Odd failures of graphic display

[Bruce L. Bachelder]

Good morning group,

I've been working on a program for use in a psychological research project. The program displays 15 to 25 circles, one at a time, in various sizes and requires a response to indicate the relative size. You input a digit from 1 to 5 to indicate if you think the specific circle is the smallest (1), largest (5), or somewhere in between (2-4). When the answer is right the program displays "That was right" under the circle; when wrong, "That was wrong."

I finally achieved reliable presentation of circles and reliable taking and scoring of responses, etc. The problem is, the graphic displays fail in odd and, so far, unpredictable ways. Here are some of the specifics:

1) Instead of the Times New Roman font, it sometimes drops back to the default font, but only for one display, the next circle or instruction will be Times New Roman again.
2) Sometimes it totally omits some, but not all printed lines of instructions.
3) Sometimes it omits a circle, but displays subsequent circles normally.
4) Sometimes it omits the "right" or "wrong" text, but works right on subsequent trials.

Some runs of the program (15 to 25 circle presentations) are flawless; but other runs may have multiple failures of display. The display failures happen with multiple runs of the same compiled code, differently on different runs. They also happen with new compiles with no code changes.

I have begun to think there may be issues with Vista and PBCC, but I've found no mention of it on the forums.

I'm using PBCC, ver 4.04 and the latest Vista upgrades (Home Premium).

TIA,

Bruce

[Mel Bishop]

Have you tried it on a different machine running xp or maybe 2k?

Failing that, post a stripped down version of your program and let's take a look at it.

[Kev Peel]

We (or PB support) need to see some code to help you.

You need to do what Mel suggested and supply a cut-down version that exhibits the problem.

[Petr Schreiber jr]

Hi Bruce,

which functions are you using for rendering?
If those are PB GRAPHIC commands, you might workaround it using double buffer mode.

That means attach window as:

GRAPHIC ATTACH hWnd, 0, REDRAW ' <-- That redraw is interesting

' -- Put your drawing code here

GRAPHIC REDRAW ' <-- Update whole screen at once

This applies at least for the graphic primitives rendering, not sure about text.
Sure thing is this way is a lot faster than the immediate drawing.

I experienced similar problem when working on turtle graphic for PB with Emil Menzel, using this approach fixed it.

I could not put together reliable ( = always manifesting problematic behaviour ) way to demonstrate problem, as it happens from time to time on same code.

That is the reason I did not dispatched any report to PB support yet about this issue.


Petr

[Petr Schreiber jr]

Here is one more sample,

Single buffer ( sometimes not drawn completely, don't know why ):

#COMPILE EXE
#DIM ALL

FUNCTION PBMAIN () AS LONG
  REGISTER i AS LONG
  
  ' Create and show a Graphic window on screen
  LOCAL hWin AS DWORD

  GRAPHIC WINDOW "Lines : SINGLE", 300, 300, 480, 480 TO hWin
  GRAPHIC ATTACH hWin, 0
  GRAPHIC CLEAR %BLACK
  GRAPHIC WIDTH 2
  FOR i = 0 TO 479 STEP 5
    GRAPHIC LINE (0, i) - (479, i), %WHITE
  NEXT

  SLEEP 5000  ' show it for 5 seconds, then end

END FUNCTION

Double buffer mode, always rendered completely:

#COMPILE EXE
#DIM ALL

FUNCTION PBMAIN () AS LONG
  REGISTER i AS LONG
  
  ' Create and show a Graphic window on screen
  LOCAL hWin AS DWORD

  GRAPHIC WINDOW "Lines : DOUBLE", 300, 300, 480, 480 TO hWin
  GRAPHIC ATTACH hWin, 0, REDRAW
  GRAPHIC CLEAR %BLACK
  GRAPHIC WIDTH 2
  FOR i = 0 TO 479 STEP 5
    GRAPHIC LINE (0, i) - (479, i), %WHITE
  NEXT

  GRAPHIC REDRAW
  
  SLEEP 5000  ' show it for 5 seconds, then end

END FUNCTION

Petr

[Bruce L. Bachelder]

Hi Petr, Mel, and Kev,

Thank you for the quick replies. I'll try "double buffer" first because it looks like the simplest solution if it works. Failing that, I'll work up some stripped down code to demonstrate the effect.

