diff --git a/oo.lua b/class.lua
similarity index 96%
rename from oo.lua
rename to class.lua
index 89491eb..58a2314 100644
--- a/oo.lua
+++ b/class.lua
@@ -5,7 +5,7 @@
 	todo: collect some stats on classes/optional global class registry
 ]]
 
-function class(inherits)
+return local function class(inherits)
 	local c = {}
 	c.__mt = {__index = c}
 	--handle single inheritence
diff --git a/colour.lua b/colour.lua
new file mode 100644
index 0000000..9860120
--- /dev/null
+++ b/colour.lua
@@ -0,0 +1,62 @@
+--[[
+	colour handling stuff
+
+	feel free to alias to colour
+]]
+
+local bit = require("bit")
+local band, bor = bit.band, bit.bor
+local lshift, rshift = bit.lshift, bit.rshift
+
+local colour = {}
+
+function colour.packRGB(r, g, b)
+	local br = lshift(band(0xff, r * 255), 16)
+	local bg = lshift(band(0xff, g * 255), 8)
+	local bb = lshift(band(0xff, b * 255), 0)
+	return bor( br, bg, bb )
+end
+
+function colour.packARGB(r, g, b, a)
+	local ba = lshift(band(0xff, a * 255), 24)
+	local br = lshift(band(0xff, r * 255), 16)
+	local bg = lshift(band(0xff, g * 255), 8)
+	local bb = lshift(band(0xff, b * 255), 0)
+	return bor( br, bg, bb, ba )
+end
+
+function colour.packRGBA(r, g, b, a)
+	local br = lshift(band(0xff, r * 255), 24)
+	local bg = lshift(band(0xff, g * 255), 16)
+	local bb = lshift(band(0xff, b * 255), 8)
+	local ba = lshift(band(0xff, a * 255), 0)
+	return bor( br, bg, bb, ba )
+end
+
+function colour.unpackARGB(argb)
+	local r = rshift(band(argb, 0x00ff0000), 16) / 255.0
+	local g = rshift(band(argb, 0x0000ff00), 8)  / 255.0
+	local b = rshift(band(argb, 0x000000ff), 0)  / 255.0
+	local a = rshift(band(argb, 0xff000000), 24) / 255.0
+	return r, g, b, a
+end
+
+function colour.unpackRGBA(rgba)
+	local r = rshift(band(rgba, 0xff000000), 24) / 255.0
+	local g = rshift(band(rgba, 0x00ff0000), 16) / 255.0
+	local b = rshift(band(rgba, 0x0000ff00), 8)  / 255.0
+	local a = rshift(band(rgba, 0x000000ff), 0)  / 255.0
+	return r, g, b, a
+end
+
+function colour.unpackRGB(rgb)
+	local r = rshift(band(rgb, 0x00ff0000), 16) / 255.0
+	local g = rshift(band(rgb, 0x0000ff00), 8)  / 255.0
+	local b = rshift(band(rgb, 0x000000ff), 0)  / 255.0
+	local a = 1.0
+	return r, g, b, a
+end
+
+--todo: hsl, hsv, other colour spaces
+
+return colour
\ No newline at end of file
diff --git a/functional.lua b/functional.lua
index 085900e..b1e2012 100644
--- a/functional.lua
+++ b/functional.lua
@@ -1,29 +1,25 @@
 --[[
 	functional programming facilities
+
+	notes:
+		be careful about creating closures in hot loops.
+		this is this module's achilles heel - there's no special
+		syntax for closures so it's not apparent that you're suddenly
+		allocating at every call
+
+		reduce has a similar problem, but at least arguments
+		there are clear!
+
+	optional:
+		set BATTERIES_FUNCTIONAL_MODULE to a table before requiring
+		if you don't want this to modify the global `table` table
 ]]
 
---collect all keys of a table into a sequence
-function table.keys(t)
-	local r = {}
-	for k,v in pairs(t) do
-		table.insert(r, k)
-	end
-	return r
-end
-
---collect all values of a table into a sequence
---(shallow copy if it's already a sequence)
-function table.values(t)
-	local r = sequence:new()
-	for k,v in pairs(t) do
-		table.insert(r, v)
-	end
-	return r
-end
+local _table = BATTERIES_FUNCTIONAL_MODULE or table
 
 --simple sequential iteration, f is called for all elements of t
 --f can return non-nil to break the loop (and return the value)
-function table.foreach(t, f)
+function _table.foreach(t, f)
 	for i,v in ipairs(t) do
 		local r = f(v, i)
 		if r ~= nil then
@@ -35,7 +31,7 @@ end
 --performs a left to right reduction of t using f, with o as the initial value
 -- reduce({1, 2, 3}, f, 0) -> f(f(f(0, 1), 2), 3)
 -- (but performed iteratively, so no stack smashing)
-function table.reduce(t, f, o)
+function _table.reduce(t, f, o)
 	for i,v in ipairs(t) do
 		o = f(o, v)
 	end
@@ -44,7 +40,7 @@ end
 
