blizzard (86 tokens)

(207 chars)

p = {}
_set_fps"60"
::_::
cls"13"
for i=1,256 do
	local r,_ENV=rnd,p[i] or add(p,{})
	if(x) x,y = x%136-3*z-sin(y/64),y%136+2*z line(x,y,x+2*z,y-2*z,6+z) else x,y,z=r"128",r"128",0.2+r"0.9"
end
flip()
goto _

slightly more readable version:

particles = {}
for i=1,256 do --init
	local r,_ENV=rnd,add(particles,{}) --rnd() needs to be redefined under _ENV
	x, y, z=r(128), r(128), 0.2+r(0.9)
end

function _update60() end --just here for the FPS

function _draw() --process
	cls(13)
	foreach(particles, function(p)
		local _ENV=p --but for some reason line() does not
		x, y = x%136-3*z-sin(y/64), y%136+2*z
		line(x,y,x+2*z,y-2*z,6+z)
	end)
	extcmd("set_title",stat(7).."FPS "..stat(1).." CPU ") --debug shtuff
end

some notes and tricks:

• utilizing _ENV is a very obtuse and complicated feature, but saving a token with every variable (x = y, 3T) vs. (scope.x = y, 4T) adds up in the long run.

• this uses about 0.14 CPU for 256 particles. i had initially tried to use static shapes with print() but that ended up using more than twice as much - it's a very cycle heavy command. i think the more dynamic shapes of line() came out better in the long run.

• order of operations is still a strange concept to me. apparently modulo comes first.

• the code in this comment doesn't utilize it, but every function you can call with a single, static value (cls(13), rnd(0.9), etc) can be called with an auto-converted string - () can be substituted for "" to save another token. (rnd(128), 3T) vs. (rnd"128", 2T)

• ARRAYS START AT 1. DON't go "for i=0,255". GRIEVOUS ERROR

Looks amazing! Great job

I feel there is a story behind your last comment about arrays haha

CPU-efficient version (without _ENV):

particles = {}
for i=1,256 do --init
	add(particles,{rnd"128",rnd"128",0.2+rnd"0.9"})
end

function _update60() end --just here for the FPS

function _draw() --process
	cls(13)
	for i=1,256 do
		local p=particles[i]
		local x,y,z = p[1],p[2],p[3]
		p[1],p[2] = x%136-3*z-sin(y/64),y%136+2*z 
		line(x,y,x+2*z,y-2*z,6+z)
	end
	extcmd("set_title",stat(7).." FPS | "..stat(1).." CPU | "..#particles) --debug shtuff
end

• doing all the stuff the original version does probably saves on tokens, but it also spends 50% more CPU time - this version runs consistently at ~0.1042.

• foreach() is not faster over for i=1,256.

• unpack() is not faster over x,y,z = p[1],p[2],p[3].

• _ENV is certainly not faster. just in general.

• you can add() a pre-compiled table

• you can't assign p and its sub-variables on the same line

• slightly more token expensive (86 vs. 117)

• since particles are only initialized once, there's no fear of accidentally putting for i=0,255 and creating an infinite number of zero-index particles that are never drawn