David Cantrell wrote:
perltie.pod in the latest 5.8 still talks about MLDBM instead of the
newer shinier more capable DBM::Deep. This rectifies that:
Thanks, applied to bleadperl as change #23410.
--- perltie.pod 2004-10-21 10:02:39.0 +0100
+++ perltie-new.pod 2004-10-21
Stas Bekman wrote:
--- pod/perlsec.pod.orig2004-10-19 20:28:19.0 -0400
+++ pod/perlsec.pod 2004-10-19 21:44:06.521585300 -0400
@@ -154,7 +154,11 @@
Or you may be able to use the following Cis_tainted() function.
sub is_tainted {
-return ! eval { eval(# .
[EMAIL PROTECTED] wrote:
This patch simply adds the Perl Debugger Pocket Reference (O'Reilly) to the
perlfaq2.pod
Please excuse the shameless self-advertising, but since no-one else had added it
yet...
Well, given that it documents a utility that can only be found
in the core, I think
Orton, Yves wrote:
This is IMO both bad advice and wrong. Its bad advice because Data::Dumper
defaults to Purity(0) so the dumps it produces are NOT guaranteed to be
eval'able, and its wrong because you don't need to turn off strictures to
eval it. Better to say that if you are using
Dan Kogai wrote:
Porters,
I just released Encode 2.05.
Just integrated into bleadperl. (#23384)
Russ Allbery wrote:
There isn't any
requirement that Pod::Man be used to generate regular manual pages; it
generates *roff output using the man macros, but is otherwise pretty
generic.
I agree with this. I sometimes write postscript documents by converting
them from pod with pod2man $* |
Peter Kay wrote:
I was using Data::Dumper to write out a (large) data structure to a
file, and then attempting to read it back in with eval. But, it didn't
work!
I finally figured it out: use strict;
When I did 'no strict' before the eval, it worked again.
If you update the Dumper
[EMAIL PROTECTED] wrote:
Yitzchak,
That works just fine - thanks muchly :-)
I just had to swap the line (in MY patch) which read:
if ($Config{useithreads} $ENV{PERL5DB_THREADED}) {
to read:
if ($ENV{PERL5DB_THREADED}) {
I attach the improved (third try) patch FYI.
I built successfull bleadperl on a linux ppc.
I've only one test failure, but I suspect it comes from the unsual setup
of this machine (automounted nfs home dirs).
Failed TestStat Wstat Total Fail Failed List of Failed
Konovalov, Vadim wrote:
attached is continuation of previous one.
it also makes win32/Makefile and win32/makefile.mk closer.
Thanks, applied as #23363.
[EMAIL PROTECTED] (via RT) wrote:
The use of local $^D still works to
locally turn on some debug but
you get a lot of stuff as well:
Fixed, the brute-force way, by change #23365 to blead.
Dintelmann, Peter wrote:
I am not sure I like the carp().
This should either be fatal or silent and return false.
Personally I prefer the silent mode.
--- /opt/perl/lib/5.8.5/sun4-solaris-thread-multi/IO/Poll.pmThu Sep 9
14:45:54 2004
+++ /var/tmp/Poll.pmMon Oct 11
Rick Delaney wrote:
The following patch changes the caching placeholder to
PL_sv_placeholder which apart from being a better name is also better
because now
keys %INC;
should only return the names of files successfully loaded.
I don't think that's better. If we are
[EMAIL PROTECTED] wrote:
Otherwise, perhaps
skip the use threads (and complain if there's a later attempt) unless
$ENV{PERL5DB_THREADED} is true (and have perl.c set PERL5DB_THREADED if
you say -dt).
Hmmm, sounds ok, a quick workaround would be to 'require threads' at
the correct position
Tels wrote:
attached is a patch to sync blead to the upcoming BigInt v1.73. I would like
to get it in to get some testing exposure.
Thanks, applied as #23359.
Vadim Konovalov wrote:
I did modifications so my patch should fit 592.
All test passed.
For explanations on what is going on please see my previous letters
with similar subject.
Thanks, applied to bleadperl as change #23360.
