@_ autovivifies hash slices

p5pRT commented 18 years ago

Migrated from rt.perl.org#37441 (status was 'open')

Searchable as RT37441$

p5pRT commented 18 years ago

From @kenahoo

This is a bug report for perl from ken@mathforum.org\, generated with the help of perlbug 1.34 running under perl v5.8.1.

As the following three one-liners demonstrate\, perl is autovivifying some hash entries in a case it seems like it ought not to:

% perl -le 'sub foo{} foo(@x{'a','b'}); print for keys %x'
b
a
% perl -le 'sub foo{} foo($x{a},$x{b}); print for keys %x'
% perl -le 'sub foo{$_=5 for @_} foo($x{a},$x{b}); print for keys %x'
a
b

The main difference is between the first two examples\, which my intuition tells me ought to be equivalent. The third example shows that even when not using slices (and thus not autovivifying unnecessarily)\, @_ still manages to create read/write aliases to the original hash entries.

Credit goes to Thomas L. Shinnick\, who figured out why File::Spec->tmpdir was autovivifying an undef value for $ENV{TMPDIR}.

-Ken

Perl Info

``` Flags: category=core severity=medium --- Site configuration information for perl v5.8.1: Configured by root at Fri Sep 12 19:46:46 PDT 2003. Summary of my perl5 (revision 5.0 version 8 subversion 1 RC3) configuration: Platform: osname=darwin, osvers=7.0, archname=darwin-thread-multi-2level uname='darwin hampsten 7.0 darwin kernel version 6.0: fri jul 25 16:58:41 pdt 2003; root:xnu-344.frankd.rootsxnu-344.frankd~objrelease_ppc power macintosh powerpc ' config_args='-ds -e -Dprefix=/usr -Dccflags=-g -pipe -Dldflags=-Dman3ext=3pm -Duseithreads -Duseshrplib' hint=recommended, useposix=true, d_sigaction=define usethreads=define use5005threads=undef useithreads=define usemultiplicity=define useperlio=define d_sfio=undef uselargefiles=define usesocks=undef use64bitint=undef use64bitall=undef uselongdouble=undef usemymalloc=n, bincompat5005=undef Compiler: cc='cc', ccflags ='-g -pipe -pipe -fno-common -DPERL_DARWIN -no-cpp-precomp -fno-strict-aliasing -I/usr/local/include', optimize='-Os', cppflags='-no-cpp-precomp -g -pipe -pipe -fno-common -DPERL_DARWIN -no-cpp-precomp -fno-strict-aliasing -I/usr/local/include' ccversion='', gccversion='3.3 20030304 (Apple Computer, Inc. build 1495)', gccosandvers='' intsize=4, longsize=4, ptrsize=4, doublesize=8, byteorder=4321 d_longlong=define, longlongsize=8, d_longdbl=define, longdblsize=8 ivtype='long', ivsize=4, nvtype='double', nvsize=8, Off_t='off_t', lseeksize=8 alignbytes=8, prototype=define Linker and Libraries: ld='MACOSX_DEPLOYMENT_TARGET=10.3 cc', ldflags ='-L/usr/local/lib' libpth=/usr/local/lib /usr/lib libs=-ldbm -ldl -lm -lc perllibs=-ldl -lm -lc libc=/usr/lib/libc.dylib, so=dylib, useshrplib=true, libperl=libperl.dylib gnulibc_version='' Dynamic Linking: dlsrc=dl_dyld.xs, dlext=bundle, d_dlsymun=undef, ccdlflags=' ' cccdlflags=' ', lddlflags='-bundle -undefined dynamic_lookup -L/usr/local/lib' Locally applied patches: RC3 --- @INC for perl v5.8.1: /sw/lib/perl5/5.8.1/darwin-thread-multi-2level /sw/lib/perl5/5.8.1 /sw/lib/perl5 /sw/lib/perl5/darwin /Users/ken/lib/darwin-thread-multi-2level /Users/ken/lib /System/Library/Perl/5.8.1/darwin-thread-multi-2level /System/Library/Perl/5.8.1 /Library/Perl/5.8.1/darwin-thread-multi-2level /Library/Perl/5.8.1 /Library/Perl /Network/Library/Perl/5.8.1/darwin-thread-multi-2level /Network/Library/Perl/5.8.1 /Network/Library/Perl . --- Environment for perl v5.8.1: DYLD_LIBRARY_PATH (unset) HOME=/Users/ken LANG (unset) LANGUAGE (unset) LC_ALL=C LD_LIBRARY_PATH (unset) LOGDIR (unset) PATH=/sw/bin:/sw/sbin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/bin:/ usr/local/teTeX/bin/powerpc-apple-darwin-current:/usr/local/bin:/opt/ local/bin:/Users/ken/bin:/Developer/Tools:.:/usr/X11R6/bin PERL5LIB=/sw/lib/perl5:/sw/lib/perl5/darwin:/Users/ken/lib PERL_BADLANG (unset) SHELL=/bin/tcsh ```

p5pRT commented 12 years ago

From @doy

This is still true in 5.16.

p5pRT commented 12 years ago

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

p5pRT commented 12 years ago

From @demerphq

On 21 June 2012 19:47\, Jesse Luehrs via RT \perlbug\-followup@perl\.org wrote:

This is still true in 5.16.

But why is it a bug? Afaik it is supposed to happen.

sub one { $_=1 for @_ }

one(@foo{qw(x y z)});

print $_\,":"\, $foo{$_}\, $/ for keys %foo;

__END__

y:1 x:1 z:1

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

p5pRT commented 12 years ago

From @doy

Sure\, it's clearly necessary in that case. But:

Sure, it's clearly necessary in that case. But:

sub nop { }
my (%foo, %bar);
nop(@foo{'a', 'b'});
nop($bar{'a'}, $bar{'b'});
say "foo:";
say for sort keys %foo;
say "bar:";
say for sort keys %bar;

__END__
foo:
a
b
bar:

I don't see a reason why %foo and %bar should end up being any different there.

-doy

p5pRT commented 12 years ago

From @demerphq

Ok\, I see. That makes sense. Maybe rename the bug?

Yves

p5pRT commented 11 years ago

From @jkeenan

@doy @demerphq

Dicussion in this RT petered out about a year ago. Do you see any way we could move this ticket toward resolution?

Thank you very much. Jim Keenan

p5pRT commented 11 years ago

From @cpansprout

This is on my to-do list\, but rather low. Please leave it open if you will.

--

Father Chrysostomos

p5pRT commented 11 years ago

From @cpansprout

In case someone wants to beat me to it\, it’s matter of extending defelem magic (search for PERL_MAGIC_defelem in pp_hot.c) to hslice and aslice.

--

Father Chrysostomos

p5pRT commented 11 years ago

From ruz@bestpractical.com

On Tue\, Jun 18\, 2013 at 8:40 AM\, Father Chrysostomos via RT \< perlbug-followup@perl.org> wrote:

In case someone wants to beat me to it\, it’s matter of extending defelem magic (search for PERL_MAGIC_defelem in pp_hot.c) to hslice and aslice.

As far as I recall similar situation applies to for loops:

perl -E '1 for @h{qw(1 2 3)}; say %h' 132

But it spreads over more operations:

perl -E '1 for $h{a}\, $h{b}; say %h' ab

Sorry\, if it's not releated.

Perl / perl5