Project progress: 5% to 10%

sub_C0D4A contains a lot more code than the simple division part we copied from Revs yesterday, so document it as GetAngleFromCoords.

So it looks like the weird tan/sin/cos lookup table at L3D02 is used in sub_C565F to calculate a hypotenuse.

I worked out the maths last night on the sofa by drawing it out on my iPad:

and here are my notes:

    |`.       (1)                |`.                     (2)
    |t `.                        |t `.
    |    `. a                    |    `. a      h
  a |      `.                  a |      `.   ,´
    |      p `.                  |        `.´
    | p ___--´                   |    ___--´`.  c
    +--´                         +--´---------`+
                                     b
  
                          __+
                     __--´   \         (3)
                __--´        |  c
           __--´              \
      __--´ 90-p           90 |
    +´------------------------+
                  b
  
  
  
  t = theta, our triangle's angle
  p = phi
  a, b = sides of the triangle (e.g. coordinates)
  
  We want to find h
  
  Extend circle arc (1) into a triangle (2), and for small angles we get
  
  h =~ a + c
  
  So what is c?
  
  Focus in on right-angled triangle (3)
  
  For small theta, the angle in the bottom-right tends towards a right angle
  
          180 - theta
  phi  =  -----------  =  90 - theta/2
               2
  
  So:
  
  90 - phi = 90 - (90 - theta/2)
           = theta/2
  
  From (3):
  
  tan(90 - phi) = c / b
  
  So
  
  c = b * tan(90 - phi)
    = b * tan(theta / 2)
  
  So 
  
  h =~ a + c
    =~ a + b * tan(theta / 2)
  
  So the table contains the hypotenuse for angle theta
  and triangle sides a and b

This looks pretty good, so document all this as GetHypotenuse and rename L3D02 to tanHalfAngle.

sub_C1C6C is close to MultiplyCoords in Revs, so copy over the relevant commentary.

Go through and document all JSR calls to the Revs-based routines above.

Lay out zero page and workspaces so each variable has its own comment.

Identified GetRandomNumber routine - it appears to be a linear feedback shift register with EOR feedback, see NES Dev.

Really starting to feel like I'm unpeeling a Geoff Crammond game now, with such a clear development path from Revs.

Documented GetRandomNumber.

sub_C36AD prints text on-screen, takes a token argument in X -> PrintTextToken.

There's a token lookup table at L5784.

Tokens end in &FF, so insert token0, token1 labels after each &FF and replace lookups with calculated offsets to tokens.

Turns out tokens are recursive, with characters 200+token expanding into recursive tokens.

>>> I have now broken through the 10% barrier <<<