Bruce

[Bruce L. Bachelder]

Have you tried it on a different machine running xp or maybe 2k?

Failing that, post a stripped down version of your program and let's take a look at it.

An excellent idea, but I don't have access to a different machine right now (we're traveling in a motorhome). I've sent the program to my project colleague to run on his XP machines, but at least two servers refused the .exe attachment. At this point, I haven't heard from him that he's received it or run it.

I'm working on a stripped down version. Very stripped down seems to eliminate the problem. A bit more of the original code seems to bring it back big time, but I'm still studying the confusing results.

Thanks,
Bruce

[Guy Dombrowski]

Graphic not always rendered

Bruce,
I had the same problem win WinXP and using double buffers improved the display but did not cure the problem all the time.

I got it to run correctly only by adding SLEEP 25 after the GRAPHIC ATTACH command

[Mel Bishop]

You might try sending the entire file to PB support and let them take a look at it. They have a reputation for maintaining privacy.

[Bruce L. Bachelder]

Bruce,
I had the same problem win WinXP and using double buffers improved the display but did not cure the problem all the time.

I got it to run correctly only by adding SLEEP 25 after the GRAPHIC ATTACH command

Hi Guy,

Very curious cure. I'll try it.

Bruce

[Bruce L. Bachelder]

Hi Guy,

Very curious cure. I'll try it.

Bruce

I thought the problem was cured. It ran many times flawlessly, then started failing to display the circles.

I'll double check to make sure I implemented Guy's suggested correctly and completely, but at this point it doesn't seem to have solved the problem.

Thanks,
Bruce

[Guy Dombrowski]

Sleep

Bruce,

Depending on the speed of your video card, you may have to increase the sleep period until it work all the time.
I have 5 different machines and they all need differents timing. Very bizarre..

[jcfuller]

I had a similar problem where a bitmap being saved was not working correctly and José offered this function to call for the delay needed. Not sure if it's relevant in this context??

James

SUB TLB_DoEvents (BYVAL hWnd AS DWORD)
   LOCAL Msg AS tagMsg
   WHILE PeekMessage(Msg, %NULL, %NULL, %NULL, %PM_REMOVE)
      IF ISFALSE IsDialogMessage(hWnd, Msg) THEN
         TranslateMessage Msg
         DispatchMessage Msg
      END IF
   WEND
END SUB

[Bruce L. Bachelder]

Bruce,

Depending on the speed of your video card, you may have to increase the sleep period until it work all the time.
I have 5 different machines and they all need differents timing. Very bizarre..

Hello Guy et al.,

Thanks for the additional comment. I'll try some more values. I just bought this computer 6 months or so ago so it probably has a faster card than the previous generation computer (inexpensive laptop) I originally worked on.

Some family illnesses have kept me away from the desk, but I probably will get to spend some more time on the problem today.

I've also been stripping down some code to share and test. It turns out it isn't hard to generate the problem.

Bruce

[Bruce L. Bachelder]

I had a similar problem where a bitmap being saved was not working correctly and José offered this function to call for the delay needed. Not sure if it's relevant in this context??

James

Hi James et al.

I'm such a novice at this stuff I don't know if it is relevant or not. However, I'm happy to have your suggestion and have tried to understand what the code does. Do you think the function was just to delay? If so, SLEEP would seem to be equivalent. I may even try WAITKEY$ at strategic spots and see if that helps.

Thanks for the contribution. I am managing to understand this stuff better and better and this forum has been helpful in several ways. Years ago a colleague of mine not only recommended I use PowerBASIC (for DOS those days) but he commented how helpful this group is. Unfortunately, he died a few months ago or I'd be writing him to thank him again for the excellent tips.

