Agnus clocksignal (CLK):
PAL = 28.37516 MHz ; NTSC = 28.63636 MHz

Paula clockcycle period: T = 1/(CLK/8)
PAL = 281.937 ns   ; NTSC = 279.365 ns

C-2 frequency   : 261.632 Hz
C-2 period      : 428 (this is the value written to AUDxPER of Paula)

C-2 sample rate : F = 1/(428*T)
PAL = 8287 Hz      ; NTSC = 8363 Hz

-------------------------------------------------------------------------------

Speed     = Number of ticks per line (defaults to 6)
1 Beat    = Time it takes to play 4 lines at speed 6 (i.e 4*6=24 ticks)
            (defaults to 125)
Tempo     = Decides the time per tick, measured in BPM (Beats Per Minute)

Time/tick = 60/(4*6*tempo) s = 2.5/tempo s

NOTE: As the calculation of time per tick doesn't take into account
      current speed it isn't a real BPM value (at least not if you stick
      to the above definition of 1 beat)

-------------------------------------------------------------------------------

When a sample is specified/changed:

If the new sample is a looped one and there is no new note, the old
note should be retriggered (at the same pitch that the old one was
playing) starting at the loopstartpoint of the new sample. The tricky
thing is that the retriggering shouldn't happen until the previous
sample is done with it's current loop (or just done if it wasn't
looped; if the sample already had all its samples played the
retriggering is done at once). If a new note comes inbetween, the
retriggering is never to be done.

The above behaviour is a sideeffect of the way Amiga moduleplayers
handle looping with Paula.  If you really want to understand why it
works like this, read the ProTracker sources.

Very few modules depend on this feature... "Behind the Walls" does.

This is a bit tricky to implement under VoxWare, so we'll deal with it
later (maybe, most probably never :-)

-------------------------------------------------------------------------------

Some effects in S3M's are not PT-compatible, even though the docs for
ST3 says so. The problem is really that the gory details are not
covered (for MODs there is the PT replay-source to follow).

These differences are mostly subtle. Examples are volumeslides with
both up and down specified at the same time (PT gives UP preference,
and ST3 DOWN). Other things like if a effect should be processed X or
X-1 times also differ. This makes some songs sound a bit weird at
times. I have tried to fix the problem as good as my ears will allow
in the absence of solid replay-code.

This may very well be true for some of the other non-MOD formats too.