Alexey Tourbin wrote:
Hello,
I would like to propose the following two changes for fixin() function:
- make use of File::Temp in order to make .new and .bak
filenames less predicatble
Given that MakeMaker must work with older perls, and that File::Temp
has been added in perl
Dave Mitchell wrote:
op_free() does a OP_REFCNT_LOCK before decrementing it, but none of
the code that increments it appears to do so.
I'll try to look into this properly when I get more time.
(P5Pers: is there any reason not to just wrap the OpREFCNT_inc/OpREFCNT_dec
macrcos with
Konstantin Stopani (via RT) wrote:
this code gives segmentation fault on my Fedora Core 2/linux-2.6.8.1
#!/usr/bin/perl -w
use strict;
use warnings;
use threads;
use threads::shared;
use encoding utf8; # no segfault if remove
my $t = threads-new(\fn);
sub fn {
print hello\n;
kaminsky (via RT) wrote:
Whan using the syntax
warnings::warn($obj, $msg)
Where $obj is an object with overloaded stringification, the call
fails. The reason is that the default stringification is assumed
for the purpose of determining the class name of the object. I
attach an
Abe Timmerman wrote:
Hi,
I like having the test_harness target for manual builds, it saves time.
So here is a patch.
Thanks, applied as 23358.
Dan Kogai wrote:
Porters,
I have released Encode 2.03 at last; The code was done about 10 days
ago but I needed a tester to verify 'piconv vs. Win32' case. Thanks,
Steve.
BTW, you forgot to look at (and integrate, if you agree) the following
change :
Change 23326 on 2004/09/20 by
Abe Timmerman wrote:
From the smokelog:
Extending failures with harness:
../lib/integer.t ../t/io/dup.t ../t/io/open.t ../t/op/pack.t
# Failed test (../lib/integer.t at line 49)
# got: '-4291708920'
# expected: '-9223372036854775808'
Looks like %Config thinks
Boris Zentner (via RT) wrote:
lisa:~ borisz$ perl -MData::Dumper -le '%h = (1 = 2, a=b, c = 2);
%h = reverse ( %x = reverse %h); print Dumper({x = \%x, h =\%h});'
$VAR1 = {
'h' = {
'' = 2
},
'x' = {
'b' =
Craig A. Berry wrote:
There's a VMS-only test that started failing after the recent upgrading
of File::Spec. The new version of File::Spec::VMS-canonpath correctly
allows a directory name of six zeros when it does not occur in the root
position. In other words, the six zeros should only be
Vadim Konovalov wrote:
Please disregard my previous dynaloader patch and instead use current
one, it should be better WRT $ENV{PERL_BUILD_EXPAND_CONFIG_VARS}.
Thanks, applied as #23348 to bleadperl.
(Hmm, looking at those templates and substitutions, I was thinking about
making TT a part of
Jim Cromie (via RT) wrote:
patch (to be attached shortly), enhances B::Showlex::newlex
(added in #22820) to produce a more readable output, and adds Pod to
document its use (with a flag).
Thanks, applied as #23350 to bleadperl (with wording nits to the POD.)
Konovalov, Vadim wrote:
Actually, as long as same thing should be performed for XS_loader and may be
some other files (lib_pm.pl), it could be reasonable to move subroutine
'expand_os_specific' somewhere (may be to lib/Devel/* directory)?
That's right. Probably a .pl file; I don't know if it
Matthew Harris wrote:
I can enclose a variable in braces to separate it from surrounding
text:
$ perl -we '$foo=hi; print ${foo}\n'
hi
And I can use $::foo to refer to a name in the 'main' package:
$ perl -we '$::foo=hi; print $::foo\n'
hi
But if I try to do both together,
Tels wrote:
Now there is only one more failing tests for me:
ext/B/t/optree_specials.t
I have two, here :
Failed Test Stat Wstat Total Fail Failed List of Failed
---
../ext/B/t/optree_specials.t
Steve Peters via RT wrote:
OK, at least it seems that we're in agreement that this behavior is
operating system dependent and that we don't want to change it since it
could break things for Perl users on environments where an open() on a
directory works.
