Re: [log4perl-devel] Change in behavior when upgrading to 1.26 and perl 5.10.0
::Log4perl::caller_depth = $Log::Log4perl::caller_depth
+ 1;
$h = Log::Log4perl-get_logger();
$h-debug(log msg);
}
1;
x.conf
==
log4perl.logger=ERROR
log4perl.logger.Server = INFO
log4perl.logger.DBIx.Log4perl=DEBUG, X1
log4perl.appender.X1=Log::Log4perl::Appender::File
log4perl.appender.X1.filename=dbix.log
log4perl.appender.X1.mode=append
log4perl.appender.X1.utf8 = 1
log4perl.appender.X1.umask = sub { 0002 }
log4perl.appender.X1.layout=Log::Log4perl::Layout::PatternLayout
log4perl.appender.X1.layout.ConversionPattern=%d %p %F{1}:%L %M - %m%n
then run
perl -I/dir_where_DBIx_dir_is 1.pl
nothing comes out in log. Change the depth after get_logger and it
works.
Martin
--
Martin J. Evans
Easysoft Limited
http://www.easysoft.com
--
Return on Information:
Google Enterprise Search pays you back
Get the facts.
http://p.sf.net/sfu/google-dev2dev
___
log4perl-devel mailing list
log4perl-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/log4perl-devel
--
This SF.Net email is sponsored by the Verizon Developer Community
Take advantage of Verizon's best-in-class app development support
A streamlined, 14 day to market process makes app distribution fast and easy
Join now and get one step closer to millions of Verizon customers
http://p.sf.net/sfu/verizon-dev2dev
___
log4perl-devel mailing list
log4perl-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/log4perl-devel
Re: [log4perl-devel] Change in behavior when upgrading to 1.26 and perl 5.10.0
Mike Schilli wrote:
Thanks for the detailed description, here's what's happening:
This change in behavior was caused by a patch introduced with
Log-Log4perl-1.19 in
http://github.com/mschilli/log4perl/commit/35d86ae53859424ea3ac357eaf0f15d9e69f4bed
in September 2008. A detailed description of the issue at the time is
available in
http://rt.cpan.org/Ticket/Display.html?id=38356
on the request tracker. It fixed problems with the caller_depth, but it
also introduced a change to get_logger() which bumped up the caller()
level to obtain the category (aka package name) of the calling package.
This is probably a good opportunity to rethink how Log4perl wrapper
classes should be implemented.
To implement a L4p wrapper class correctly, it needs to provide the
following four methods:
1 Wrap::get_logger()
2 Wrap::get_logger( $package )
3 Wrap-get_logger()
4 Wrap-get_logger( $package )
Calls #1 and #3 are supposed to obtain a logger with the category of the
calling package, not the category of the wrapper package. This is
especially difficult in case #3, as Log4perl doesn't know if Wrap is
a Log4perl wrapper or if it's the #2 signature where $package is an
application package called Wrap.
Another complication is that we don't know how Wrap is implemented.
Is it relaying the get_logger() call to Log4perl by implementing its own
get_logger function? Or is it simply inheriting from Log4perl?
I've put together a fix to resolve this, please check out the
documentation:
http://github.com/mschilli/log4perl/commit/8ad3fbae60a4667aba848eb545c66339aeff161a
As this might break backward compatibility with earlier versions of
Log4perl, I'm interested in hearing what you think about it, feedback
definitely welcome. If I don't hear anything back, it'll go out with the
next release.
You can play around with the new implementation, here's the tarball:
http://github.com/mschilli/log4perl/tarball/wrapper_fix
Also, I've opened a bug on RT to track this issue:
https://rt.cpan.org/Ticket/Display.html?id=52913
Give it a spin!
-- Mike
Mike Schilli m...@perlmeister.com
Thank you Mike. I've tried this now and it appears to fix the change in
behaviour I reported and does not appear to break anything else I can
see. I am now using the tarball you provided and will report if I see
anything. I will also update the rt later today.