 --maps a sequence {a, b, c} -> {f(a), f(b), f(c)}
 -- (automatically drops any nils due to table.insert, which can be used to simultaneously map and filter)
-function table.map(t, f)
+function _table.map(t, f)
 	local r = {}
 	for i,v in ipairs(t) do
 		local mapped = f(v, i)
@@ -55,8 +51,25 @@ function table.map(t, f)
 	return r
 end
 
+--maps a sequence inplace, modifying it {a, b, c} -> {f(a), f(b), f(c)}
+-- (automatically drops any nils, which can be used to simultaneously map and filter,
+--	but this results in a linear table.remove so "careful" for big working sets)
+function _table.remap(t, f)
+	local i = 1
+	while i <= #t do
+		local mapped = f(t[i])
+		if mapped ~= nil then
+			t[i] = mapped
+			i = i + 1
+		else
+			table.remove(t, i)
+		end
+	end
+	return t
+end
+
 --filters a sequence
-function table.filter(t, f)
+function _table.filter(t, f)
 	local r = {}
 	for i,v in ipairs(t) do
 		if f(v, i) then
@@ -67,7 +80,7 @@ function table.filter(t, f)
 end
 
 --partitions a sequence based on filter criteria
-function table.partition(t, f)
+function _table.partition(t, f)
 	local a = {}
 	local b = {}
 	for i,v in ipairs(t) do
@@ -84,7 +97,7 @@ end
 --iteration limited by min(#t1, #t2)
 --function receives arguments (t1, t2, i)
 --nil results ignored
-function table.zip(t1, t2, f)
+function _table.zip(t1, t2, f)
 	local ret = {}
 	local limit = math.min(#t2, #t2)
 	for i=1, limit do
@@ -101,9 +114,9 @@ end
 --return a copy of a sequence with all duplicates removed
 --	causes a little "extra" gc churn; one table and one closure
 --	as well as the copied deduped table
-function table.dedupe(t)
+function _table.dedupe(t)
 	local seen = {}
-	return table.filter(t, function(v)
+	return _table.filter(t, function(v)
 		if seen[v] then
 			return false
 		end
@@ -113,15 +126,15 @@ function table.dedupe(t)
 end
 
 --append sequence t2 into t1, modifying t1
-function table.append_inplace(t1, t2)
-	table.foreach(t2, function(v)
+function _table.append_inplace(t1, t2)
+	for i,v in ipairs(t2) do
 		table.insert(t1, v)
-	end)
+	end
 	return t1
 end
 
 --return a new sequence with the elements of both t1 and t2
-function table.append(t1, t2)
+function _table.append(t1, t2)
 	local r = {}
 	append_inplace(r, t1)
 	append_inplace(r, t2)
@@ -133,7 +146,7 @@ end
 -----------------------------------------------------------
 
 --true if any element of the table matches f
-function table.any(t, f)
+function _table.any(t, f)
 	for i,v in ipairs(t) do
 		if f(v) then
 			return true
@@ -143,7 +156,7 @@ function table.any(t, f)
 end
 
 --true if no element of the table matches f
-function table.none(t, f)
+function _table.none(t, f)
 	for i,v in ipairs(t) do
 		if f(v) then
 			return false
@@ -153,7 +166,7 @@ function table.none(t, f)
 end
 
 --true if all elements of the table match f
-function table.all(t, f)
+function _table.all(t, f)
 	for i,v in ipairs(t) do
 		if not f(v) then
 			return false
@@ -163,7 +176,7 @@ function table.all(t, f)
 end
 
 --counts the elements of t that match f
-function table.count(t, f)
+function _table.count(t, f)
 	local c = 0
 	for i,v in ipairs(t) do
 		if f(v) then
@@ -174,7 +187,7 @@ function table.count(t, f)
 end
 
 --true if the table contains element e
-function table.contains(t, e)
+function _table.contains(t, e)
 	for i, v in ipairs(t) do
 		if v == e then
 			return true
@@ -184,48 +197,53 @@ function table.contains(t, e)
 end
 
 --return the numeric sum of all elements of t
-function table.sum(t)
-	return table.reduce(t, function(a, b)
+function _table.sum(t)
+	return _table.reduce(t, function(a, b)
 		return a + b
 	end, 0)
 end
 
 --return the numeric mean of all elements of t
-function table.mean(t)
+function _table.mean(t)
 	local len = #t
 	if len == 0 then
 		return 0
 	end
-	return table.sum(t) / len
+	return _table.sum(t) / len
 end
 
 --return the minimum and maximum of t in one pass
-function table.minmax(t)
-	local a = table.reduce(t, function(a, b)
-		a.min = a.min and math.min(a.min, b) or b
-		a.max = a.max and math.max(a.max, b) or b
-		return a
-	end, {})
-	if a.min == nil then
-		a.min = 0
-		a.max = 0
+--or zero for both if t is empty
+--	(would perhaps more correctly be math.huge, -math.huge
+--	 but that tends to be surprising/annoying in practice)
+function _table.minmax(t)
+	local max, min
+	for i,v in ipairs(t) do
+		min = not min and v or math.min(min, v)
+		max = not max and v or math.max(max, v)
 	end
-	return a.min, a.max
+	if min == nil then
+		min = 0
+		max = 0
+	end
+	return min, max
 end
 