Yep.
Two points brought up
[EMAIL PROTECTED] (via RT) wrote:
# 13 characters of junk are printed.
# I expect nothing to be printed.
$data = 13 characters;
$data = undef;
open(MEM,'',\$data) or die Fail: $!\n;
while (MEM) {
print;
}
I committed the following fix to the development branch of perl. It's
not
Jay Hannah wrote:
You can change $Sys::Syslog::host and the package will happily go
connect (TCP or UDP) to a remote server. Works great. However, here is
no interface to change $Sys::Syslog::host, and the feature isn't
documented.
Instead of changing openlog(), connect(), connect_udp(),
[EMAIL PROTECTED] (via RT) wrote:
The folling one-liner:
perl -le '@a = qw/2 3/; @a = (new, sort @a); print @a'
Should print new 2 3, and indeed it does on 5.8.2 and previous
versions. However, on version 5.8.4, it prints 2 3.
This bug (a misoptimisation) has been fixed in 5.8.5.
Steve Peters via RT wrote:
This seems to be OK in Perl 5.8.5. Has this fix been placed into the
Perl 5.9 releases?
It appears to have been fixed in bleadperl as well, but the regression
test seems to be missing.
--
Cruel. Her nature. Curious mice never squeal. Seem to like it.
-- Ulysses
Merijn Broeren wrote:
After much hunting we found several issues with perl.c and handling of
the environment variables. We have Perl 5.8 core dumping when run from
make in a 'safe' environment, where perl 5.6 was working happily.
Running 5.8 in a normal environment from make, or in the safe
Ronald J Kimball wrote:
On Wed, Sep 22, 2004 at 06:12:12PM -, gyepi @ praxis-sw. com wrote:
In the following paragraph, found in perlop, the second instance of
-bareword should read +bareword
Unary - performs arithmetic negation if the operand is numeric. If
the
Andy Lester wrote:
perlsec says
For efficiency reasons, Perl takes a conservative view of whether data
is tainted. If an expression contains tainted data, any subexpression
may be considered tainted, even if the value of the subexpression is
not itself affected by the
John Peacock wrote:
What kind of behaviour change can be expected from this patch ?
It only affects modules built with this particular XSUB.h, and at least
for all of the core modules that do so, all tests pass. It is intended
to be completely transparent to the XS module built with this
Robert via RT wrote:
perl -e 'sub _ { }'
5.8.3 - OK
blead - SEGV
Not for me ; are you using threads ?
#0 0x08081563 in Perl_op_free ()
#1 0x080ccd6f in Perl_leave_scope ()
#2 0x080888c5 in Perl_newATTRSUB ()
#3 0x0807fc97 in Perl_yyparse ()
#4 0x08060fb3 in S_parse_body ()
Sam Vilain (via RT) wrote:
h2ph sys/sysmacros.h incorrectly translates this:
# define makedev(major, minor) { unsigned int) (major)) 8) \
| ((unsigned int) (minor))), 0 }
to this:
eval \n#line 10 sys/sysmacros.ph\n . 'sub makedev {
Dave Mitchell wrote:
After debugging, the problem seems to be in some incompatibility
between Carp and Safe.
...
Carp will sometimes do:
{ local $@; require Carp::Heavy; }
Since require isn't allowed within the compartment, Perl raises an exception;
control is unwound to the eval
Steve Hay wrote:
Turning off CRLF translation without switching to byte semantics is, of
course, achieved by ':raw:encoding(utf16le)'! The attached patch fixes
things.
Thanks, applied as #22827.
[EMAIL PROTECTED] (via RT) wrote:
The '$header' variable in the parameter list at line 120 in
lib/h2xs.t is incorrectly placed (it should be the last one). It will
cause the h2xs test to fail as follows:
...
Below is a patch to fix it.
