It seems that 'autoflush' doesn't mean what it usually does in
Catalyst::Log::Log4perl. It usually means that a write should happen
immediately, and not be buffered.
That isn't the behaviour in Catalyst::Log::Log4perl, either when set using
the log4perl conf, or when using the option to new(). The option to new()
disables 'abort', which I think is a misunderstanding. The logging functions
always simply push to the log4perlstack, which is only flushed after the
request has been mostly handled.
In order to get autoflush working as expected, I have had to create a
derived class and override _log():
sub _log {
my $self = shift;
$self->SUPER::_log(@_);
$self->_flush if scalar(caller(1)) =~ /^MyApp(::|$)/;
}
perl -v: v5.8.8 built for i486-linux-gnu-thread-multi
uname -a: Linux PC-5023452 2.6.22-14-generic #1 SMP Fri Feb 1 04:59:50
UTC 2008 i686 GNU/Linux
Catalyst::Log::Log4perl version: 1.0
I've raised a bug in RT for this:
http://rt.cpan.org/Public/Bug/Display.html?id=35221
David,
Could you throw a test case in with this (either on the RT bug or in
this mail) and I'll get a patch put in place?
Hello Jay,
Which of the autoflushes should I test against: the parameter to
new(), or the option in a log4perl configuration file?
Do you (or others) agree that the autoflush option to new() is a misnomer?
If not, wouldn't my test simply prove that the code doesn't do what it
doesn't say it does?
David
_______________________________________________
List: [email protected]
Listinfo: http://lists.scsys.co.uk/cgi-bin/mailman/listinfo/catalyst
Searchable archive: http://www.mail-archive.com/[email protected]/
Dev site: http://dev.catalyst.perl.org/