-function table.max(t)
-	local min, max = table.minmax(t)
+--return the maximum element of t or zero if t is empty
+function _table.max(t)
+	local min, max = _table.minmax(t)
 	return max
 end
 
-function table.min(t)
-	local min, max = table.minmax(t)
+--return the minimum element of t or zero if t is empty
+function _table.min(t)
+	local min, max = _table.minmax(t)
 	return min
 end
 
 --return the element of the table that results in the greatest numeric value
 --(function receives element and key respectively, table evaluated in pairs order)
-function table.find_best(t, f)
+function _table.find_best(t, f)
 	local current = nil
 	local current_best = -math.huge
 	for k,e in pairs(t) do
@@ -239,14 +257,15 @@ function table.find_best(t, f)
 end
 
 --return the element of the table that results in the value nearest to the passed value
-function table.find_nearest(t, f, v)
-	return table.find_best(t, function(e)
+--todo: optimise as this generates a closure each time
+function _table.find_nearest(t, f, v)
+	return _table.find_best(t, function(e)
 		return -math.abs(f(e) - v)
 	end)
 end
 
 --return the first element of the table that results in a true filter
-function table.find_match(t, f)
+function _table.find_match(t, f)
 	for i,v in ipairs(t) do
 		if f(v) then
 			return v
@@ -254,3 +273,5 @@ function table.find_match(t, f)
 	end
 	return nil
 end
+
+return _table
diff --git a/init.lua b/init.lua
index ce20613..3551045 100644
--- a/init.lua
+++ b/init.lua
@@ -1,35 +1,92 @@
 --[[
-	core modules
+	batteries for lua
 
 	if required as the "entire library" (ie by this file), puts everything into
-	global namespace as it'll presumably be commonly used
+	global namespace by default as it'll presumably be commonly used
 
-	if not, several of the modules work as "normal" modules and return a table
+	if not, several of the modules work as normal lua modules and return a table
 	for local-friendly use
+
+	the others that modify some global table can be talked into behaving as normal
+	lua modules as well by setting appropriate globals prior to inclusion
+
+	you can avoid modifying any global namespace by setting
+
+		BATTERIES_NO_GLOBALS = true
+
+	before requiring, then everything can be accessed as eg
+
+		batteries.table.stable_sort
 ]]
 
 local path = ...
-local function relative_file(p)
-	return table.concat({path, p}, ".")
+local function require_relative(p)
+	return require(table.concat({path, p}, "."))
 end
 
-require(relative_file("oo"))
+if BATTERIES_NO_GLOBALS then
+	--define local tables for everything to go into
+	BATTERIES_MATH_MODULE = {}
+	BATTERIES_TABLE_MODULE = {}
+	BATTERIES_FUNCTIONAL_MODULE = {}
+end
 
-require(relative_file("math"))
+local _class = require_relative("class")
 
-require(relative_file("table"))
-require(relative_file("stable_sort"))
+local _math = require_relative("math")
 
-require(relative_file("functional"))
-sequence = require(relative_file("sequence"))
-unique_mapping = require(relative_file("unique_mapping"))
+local _table = require_relative("table")
+local _stable_sort = require_relative("stable_sort")
 
-vec2 = require(relative_file("vec2"))
-vec3 = require(relative_file("vec3"))
-intersect = require(relative_file("intersect"))
+local _functional = require_relative("functional")
+local _sequence = require_relative("sequence")
 
-state_machine = require(relative_file("state_machine"))
+local _vec2 = require_relative("vec2")
+local _vec3 = require_relative("vec3")
+local _intersect = require_relative("intersect")
 
-async = require(relative_file("async"))
+local _unique_mapping = require_relative("unique_mapping")
+local _state_machine = require_relative("state_machine")
 
-manual_gc = require(relative_file("manual_gc"))
+local _async = require_relative("async")
+
+local _manual_gc = require_relative("manual_gc")
+
+local _colour = require_relative("colour")
+
+--export globally if required
+if not BATTERIES_NO_GLOBALS then
+	class = _class
+	sequence = _sequence
+	
+	vec2 = _vec2
+	vec3 = _vec3
+	intersect = _intersect
+
+	unique_mapping = _unique_mapping
+	state_machine = _state_machine
+	async = _async
+	manual_gc = _manual_gc
+
+	--support both spellings
+	colour = _colour
+	color = _colour
+end
+
+--either way, export to package registry
+return {
+	class = _class,
+	math = _math,
+	table = _table,
+	stable_sort = _stable_sort,
+	functional = _functional,
+	sequence = _sequence,
+	vec2 = _vec2,
+	vec3 = _vec3,
+	intersect = _intersect,
+	unique_mapping = _unique_mapping,
+	state_machine = _state_machine,
+	async = _async,
+	manual_gc = _manual_gc,
+	colour = _colour,
+}
diff --git a/math.lua b/math.lua
index 3ad0a1a..7c9def4 100644
--- a/math.lua
+++ b/math.lua
@@ -1,11 +1,15 @@
 --[[
 	extra mathematical functions
+
+	optional:
+		set BATTERIES_MATH_MODULE to a table before requiring
+		if you don't want this to modify the global `math` table
 ]]
 
