Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
tc39_study
Search
Daiki Ihara
October 09, 2019
Programming
1
770
tc39_study
Temporal
BinaryAST
Daiki Ihara
October 09, 2019
Tweet
Share
More Decks by Daiki Ihara
See All by Daiki Ihara
ArrayBufferとBinary
sasurau4
1
640
CLIから見るAngular, React, Vue
sasurau4
1
960
Metro Bundler in Web
sasurau4
0
980
reading-mtc2018-web.pdf
sasurau4
1
560
Other Decks in Programming
See All in Programming
CSC307 Lecture 03
javiergs
PRO
1
490
AIエージェント、”どう作るか”で差は出るか? / AI Agents: Does the "How" Make a Difference?
rkaga
4
2k
CSC307 Lecture 05
javiergs
PRO
0
500
AI Agent Tool のためのバックエンドアーキテクチャを考える #encraft
izumin5210
6
1.8k
今こそ知るべき耐量子計算機暗号(PQC)入門 / PQC: What You Need to Know Now
mackey0225
3
370
フルサイクルエンジニアリングをAI Agentで全自動化したい 〜構想と現在地〜
kamina_zzz
0
400
Basic Architectures
denyspoltorak
0
670
LLM Observabilityによる 対話型音声AIアプリケーションの安定運用
gekko0114
2
430
Unicodeどうしてる? PHPから見たUnicode対応と他言語での対応についてのお伺い
youkidearitai
PRO
1
2.5k
登壇資料を作る時に意識していること #登壇資料_findy
konifar
4
1.1k
プロダクトオーナーから見たSOC2 _SOC2ゆるミートアップ#2
kekekenta
0
210
[KNOTS 2026登壇資料]AIで拡張‧交差する プロダクト開発のプロセス および携わるメンバーの役割
hisatake
0
270
Featured
See All Featured
Imperfection Machines: The Place of Print at Facebook
scottboms
269
14k
The Spectacular Lies of Maps
axbom
PRO
1
520
Sharpening the Axe: The Primacy of Toolmaking
bcantrill
46
2.7k
HU Berlin: Industrial-Strength Natural Language Processing with spaCy and Prodigy
inesmontani
PRO
0
210
Visualizing Your Data: Incorporating Mongo into Loggly Infrastructure
mongodb
49
9.9k
Tell your own story through comics
letsgokoyo
1
810
Git: the NoSQL Database
bkeepers
PRO
432
66k
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
47
7.9k
The Cost Of JavaScript in 2023
addyosmani
55
9.5k
Leveraging LLMs for student feedback in introductory data science courses - posit::conf(2025)
minecr
0
140
Navigating Weather and Climate Data
rabernat
0
100
RailsConf & Balkan Ruby 2019: The Past, Present, and Future of Rails at GitHub
eileencodes
141
34k
Transcript
tc39_study https://github.com/sasurau4/slides
ABOUT ME { "id": "sasurau4", "name": "Daiki Ihara", "description": "Front-end engineer at CureApp, Inc.", "techs": { "work": ["React Native", "TypeScript", "JavaScript", "MongoDB"], "subwork": ["React Native", "React", "TypeScript"], "hobby": ["React", "TypeScript", "JavaScript", "Deno"] },
"homepage": "https://sasurau4.github.io/profile/" }
What is temporal? Provides standard objects and functions for working with dates and times. This proposal is currently stage 2.
Motivation Fixing JavaScript Date java.util.Date― その悲劇と歴史
Principles All temporal APIs are non-mutating. All temporal objects are effectively immutable. All date values are based on the Proleptic Gregorian Calendar. All time-of-day values are based on a standard 24-hour clock. Leap seconds are not represented.
Specification The documentation is out of date.
Try it with polyfill node --experimental-modules --no-warnings --icu-data-dir ./node_modules/full-icu/ -r ./lib/initialise.js For more details, https://github.com/tc39/proposal-temporal/tree/main/docs
Results > const day = Temporal.Date(2019, 1, 1) undefined > day.day 1 > day.toString() '2019-01-01'
> let d1 = Temporal.DateTime(2019, 2, 1) undefined > d1.toLocaleString() '2019/2/1 0:00:00' > d1.daysInMonth 28 > let onem = Temporal.Duration(0, 1) undefined > onem.toString() 'P1M'
What is binaryAST? Startup performance degrades with larger JS payloads, even if only a fraction of the code is actually executed. Parsing time is a significant component, taking more CPU time than bytecode / initial JIT code generation. For example, on a powerful laptop, Chrome spends 10% to 15% of CPU time parsing JS while loading facebook.com. We propose a new over-the-wire format for JS that is a binary encoding of an AST. We believe this new format would allow for drastically faster parsing. Moreover, web developers are well positioned to adopt a new format as they have embraced build tooling.
Motivation from https://blog.cloudflare.com/binary-ast/
prototype of binAST from https://github.com/binast/binjs-ref As websites become more sophisticated, the amount of JavaScript source code keeps increasing. While depending upon a large JavaScript codebase won't prevent a website from working, it will cause websites to start slowly – often unacceptably slow. This is due to two bottlenecks: parsing and bytecode compiling JavaScript. Unfortunately, browsers have pretty much reached efficiency peak for both operations. We (Mozilla, Bloomberg, Facebook,
CloudFlare) are currently working on a domain-specific encoding for JavaScript, called "BinAST" (short for "JavaScript Binary AST"). The JavaScript Binary AST is designed to break the bottleneck.
Result
Details https://blog.cloudflare.com/binary-ast/
Demo Try it with Firefox nightly! https://serve-binjs.that-test.site
The End