Re: Reference sub name

[martin]

The caller function will let you do that.

Example:

sub foo {
    my $calling_sub = (caller 1)[3];
    $calling_sub = 'the main block' if !defined $calling_sub;
    print "called from: $calling_sub\n";
}

sub bar {
    foo();
}

bar();  # prints: called from: main::bar
foo();  # prints: called from: the main block

Personally, when I need additional log information depending on context, I prefer to use explicit parameters for the logging function. This way, I have more control on the logging protocol, and the individual context tags at each logger call make it easy to map log messages back to the related pieces of code without exposing program internals in the logs. I do not depend on a one-to-one relationship between subroutine names and things worth while to log, either.

Example:

# using Sys::Syslog log levels
sub open_db {
    my $dbname = shift;
    # ...
    if (defined $dbh) {
        log_this(LOG_NOTICE, 'OPEN_DB', $dbname, 'db connection establ
+ished');
    }
    else {
        log_this(LOG_ERR, 'OPEN_DB_NOK', $dbname, 'db connection faile
+d');
    }
}

[BrowserUk]

... without exposing program internals in the logs.

Why? Who is reading your logs and what do you think that they might discover?

(That they could just as easily discover by reading the sources which they presumably would also have access to.)

[martin]

Exposing internals, in some cases, is not only a matter of secrecy. If your logs are not mainly intended for the purpose of debugging code, but for keeping track of any kind of events, internals might be irrelevant, needlessly verbose, and inconveniently varying with software versions. In any case simpler log file contents make log file evaluation simpler.

The original post mentioned a logging interface already being in place, which supports the notion of a more general attitude.