Thanks again
Martin
--
Martin J. Evans
Easysoft Limited
http://www.easysoft.com
On Wed, 9 Dec 2009, Martin Evans wrote:
Mike Schilli wrote:
On Wed, 9 Dec 2009, Martin Evans wrote:
Since the upgrade we are getting no logging in one of the files we
expected to get it. After a bit of searching around I discovered the
module we were using was doing this:
local $Log::Log4perl::caller_depth = $Log::Log4perl::caller_depth + 1;
$h{logger} = Log::Log4perl-get_logger();
and the fix was to change the order of those lines.
Thanks for reporting this, although I have a hard time imaginining how
increasing the caller_depth and then getting a logger would be different
from what you'd get if you did it in the reverse order. Not to mention
that it's puzzling why this would change the logging behavior, as
caller_depth is used mainly for cosmetic reasons in certain features
of the pattern layout.
Can you provide a snippet of code that reproduces the problem in full?
That would really help track down the root of the problem.
Thanks!
-- Mike
Mike Schilli
m...@perlmeister.com
It was ending up with main and the category in the following code
instead of DBIx::Log4perl.
sub get_logger { # Get an instance (shortcut)
##
# get_logger() can be called in the following ways:
#
# (1) Log::Log4perl::get_logger() = ()
# (2) Log::Log4perl-get_logger() = (Log::Log4perl)
# (3) Log::Log4perl::get_logger($cat) = ($cat)
#
# (5) Log::Log4perl-get_logger($cat) = (Log::Log4perl, $cat)
# (6) L4pSubclass-get_logger($cat) = (L4pSubclass, $cat)
# Note that (4) L4pSubclass-get_logger() = (L4pSubclass)
# is indistinguishable from (3) and therefore can't be allowed.
# Wrapper classes always have to specify the category explicitely.
my $category;
if(@_ == 0) {
# 1
$category = scalar caller($Log::Log4perl::caller_depth);
} elsif(@_ == 1) {
# 2, 3
if($_[0] eq __PACKAGE__) {
# 2
$category = scalar caller($Log::Log4perl::caller_depth);
} else {
$category = $_[0];
}
} else {
# 5, 6
$category = $_[1];
}
# Delegate this to the logger module
return Log::Log4perl::Logger-get_logger($category);
}
Here is an example:
1.pl
use Log::Log4perl qw(get_logger :levels);
use DBIx::Log4perl;
Log::Log4perl-init_and_watch('x.conf', 60);
my $a = DBIx::Log4perl
[log4perl-devel] Why is Log::Log4perl opening unreferenced log files and can I stop it
Hi, I have a number of daemon processes using Log::Log4perl and tens of modules shared between them also using Log::Log4perl. All the log4perl configuration is in a single file for convenience (as most modules in it are common). Each perl module has its own log entry and logs to a different file. The problem is there are ALOT of different log files defined and Log4perl seems to open every file mentioned in the config file even though only a handful might be used. e.g., config: log4perl.logger.XXX.module1 = INFO, MONE log4perl.appender.MONE=Log::Dispatch::File log4perl.appender.MONE.filename=/tmp/file1.log log4perl.logger.XXX.module2 = INFO, MTWO log4perl.appender.MTWO=Log::Dispatch::File log4perl.appender.MTWO.filename=/tmp/file2.log . . . etc and daemon1 uses module XXX::module1 but not XXX.module2, the /tmp/file2.log is opened in daemon1 even though it is never going to be logged to. This is rather annoying since it is using up my open file descriptors etc. Using lsof on a single daemon shows dozens of log files open even though there is NO chance it will log anything to them. Is there any way to stop this? Thanks. Martin -- Martin J. Evans Easysoft Limited http://www.easysoft.com -- This SF.net email is sponsored by Sprint What will you do first with EVO, the first 4G phone? Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first ___ log4perl-devel mailing list log4perl-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/log4perl-devel
[log4perl-devel] logwarn not calling warn if L::L's init not called
