David Golden (dag...@cpan.org)

Of course!.  That could probably be better documented.  Not being an
IPC expert, I wouldn't have expected it except for actual pipes.
Grepping for SIGPIPE in blead's pod/ directory did reveal this in
perl581delta:

> If a socket gets closed by the server while printing to it, the client
> now gets a SIGPIPE. While this new feature was not planned, it fell
> naturally out of PerlIO changes, and is to be considered an accidental
>  feature.

But almost all the other references are in relation to pipes, not
filehandles or sockets.

Maybe it could be added at least to Socket.pm or perlfunc for
'socket'.  perlfunc for 'print' would be another good place to note
that printing to a closed socket or pipe will throw SIGPIPE.

-- David

Carp frequently gets loaded very early, before tools that want to
override caller().  Previously, caller() was only in Carp::Heavy,
which was only loaded on demand (thus after any CORE::GLOBAL::caller
override).  This patch unbreaks anything expecting the old behavior.
---
 lib/Carp.pm |   15 +++++++++------
1 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/lib/Carp.pm b/lib/Carp.pm
index 69d5c1f..0826016 100644
--- a/lib/Carp.pm
+++ b/lib/Carp.pm
@@ -43,7 +43,7 @@ sub longmess {
     # number of call levels to go back, so calls to longmess were off
     # by one.  Other code began calling longmess and expecting this
     # behaviour, so the replacement has to emulate that behaviour.
-    my $call_pack = caller();
+ my $call_pack = exists $CORE::GLOBAL::{caller} ? $CORE::GLOBAL::{caller}->() : caller();
     if ($Internal{$call_pack} or $CarpInternal{$call_pack}) {
       return longmess_heavy(@_);
     }
@@ -55,7 +55,7 @@ sub longmess {

sub shortmess {
     # Icky backwards compatibility wrapper. :-(
-    local @CARP_NOT = caller();
+ local @CARP_NOT = exists $CORE::GLOBAL::{caller} ? $CORE::GLOBAL::{caller}->() : caller();
     shortmess_heavy(@_);
};

@@ -70,7 +70,7 @@ sub caller_info {
   my %call_info;
   @call_info{
     qw(pack file line sub has_args wantarray evaltext is_require)
-  } = caller($i);
+ } = exists $CORE::GLOBAL::{caller} ? $CORE::GLOBAL::{caller}->($i) : caller($i);
   
   unless (defined $call_info{pack}) {
     return ();
@@ -149,7 +149,8 @@ sub long_error_loc {
   my $i;
   my $lvl = $CarpLevel;
   {
-    my $pkg = caller(++$i);
+    ++$i;
+ my $pkg = exists $CORE::GLOBAL::{caller} ? $CORE::GLOBAL::{caller}->($i) : caller($i);
     unless(defined($pkg)) {
       # This *shouldn't* happen.
       if (%Internal) {
@@ -224,8 +225,10 @@ sub short_error_loc {
   my $i = 1;
   my $lvl = $CarpLevel;
   {
-    my $called = caller($i++);
-    my $caller = caller($i);
+
+ my $called = exists $CORE::GLOBAL::{caller} ? $CORE::GLOBAL::{caller}->($i) : caller($i);
+    $i++;
+ my $caller = exists $CORE::GLOBAL::{caller} ? $CORE::GLOBAL::{caller}->($i) : caller($i);

     return 0 unless defined($caller); # What happened?
     redo if $Internal{$caller};

Module loading in general.  Otherwise, 'use Scalar::Util qw/refaddr/'.
The question is whether refaddr is something that really *must* be
there when @INC is wrong/empty or the dirs are missing/renamed.  Seems
a bit esoteric to me, but if someone insists that something replace
reference stringification...

-- David

Fortunately, you'd still be able to do so in pre-5.10 perl.  :-)

Making reference stringification a warning wouldn't stop quick and
dirty debugging, which is why I say that would, at a minimum, be an
improvement on the current situation.

I might be in favor of refaddr() moving into core from Scalar::Util so
it's always available even if module loading is borked.

-- David

That "idiom" is an accident waiting to happen because of overloading.
It should at least warn.  But I actually think it should be fatal
under strictures -- that's the whole point of strictures.  You can
turn them off and be unsafe, but you need to know what you're doing.
Now that Scalar::Util is core, refaddr() is available and should
always be used to get a reference address.  (refaddr should *not*
warn/die, because it's explicit.)

-- David