$30 off During Our Annual Pro Sale. View Details »
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Monads you've already put in production (withou...
Search
Tejas Dinkar
October 10, 2014
Technology
1
1.2k
Monads you've already put in production (without knowing it)
Tejas Dinkar
October 10, 2014
Tweet
Share
More Decks by Tejas Dinkar
See All by Tejas Dinkar
Quick Wins for Page Speed
gja
0
140
Progressive Web Apps In Clojure(Script)
gja
4
2.4k
Lightning - Monads you already use (without knowing it)
gja
1
410
Native Extensions Served 3 Ways
gja
0
360
Other Decks in Technology
See All in Technology
今からでも間に合う!速習Devin入門とその活用方法
ismk
1
750
Haskell を武器にして挑む競技プログラミング ─ 操作的思考から意味モデル思考へ
naoya
6
1.6k
AI-DLCを現場にインストールしてみた:プロトタイプ開発で分かったこと・やめたこと
recruitengineers
PRO
2
150
ログ管理の新たな可能性?CloudWatchの新機能をご紹介
ikumi_ono
1
840
AIプラットフォームにおけるMLflowの利用について
lycorptech_jp
PRO
1
170
シニアソフトウェアエンジニアになるためには
kworkdev
PRO
3
170
AWS CLIの新しい認証情報設定方法aws loginコマンドの実態
wkm2
6
750
2025年 開発生産「可能」性向上報告 サイロ解消からチームが能動性を獲得するまで/ 20251216 Naoki Takahashi
shift_evolve
PRO
1
200
「図面」から「法則」へ 〜メタ視点で読み解く現代のソフトウェアアーキテクチャ〜
scova0731
0
310
マイクロサービスへの5年間 ぶっちゃけ何をしてどうなったか
joker1007
12
6k
Lessons from Migrating to OpenSearch: Shard Design, Log Ingestion, and UI Decisions
sansantech
PRO
1
140
re:Invent2025 コンテナ系アップデート振り返り(+CloudWatchログのアップデート紹介)
masukawa
0
390
Featured
See All Featured
The Cost Of JavaScript in 2023
addyosmani
55
9.4k
Connecting the Dots Between Site Speed, User Experience & Your Business [WebExpo 2025]
tammyeverts
10
730
Practical Orchestrator
shlominoach
190
11k
Producing Creativity
orderedlist
PRO
348
40k
Visualizing Your Data: Incorporating Mongo into Loggly Infrastructure
mongodb
48
9.8k
Context Engineering - Making Every Token Count
addyosmani
9
520
Principles of Awesome APIs and How to Build Them.
keavy
127
17k
Agile that works and the tools we love
rasmusluckow
331
21k
The Cult of Friendly URLs
andyhume
79
6.7k
実際に使うSQLの書き方 徹底解説 / pgcon21j-tutorial
soudai
PRO
196
70k
Keith and Marios Guide to Fast Websites
keithpitt
413
23k
Visualization
eitanlees
150
16k
Transcript
Monads you are already using in prod Tejas Dinkar nilenso
about.me • Hi, I’m Tejas • Nilenso: Partner • twitter:
tdinkar • github: gja
Serious Pony
Online Abuse
Trouble at the Koolaid Point http://seriouspony.com/trouble-at-the-koolaid-point/ https://storify.com/adriarichards/telling-my-troll-story-because- kathy-sierra-left-t
If you think you understand Monads, you don't understand Monads.
None
This talk is inaccurate and will make a mathematician cry
None
Goal of this talk For you to say “Oh yeah,
I’ve used that hack”
None
Monads • Programmable Semicolons • Used to hide plumbing away
from you • You can say Monads in almost any sentence and people will think you are smart
None
Values Value
Monads Value Box
Mysore Masala Monad M onad Value
Monads Value Box
Monads • Monads define two functions • return takes a
value and puts it in a box • bind takes a box & function f, returning f(value) • it is expected that the function returns a box
Value Value Another Value Value Function return bind
Our Function Signatures Value f(value)
Some math (√4) + 5
Some math (√4) + 5 3 or 7!
Value 4
Monad [4]
[alive, dead]
ruby! x = [1, 2, 3] y = x.map {
|x| x + 1 } # y = [2, 3, 4]
return Value Value return
return def m_return(x) [x] end # m_return(4) => [4]
The functions Value f(value)
Square Root fn def sqrt(x) s = Math.sqrt(x) [s, -s]
end # sqrt(4) => [2, -2]
Increment Fn def inc_5(x) [x + 5] end # inc_5(1)
=> [6]
Bind Functions Another Value Value Function bind
Bind Function x = m_return(4) y = x.????? { |p|
sqrt(p) } # I want [-2, 2]
Bind Function x = m_return(4) y = x.map {|p| sqrt(p)
} # y => [[2, -2]] # ^—— Box in a box?
