dlal66 has asked for the wisdom of the Perl Monks concerning the following question:

Dear Monks, I have a perfectly working perl/Tk script that we use for network management. It runs on Centos 5.6 and perl version is 5.8.8 I installed a new PC running Centos 5.7 and perl 5.8.8. I made sure all perl modules the script needs are installed. But I keep on getting the following error on the new machine: 
Unable to create sub named "" at /usr/lib/perl5/site_perl/5.8.8/i386-l
+inux-thread-multi/AnyEvent.pm line 1488.

[download]
The lines in the script relevant to this are: 
my $inotify = new Linux::Inotify2 or die "unable to create new inotify
+ object: $!";

# add watchers --watch the snmptrapd.log file
$watchObject = $inotify->watch ("/var/log/snmptrapd.log", IN_MODIFY, \
+&processTrap);

$inotify_w = AnyEvent->io (
                           fh => $inotify->fileno,
                           poll => 'r',
                           cb => sub {
                                      $inotify->poll ;
                                     },
                          );

MainLoop;

[download]
The sub processTrap is defined.If I comment the 
$inotify_w= AnyEvent->io (...

[download]
line, the code does not give an error and quit. But obviously I need that line. Any wisdom is appreciated...

Replies are listed 'Best First'.
Re: Inotify2 + AnyEvent
by zentara (Cardinal) on Oct 04, 2011 at 17:20 UTC
    Just a guess from knowing how finicky calling subs by reference sometimes can be. Instead of 
    $watchObject = $inotify->watch (
  "/var/log/snmptrapd.log", IN_MODIFY, \&processTrap);

    [download]
    try 
    $watchObject = $inotify->watch (
  "/var/log/snmptrapd.log", IN_MODIFY,  &processTrap()  );
or
  "/var/log/snmptrapd.log", IN_MODIFY,  processTrap()  );
or
  "/var/log/snmptrapd.log", IN_MODIFY,  sub{ processTrap() }  );

    [download]
    anyways, experiment around with the way you call that sub. In the Ionotify2 docs, the examples show .... sub{ } notation, and I know Tk can get very fussy about the way subs are called. One tip, if you use &, include the ().
    I'm not really a human, but I play one on earth.
    Old Perl Programmer Haiku ................... flash japh
[reply]
[d/l]
[select]

      I'd use \&processTrap. The sub { procesTrap() } should also work, but I don't recommend using that with Tk because I find that it can leak memory (and I don't know if Inotify2 avoids that problem). The other two things you suggested don't pass in a reference to code to be called later; they just call the code right then and then pass in whatever that code returns.

      - tye        

[reply]
[d/l]
[select]
        Thank you gentlemen for replying. I tried the suggestions provided but they did not work. So I checked and saw that the script was working on the machine that had AnyEvent version 5.34 installed. I checked the version of AnyEvent on the machine that has the issue and saw that it was running version 7.02 of AnyEvent. So I downloaded the version 5.34 of AnyEvent and after installion, my application is running fine on the new machine. So I guess the version 7.02 introduced a bug. Thanks again for your replies.
[reply]
Re: Inotify2 + AnyEvent
by rcaputo (Chaplain) on Oct 04, 2011 at 16:19 UTC
[reply]
Re: Inotify2 + AnyEvent
by zentara (Cardinal) on Oct 05, 2011 at 09:44 UTC
    To amplify with a code example, look at this Tk code. When the code ref for the button callback is left hanging naked, it gets executed at creation time. I suspected that is what was happening in your Ionotify2 code, the naked coderef was been acted upon at creation time, giving the error can't find "". 
    #!/usr/bin/perl
use warnings;
use strict;
use Tk;

my $mw = new MainWindow;
$mw->geometry("400x200");

my $button = $mw->Button(-text    => "Sub test exit",
                     
                     # this naked coderef is incorrect, as it causes
                     # the callback to be executed at button creation
                     # time, instead of later at button press
                         -command => \&callback(time),
                     
                     #  below is correct method, put [] around coderef
                     #    -command => [ \&callback, time ],
                         )->pack();

MainLoop;

sub callback{
  my $in = shift;
  print "$in exiting\n";
  Tk::exit;
}
__END__

    [download]
    I'm not really a human, but I play one on earth.
    Old Perl Programmer Haiku ................... flash japh
[reply]
[d/l]

      No, \&callback(time) is not a "naked code ref". Yes, the similar-looking \&callback is a code ref, but adding (time) on the end changes the parsing to \(&callback(time)), that is, a call to callback(time) but ignoring prototypes and then returning reference(s) to the returned scalar value(s).

      - tye        

[reply]
[d/l]
[select]
        I know you are smarter than me about this stuff, but even without adding (time), to the coderef 
        # -command => \&callback(time),
 -command => \&callback,

        [download]
        still executes the sub callback at button creation time, causing a pre-mature exit, and crash.
        I'm not really a human, but I play one on earth.
        Old Perl Programmer Haiku ................... flash japh
[reply]
[d/l]

Back to Seekers of Perl Wisdom