PB9 Anomoly (Bug?)

[Michael Mattias]

Hmm, workaround code + explanation in SDK kind of explains why the workaround actually "works" (I hate it when something 'works' but I don't know why)....

The WM_MOUSEWHEEL message is sent to the focus window when the mouse wheel is rotated. The DefWindowProc function propagates the message to the window's parent. There should be no internal forwarding of the message, since DefWindowProc propagates it up the parent chain until it finds a window that processes it.

I guess EXIT FUNCTION in the subclass proc cancels the (PB) default forwarding.

I think it also tells me I want to do some testing, since I usually don't do squat with that message. But since I've never had a problem, I must be missing something here... or maybe I just haven't "wheeled" with a 9x-compiled application yet...

YIKES.. I'm writing my new app in 9x.... so it must be REALLY time to test.

[Gösta H. Lovgren-2]

Nice workaround Dave. The Life program (I am/was working/playing/diddling (see link above) on now has the "Mouse Wheel Exit" feature removed. {grin}

==================================================
"The object of war is not to die for your country
but to make the other ******* die for his."
General George Patton (1885-1945)
==================================================

[William Martin]

I did send the code to support. Preliminary response is that they can replicate it, and it does look like a genuine bug. R&D is chewing on it and will decide what to do.

Bill

[Dave Biggs]

Not sure if this is a feature or an issue Should be sent to support for a ruling.

Meanwhile you can avoid the premature closing of the Graphic Window by capturing the wheel..

#COMPILE EXE
#DIM ALL
#INCLUDE "WIN32API.inc"
 
%WinPelHt = 300
%WinPelWd = 400
 
GLOBAL GrStaticProc AS DWORD
 
FUNCTION GrProc(BYVAL hWnd AS DWORD, BYVAL wMsg AS DWORD, BYVAL wParam AS DWORD, BYVAL lParam AS LONG) AS LONG
  SELECT CASE wMsg
 
    CASE %WM_MOUSEWHEEL
      EXIT Function 
 
  END SELECT
 FUNCTION = CallWindowProc(GrStaticProc, hWnd, wMsg, wParam, lParam)
END FUNCTION
'------------------/
 
FUNCTION PBMAIN () AS LONG
    GLOBAL hWin AS DWORD   'handle for a graphics window to use for console output.
    DIM hStatic AS LONG
    DIM I       AS INTEGER
    DIM W       AS SINGLE
 
    GRAPHIC WINDOW "PB/DLL Optimization Routine", 0, 0, CLNG(%WinPelWd), CLNG(%WinPelHt) TO hWin
     hStatic = GetWindow(hWin, %GW_CHILD)               ' Retrieve handle (static child of of graphic window)
     GrStaticProc = SetWindowLong(hStatic, %GWL_WNDPROC, CODEPTR(GrProc)) ' Subclass to catch mouse messages etc
    GRAPHIC ATTACH hWin, 0
    GRAPHIC FONT "Lucinda Console",10,1
 
    FOR i = 1 TO 10
      GRAPHIC GET CLIENT TO W, W                        ' Has window closed? - exit For Next Loop
      IF W = 0 THEN GOTO CLEANUP
      SLEEP 1000
      GRAPHIC PRINT ("I =" & STR$(i))
    NEXT i
    MSGBOX "Waiting for key..."
 
CLEANUP:
  SetWindowLong hStatic, %GWL_WNDPROC, GrStaticProc
  GRAPHIC WINDOW END
END FUNCTION
'------------------/

Some discussion of mousing around in a graphic window here:

http://www.powerbasic.com/support/pbforums/showthread.php?t=38632

[Gösta H. Lovgren-2]

Bill,

The program terminates here as well ('PBWIN 9.00 - WinApi 05/2008 - XP Pro SP3). The same thing happens (mouse wheel exits immediately and quietly) in a graphics program I have been playing with (http://www.powerbasic.com/support/pb...ad.php?t=38601). Dominique pointed it out.

After fooling with the program and getting nowhere, I just decided to make it a feature ("Mouse Wheel Exits Autiomatically") and move on.

============================
"Dancing is silent poetry."
Simonides (556-468bc)
============================

[Cliff Nichols]

Not sure what it means but Pedro's post of

ADDED LATER ON:
I closed the window while counting. The window closed and the program ran silently until completion (it shows the messagebox when done counting).

Until I can test with XP myself (guessing the same results), I ran your test on VISTA, and both the PB8 and PB9 compiled, and no scroll, but just close the window, they completed and then closed as you would expect (except they typical "I closed, Why is it still counting????" sort of thing)

May help, may not, but its a piece of info we did not have before

[Pedro Ramirez]

No problem here. I spin the wheel up and down, press it. also all other mouse buttons. it works to completion.