Hi, Is this really the intended behaviour: perl -w -le 'use strict;use warnings;use Log::Log4perl qw(get_logger); my $lh = get_logger(BET::Data::Remove); $lh-debug(fred);$lh-logwarn(warning from l:l); warn(warning);' Log4perl: Seems like no initialization happened. Forgot to call init()? warning at -e line 1. i.e., if something does not call init for Log::Log4perl but has a log handle when logwarn is called a warn does not happen? So substituting warn with logwarn does not always warn! I spent some time debugging a problem this morning only to discover this. I find this most worrying as I'd expect the warn to happen whatever. If you substitute logdie for die it seems to work. perl -MLog::Log4perl -le 'print $Log::Log4perl::VERSION;' 1.33 This is perl, v5.10.1 (*) built for i686-linux-gnu-thread-multi Martin -- Martin J. Evans Easysoft Limited http://www.easysoft.com -- All the data continuously generated in your IT infrastructure contains a definitive record of customers, application performance, security threats, fraudulent activity, and more. Splunk takes this data and makes sense of it. IT sense. And common sense. http://p.sf.net/sfu/splunk-novd2d ___ log4perl-devel mailing list log4perl-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/log4perl-devel
Re: [log4perl-devel] logwarn not calling warn if L::L's init not called
On 21/11/11 04:46, Mike Schilli wrote:
On Tue, 1 Nov 2011, Martin J. Evans wrote:
So substituting warn with logwarn does not always warn!
That's an interesting case. The way it's implemented right now is that
logwarn() will only call warn() (along with other log4perl actions) if
the log level is greater or equal than WARN.
logdie(), on the other hand, will call die() unconditionally, because
it's an action, not only a message.
-- -- Mike
So simply substituting warn with logwarn can change the way your program works.
It is not for me to tell you how Log::Log4perl should work but I find this
behaviour unacceptable as my program will behave differently depending on the
log level. Logging should be logging, not changing the way my program runs.
Example:
$ perl -le '$SIG{__WARN__} = sub {CORE::die Warning:\n, @_, \n}; use
Log::Log4perl qw(get_logger); my $lh =
get_logger(BET::Data::Remove);warn(fred); print ok;'
Warning:
fred at -e line 1.
Change the warn to logwarn:
$ perl -le '$SIG{__WARN__} = sub {CORE::die Warning:\n, @_, \n}; use
Log::Log4perl qw(get_logger); my $lh =
get_logger(BET::Data::Remove);$lh-logwarn(fred); print ok;
ok
Would you consider changing this? If not, I would be most grateful if you could
you point me at the place where I could change this behaviour or tell my how I
could override it.
Martin
Hi,
Is this really the intended behaviour:
perl -w -le 'use strict;use warnings;use Log::Log4perl qw(get_logger); my
$lh = get_logger(BET::Data::Remove);
$lh-debug(fred);$lh-logwarn(warning from l:l); warn(warning);'
Log4perl: Seems like no initialization happened. Forgot to call init()?
warning at -e line 1.
i.e., if something does not call init for Log::Log4perl but has a log handle
when logwarn is called a warn does not happen?
I spent some time debugging a problem this morning only to discover this. I
find this most worrying as I'd expect the warn to happen whatever. If you
substitute logdie for die it seems to work.
perl -MLog::Log4perl -le 'print $Log::Log4perl::VERSION;'
1.33
This is perl, v5.10.1 (*) built for i686-linux-gnu-thread-multi
Martin
--
Martin J. Evans
Easysoft Limited
http://www.easysoft.com
--
All the data continuously generated in your IT infrastructure
contains a definitive record of customers, application performance,
security threats, fraudulent activity, and more. Splunk takes this
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d
___
log4perl-devel mailing list
log4perl-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/log4perl-devel
