[4.51 x86] GrabSprite crashes in 16-bit Screen

[kenmo]

I've been having some GrabSprite() issues, and it seems that it crashes with an Invalid Memory Access in any 16-bit screen. 32-bit works fine. This happens whether my desktop display is set to 32 or 16.

(tested only in PB 4.51, 32-bit XP Pro)

Code: Select all

; Try 16-bit and 32-bit...

ScreenW.i = 640
ScreenH.i = 480
;ScreenD.i =  32
ScreenD.i =  16

If InitSprite()
  If OpenScreen(ScreenW, ScreenH, ScreenD, "GrabSprite Test")
  
  
    ; Comment out any GrabSprite() you'd like...
  
    GrabSprite(0, 0, 0, ScreenW, ScreenH) ; Before ClearScreen & FlipBuffers
    
    ClearScreen(#Red)
    GrabSprite(0, 0, 0, ScreenW, ScreenH) ; After ClearScreen, Before FlipBuffers
    
    FlipBuffers()
    GrabSprite(0, 0, 0, ScreenW, ScreenH) ; After ClearScreen & FlipBuffers
    Delay(1000)
    
    ClearScreen(#Green)
    GrabSprite(0, 0, 0, ScreenW, ScreenH) ; Another ClearScreen
    
    FlipBuffers()
    GrabSprite(0, 0, 0, ScreenW, ScreenH) ; Another FlipBuffers
    Delay(1000)
    
  Else
    Debug "No screen."
  EndIf
Else
  Debug "No sprite."
EndIf

[kenmo]

EDIT: sorry for bumping

[DoubleDutch]

I wouldn't be bumping bugs if I were you... (I don't think they like it).

[Psychophanta]

May be Fred can not reproduce it.
It works here with 16 and 32 bitplanes, no error is reported!

[gnozal]

Works well here too !
(PB 4.51 x86, Win XP, GeForce 6150LE)

[kenmo]

Really?

16-bit crashes (always with Invalid Memory Access at address 52) on any of those GrabSprite() calls for me.

I'm on XP Pro SP3, on a Toshiba laptop with cheap Intel integrated graphics.

(I changed the code above so that it tries 16-bit first, in case anyone was running it without uncommenting the 16-bit line.)

[nco2k]

no crash sorry. i guess its because of the intel igp.

c ya,
nco2k

[Psychophanta]

Tested also now on a crappy benq laptop with VIA/S3G Unichrome pro IGP, and it works.
I doubt the fault is from the igp. Have you tried with a clean PB4.51 installing?

[kenmo]

Just uninstalled and reinstalled PB 4.51, same thing happens.

But I did try it with different subsystems:

opengl: works fine, no crash
nt4: can't open 640x480x16 screen (never used nt4 before)
directx7: works fine, no crash

So my new questions are:
what is the default subsystem used when none is specified, and
what would be the disadvantages of compiling with the dx7 subsystem?

[nco2k]

I doubt the fault is from the igp

intel igps are the biggest pain in the a**. there are a lot of issues with those, not only in pb.

what is the default subsystem used when none is specified

by default DirectX9 is used (since pb 4.40).

what would be the disadvantages of compiling with the dx7 subsystem?

no aero, higher cpu usage etc.

just out of curiosity, have you tried with pb 4.30?

c ya,
nco2k

[kenmo]

just out of curiosity, have you tried with pb 4.30?

Is dx7 the default in 4.30? Because I just installed and tried it, works fine.
Edit: Oh, okay, you said dx9 has been default since 4.40.

Specifying "directx9" in 4.30 crashes it, so it looks like dx9 + my computer + 16-bit is a bad combination.

(Also, "nt4" works fine in 4.30. But in 4.51, the code above outputs "No screen." implying OpenScreen returns null, despite the screen flickering black for a moment as if it successfully opened. Is this a separate issue??)

Odd stuff going on.