-local bit = require("bit")
+local _math = BATTERIES_MATH_MODULE or math
 
 --wrap v around range [lo, hi)
-function math.wrap(v, lo, hi)
+function _math.wrap(v, lo, hi)
 	local range = hi - lo
 	local relative = v - lo
 	local relative_wrapped = relative % range
@@ -15,51 +19,51 @@ function math.wrap(v, lo, hi)
 end
 
 --clamp v to range [lo, hi]
-function math.clamp(v, lo, hi)
+function _math.clamp(v, lo, hi)
 	return math.max(lo, math.min(v, hi))
 end
 
-function math.clamp01(v)
-	return math.clamp(v, 0, 1)
+function _math.clamp01(v)
+	return _math.clamp(v, 0, 1)
 end
 
 --round v to nearest whole
-function math.round(v)
+function _math.round(v)
 	return math.floor(v + 0.5)
 end
 
 --round v to one-in x
 -- (eg x = 2, v rounded to increments of 0.5)
-function math.to_one_in(v, x)
-	return math.round(v * x) / x
+function _math.to_one_in(v, x)
+	return _math.round(v * x) / x
 end
 
 --round v to a given decimal precision
-function math.to_precision(v, decimal_points)
-	return math.to_one_in(v, math.pow(10, decimal_points))
+function _math.to_precision(v, decimal_points)
+	return _math.to_one_in(v, math.pow(10, decimal_points))
 end
 
 --0, 1, -1 sign of a scalar
-function math.sign(v)
+function _math.sign(v)
 	if v < 0 then return -1 end
 	if v > 0 then return 1 end
 	return 0
 end
 
 --linear interpolation between a and b
-function math.lerp(a, b, t)
+function _math.lerp(a, b, t)
 	return a * (1.0 - t) + b * t
 end
 
 --classic smoothstep
 --(only "safe" for 0-1 range)
-function math.smoothstep(v)
+function _math.smoothstep(v)
 	return v * v * (3 - 2 * v)
 end
 
 --classic smootherstep; zero 2nd order derivatives at 0 and 1
 --(only safe for 0-1 range)
-function math.smootherstep(v)
+function _math.smootherstep(v)
 	return v * v * v * (v * (v * 6 - 15) + 10)
 end
 
@@ -125,7 +129,7 @@ local sparse_primes_1k = {
 	6689, 7039, 7307, 7559, 7573, 7919,
 }
 
-function math.first_above(v, t)
+function _math.first_above(v, t)
 	for _,p in ipairs(t) do
 		if p > v then
 			return p
@@ -134,79 +138,30 @@ function math.first_above(v, t)
 	return t[#t]
 end
 
-function math.next_prime_1k(v)
-	return math.first_above(v, primes_1k)
+function _math.next_prime_1k(v)
+	return _math.first_above(v, primes_1k)
 end
 
-function math.next_prime_1k_sparse(v)
-	return math.first_above(v, sparse_primes_1k)
-end
-
---color handling stuff
-local band, bor = bit.band, bit.bor
-local lshift, rshift = bit.lshift, bit.rshift
-
-function math.colorToRGB(r, g, b)
-	local br = lshift(band(0xff, r * 255), 16)
-	local bg = lshift(band(0xff, g * 255), 8)
-	local bb = lshift(band(0xff, b * 255), 0)
-	return bor( br, bg, bb )
-end
-
-function math.colorToARGB(r, g, b, a)
-	local ba = lshift(band(0xff, a * 255), 24)
-	local br = lshift(band(0xff, r * 255), 16)
-	local bg = lshift(band(0xff, g * 255), 8)
-	local bb = lshift(band(0xff, b * 255), 0)
-	return bor( br, bg, bb, ba )
-end
-
-function math.colorToRGBA(r, g, b, a)
-	local br = lshift(band(0xff, r * 255), 24)
-	local bg = lshift(band(0xff, g * 255), 16)
-	local bb = lshift(band(0xff, b * 255), 8)
-	local ba = lshift(band(0xff, a * 255), 0)
-	return bor( br, bg, bb, ba )
-end
-
-function math.ARGBToColor(argb)
-	local r = rshift(band(argb, 0x00ff0000), 16) / 255.0
-	local g = rshift(band(argb, 0x0000ff00), 8)  / 255.0
-	local b = rshift(band(argb, 0x000000ff), 0)  / 255.0
-	local a = rshift(band(argb, 0xff000000), 24) / 255.0
-	return r, g, b, a
-end
-
-function math.RGBAToColor(rgba)
-	local r = rshift(band(rgba, 0xff000000), 24) / 255.0
-	local g = rshift(band(rgba, 0x00ff0000), 16) / 255.0
-	local b = rshift(band(rgba, 0x0000ff00), 8)  / 255.0
-	local a = rshift(band(rgba, 0x000000ff), 0)  / 255.0
-	return r, g, b, a
-end
-
-function math.RGBToColor(rgb)
-	local r = rshift(band(rgb, 0x00ff0000), 16) / 255.0
-	local g = rshift(band(rgb, 0x0000ff00), 8)  / 255.0
-	local b = rshift(band(rgb, 0x000000ff), 0)  / 255.0
-	local a = 1.0
-	return r, g, b, a
+function _math.next_prime_1k_sparse(v)
+	return _math.first_above(v, sparse_primes_1k)
 end
 
