Mixing Float and Integer Inequalities?

[skywalk]

I know this has been mentioned before, but I just got burned by this.

Code: Select all

#CONST_DOUBLE = 0.9
Define.i x = 0
; BAD: Force Integer comparison!
If x >= #CONST_DOUBLE
  Debug 1
Else
  Debug 2
EndIf
; OK: Force Float comparison!
If #CONST_DOUBLE <= x
  Debug 1
Else
  Debug 2
EndIf

EDIT: Thanks to luis for the author of my current quote line.

[luis]

The nice thing about standards is there are so many to choose from. ~ unknown

It's by Andrew Tanenbaum, the author of the MINIX OS.
And he's so right.

[Bitblazer]

Amoeba will rule us all

[mk-soft]

Small trick

Code: Select all

#CONST_DOUBLE = 0.9
Define.i x = 0
; Force Integer to float comparison!
If 0.0+x >= #CONST_DOUBLE
  Debug 1
Else
  Debug 2
EndIf

[skywalk]

Yes, I used that in the past with macros, but my point was, I did not remember the left side was an integer. It would have been a painful bug if it went to release.
The manual should stress the casting effects, just as it explains order of operations.

[luis]

It should preferably be fixed instead of just mentioning it in the manual.
It's a bug using common sense, and even if sometimes with PB what should be normally considered a bug it's not considered as such since the compiler has its limitations, in this case even Fred agreed.

[skywalk]

Haha yeah, I am simply posting as a sort of muscle memory trick to NEVER get burned again in this case.
I voted for Float elevation in the many prior posts on this topic.
The speed hit does not compare to the propagation of potential logic bugs.