Bind Function x = m_return(4) y = x.mapcat {|p| sqrt(p)
} # y => [2, -2]
Putting it together m_return(4) .mapcat {|p| sqrt(p)} .mapcat {|p| inc_5(p)}
# => [3, 7]
You have invented the List Monad, used to model non-determinism
Congrats
Turtles all the way down
A small constraint • Let’s do a bit of a
self imposed constraint on this • Functions must return either 0 or 1 elements • (we’ll only model positive integers here)
return - stays the same
bind - stays the same x = m_return(4) y =
x.mapcat { |p| inc_5(p) } # y => 9
Square Root Fn def sqrt(x) if (x < 0) return
[] #error else [Math.sqrt(x)] end end # sqrt(4) => [2] # sqrt(-1) => []
Describe in English There is a list passed to each
step Maybe this list has just one element, or Maybe it has none
None
The Maybe Monad • The intent is to short circuit
computation • The value of the `box’ is None, or Just(Value) • You can think of it as a type-safe nil / null
try def try(x, f) if x == nil return f(x)
else return nil end end # 4.try { |x| x + 5 } => 9 # nil.try {|x| x + 5 } => nil
None
Let’s start over • The Monad Laws • Left Identity
• Right Identity • Associativity
Left Identity m_return(a).bind(f) == f(a)
Right Identity m.bind(m_return) == m
Associativity m.bind(f).bind(g) == m.bind(x -> f(x).bind(g))
Store Computation
The State Monad • Rest of the world - State
Machine (sorta) • The value inside the box f(state) => [r new-state] • Particularly useful in pure languages like Haskell • Let’s build a stack
The functions Value f(value)
The functions (f(value) state) [new-value, new-state]
push def push(val) lambda { |state| new_state = state.push(val) [value,
new_state] } end
pop def pop() lambda { |state| val = state.pop() [val,
state] } end
def double_top() lambda { |state| top = state.pop() [2 *
top, state.push(2*top)] } end double_top
return def m_return(x) lambda { |state| [x, state] } end
bind def bind(mv, f) lambda { |state| v, temp_state =
mv(state) state_fn = f(v) state_fn(temp_state) } end
example # Not working code ! m_return(4) .bind(a -> push(a))
.bind(b -> push(b + 1)) .bind(c -> double_top()) .bind(d -> sum_top2()) .bind(e -> pop())
None
Associativity m.bind(f).bind(g) == m.bind(x => f(x).bind(g))
turn this # Not working code ! m_return(4) .bind(a ->
push(a)) .bind(b -> push(b + 1)) .bind(c -> double_top()) .bind(d -> sum_top2()) .bind(e -> pop())
into this m_return(4) .bind(a -> push(a) .bind(b -> push(b +
1) .bind(c -> double_top() .bind(d -> sum_top() .bind(e -> pop())))))
done with ruby
imagine # Not working code state_monad { a <- m_return(4)
b <- push(a) c <- push(b + 1) d <- double_top() e <- sum_top2() pop() }
Back to List m_return(4) .mapcat {|p| sqrt(p)} .mapcat {|p| inc_5(p)}
# => [3, 7]
Back to List m_return(4) .mapcat {|a| sqrt(a) .mapcat {|b| inc_5(b)}}
# => [3, 7]
Back to List list_monad { a <- m_return(4) b <-
sqrt(a) c <- inc_5(b) c }
On to Clojure • this is an example from clojure.net
• the state is a vector containing every function we’ve called so far
(defn inc-s [x] (fn [state] [(inc x) (conj state :inc)]))
in clojure (defn inc-s [x] (fn [state] [(inc x) (conj
state :inc)])) (defn do-things [x] (domonad state-m [a (inc-s x) b (double-s a) c (dec-s b) d (dec-s c)] d)) ! ((do-things 7) []) => [14 [:inc :double :dec :dec]]
state monad in Clojure (defmonad state-m "Monad describing stateful computations.
The monadic values have the structure (fn [old-state] [result new-state])." [m-result (fn m-result-state [v] (fn [s] [v s])) m-bind (fn m-bind-state [mv f] (fn [s] (let [[v ss] (mv s)] ((f v) ss)))) ])
state monad in Haskell inc = state (\st -> let
st' = st +1 in (st’,st')) inc3 = do x <- inc y <- inc z <- inc return z
Finally, IO
IOMonad • rand-int(100) is non deterministic !
ay-yo
IOMonad • rand-int(100) is non deterministic • rand-int(100, seed =
42) is deterministic • monadic value: f(world) => [value, world-after-io]
IOMonad • puts() just `appends to a buffer’ in the
real world • How does gets() return different strings? • gets() returns a fixed value based on the `world’
Image Credits http://www.myfoodarama.com/2010/11/masala- dosa.html http://www.clojure.net/2012/02/10/State/ http://www.cafepress.com/ +no_place_like_home_ruby_slippers_3x5_area_rug, 796646161 http://www.netizens-stalbans.co.uk/installs-and- upgrades.html.htm
http://www.hpcorporategroup.com/what-is-the-life- box.html
Thank You MANY QUESTIONS? VERY MONAD SO FUNCTIONAL Y NO
CLOJURE?
[email protected]
@tdinkar WOW WOW WOW MUCH EASY SUPER SIMPLE