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
Haunted House of Angular unit testing
Search
Sponsored
·
SiteGround - Reliable hosting with speed, security, and support you can count on.
→
Yosuke Kurami
May 10, 2017
Programming
0
490
Haunted House of Angular unit testing
How to reduce execution time Angular Karma testing
Yosuke Kurami
May 10, 2017
Tweet
Share
More Decks by Yosuke Kurami
See All by Yosuke Kurami
TypeScript LSP の今までとこれから
quramy
1
1.6k
フロントエンドテストの育て方
quramy
12
3.7k
App Router 悲喜交々
quramy
8
690
上手に付き合うコンポーネントテスト
quramy
6
2.3k
Patched fetch did not work
quramy
6
750
GraphQL あるいは React における自律的なデータ取得について
quramy
18
5.7k
Next.js App Router
quramy
15
3.8k
Fragment Composition of GraphQL
quramy
17
4.5k
reg-viz VRT tools
quramy
4
1.6k
Other Decks in Programming
See All in Programming
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
380
The Past, Present, and Future of Enterprise Java
ivargrimstad
0
600
要求定義・仕様記述・設計・検証の手引き - 理論から学ぶ明確で統一された成果物定義
orgachem
PRO
1
180
責任感のあるCloudWatchアラームを設計しよう
akihisaikeda
3
180
Raku Raku Notion 20260128
hareyakayuruyaka
0
350
Best-Practices-for-Cortex-Analyst-and-AI-Agent
ryotaroikeda
1
110
FOSDEM 2026: STUNMESH-go: Building P2P WireGuard Mesh Without Self-Hosted Infrastructure
tjjh89017
0
170
開発者から情シスまで - 多様なユーザー層に届けるAPI提供戦略 / Postman API Night Okinawa 2026 Winter
tasshi
0
210
例外処理とどう使い分ける?Result型を使ったエラー設計 #burikaigi
kajitack
16
6.1k
そのAIレビュー、レビューしてますか? / Are you reviewing those AI reviews?
rkaga
6
4.6k
なるべく楽してバックエンドに型をつけたい!(楽とは言ってない)
hibiki_cube
0
140
AIで開発はどれくらい加速したのか?AIエージェントによるコード生成を、現場の評価と研究開発の評価の両面からdeep diveしてみる
daisuketakeda
1
2.5k
Featured
See All Featured
Joys of Absence: A Defence of Solitary Play
codingconduct
1
290
WENDY [Excerpt]
tessaabrams
9
36k
Statistics for Hackers
jakevdp
799
230k
Believing is Seeing
oripsolob
1
57
Leveraging Curiosity to Care for An Aging Population
cassininazir
1
160
The Power of CSS Pseudo Elements
geoffreycrofte
80
6.2k
XXLCSS - How to scale CSS and keep your sanity
sugarenia
249
1.3M
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
47
7.9k
Groundhog Day: Seeking Process in Gaming for Health
codingconduct
0
94
Utilizing Notion as your number one productivity tool
mfonobong
3
220
Measuring Dark Social's Impact On Conversion and Attribution
stephenakadiri
1
130
Building Applications with DynamoDB
mza
96
6.9k
Transcript
Haunted House of Angular unit testing by @Quramy #ng_sake 2017.05.10
About me
ͦΕڈͷࢣͷ͜ͱ…
feature branch࡞ͬͯͶɺ ŝƄŕŰŲƄʹcommitͨ͠ΜͰ͢Α
ͦͨ͠ΒCI͕Γग़ͯ͠Ͷɺ
ͰͶɺ ͳʔΜ͔༷ࢠ͕͓͔͍͠
มͩͳʔɺා͍ͳʔ
ڪΔڪΔCIͷཧը໘Λ ։͍ͯΈΔͱ…
ςετ͕ऴΘΒͳ͍ΜͰ͢ Α͓͓͓͓͒͒͒͒͒
Angular
Scale of our product: ~130,000 loc ~30 angular modules ~700
components ~1,000 cases
1. Docker pull, run container 2. Webpack bundle 3. Run
Karma (with capture screenshots) 4. Upload screenshot images to AWS S3 Workflow of unit test on CI 1 2 3 4
Question: workflow࣮ߦʹͲΕ͘Βֻ͍͔Δͷ͔
Changes of duration ઢ: CIͰͷ୯ମςετ࣮ߦ࣌ؒ [msec]: ࠨ࣠, ੨ઢ: masterͷιʔείʔυߦ
[loc]: ӈ࣠
Changes of duration ઢ: CIͰͷ୯ମςετ࣮ߦ࣌ؒ [msec]: ࠨ࣠, ੨ઢ: masterͷιʔείʔυߦ
[loc]: ӈ࣠ 1 upgrade ng 4 3 2
ɹͰԿΛͨ͠ͷ͔?
Today’s theme: ୯ମςετߴԽςΫχοΫ (ςετίʔυͷॻ͖ํͱ͔Ұઆ໌͠ͳ͍Α!)
