|
@tonycoz - Status changed from 'open' to 'resolved' |
From @craigberryOn Sat, Feb 16, 2013 at 12:46 PM, Kent Fredric via RT
If you're referring to the snippets of eccentricity in mg.c and hv.c, I'm actually not sure what to do about this change on VMS. ok 157 - setting $0 does not break %ENV What's happening here is that when we set an environment variable to When we retrieve such a value from Perl, we correctly recognize that $ perl -e "$ENV{XYZ}='foo'; $ENV{XYZ}=undef; print qq/exists but but the env_is() subroutine in magic.t is retrieving that null byte by |
Migrated from rt.perl.org#114504 (status was 'resolved')
Searchable as RT114504$
The text was updated successfully, but these errors were encountered: