Please do not reply directly to this email. All additional comments should be made in the comments box of this bug report.
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=185542 Summary: perlbug #38657: Using import() with arguments with -d: broke in 5.8.8, was okay in 5.8.7 Product: Red Hat Certified Stacks Version: LAMPv1 Platform: All URL: http://rt.perl.org/rt3/index.html?q=38657 OS/Version: Linux Status: NEW Severity: normal Priority: normal Component: perl AssignedTo: [EMAIL PROTECTED] ReportedBy: [EMAIL PROTECTED] QAContact: [EMAIL PROTECTED] CC: [email protected],[EMAIL PROTECTED] +++ This bug was initially created as a clone of Bug #183553 +++ Description of problem: Copied from perlbug #38657: http://rt.perl.org/rt3/index.html?q=38657 : Using import() with arguments with -d: broke in 5.8.8, was okay in 5.8.7. $ mkdir -p lib/Devel $ echo 'package Devel::Foo; sub import { print "import(@_)" } sub \ DB::DB { } 1' > lib/Devel/Foo.pm $ /tmp/jhi/p587/bin/perl -wIlib -d:Foo -le 1 import(Devel::Foo) $ /tmp/jhi/p588/bin/perl -wIlib -d:Foo -le 1 import(Devel::Foo) $ /tmp/jhi/p587/bin/perl -wIlib -d:Foo=bar -le 1 import(Devel::Foo bar) $ /tmp/jhi/p588/bin/perl -wIlib -d:Foo=bar -le 1 Can't find string terminator ";" anywhere before EOF. $ Version-Release number of selected component (if applicable): perl-5.8.8 How reproducible: 100% Steps to Reproduce: See above -- Additional comment from [EMAIL PROTECTED] on 2006-03-01 16:53 EST -- I have reproduced this problem, and confirmed the code above used to work on perl-5.8.6 . Suggested fix: --- From: "Rafael Garcia-Suarez" <[EMAIL PROTECTED]> To: [email protected] Date: 2006-03-01 16:16 On 3/1/06, via RT jhi @ ugli. hut. fi <[EMAIL PROTECTED]> wrote: > $ /tmp/jhi/p587/bin/perl -wIlib -d:Foo=bar -le 1 > import(Devel::Foo bar) > $ /tmp/jhi/p588/bin/perl -wIlib -d:Foo=bar -le 1 > Can't find string terminator ";" anywhere before EOF. Bad news, guys. This patch solves it : ==== //depot/perl/perl.c#736 - /home/rafael/p4blead/perl.c ==== --- /home/rafael/tmp/tmp.9616.0 2006-03-01 22:18:07.000000000 +0100 +++ /home/rafael/p4blead/perl.c 2006-03-01 22:18:04.000000000 +0100 @@ -3031,7 +3031,7 @@ Perl_moreswitches(pTHX_ char *s) sv_catpv(sv, start); else { sv_catpvn(sv, start, s-start); - Perl_sv_catpvf(aTHX_ sv, " split(/,/,q%c%s%c)", 0, ++s, 0); + Perl_sv_catpvf(aTHX_ sv, " split(/,/,q(%s))", ++s); } s += strlen(s); my_setenv("PERL5DB", SvPV_nolen_const(sv)); That means that using \0 as a q() delimiter no longer works. --- Now testing this. -- Additional comment from [EMAIL PROTECTED] on 2006-03-01 17:08 EST -- Actually, the above fix was for bleadperl; I think it's probably better to revert to 5.8.7's code for the above, which was: --- /* We now allow -d:Module=Foo,Bar */ while(isALNUM(*s) || *s==':') ++s; if (*s != '=') sv_catpv(sv, start); else { sv_catpvn(sv, start, s-start); sv_catpv(sv, " split(/,/,q{"); sv_catpv(sv, ++s); sv_catpv(sv, "})"); } --- -- Additional comment from [EMAIL PROTECTED] on 2006-03-01 17:11 EST -- Hold on guys, the patch I posted on P5P is just here to demonstrate the cause of the problem, not to be applied. q\0foo\0 ought to be equivalent to 'foo'. It was working in previous perls. Still looking for a proper fix (sorry for haven't been clear enough) -- Additional comment from [EMAIL PROTECTED] on 2006-03-01 17:21 EST -- Yes, I noticed that - I'm not applying anything yet - I'm also still investigating - many thanks, Rafael -- Additional comment from [EMAIL PROTECTED] on 2006-03-01 18:07 EST -- Actually, under perl-5.8.0, perl-5.8.5, perl-5.8.6, and perl-5.8.7, 'q\0foo\0' does NOT work (I have tested all versions) - I get the same result : $ perl -e '$s=q\0foo\0;' Number found where operator expected at -e line 1, near "q\0foo\0" syntax error at -e line 1, near "q\0foo\0" Execution of -e aborted due to compilation errors. This is because '\' is a legal delimiter for q(), and the parse breaks on the last '0;' . $ perl -e '$s=q\0foo\; print $s,"\n";' 0foo But if I actually create a file with 'q{0x0}foo{0x0}', it DOES work OK on all versions: $ cat tq.pl #!/usr/bin/perl $s=q0foo0; print "$s\n"; $ ./tq.pl q0foo0 $ tr '0' '\0' < tq.pl > tq0.pl $ od -c tq0.pl 0000000 # ! / u s r / b i n / p e r l \n 0000020 $ s = q \0 f o o \0 ; \n p r i n t 0000040 " $ s \ n " ; \n 0000051 $ ./tq0.pl foo So q\0foo\0 with real, unescaped 0x0 chars DOES work OK on all versions . I think it is just that 0x0 is a bad choice for a '%c' in Perl_sv_catpvf, as it terminates the C string, and the interpreter doesn't get the rest of the expression . So I think the first fix posted is OK . If we really must make Perl_sv_catpvf handle 0x0 chars OK in expressions, that is a different problem - I think it is debatable whether it is a problem worth fixing. -- Additional comment from [EMAIL PROTECTED] on 2006-03-01 18:14 EST -- Actually it works, yes, but due to shell quoting, you need to be clever : $ perl -le 'print eval "q\0foo\0"' foo and that was what I meant. The problem was with the setenv, a bit later, because setenv expects \0-terminated strings. Fixed upstream now. Looks like my first patch was correct after all :) -- Additional comment from [EMAIL PROTECTED] on 2006-03-01 18:47 EST -- fixed with perl-5.8.8-4, now in rawhide -- Additional comment from [EMAIL PROTECTED] on 2006-03-01 22:22 EST -- No, now it is in rawhide (moving now). I'm guessing that Jason has been very careful about testing this change like many previous changes that I've seen him do. Great work Jason. -- Configure bugmail: https://bugzilla.redhat.com/bugzilla/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug, or are watching someone who is. -- Fedora Extras Perl SIG http://www.fedoraproject.org/wiki/Extras/SIGs/Perl Fedora-perl-devel-list mailing list [email protected] https://www.redhat.com/mailman/listinfo/fedora-perl-devel-list
