>. and <. Operators in Propeller Tool Inconsistent

RbtsEvrywhr_Riley · 2025-07-02 22:37

We've discovered an inconsistency between how >. and <. compare constants vs how they compare variables. Here's a code snippet.

Pub FMathAngle(theta) : temp

   temp := theta

   if (temp & $8000_0000) ' negative

    if temp <. -180.0 '(fcmpi(temp, LESSTHAN, -180.0))
     repeat
       temp := temp +. 360.0 'fadd(temp, 360.0)
     until temp >. -180.0 '(fcmpi(temp, MORETHAN, -180.0))

   else

    if temp >. 180.0 '(fcmpi(temp, MORETHAN, 180.0))
     repeat
       temp := temp -. 360.0 ':= fadd(temp, -360.0)
     until temp <. 180.0 '(fcmpi(temp, LESSTHAN, 180.0))

   return temp

This will NEVER work because >. and <. do not evaluate correctly. To wit if a variable is being compared with a constant they'll want to evaluate false regardless

So for example FMathAngle(400.0) will always return 400.0 instead of in this case, 40.0

Has anyone else seen this?

Rayman · 2025-07-02 23:25

This is with Prop Tool, right?

The prop tool compiler is old and new features like floats are buggy there.

Try with PNut, or Flexprop, or Spin Tools, etc.

Wuerfel_21 · 2025-07-03 02:04

In case of compiler bugs
- sanity check against the other Spin2 implementations
- whittle the problem down to the shortest possible compilable stand-alone program that reproduces obviously broken behaviour

Unsolicited code review: Question yourself wehter you really want to be storing angles as floating-point degrees. Very likely the answer is no and you probably want to be using binary angles instead. (where this sort of wrapping operation is either not needed or a simple bitwise AND). There are cases for float angles over binary angles in cases where you care about extreme precision at acute angles 0°~1° but nowhere else. Anything where an angle could possibly wrap around multiple times will be faster and higher-precision with binary angles.

Genetix · 2025-07-03 02:28

Rayman,

That sucks to know that Prop Tool is out of date but thanks for letting those of us who use that it is.

RbtsEvrywhr_Riley · 2025-07-03 02:33

Regarding refactoring floating-point degrees vs binary angles: yeah, there's some math library conversations happening in the shop right now, but ease of debugging is currently winning out over refactoring various math libs.

Also yeah, I'll put a clean bug report together when there's time this just kinda came up right at the end of the day and I figured I'd just see if anybody already knows about it.

JonnyMac · 2025-07-03 02:56

That sucks to know that Prop Tool is out of date but thanks for letting those of us who use that it is.

It's frustrating because I really like the Alt-select and paste-in-place feature, the bookmarks, and the Summary view in the main window.

>. and <. Operators in Propeller Tool Inconsistent

Comments