 --angle handling stuff
-function math.normalise_angle(a)
-	return math.wrap(a, -math.pi, math.pi)
+function _math.normalise_angle(a)
+	return _math.wrap(a, -math.pi, math.pi)
 end
 
-function math.relative_angle(a1, a2)
-	a1 = math.normalise_angle(a1)
-	a2 = math.normalise_angle(a2)
-	return math.normalise_angle(a1 - a2)
+function _math.relative_angle(a1, a2)
+	a1 = _math.normalise_angle(a1)
+	a2 = _math.normalise_angle(a2)
+	return _math.normalise_angle(a1 - a2)
 end
 
 --geometric rotation multi-return
-function math.rotate(x, y, r)
+function _math.rotate(x, y, r)
 	local s = math.sin(r)
 	local c = math.cos(r)
 	return c * x - s * y, s * x + c * y
 end
+
+return _math
diff --git a/readme.md b/readme.md
index 6b4b51b..112d677 100644
--- a/readme.md
+++ b/readme.md
@@ -1,5 +1,38 @@
-# lua-core
+# batteries
 
 Core dependencies for making games with lua, especially with [love](https://love2d.org).
 
-Definitely needs a catchier name, but does a lot of useful stuff to get projects off the ground faster.
+Does a lot to get projects off the ground faster, filling out lua's standard library and providing implementations of common algorithms and data structures useful for games.
+
+# Module Overview
+
+- `class` - single-inheritance oo in a single function
+- `math` - mathematical extensions
+- `table` - table handling extensions
+- `stable_sort` - a stable sorting algorithm that is also faster than table.sort under luajit
+- `functional` - functional programming facilities. `map`, `reduce`, `any`, `match`, `minmax`, `mean`...
+- `sequence` - an oo wrapper on sequential tables so you can do `t:insert(i, v)` instead of `table.insert(t, i, v)`. Also supports the functional interfance.
+- `vec2` - 2d vectors with method chaining, garbage saving interface
+- `vec3` - 3d vectors as above
+- `intersect` - 2d intersection routines, a bit sparse at the moment
+- `unique_mapping` - generate a unique mapping from arbitrary lua values to numeric keys - essentially making up a consistent ordering for unordered data. niche, but useful for optimising draw batches for example, as you can't sort on textures without it.
+- `state_machine` - finite state machine implementation with state transitions and all the rest. useful for game states, ai, cutscenes...
+- `async` - async operations as coroutines. 
+- `manual_gc` - get gc out of your update/draw calls. really good when trying to get accurate profiling information. requires you to think a bit about your garbage budgets though.
+- `colour` - colour conversion routines. can also be spelled `color` if you're into that.
+
+# PRs
+
+Pull requests are welcome. If you have something "big" to add please get in touch before starting work, but I'm quite open minded!
+
+# Globals?
+
+By default `batteries` will modify builtin lua tables (such as `table` and `math`) as it sees fit, as this eases consumption later on - you don't have to remember if say, `table.remove_value` is built in to lua or not. As the intended use case is fairly pervasive, required early, and "fire and forget", this sits with me just fine.
+
+If however, you'd prefer to require things locally, you can (rather ironically) set a few globals at boot time as documented in each module to change this behaviour, or set `BATTERIES_NO_GLOBALS = true` to make none of them modify anything global. If you really want to you can undefine the behaviour-changing globals after the module is required, as the results are cached.
+
+I'd strongly recommend that if you find yourself defining the above, stop and think why/if you really want to avoid globals for a library intended to be commonly used across your entire codebase!
+
+Some folks will have good reasons, which is why the functionality is present!
+
+Others may wish to reconsider and save themselves typing batteries a few hundred times :)
diff --git a/sequence.lua b/sequence.lua
index e24afb1..80fc8d6 100644
--- a/sequence.lua
+++ b/sequence.lua
@@ -7,6 +7,7 @@
 	in that case, you can still use the table methods that accept a table
 	first as method calls.
 ]]
+
 local sequence = {}
 
 sequence.mt = {__index = sequence}
@@ -20,52 +21,59 @@ function sequence:new(t)
 	return setmetatable(t or {}, sequence.mt)
 end
 
---import table functions to sequence as-is
-sequence.join = table.concat                --alias
+--alias
+sequence.join = table.concat
 