Bruce

[jcfuller]

Hi James et al.

I'm such a novice at this stuff I don't know if it is relevant or not. However, I'm happy to have your suggestion and have tried to understand what the code does. Do you think the function was just to delay? If so, SLEEP would seem to be equivalent. I may even try WAITKEY$ at strategic spots and see if that helps.
Bruce

Bruce,
No. It's not the same. Sleep lets other processes have a time slice. It may not help within the same process. I had tried sleep with the problem I had and it did not work but this call solved the problem.

James

[Bruce L. Bachelder]

We (or PB support) need to see some code to help you.

You need to do what Mel suggested and supply a cut-down version that exhibits the problem.

Hi Kev et al.,

Here's the cut-down code. I have compiled it and run it over and over. I have seen it:

1) run perfectly
2) omit the first of two lines of "instructions"
3) omit all instructions
4) revert to the default font
5) then run perfectly several consecutive times

#COMPILE EXE
#DIM ALL

DECLARE SUB DisplayCircle
DECLARE SUB CreateGraphicWindow
DECLARE SUB PresentInstructions

GLOBAL hWin AS DWORD

FUNCTION PBMAIN () AS LONG

    CONSOLE SET LOC 500, 300   'Get the window out of the way of the circle
    CreateGraphicWindow
    PresentInstructions
    GRAPHIC CLEAR
    GRAPHIC PRINT "Press any key to display a circle."
    CONSOLE SET FOCUS
    WAITKEY$
    DisplayCircle
END FUNCTION

SUB CreateGraphicWindow

    DIM hWin AS GLOBAL DWORD
    GRAPHIC WINDOW "", 0, 0, 1278, 750 TO hWin
    GRAPHIC ATTACH hWin, 0
    GRAPHIC FONT "Times New Roman", 16
END SUB

SUB PresentInstructions

    GRAPHIC PRINT "This is standing in for instructions"
    GRAPHIC PRINT "When finished, press any key."
    CONSOLE SET FOCUS
    WAITKEY$
    
END SUB
SUB DisplayCircle


    GRAPHIC ELLIPSE (50, 50) - (100, 100)

    GRAPHIC SET POS (50, 120)
    GRAPHIC PRINT "Press any key to finish."


    CONSOLE SET FOCUS
    WAITKEY$


END SUB

My next step is to try some of the group's suggestions on this code.

Bruce

[Guy Dombrowski]

Code

Bruce,

I just downloaded your code and sure enough, I have the same problem you have.

I just added a START: line on top and a GOTO START at the bottom to run it quickly many times and I got it to run all the time with a SLEEP 5 after the GRAPHIC ATTACH command.

My machine is a 6000 mhz AMD X2 and I have a fast Nvidia
8800GT card with WinXP SP2.
On my slowest machines I needed SLEEP 50 to fix it

[Bruce L. Bachelder]

Bruce,

quickly many times and I got it to run all the time with a SLEEP 5 after the GRAPHIC ATTACH command.


On my slowest machines I needed SLEEP 50 to fix it

Hi Guy et al.

Great news! My colleague, even less a programmer than I, will be pleased that we are over this challenge. He is eager to get back collecting data again.

I'll start testing here with the sample code, then move to the program, itself. I'll let you know what value is required here.

I'm running an HP Pavilion dv6707us with AMD 64 Athlon X2 with NVIDIA GeForce Go 7150M graphics. Maybe this information and yours will be useful to someone more knowledgeable than I am.

Thank you very much.

Bruce

[Bruce L. Bachelder]

Hi Guy et al.

I'll start testing here with the sample code, I'll let you know what value is required here.

Bruce

Hi Guy et al.,

SLEEP 5 and 10 failed; 20 seems to have fixed it.

Now, I'll see if I can fix the program, itself.

Thank you,
We can get moving here, again.

Bruce