RE: GTK Progress bar: assistance pls.

[zhangLance <zhang3437 hotmail com>]

/*MYSTERIOUS CODE [FOR ME] -- WORKS !! :-)*/
while (gtk_events_pending ())
{
     g_print("[.]");
     gtk_main_iteration ();
}

Vijay:

Hi, all. How convinient! I was looking for it. Thanks!

I do understand the reason though. You're running a single thread program, instruction are run in a sequence. 
Without that fraction of code, time_out callback which you registered will be call when you exit the 
iteration and returned control to the GTK system. I am new to GTK, but I guess the gtk_main() has a logic 
like this:

pseud-code

void gtk_main()
{
      while(TRUE){
          // any portion of windows need to be paint?
          // any timer event, keyboard event, mouse event ....?
          //
          if( gtk_events_pending () ){
               gtk_main_iteration ();
               
               if ( gtk_main_quit_is_called )
                   break;
          }else if( there is any registered idle action)
              handle_idle_action ()
          else
              sleep (200); // no need to keep the system busy do nothing.
     }
}

while gtk_main_iteration () essentially retrieve an event from events queue, translate it , dispatch it ....


If you want to update user interface (respond to GTK signal/events) while performing a very large job without 
feeling much delay, you either do the job in a thread in which case the operating system will guarantee both 
your main thread and the work thread will have a share of cpu time, or you poll for pending events once in a 
while and give the GTK system a chance to handle them, this is essentially what the above code fraction do.

HTH.