 --sorting default to stable if present
 sequence.sort = table.stable_sort or table.sort
 
---import functional interface to sequence in a sequence preserving way
+--(handle functional module delegation correctly)
+local _func = BATTERIES_FUNCTIONAL_MODULE or table
+
+--import functional interface to sequence in a type preserving way
 function sequence:keys()
-	return sequence:new(table.keys(self))
+	return sequence:new(_func.keys(self))
 end
 
 function sequence:values()
-	return sequence:new(table.values(self))
+	return sequence:new(_func.values(self))
 end
 
 function sequence:foreach(f)
-	return table.foreach(self, f)
+	return _func.foreach(self, f)
 end
 
 function sequence:reduce(f, o)
-	return table.foreach(self, f, o)
+	return _func.foreach(self, f, o)
 end
 
 function sequence:map(f)
-	return sequence:new(table.map(self, f))
+	return sequence:new(_func.map(self, f))
+end
+
+function sequence:remap(f)
+	return _func.remap(self, f)
 end
 
 function sequence:filter(f)
-	return sequence:new(table.filter(self, f))
+	return sequence:new(_func.filter(self, f))
 end
 
 function sequence:partition(f)
-	local a, b = table.partition(self, f)
+	local a, b = _func.partition(self, f)
 	return sequence:new(a), sequence:new(b)
 end
 
 function sequence:zip(other, f)
-	return sequence:new(table.zip(self, other, f))
+	return sequence:new(_func.zip(self, other, f))
 end
 
 function sequence:dedupe()
-	return table.dedupe(self)
+	return _func.dedupe(self)
 end
 
 function sequence:append_inplace(other)
-	return table.append_inplace(self, other)
+	return _func.append_inplace(self, other)
 end
 
 function sequence:append(other)
@@ -73,7 +81,7 @@ function sequence:append(other)
 end
 
 function sequence:copy(deep)
-	return sequence:new(table.copy(self, deep))
+	return sequence:new(_func.copy(self, deep))
 end
 
 return sequence
\ No newline at end of file
diff --git a/stable_sort.lua b/stable_sort.lua
index 09e9a8d..b0ac601 100644
--- a/stable_sort.lua
+++ b/stable_sort.lua
@@ -1,6 +1,6 @@
 -- stable sorting routines for lua
 --
---	modifies the global table namespace so you don't have
+--	by default modifies the global table namespace so you don't have
 --	to re-require it everywhere.
 --
 --		table.stable_sort
@@ -10,6 +10,11 @@
 --		table.insertion_sort
 --			an insertion sort, should you prefer it
 --
+--	alternatively, adds them to either BATTERIES_SORT_MODULE or
+--	BATTERIES_TABLE_MODULE if defined
+--
+--	also returns just the interface above as a module either way
+--
 
 --this is based on MIT licensed code from Dirk Laurie and Steve Fisher
 --license as follows:
@@ -36,14 +41,15 @@
 	DEALINGS IN THE SOFTWARE.
 ]]
 
--- (modifications by Max Cahill 2018)
+-- (modifications by Max Cahill 2018, 2020)
 
 local _sort_core = {}
 
---tunable size for
-_sort_core.max_chunk_size = 24
+--tunable size for insertion sort "bottom out"
+_sort_core.max_chunk_size = 32
 
-function _sort_core.insertion_sort_impl(array, first, last, less)
+--insertion sort on a section of array
+function _sort_core._insertion_sort_impl(array, first, last, less)
 	for i = first + 1, last do
 		local k = first
 		local v = array[i]
@@ -59,7 +65,8 @@ function _sort_core.insertion_sort_impl(array, first, last, less)
 	end
 end
 
-function _sort_core.merge(array, workspace, low, middle, high, less)
+--merge sorted adjacent sections of array
+function _sort_core._merge(array, workspace, low, middle, high, less)
 	local i, j, k
 	i = 1
 	-- copy first half of array to auxiliary array
@@ -91,24 +98,25 @@ function _sort_core.merge(array, workspace, low, middle, high, less)
 	end
 end
 
-
-function _sort_core.merge_sort_impl(array, workspace, low, high, less)
+--implementation for the merge sort
+function _sort_core._merge_sort_impl(array, workspace, low, high, less)
 	if high - low <= _sort_core.max_chunk_size then
-		_sort_core.insertion_sort_impl(array, low, high, less)
+		_sort_core._insertion_sort_impl(array, low, high, less)
 	else
 		local middle = math.floor((low + high) / 2)
-		_sort_core.merge_sort_impl(array, workspace, low, middle, less)
-		_sort_core.merge_sort_impl(array, workspace, middle + 1, high, less)
-		_sort_core.merge(array, workspace, low, middle, high, less)
+		_sort_core._merge_sort_impl(array, workspace, low, middle, less)
+		_sort_core._merge_sort_impl(array, workspace, middle + 1, high, less)
+		_sort_core._merge(array, workspace, low, middle, high, less)
 	end
 end
 