I've applied your patches to the development version of
Thorvaldur Gunnlaugsson wrote:
#!/usr/local/bin/perl -n
$[=1;
BEGIN{
eval 'print $[, in eval'; # prints 0
Contrary to what I mistakenly said in my previous message, $[ is *not*
propagated into eval(). That's consistent with what perlvar implies
(the value of $[ is not seen from other
Jarkko Hietaniemi wrote:
$CPAN/authors/id/J/JH/JHI/Time-HiRes-1.57.tar.gz
Thanks, now applied as #22673 to bleadperl.
Steve Hay wrote:
...
../ext/B/t/optree_samples.t.FAILED 11
Most probably some difference in the output of B::Concise; can you
post the complete output of this test, so we can adjust it
accordingly ?
Ton Hospel wrote in perl.perl5.porters :
Current perl behaviour is proper I think. It's a closure, and at the
moment sub n_factorial is defined $prod is actually still undef. It only
later when the body of the BEGIN runs gets its final value.
Though it might be nice to update the docs with
don't cope with overloaded objects correctly regarding UTF-8 encoding.
(bug #27658.)
http://groups.google.com/groups?selm=rt-3.0.8-27658-82032.6.02101148715519%40perl.org
About this summary
This summary was written by Rafael Garcia-Suarez, from Lyon, France, and
it's likely
H.Merijn Brand wrote in perl.perl5.porters :
... //depot/perl/makedef.pl#156 edit
That would still not cover the AIX failures, I guess?
Only the build failures. No clue -yet- about the miniperl panic.
Nick Ing-Simmons wrote in perl.perl5.porters :
NOT TO BE APPLIED
--- pp.c(revision 3350)
+++ pp.c(working copy)
@@ -797,6 +797,7 @@ PP(pp_undef)
GV* gv = CvGV((CV*)sv);
cv_undef((CV*)sv);
CvGV((CV*)sv) = gv;
+ GvCVGEN(gv) =
Steve Hay wrote:
It changes splitpod to keep track of how many levels deep in =over ...
=back sections it is, and only splits on =item's that are one level deep.
..
OK, final patch attached.
Thanks, applied as #22494.
Quoting Tels [EMAIL PROTECTED]:
is this section in perltodo still relevant for the v5.8.x track or at all?
I had more or less the intent to remove perltodo.pod completely
and put a TODO list for 5.10 in blead. Now, perltodo.pod is not
completely without any good ideas...
Chip Salzenberg wrote:
According to Rafael Garcia-Suarez:
As Chip pointed out the current conversion does not the right thing
for upper bytes. IIUC if we convert properly bytes to UTF8 wrt the C
locale (unless the user specified otherwise) this problem goes away.
Constants are supposed
Steve Hay wrote:
Running [dn]make then [dn]make distclean on Win32 still doesn't
quite put you back where you started, which could affect Test::Smoke.
Various .pm files that were either generated by _pm.PL files or else
copied from somewhere get left behind (including a whole Encode/
OK. What happened ? I thought change #22441 fixed the problem (that appeared
between changes 22427 and 22436 if I read my logs correctly)
Steve Hay wrote:
Automated smoke report for 5.9.1 patch 22476
TANGAROA.uk.radan.com: x86 Family 15 Model 2 Stepping 4, GenuineIntel (x86/1 cpu)
on
Stas Bekman wrote:
I find myself using the following macro in my C/XS code, again and again:
#ifdef USE_ITHREADS
#define cop_file PL_curcop-cop_file
#else
#define cop_file SvPVX(GvSV(PL_curcop-cop_filegv))
#endif
Does it make sense to introduce a public API macro (of course using an
Dave Mitchell wrote:
On Tue, Jan 20, 2004 at 10:06:09PM +0100, Rafael Garcia-Suarez wrote:
Rafael Garcia-Suarez wrote:
Hmm, that probably a reason for forbidding my $x:unique too.
(currently it looks :unique up in user attributes, or better said,
it wastes time.)
Yuck.