Windows XP Pro in Spanish / SP3
Graphic Card: NVidia GeForce 6200 256Mb

ADDED LATER ON:
I closed the window while counting. The window closed and the program ran silently until completion (it shows the messagebox when done counting).

[Cliff Nichols]

Now that I am home, I gave the same test again on a Vista machine, and without recompiling. Then I recompiled and tried again and the same results

Compiled in 8.04 on XP Pro SP3 ---> No crash
Compiled in 9.00 on XP Pro SP3 ---> No crash ---> But Window disappears and messagebox appears on schedule.

So the program is not crashing, just disappearing from sight. I would look closer at what messages are you getting and what of those messages are you not handling and letting the OS handle it?

And now that I have a moment, I figured pretty interesting results, so I would throw my error handler at it, and all I got was no error. (Since I am still working on the handler, so I figured something I had not thought of yet, so I had an idea to add, BUTTTTT.....)

Reading the docs, I see nothing about callbacks, and from VERY limited knowledge of (what I thought was basically a picturebox on a dialog, but was wayyyy offfff....its more like a VB Form that I could print text to kinda thing)

Unless someone has an idea, (like Ricard said) I would submit to [email protected] and see what they say?

Seems odd to me to not be a "CRASH" (whatever that may actually mean), but more "Does not work as expected" (aka, it runs fine, but the "Dialog" disappears???)

[Richard Angell]

Ms

I did not try with a wheel mouse. If that is the ingredient here that triggers then send the code to [email protected]

[Michael Mattias]

Gee, doesn't the way this program is terminating sound like my aforementioned 'stack fault'?

But if it is, ain't nobody outside Venice FL gonna solve it, since demo program does not have but token access to the stack.

[Iain Johnstone]

Crashes for me with PB9 and XP S2. The program also disappears from Task Manager/Processes as soon as it goes off the screen, so it is not hanging around.

Iain Johnstone

[Michael Mattias]

>In my case, the toy program I posted does not produce an error message under XP - it >just quietly terminates

Yes, that is what I got.

But I would have liked to have gotten the 'full blown "WIndows has detected...."' screen. Over the weekend I would have seen a stack fault error and would not have had to code in the TRACE stuff - because as soon as I would have seen "stack fault" I would have known what the problem was. (Or what I was sure it "had to be"). (Which it was).

Fortunatately TRACE told me I had.... a stack fault problem, so at least I was able to find and fix it.

[William Martin]

Michael:

In my case, the toy program I posted does not produce an error message under XP - it just quietly terminates. The DLL version in which I originally encountered the problem did cause Excel to produce a Windows error message though.

But the toy program under Vista does cause Windows to produce an error message when it terminates.

Bill

[Cliff Nichols]

Compiled in 8.04 on XP Pro SP3 ---> No crash
Compiled in 9.00 on XP Pro SP3 ---> No crash ---> But Window disappears and messagebox appears on schedule.

So the program is not crashing, just disappearing from sight. I would look closer at what messages are you getting and what of those messages are you not handling and letting the OS handle it?

[jcfuller]

What type of mouse? Mine is a ps2 and I can't get it to gpf using the scroll wheel.

James

[Michael Mattias]

I don't get a GPF message, but the program ends prematurely without the 'Waiting for key" message box when I scrolll the wheel.. so the system must have terminated it.

Do I need to have some system setting to get GPF messages? I ask because over the weekend, too, I did not get a message when I know I should have gotten a stack fault GPF.

Windows/XP Pro SP3 all updates via Win/9x IDE compile/run.

Except today I DID get GPF message boxes from Lotus (it does not like the so-called "standard richtext" created by MS-Word. But it likes what Wordpad creates)

MCM

[Conrad Hoffman]

Yup, terminates on XP using ver 9.0 just as you describe. I replaced the sleep with a different delay to no avail. I wonder if it has to do with not handling the messages?

[Rodney Hicks]

Running Vista I get your GPF. Have to click on the window first.
Perhaps because there's no message handling outside the inherent graphics?

And it works for me in PBWin 8.04

[William Martin]

And just for good measure, I copied the two EXE files over to a Vista computer and ran them there. The PB8 version runs to completion, while the PB9 version crashes when I spin the mouse wheel, and says "Windows has detected an error" along with a promise to inform me if Microsoft figures out the problem.

Bill

[William Martin]

Works as expected here, no problems. WinME, PBWin 9.0

Maybe you need to re-install 9.0 or have an AV that is interfering ...

Just so I fully understand. You are spinning your mouse wheel while the graphic window is counting up? And it still continues to count up to the full '10' without aborting mid count?

Disabling my AV seems to have no effect. And I've already installed PB9 twice with the same effect.

Bill