-function default_less(a, b)
+--default comparison; hoisted for clarity
+local function default_less(a, b)
 	return a < b
 end
 
 --inline common setup stuff
-function _sort_core.sort_setup(array, less)
+function _sort_core._sort_setup(array, less)
 	--default less
 	less = less or default_less
 	--
@@ -127,28 +135,35 @@ end
 
 function _sort_core.stable_sort(array, less)
 	--setup
-	local trivial, n, less = _sort_core.sort_setup(array, less)
+	local trivial, n, less = _sort_core._sort_setup(array, less)
 	if not trivial then
 		--temp storage; allocate ahead of time
 		local workspace = {}
 		local middle = math.ceil(n / 2)
 		workspace[middle] = array[1]
 		--dive in
-		_sort_core.merge_sort_impl( array, workspace, 1, n, less )
+		_sort_core._merge_sort_impl( array, workspace, 1, n, less )
 	end
 	return array
 end
 
 function _sort_core.insertion_sort(array, less)
 	--setup
-	local trivial, n, less = _sort_core.sort_setup(array, less)
+	local trivial, n, less = _sort_core._sort_setup(array, less)
 	if not trivial then
-		_sort_core.insertion_sort_impl(array, 1, n, less)
+		_sort_core._insertion_sort_impl(array, 1, n, less)
 	end
 	return array
 end
 
+_sort_core.unstable_sort = table.sort
+
 --export sort core
-table.insertion_sort = _sort_core.insertion_sort
-table.stable_sort = _sort_core.stable_sort
-table.unstable_sort = table.sort
+
+local _table = BATTERIES_SORT_MODULE or BATTERIES_TABLE_MODULE or table
+
+_table.insertion_sort = _sort_core.insertion_sort
+_table.stable_sort = _sort_core.stable_sort
+_table.unstable_sort = _sort_core.unstable_sort
+
+return _sort_core
diff --git a/table.lua b/table.lua
index 4f279b8..0ebedae 100644
--- a/table.lua
+++ b/table.lua
@@ -1,36 +1,52 @@
 --[[
 	extra table routines
+
+	optional:
+		set BATTERIES_TABLE_MODULE to a table before requiring
+		if you don't want this to modify the global `table` table
 ]]
 
+local _table = BATTERIES_TABLE_MODULE or table
+
+--apply prototype to module if it isn't the global table
+if BATTERIES_TABLE_MODULE ~= table then
+	setmetatable(BATTERIES_TABLE_MODULE, {
+		__index = table,
+	})
+end
+
+--alias
+_table.join = _table.concat
+
 --return the back element of a table
-function table.back(t)
+function _table.back(t)
 	return t[#t]
 end
 
 --remove the back element of a table and return it
-function table.pop(t)
+function _table.pop(t)
 	return table.remove(t)
 end
 
 --insert to the back of a table
-function table.push(t, v)
+function _table.push(t, v)
 	return table.insert(t, v)
 end
 
 --remove the front element of a table and return it
-function table.shift(t)
+function _table.shift(t)
 	return table.remove(t, 1)
 end
 
 --insert to the front of a table
-function table.unshift(t, v)
+function _table.unshift(t, v)
 	return table.insert(t, 1, v)
 end
 
 --find the index in a sequential table that a resides at
 --or nil if nothing was found
 --(todo: consider pairs version?)
-function table.index_of(t, a)
+function _table.index_of(t, a)
 	if a == nil then return nil end
 	for i,b in ipairs(t) do
 		if a == b then
@@ -42,8 +58,8 @@ end
 
 --remove the first instance of value from a table (linear search)
 --returns true if the value was removed, else false
-function table.remove_value(t, a)
-	local i = table.index_of(t, a)
+function _table.remove_value(t, a)
+	local i = _table.index_of(t, a)
 	if i then
 		table.remove(t, i)
 		return true
@@ -53,8 +69,8 @@ end
 
 --add a value to a table if it doesn't already exist (linear search)
 --returns true if the value was added, else false
-function table.add_value(t, a)
-	local i = table.index_of(t, a)
+function _table.add_value(t, a)
+	local i = _table.index_of(t, a)
 	if not i then
 		table.insert(t, a)
 		return true
@@ -70,7 +86,7 @@ local function _random(min, max, r)
 end
 
 --pick a random value from a table (or nil if it's empty)
-function table.pick_random(t, r)
+function _table.pick_random(t, r)
 	if #t == 0 then
 		return nil
 	end
@@ -78,7 +94,7 @@ function table.pick_random(t, r)
 end
 
 --shuffle the order of a table
-function table.shuffle(t, r)
+function _table.shuffle(t, r)
 	for i = 1, #t do
 		local j = _random(1, #t, r)
 		t[i], t[j] = t[j], t[i]
@@ -87,7 +103,7 @@ function table.shuffle(t, r)
 end
 
 --reverse the order of a table
-function table.reverse(t)
+function _table.reverse(t)
 	for i = 1, #t / 2 do
 		local j = #t - i + 1
 		t[i], t[j] = t[j], t[i]
@@ -95,28 +111,57 @@ function table.reverse(t)
 	return t
 end
 