Do you want
Stas Bekman wrote:
Nicholas Clark wrote:
On Mon, Dec 22, 2003 at 06:07:54PM -0800, Stas Bekman wrote:
what would it take to get these changes back into maint? [EMAIL PROTECTED]
still produces the dreaded 'Attempt to free unreferenced scalar' on the
code below. I won't be around till Dec
LAUN Wolfgang wrote:
[EMAIL PROTECTED] (via RT) [EMAIL PROTECTED] writes:
I'm getting a segmentation fault when using a complicated regex inside
a map. Consider the following program:
my $MAX = shift || 100;
my $re = qr /^(1+)(??{(?:$1){ . (length ($1) - 1) . } })$/;
map
[EMAIL PROTECTED] (via RT) wrote:
perl -wle 'print $? = $? ^ 3'
Argument ^C isn't numeric in scalar assignment at -e line 1.
0
...
I think that the magical purely numeric variables should start with the
integer flag set. Just adding pIOK should probably be enough.
But magical variables
Marcus Holland-Moritz wrote:
You mean that Windows has setuid() and no getuid() ? how weird.
No. I mean: because Windows hasn't got setuid() and setgid(),
it doesn't make sense to do the cleanup. POSIX.xs does
#if defined (WIN32)
# undef setuid
# undef setgid
# define setuid(a)
H.Merijn Brand wrote:
On Wed 17 Dec 2003 15:41, Dorner Thomas [EMAIL PROTECTED] wrote:
OK, I've 4 patches:
I applied patches 2, 3 and 4, following Merijn's advice of leaving off 1
if it can be done via the hintfile :
(change #21938 to bleadperl)
Number 1 is a bit awkward as it's a patch
Elizabeth Mattijsen wrote:
I found the description of the functionality of delete() very concise
and not at all explaining what delete actually does. I hope this
documentation patch is an improvement, and possibly an eyeopener for
some.
Thanks, applied as #21912.
Tassilo von Parseval wrote:
This goes beyond strange. No matter how hard I try, I can't make a tied
stash trigger FETCH or STORE. All I get are those 'Can't upgrade that
kind of scalar' messages. So either these errors (note that they are
marked internal in perldiag) have to go away (and
Tassilo von Parseval wrote:
On Mon, Dec 08, 2003 at 06:29:43PM + Zsban Ambrus wrote:
DESTROY error message missing from perldiag pod
I've added it as change #21877 :
//depot/perl/pod/perldiag.pod#365 (text)
@@ -1285,6 +1285,12 @@
long for Perl to handle. You have to be
Nicholas Clark wrote:
On Sun, Nov 30, 2003 at 05:45:21PM -0600, Andy Lester wrote:
I had to apply the patch below to the Test::Harness distribution
to make the tests happy.
(I don't know enough about prove to adjust the tests).
Thanks, Rafael. I'll make the necessary adjustments. Now
Fergal Daly wrote:
With blead, 5.8.2 and probably all perls that have weak refs
perl -MScalar::Util=weaken -e '$a=\h;weaken($a)'
Modification of a read-only value attempted at -e line 1
Any good reason for this? It's not documented and it's a pain.
If you want to weaken refs of
Alan Burlison wrote:
I don't think you would want it, the changes are pretty specific to the
Solaris build process:
#* umask set to 022
#* install locations modified to be relative to $ROOT
a bit like installperl's --destdir=/foo option.
#* check for 'make test' being run
H.Merijn Brand wrote:
If you regard 'info' files as documentation, I fully disagree. One has to be
an emacs user to be able to read that, and even then, finding the info you
need takes too much time. man or pod, all info is ready for immediate
trashcanning and can IMHO *never* be marked good
Warren L Dodge (via RT) wrote:
I just built perl-5.8.1 and had the following error five times during the make
test
perl: relocation error:
./lib/auto/NDBM_File/NDBM_File.so: undefined symbol: dbm_open
In looking at NDBM_File.so with ldd I see
libgdbm.so.3 =
If I look at the recent smoke reports for 5.6.2, I see that Solaris and *BSD
fail during the make test-prep stage.
The only change was the upgrade of ExtUtils::MakeMaker to 6.20.
Before I revert it, could I see an example of what's going on on your
platforms ?
Thanks.
