Version 2.2.0 RC2 Arm Linux using calculations in function call?

[jdp]

Hi all. I have been using version 2.1.0 on Arm Linux on the Raspberry Pi without issues. I have now build a new 8GB Dev Pi and installed first 2.3.0 Trunk and now 2.2.0 RC2. In both versions, I have issues with some of the BGRA Controls. I did log an issue on the developer's Git page.

https://github.com/bgrabitmap/bgracontrols/issues/113

I have done some investigation and it seems the issue is with the calculation inside the function call. The datatype the function arc is looking for is a single. Yet the code Height / 2 is a double. If I leave the code like this it seems that the receiving function gets 0 and then it draws the control at the wrong place. But if I change the code to use a variable declared as a single and then I do the calculation before the time using round to ignore the decimal point and use the variable in the call then the control works.

Code: Pascal  [Select][+][-]

1. temp_height : single;
2. begin
3.  
4. temp_height := round(Height / 2);
5.  
6. FreeAndNil(FBitmap);
7. FBitmap := TBGRABitmap.Create(Width , Height);
8.  
9. FBitmap.Canvas2D.beginPath;
10. //FBitmap.Canvas2D.arc(Width / 2, (Height / 2) , Height / 2.5, 0, pi * 2, False);
11. FBitmap.Canvas2D.arc((Width / 2), temp_height , Height / 2.5, 0, pi * 2, False);
12. FBitmap.Canvas2D.fillStyle(Color);
13. FBitmap.Canvas2D.fill;

So as this code works fine in version 2.1.0 on Arm Linux on the Pi is there something that changed or is there some setting I need to set to make all the old code still work?

[jdp]

So I found a way around it. But still wondering why it is doing this.

[rvk]

The strange thing is that the first parameter (Width / 2) and third parameter (Height / 2.5) do not suffer the same effect.

Is Height at the end of that procedure still the correct value?
(so after the FBitmap.Canvas2D.fill line)