+--collect all keys of a table into a sequential table
+--(useful if you need to iterate non-changing keys often and want an nyi tradeoff;
+--	this call will be slow but then following iterations can use ipairs)
+function _table.keys(t)
+	local r = {}
+	for k,v in pairs(t) do
+		table.insert(r, k)
+	end
+	return r
+end
+
+--collect all values of a keyed table into a sequential table
+--(shallow copy if it's already sequential)
+function _table.values(t)
+	local r = {}
+	for k,v in pairs(t) do
+		table.insert(r, v)
+	end
+	return r
+end
+
 --(might already exist depending on luajit)
-if table.clear == nil then
-	--destructively clear a numerically keyed table
-	--useful when multiple references are floating around
-	--so you cannot just pop a new table out of nowhere
-	function table.clear(t)
-		assert(type(to) == "table", "table.clear - argument 't' must be a table")
-		while t[1] ~= nil do
-			table.remove(t)
+if _table.clear == nil then
+	if _table ~= table and table.clear then
+		--import from global if it exists
+		_table.clear = table.clear
+	else
+		--remove all values from a table
+		--useful when multiple references are floating around
+		--so you cannot just pop a new table out of nowhere
+		function _table.clear(t)
+			assert(type(to) == "table", "table.clear - argument 't' must be a table")
+			local k = next(t)
+			while k ~= nil do
+				t[k] = nil
+				k = next(t)
+			end
 		end
 	end
 end
 
---overlay one table directly onto another, shallow only
-function table.overlay(to, from)
-	assert(type(to) == "table", "table.overlay - argument 'to' must be a table")
-	assert(type(from) == "table", "table.overlay - argument 'from' must be a table")
-	for k,v in pairs(from) do
-		to[k] = v
-	end
-	return to
-end
+--note: 
+--	copies and overlays are currently not satisfactory
+--
+--	i feel that copy especially tries to do too much and
+--	probably they should be split into separate functions
+--	to be both more explicit and performant, ie
+--
+--	shallow_copy, deep_copy, shallow_overlay, deep_overlay
+--
+--	input is welcome on this :)
 
 --copy a table
 --	deep_or_into is either:
@@ -128,7 +173,7 @@ end
 --	if into specified, copies into that table
 --		but doesn't clear anything out
 --		(useful for deep overlays and avoiding garbage)
-function table.copy(t, deep_or_into)
+function _table.copy(t, deep_or_into)
 	assert(type(t) == "table", "table.copy - argument 't' must be a table")
 	local is_bool = type(deep_or_into) == "boolean"
 	local is_table = type(deep_or_into) == "table"
@@ -140,7 +185,7 @@ function table.copy(t, deep_or_into)
 			if type(v.copy) == "function" then
 				v = v:copy()
 			else
-				v = table.copy(v, deep)
+				v = _table.copy(v, deep)
 			end
 		end
 		into[k] = v
@@ -148,3 +193,45 @@ function table.copy(t, deep_or_into)
 	return into
 end
 
+--overlay one table directly onto another, shallow only
+function _table.overlay(to, from)
+	assert(type(to) == "table", "table.overlay - argument 'to' must be a table")
+	assert(type(from) == "table", "table.overlay - argument 'from' must be a table")
+	for k,v in pairs(from) do
+		to[k] = v
+	end
+	return to
+end
+
+--faster unpacking for known-length tables up to 8
+--gets around nyi in luajit
+--note: you can use a larger unpack than you need as the rest
+--		can be discarded, but it "feels dirty" :)
+
+function _table.unpack2(t)
+	return t[1], t[2],
+end
+
+function _table.unpack3(t)
+	return t[1], t[2], t[3]
+end
+
+function _table.unpack4(t)
+	return t[1], t[2], t[3], t[4]
+end
+
+function _table.unpack5(t)
+	return t[1], t[2], t[3], t[4], t[5]
+end
+
+function _table.unpack6(t)
+	return t[1], t[2], t[3], t[4], t[5], t[6]
+end
+
+function _table.unpack7(t)
+	return t[1], t[2], t[3], t[4], t[5], t[6], t[7]
+end
+
+function _table.unpack8(t)
+	return t[1], t[2], t[3], t[4], t[5], t[6], t[7], t[8]
+end
diff --git a/vec2.lua b/vec2.lua
index 48d99a9..b7df938 100644
--- a/vec2.lua
+++ b/vec2.lua
@@ -5,7 +5,7 @@
 --[[
 	notes:
 
-	depends on a class() function as in oo.lua
+	depends on a class() function as in batteries class.lua
 
 	some methods depend on math library extensions
 
diff --git a/vec3.lua b/vec3.lua
index f66de1e..5371dbe 100644
--- a/vec3.lua
+++ b/vec3.lua
@@ -5,7 +5,7 @@
 --[[
 	notes:
 
-	depends on a class() function as in oo.lua
+	depends on a class() function as in batteries class.lua
 
 	some methods depend on math library extensions