1. Good bye Phantom, welcome to Nightmare
Unit testing on CI • Angularͷ୯ମςετͰKarmaΛ༻͍Δ • Karmaͷಈ࡞ʹϒϥβ͕ඞཁ • CIͰheadlessʹಈ࡞ͤ͞ͳͯ͘ͳΒͳ͍
PhantomJS • Prjॳظkarma-phantomjs-launcher Λར༻ • PhantomJSࣗମɺChromeͱൺΔͱ͍ • ࠷େͷܽͦͷෆ҆ఆੑɻ “ྶ”ͱ͔ݴ͏ׂʹͨΒࢮ͵ɻ
PhantomJS -> hangs up
Nightmare • karma-nightmare[1] ʹΓ͑ • Nightmare: Electronϕʔεͷscraping tool • ҎલPhantomJSϕʔεͰ͕͋ͬͨɺੑೳվ
ળͷͨΊʹElectronϕʔεʹΓସ͑ΒΕͨ[2] <>https://github.com/bokuweb/karma-nightmare <>https://github.com/segmentio/nightmare/issues/484#issuecomment-184519591
How effective ? • 10%~20%ͷ࣮ߦ࣌ؒݮޮՌ • Pros: • ϋϯάൃੜ͕ແ͘ͳΓɺίϯελϯτʹςε τ͕ྃ͢ΔΑ͏ʹͳͬͨ
• ಋೖ͕༰қ. karma-confͷઃఆ + ԾϑϨʔ ϜόοϑΝઃఆͰCIಈ࡞͢Δ
2. More Karma, more and more…!
Multiple cores, single thread • Karmaϒϥβ্ͰςετΛ࣮ߦ͢Δ • جຊతʹγϯάϧεϨου • ϚϧνίΞڥͰଞͷίΞ͕༡ͼ͕ͪʹ
• ͋Εɺϓϩηε͚ΕΑ͘Ͷʁ
CPU Load ϩʔΧϧ(MBP, 4 core)ͰͷCPUෛՙ ۀ ۀ
ͦ͏ɺզʑʹ ۀ͕Γ͍ͯͳ͔ͬͨͷͩ
Split test bundles • test bundleΛmodule୯ҐͰׂ, bundle୯ҐͰKarmaϓϩηεΛىಈ /* test.ts */
const context = require.context('./', true, /\.spec\.ts$/); context.keys().map(context); /* test-a.ts */ const context = require.context('./module-a', true, /\.spec\.ts$/); context.keys().map(context); /* test-b.ts */ const context = require.context('./module-b', true, /\.spec\.ts$/); context.keys().map(context);
How effective ? • 50%~60%ͷ࣮ߦ࣌ؒݮޮՌ • Cons: • ʮCPUΛ͑Δ͚ͩ͏ʯઓུͷͨΊɺ CI(Wercker)ͷՔಇঢ়گ࣍ୈͳ෦ΞϦ
• ಋೖোนߴΊɻbundle࡞ॲཧΛΠδΔͨΊɺ angular-cliͷ߹ཁng eject
3. Zombie compiler
What does make tests slow? JiT Compiler. CPU profile with
~300 components, 15 test cases
What TestBed does? it(…)ͷɺFooModuleͷશcomponent͕compile͞ΕΔ beforeEach(async(() => { TestBed.configureTestingModule({ imports: [
FooModule, ], }).compileComponents(); })); it('should ...', () => { /* ... /* }); .component.spec.tsͰimportsΛॻ͘ͱ…
Avoid to compile components componentͷcompileճΛݮΒͤ, ͦͷߴԽ • moduleΛదͳཻʹׂͯ͠ґଘରΛখ͘͢͞Δ • componentΛcompile͠ͳ͍.
௨ৗͷclassͱͯ͠test͢Δ (isolated-component-tests) • ෳࡶͳॲཧΛcomponent͔ΒserviceҠߦ͢Δ service testͷׂ߹Λ૿͢
None
͏গ͠ૡ͍ͯΈΑ͏
TestingCompiler impl • JiT CompilerΩϟογϡػߏΛඋ͍͑ͯΔ • Compiler Lifecycleit(…)ຖͰ͋ΔͨΊɺςε τέʔεΛލ͙ͱΩϟογϡ͝ͱഁغ͞ΕΔ •
it(“should do A”) ͱ it(“should do B”) Ͱ Ωϟογϡ͕࠶ར༻͞ΕΔࣄͳ͍
ͰςετରͷఆٛͳΜͯɺ Karma࣮ߦதෆม͡ΌͶʁ ※ overrideComponentͷར༻͕ແ͍ͱ͍͏લఏͰ
Make compiler “undead” ngx-zombie-compiler[1] • ςετέʔεΛލ͍Ͱʮࢮͳͳ͍ʯcompiler • (ՄೳͳݶΓ)cacheͨ͠compile݁ՌΛ͍ճͨ͢Ίɺߴ import {
ZOMBIE_COMPILER_PROVIDERS } from "ngx-zombie-compiler"; beforeEach(() => { TestBed.configureCompiler({ providers: ZOMBIE_COMPILER_PROVIDERS, }); }); [1] https://github.com/Quramy/ngx-zombie-compiler
Demonstration normal with zombie
How effective ? • ࣮prjͰ50~60%ͷ࣮ߦ࣌ؒݮޮՌ • Pros: • ಋೖ༰қɻtest.tsͷमਖ਼ͷΈͰಈ࡞͢Δ •
Cons: • Angular 5.xͷࠒʹ௨༻͠ͳ͘ͳΔ… • ୯ମςετͷႈੑΛߟྀ͢ΔͱΩϫϞϊ…
Summary
Improvement techniques ҙ֎ͱڭՊॻʹԊͬͯΔײɿ • Nightmare < - > ߴͳ࣮ߦج൫ͷಋೖ •
Parallel Karma < - > ྻॲཧͷฒྻԽ • Zombie Compiler < - > Flyweightύλʔϯద༻ ʮ͍͔Βճ͞ͳ͍ʯ͑ɻͲ͠Ͳ͠ςετॻ͜͏ʂ
Thank you!