TL;DR

I’m updating Log::Log4perl::Tiny to fix a bug.

Back in 2010 I decided to clean up some code that I had been using for logging. It was inspired to the interface of the venerable Log::Log4perl. I liked that module a lot, but it included a lot of stuff inside and at that time I often needed to pack standalone programs with all batteries inside, so a one-file module was ideal.

That led to Log::Log4perl::Tiny. It tries to follow the bigger module as much as possible, while providing a couple of extensions of its own.

It seems that some people found it useful in time, and took the burden to report bugs as they were found. Yay free software!

The latest one reported is this issue, by JC001010. In short, it’s possible to intermix text parts in the messages to log with sub references, which will be called “just in time” and only if really needed (i.e. if the message has to be logged according to the current logging level setup). This allows skipping costly operations, like dumping data structures while debugging, because it would allow to encapsulate the costly operations in a sub reference, like this:

DEBUG 'the data structure is ', sub { Data::Dumper::Dumper($data) };

If the log level is INFO or above, nothing is printed by the above line and, more importantly, the dumping function is not called.

So well, yes, that had a problem. Some of the provided methods/functions allow using both the machinery that comes with Perl (like die, warn, or the stuff in Carp) and the one shipped with the module. Fact is that the message parts were not properly expanded in these cases, leading to some confusing messages with strange CODE strings inside. Whoops!

I uploaded a hopefully fixing trial release 1.4.1 to CPAN, let’s see how it goes with the CPANTesters and then I’ll do an official release.

Thanks Perl community!