waitbar with time estimation

Estimated remaining, elapsed time, and percentage complete.

Description

This version of waitbar is compatible with MATLAB's version. Thus, doesn't require any modification of existing functions using the waitbar. However, the property settings include with MATLAB's version will not be recognized.

This program is a heavily modified version of timebar.m by Chad English.

• MATLAB release: MATLAB 7.5 (R2007b)
• Other Files: license.txt, ; waitbar.m

[Anthony Kendall]

Very nice, though there is one bug that shows up in the provided example: the message is overwritten with the graphics handle

A fix I used is:
line 45: h = message;
becomes
line 45: h = message; message = [];

And, in the subfunction updatewaitbar:
line 143: set(hh.message,'String',message);
becomes
line 145: if ~isempty(message),set(hh.message,'String',message);end

Otherwise, it works great and is a complete drop-in replacement for waitbar. Thanks!

[William Henry Brave]

Great contribution, works seamlessly without any modification of the code, thanks...

[Arturo Serrano]

 

[skouliki katelouzos]

ok, cool looking, but this whole waitbar thing is not very useful.
the point is that updating graphics in Matlab is very time consuming.look at this:

h = waitbar(0,'Please wait...');
k=0;
tic
for i=1:1000,
    k=k+1;
waitbar(i/1000);
end
toc
k=0;
tic
for i=1:1000,
    k=k+1;
end
toc
Elapsed time is 15.628354 seconds.
Elapsed time is 0.003811 seconds.

If you have a loop that takes enough time to make reasonable the use of a timebar, then using the timebar will make the computation 4000 times slower.
Updating only test is much more efficient, so i like going for a running percentage thing.
The timebar is good looking though

[Ross Hatton]

In response to the time comparison:

Updating the graphical progress display is certainly going to take longer than an addition operation like k=k+1, but will be negligible when compared to a computationally heavy process, and can be read from across the room.

[skouliki katelouzos]

Ross, i don't want to argue much on this, i see your point.
I mean you have a point if the computation is heavy due to what's inside the loop, but not when is due to the high number of iterations.

[csb]

try this to speed up things

% pause(10e-120) % Needed for window to appear (do not know why)
drawnow

[Philipp Pruefer]

Great tool! The only prob I had was dealing with zero progress. My solution is:

    % 2.2 - Calculate the estimated time remaining
        e_mes = datestr(elap/86400,'HH:MM:SS');
        if progress~=0
            sec_remain = elap*(1/progress-1);
            r_mes = datestr(sec_remain/86400,'HH:MM:SS');
        else
            r_mes = ('--:--:--');
        end

Not sure why I had the prob only after a fresh start of Matlab though...