casting issue 6.30b6

[idle]

when a function parameter or result variable is a float, the compiler or ide barfs on bit logic & ! | << >>
it shouldn't matter as the evaluation should cast to float as the last step not before cx, cy and ct1 are integers

So this raises an error

Code: Select all

DisplayTransparentSprite(spriteNumber,cx-(ct1&2),cy-(ct1&1))

and needs to becomes this to avoid it.

Code: Select all

 Protected x.i,y.i 
 x = cx-(ct1&2) 
 y = cy-(ct1&1)
DisplayTransparentSprite(spriteNumber,x,y) 

[Fred]

It how the compiler works, if a float is expected, the whole operation is translated to floats.

[idle]

If the breaking change has taken place for the c backend maybe add a float() cast so we don't need to break up logic and use additional variables.

[Fred]

It's just the type of the parameters which changed to floats, that's why you have an error now (to allow in between pixels display)

[Michael Vogel]

Would be nice to get a simple casting feature within PB (similar to the Bool function)...

Code: Select all

Procedure.f float(foo)
	ProcedureReturn foo
EndProcedure

Debug Abs(float(c>>1))

Selfmade casting routines are not very useful, but internal Purebasic methods would do a perfect job...

Code: Select all

Procedure.i integer(foo)
	ProcedureReturn foo
EndProcedure

x.i
y.i
cx.f
cy.f
ct1.f

x = integer(cx)-(integer(ct1)&2)
y = integer(cy)-(integer(ct1)&1)

; DisplayTransparentSprite(spriteNumber,x,y)

[SMaag]

@idle
which of your variable is float?

The documentation says x,y of DisplayTransparentSprite(#Sprite, x, y) is an Integer.
There is no reason for the Compiler to switch to Float if your variables are not float.

maybe

Code: Select all

DisplayTransparentSprite(spriteNumber,Int(cx)-(ct1&2),INt(cy)-(ct1&1)

[idle]

All of them are integers, the function wants floats as input but It shouldn't be an issue, the promotion to float should only happen at the end of the evaluation when it passes the parameter to the function.

[SMaag]

All of them are integers, the function wants floats as input but It shouldn't be an issue, the promotion to float should only happen at the end of the evaluation when it passes the parameter to the function.

Now I understand. To make it clear for all!

In PB6.30 the variable type for Sprite functions like DisplaySprite() DisplayTransparentSprite() changed from Int to Float.
Now when we are using 'calculated' intergers for calling DisplaySprite, PB change all Integer operations to Float operations.
This cause an Error because the conversion from Integer to Float is done at first. So all the Integer Operations are Float operations for PB and we can not see this.
But the Float conversion should be done as last operation when passing the values to DisplaySprite() -> It should not affect the Integer Operations

[Michael Vogel]

Some internal casting options would allow to keep such things in a single line statement without using additional variables...

Code: Select all

Procedure.f float(foo); float{...}
	ProcedureReturn foo
EndProcedure
Procedure.i integer(foo); integer{...}
	ProcedureReturn foo
EndProcedure

cx.i
cy.f
ct1.d

DisplayTransparentSprite( spriteNumber, float(cx-integer(integer(ct1)&2)), float(cy-integer(integer(ct1)&2)) )