Andreas J Koenig wrote:
On 29 Oct 2003 03:50:10 -, Ilya Zakharevich (via RT) [EMAIL PROTECTED]
said:
The following file still segfaults (5.6.1 as well as 5.8.1). It works
fine with 5.005_53.
Blame analysis says that was introduced with
Change 8390 by [EMAIL PROTECTED] on
Uri Guttman wrote:
i would like to propose that my module be part of the 5.10 distribution
and so it should go under the IO::File namespace.
Disregarding the coolness factor of this module, here's why I think
it doesn't meet the criteria for being integrated into the core
currently :
* it
Slaven Rezic wrote:
How the TIEHANDLE crept in?
--- bleedperl/lib/Tie/Hash.pm Mon Oct 6 06:51:46 2003
+++ bleedperl2/lib/Tie/Hash.pmSun Oct 12 18:50:59 2003
Thanks, applied as #21439.
Tom Horsley wrote:
*** hints/powerux.sh Thu Sep 25 14:45:37 2003
--- hints/powerux.sh Thu Sep 25 14:47:27 2003
Thanks, applied to blead as #21387.
Nicholas can merge now :)
(in general, I think new developments and fixes should go into blead
first, except of course when they're specific to
Mark-Jason Dominus wrote:
perl -MO=Deparse -e 'use strict; sub foo { }'
says:
sub foo {
use strict 'refs';
}
use strict 'refs';
;
-e syntax OK
How did 'refs' get in there?
It's better to ask how were the other two left
Dave Mitchell wrote:
Does anyone know of a safe way that I can distinguish between compile-time
and run-time calls to sv_compile_2op() ?
(Maybe a silly idea)
Compare PL_curcop and PL_compiling ?
H.Merijn Brand wrote:
I've integrated all configure hintfiles ; I wonder about integrating the
module-specific hintfiles as well. (At least, I will do it for the upgraded
modules.)
Time to give HP-UX a kick again?
why not ?
H.Merijn Brand wrote:
On Tue 09 Sep 2003 10:51, Rafael Garcia-Suarez [EMAIL PROTECTED] wrote:
H.Merijn Brand wrote:
I've integrated all configure hintfiles ; I wonder about integrating the
module-specific hintfiles as well. (At least, I will do it for the upgraded
modules
Andreas J Koenig wrote:
The following patch implements a rudimentary skeleton that is able to
test some aspects of the debugger that have remained untested up to
now. There are two test scripts, tailored to catch the bug in patch
21010. I believe the framework is fairly flexible (well,
Tels wrote:
Just what I need for bigat, bigint, bignum and bigmouse. Thank you!
bigmouse ?
like the ones they grow in Disneyland ?
Orton, Yves wrote:
Perhaps Ken Williams' new CPAN dist of File::Spec could
simply be plopped
into 5.6.2?
I was looking at the diff and I don't see a reason why this would fix
Yves' build. There's no difference in file_path_is_absolute().
I was considering the upgrade anyway.
Abigail wrote:
This is a plea to remove the 'print (...) interpreted as function'
warning.
I tend to agree.
Let's look at the rational for this warning. It's to warn programmers
that if they write:
print (3 + 4) * 2;
Perl adds 3 and 4, calls print with the result as argument, and
chocolate boy wrote:
(PS. Could autobox.pm be implemented without a core
patch ? by using optimizer.pm and B::Generate for
example.)
Not using optimize.pm alone. Its peep() hook occurs
too late i.e. outside the scope in which $^H and %^H
are meaningful.
On the big TODO list is making
I'm pleased to announce that the blamelog feature of
the perl repository browser now includes hyperlinks
to the actual patches where lines were first introduced.
Nicholas Clark wrote:
Feature, IIRC. Although I can't find the message on p5p about it. IIRC
someone observed that they couldn't name identifiers in Ethiopian, because
there was an Ethiopian character similar in function to _ which wasn't in \w
IIRC this was on the perl-unicode list, and the
Michael G Schwern wrote:
Configure doesn't seem to be honoring -Dusemymalloc on OS X. I've tried
-Dusemymalloc and -Dusemymalloc=y. Both result in usemymalloc=n.
