Re^2: weird case of memory corruption? -- Tk ProgressBar bugged?

I simplified the OP example to trap (but only sometimes!!) the bagged behaviour

#!/usr/bin/perl

use Data::Dump;
#https://www.perlmonks.org/index.pl?node_id=11132441
use strict;
use warnings;
use Tk;
use Tk::ProgressBar;

my %pid=(ex1=>{as=>'temp', val=> -21.5});

my $mw=MainWindow->new();
my $tmpbar=$mw->Frame()->pack( -padx=>30 );


my $pbar = $tmpbar->ProgressBar(
        -anchor=>'s',
        -width=>100,
        -length=>500,
        -blocks=>100,
        -gap=>1,
        -resolution=>0.5,
        -variable=>\$pid{ex1}{val}, 
        -from=>-30,
        -to=>70,
    )->pack();

my $repeat;
$repeat = $mw->repeat(500,sub{  

                                $repeat->cancel() if  $pid{ex1}{val} >
+= 65;  
                                unless ( defined $pid{ex1}{val} ){
                                    print "WARNING: \$pid{ex1}{val} NO
+T FOUND!!\n";
                                    dd %pid;
                                }
                                $pid{ex1}{val}+=5;
                                
                            });


$tmpbar->Label( -textvariable => \$pid{ex1}{val} )->pack();


MainLoop;
[download]

There are no rules, there are no thumbs..
Reinvent the wheel, then learn The Wheel; may be one day you reinvent one of THE WHEELS.

Comment on Re^2: weird case of memory corruption? -- Tk ProgressBar bugged? Download Code

Replies are listed 'Best First'.
Re^3: weird case of memory corruption? -- Tk ProgressBar bugged? by Marshall (Canon) on May 12, 2021 at 11:04 UTC
I ran this code on v5.24 with good looking GUI, but with run time error: `Use of uninitialized value in numeric ge (>=) at C:\Users\xxxx\Documen +ts\PerlProjects\Monks\node_id=11132441.pl line 31. WARNING: $pid{ex1}{val} NOT FOUND!! ("ex1", { as => "temp", val => undef })` [download] I am a bit perplexed. Update: I changed: `$repeat->cancel() if $pid{ex1}{val} >= 65; # to the following... $repeat->cancel() if ($pid{ex1}{val} >= 65); and that seems to work better.` [download]	[reply] [d/l] [select]
Re^4: weird case of memory corruption? -- Tk ProgressBar bugged? by Discipulus (Canon) on May 12, 2021 at 11:28 UTC
Hello Marshall, this is exatcly the point: my code show the bug that happens only sometimes. Try to execute the code many times: `>perl tkprogressbar2.pl Use of uninitialized value in numeric ge (>=) at tkprogressbar2.pl lin +e 31. WARNING: $pid{ex1}{val} NOT FOUND!! ("ex1", { as => "temp", val => undef }) >perl tkprogressbar2.pl >perl tkprogressbar2.pl >perl tkprogressbar2.pl >perl tkprogressbar2.pl Use of uninitialized value in numeric ge (>=) at tkprogressbar2.pl lin +e 31. WARNING: $pid{ex1}{val} NOT FOUND!! ("ex1", { as => "temp", val => undef })` [download] L* There are no rules, there are no thumbs.. Reinvent the wheel, then learn The Wheel; may be one day you reinvent one of THE WHEELS.	[reply] [d/l]
Re^5: weird case of memory corruption? -- Tk ProgressBar bugged? by Marshall (Canon) on May 12, 2021 at 11:44 UTC
OK. Your point is accepted and verified by me on Perl v5.24. I actually saw this in my testing, but I couldn't believe my own eyes! The exact same source code can produce different results when run more than once. I am flabbergasted! This is true, but I don't know why or how? More Info: Updating a Windows GUI is an extremely expensive operation. If you have a loop that will do say 10,000 operations, and you want to show progress via a percent or a progress bar, calculate whether or not the progress bar will move before you update the GUI's value - do this at most 100 times (each one percent for a 100% bar) and probably less than that! The difference in performance between calling the GUI 10,000 times vs 100 times is astonishing!	[reply]