Lance








> Date: Mon, 7 Jan 2008 17:18:08 +0530
> Subject: Re: GTK Progress bar: assistance pls.
> From: vijayasarathy setsindia net
> To: dunno stoptrick com
> CC: gtk-app-devel-list gnome org
>
>
>
>
>
> Hey Carlo,
>
> Thanks man. Your idea works, but .... I am still not able to figure out
> how  it works .....
>
>
> Let me first tell you all what I have done.
>
>
> I have a button which when clicked shall start iterating from 0 to
> 1000[say], and for each iteration, does some work[counts another variable
> j for some time].
>
> Total work to be done here = 1000 * counting js
>
> Periodically I want my progress bar to keep me updated on how much of this
> above work has been completed.
>
> I compute fraction = (i+1)/1000. This much of work has been done at that
> point of time.
>
> I have written a time out function for the progress bar[which times out
> when fraction is >=0.99] which will be called every 1 second. This
> function looks at the fraction of work done, and update the fraction at
> the progress bar accordingly.
>
> Problem:
> for every iteration in i[from 1 to 1000], I kept updating value fraction,
> and passed this fraction to the progress bar inside the for loop for
> getting prgress bar updated. I always saw that only when the fraction was
> 0.99[that is when the work was complete] did control pass to progress bars
> timeout function, and so the first value to progress bar was itself 0.99.
> So, progress bar waited for operation to finish and then displays a fully
> completed picture.
>
> The problem was then solved [Thanks to everyone who contributed, special
> thanks to Carlo] and here is the code that updates progress bar while the
> work is on[and not only after the work is complete]:
>
> ****************START CODE ************************************************
> #ifdef HAVE_CONFIG_H
> #  include <config.h>
> #endif
>
> #include <gtk/gtk.h>
>
> //C Related includes
> #include<unistd.h>
> #include<stdlib.h>
>
> #include "callbacks.h"
> #include "interface.h"
> #include "support.h"
>
> typedef struct
> {
>       GtkWidget *pbar;
>       gdouble fraction;
>
> }ProgressData;
>
> void
> on_button1_clicked                     (GtkButton       *button,
>                                         gpointer         user_data)
> {
>     GtkWidget *wdgt_progress_bar;
>     ProgressData *pdata;
>     gint i = 0;
>     gint j = 0;
>     gint range = 1000000;
>
>     wdgt_progress_bar = lookup_widget(GTK_WIDGET(button),"progressbar1");
>
>     pdata = (ProgressData *)g_malloc0(sizeof(ProgressData));
>     pdata->pbar = wdgt_progress_bar;
>
>       //Call progress_timeout() every 1 second.
>     g_timeout_add (100, progress_timeout, pdata);
>
>     for(i=0;i<range;i++)
>       {
>               //Do work w1.
>               //g_print("(%d)",i);
>               while(j<100000)//To kill time, keep counting until a lakh
>                       j++;
>               j = 0;//Make j 0 for the next loop.
>
>               //Update fraction
>               pdata->fraction = (gdouble)i/(gdouble)(range);
>
>               g_print("(%f)",pdata->fraction);
>
>               /*MYSTERIOUS CODE [FOR ME] -- WORKS !! :-)*/
>               while (gtk_events_pending ())
>               {
>                       g_print("[.]");
>                       gtk_main_iteration ();
>               }
>
>
>       }//End for i = start_count to end_count
>
> }
>
> /* Update the value of the progress bar so that we get
>  * some movement */
> static gboolean progress_timeout( gpointer data)
> {
>       ProgressData *pdata = (ProgressData *)data;
>       gtk_progress_bar_set_fraction (GTK_PROGRESS_BAR (pdata->pbar),
> pdata->fraction);
>
>       if(pdata->fraction>=0.99)//If work is complete.
>               return FALSE;//Stop updating.
>       else//If work is not complete.
>               return TRUE;
> }
>
> ****************************END CODE ***************************************
>
>
> I am still not clear as to how this works.
>
> Inclusion of this piece of code is what made it work.
>
> while (gtk_events_pending ())
> {
>
>   gtk_main_iteration ();
> }
>
> 1. What does gtk_events_pending() signify ? for the code running when the
> button is clicked, is the timeout function of the progress bar a pending
> event ?
>
> 2. What is the use of gtk_main_iteration() ? From the docs, I see that it
> calls gtk_main() once. As I know when gtk_main() is called, the program
> automatically starts listening to X events.
>
>
> 3.Why should this be put into the while loop of gtk_events_pending() ?
>
> 4. When does this while loop end ? Is that random [because that is how it
> looked to me when I put values into the while loop and observed fraction
> values at which this was called]
>
> Carlo, it will be good if you can throw more light on how this code
> snippet made it work.
>
>
> Coming back to discussion with Dan, I still dont understand how you would
> say memory is being leaked when I malloc(50) and then use only less than
> that from the return value of get_text[that is what was returned from the
> text box in UI]. [By the way, memory wasted is some_size -
> strlen(something_else) :-)].
>
> I feel allocating space to variables which will receive values whose size
> it doesnt know is important becuase the program atleast deals with memory
> which it knows about. Comment on that statement Dan.
>
>
> Thanks you very much folks, this is a great place to be part of.
>
> One might wonder why g++ compiler when most things I use is C. The point
> is, there are certain modules which have been developed in C++ and I
> wanted the compiler to be consistent across code. Moreover, g++ is more
> stringent compared to gcc when it comes to usage of variables.
>
> Vijay
>
> On Mon, January 7, 2008 4:18 pm, Dan H wrote:
> > On Mon, 7 Jan 2008 14:51:11 +0530 (IST)
> > vijayasarathy setsindia net wrote:
> >
> > > IN REPLY TO DANS COMMENTS:
> > >
> > >
> > > 1. I am using the g++ compiler and g++ doesnt let a gpointer[return
> > > type of g_malloc0()] or a void pointer point a gchar or a ProgressData
> > > pointer. That is why the typecast.
> >
> > Ah, so you're using it in C++ mode. Yes, C++ requires the cast, C
> > doesn't.
> >
> > > 2. I need more explanation here:
> > >
> > >
> > > txt_start_count = (gchar
> > > > > *)gtk_entry_get_text(GTK_ENTRY(wdgt_start_count));
> > > >
> > > > Useless (possibly dangerous) cast, and you're leaking the memory
> > > > g_malloced() above.
> > >
> > >
> > > Why do you think memory is leaked ?
> >
> > Because this is, briefly, what you do in your code:
> >
> >
> > foo = malloc(some_size) foo = something_else
> >
> > This leaks some_size bytes of memory. The question is: What is the
> > purpose of g_malloc() in the first place?
> >
> > > I have already done a
> > > lookup_widget on wdgt_start_count. gtk_entry_get_text() returns a const
> > > gchar * value while mine is a gchar *, and that is why the typecast.
> >
> > Well then declare txt_start_count as a const char as well.
> >
> >
> > The typecast is both unneccessary and dangerous in the (unlikely) event
> > that some new version of GTK+, gtk_entry_get_text() returns something
> > other than a gchar *.
> >
> > You really need to get your C act together before embarking on such a
> > project.
> >
> > --D.
> > _______________________________________________
> > gtk-app-devel-list mailing list gtk-app-devel-list gnome org
> > http://mail.gnome.org/mailman/listinfo/gtk-app-devel-list
> > **************************************************************************
> > ***********
> > DISCLAIMER
> > This e-mail and any files transmitted with it are for the sole use of the
> > intended recipient(s) and may contain confidential and privileged
> > information. Any use,distribution,copying or disclosure by any other
> > person is strictly prohibited. If you have received this transmission in
> > error please notify SETS immediately either by replying to this e-mail or
> > by telephone +91-44-28205655 and then delete this e-mail. The sender does
> > not accept any responsibility for any damage which you sustain as a
> > result of software viruses or malicious programs. You should therefore
> > scan attachments (if any) prior to opening them.
> > *************************************************************************
> > ***********
>
>
> Vijayasarathy
> Senior Research Associate
> SETS
>
> ***************************************************************************
> DISCLAIMER
> This e-mail and any files transmitted with it are for the sole use of
> the intended recipient(s) and may contain confidential and privileged
> information. Any use, distribution, copying or disclosure by any other
> person is strictly prohibited. If you have received this transmission
> in error please notify SETS immediately either by replying to this e-mail
> or
> by telephone on +91-44-28205655 and then delete this e-mail.
> The sender does not accept any responsibility for any damage which
> you sustain as a result of software viruses or malicious programs.
> You should therefore scan attachments (if any) prior to opening them.
> **************************************************************************

_________________________________________________________________
Windows Live Photo gallery 数码相机的超级伴侣，轻松管理和编辑照片，还能制作全景美图！
http://get.live.cn/product/photo.html