I think the hint is getting a little too pushy.
try at your own risk :
Index: hints/rhapsody.sh
Nicholas Clark wrote:
On Mon, Jul 21, 2003 at 06:47:58PM +0100, Tim Bunce wrote:
On Sat, 19 Jul 2003 01:03:59 +0200, Marcus Holland-Moritz [EMAIL PROTECTED]
wrote:
[patch]
I've no time to study the patch, but I'm happy with the principle.
I think a lot of people won't have time
Jason Rhinelander (via RT) wrote:
The new Perl 5.8.1 warning: Possible precedence problem on bitwise %c
operator comes up incorrectly for bitwise assignment operators (as of
at least 5.8.1-RC2). The following example demonstrates:
$ perl -wle '$a=$b=1; $a ^= $b == 1'
Possible precedence
Dan Kogai wrote:
Porters,
The following is the pre-1.97 patch for Encode. It fixes one bug and
adds one feature.
$Revision: 1.96 $ $Date: 2003/06/18 09:29:02 $
! lib/Encode/Guess.pm
$Encode::Guess::NoUTFAutoGuess is added so you can turn off
automatic utf(8|16|32) guessing
H.Merijn Brand wrote:
`sh cflags optimize='' miniperlmain.o` miniperlmain.c
CCCMD = gcc -DPERL_CORE -c -DDEBUGGING -I/usr/X11R6/include -DPERL_US
E_SAFE_PUTENV -fno-strict-aliasing -Wall
miniperlmain.c: In function `main':
miniperlmain.c:62: parse error before ')' token
make:
Jarkko Hietaniemi wrote:
Are we sure that one seed computed at script start is safe-enough?
If the CGI server is up for months cannot the bad-hat deduce the seed
and tailor the exploit to match?
I think yes.
CGI programs are forked at each request.
However, if I understand correctly,
Dave Rolsky wrote:
This release adds some more docs and brings the code into sync with
bleadperl.
I haven't had any reports of problems with this code yet so I expect to
send patches to p5p real soon now.
Too late. Integrated in the core as change #19702, thanks!
Michael G Schwern [EMAIL PROTECTED] wrote:
I don't understand why the warning doesn't happen when require loads and
compiles Foo.pm. If the warning *only* happens with -c, I don't
understand why it would be desirable to work that way.
-c doesn't influence the presence of the warning.
That
Dave Rolsky [EMAIL PROTECTED] wrote:
Oh, and backport your CPAN version to bleadperl.
Ok, does that mean I have to put the our stuff back in? Does it serve
any purpose except to insure backwards incompatibility?
At the contrary, this means that it would be helpful to have the exact
same
Alex Vandiver wrote:
As discussed at http://perlmonks.org/?node_id=261206 there is a small
bug with stringification of regexes containing escaped #s with the /x
modifier. The patch below fixes it; thanks to BrowserUK for the
impetus.
As far as I can tell, the stringified form of a regexp
Steve Hay wrote:
I posted two patches to this list on March 20:
[PATCH 5.8.0 DOC] Fix missing functions when splitting perlfunc
[PATCH 5.8.0 UTIL] Fix installhtml for splitting and PM/POD conflicts
and haven't heard any more regarding them since.
Don't worry, they're not ignored,
Andreas J. Koenig wrote:
On Wed, 26 Mar 2003 13:18:37 +0100, Rafael Garcia-Suarez [EMAIL PROTECTED]
said:
So my suggestion is to have a new file lib/CPAN/bin/cpan corresponding
to the CPAN-distributed version, and a script utils/cpan.PL to read this
file to produce utils/cpan
Jarkko Hietaniemi [EMAIL PROTECTED] wrote:
I am not certain but I suspect/guess that SIGRTMIN == -1 means there
are no realtime signals at all. I think my patch suggestion to
Configure where less-than-zero signals are just simply not detected
should be a reasonable workaround. Perl doesn't
801 - 900 of 949 matches
Mail list logo
