refactor(suite): separate app composition from initialization by peter-sanderson · Pull Request #32195 · trezor/trezor-suite

peter-sanderson · 2026-09-07T17:51:10Z

Description

Make the web and desktop composition roots synchronous and return an async init service, so service wiring is independent of application startup. Create the Redux store during composition and hydrate that same store after storage preloading and the desktop handshake, preserving preload behavior without running application middleware during hydration.

🤖 LLM Test Recommendations

Summary: The changed files are concentrated in app initialization, web/desktop composition roots, and Redux store creation/hydration. The biggest risks are app startup failures, broken persisted-state/migration behavior, and regressions in TrezorConnect or bridge initialization. A small, targeted smoke suite covering initial run, database migrations, browser entry gates, desktop bridge lifecycle, and representative TrezorConnect flows provides high confidence without running the full catalog.

Changed files (22)

packages/suite-desktop-ui/src/MainDesktop.tsx
packages/suite-desktop-ui/src/createDesktopInit.tsx
packages/suite-desktop-ui/src/createSuiteDesktopCompositionRoot.ts
packages/suite-desktop-ui/src/index.tsx
packages/suite-web/src/MainWeb.tsx
packages/suite-web/src/createSuiteWebCompositionRoot.ts
packages/suite-web/src/createWebInit.tsx
packages/suite-web/src/index.ts
packages/suite-web/src/static/vite-index.ts
packages/suite/index.ts
packages/suite/src/reducers/createHydrateReduxStore.ts
packages/suite/src/reducers/createReduxStore.ts
packages/suite/src/reducers/store.ts
packages/suite/src/support/createConnectLoggerFactory.ts
packages/suite/src/support/createGetBinFilesBaseUrl.ts
packages/suite/src/support/suite/preloadStore.ts
suite/test-utils/package.json
suite/test-utils/src/StoreProviderForTests.tsx
suite/test-utils/src/index.ts
suite/test-utils/src/initStoreForTests.ts
suite/test-utils/src/renderWithStore.tsx
suite/test-utils/tsconfig.json

Recommended tests (10)

🔴 High priority (10)

suite/e2e/tests/suite/initial-run.test.ts — Verifies onboarding/analytics consent state persists across reloads and that the app reaches a connected-device state after initial run. This directly exercises the web composition root, init flow, and Redux store hydration/preload logic.
suite/e2e/tests/suite/db-migration.test.ts — Tests IndexedDB migration of persisted user settings between Suite versions, which relies on the updated store creation, hydration, and preload-store code paths.
suite/e2e/tests/suite/db-locale-migration.test.ts — Validates that language preference survives a cross-version database migration and app reload, directly exercising persisted store hydration and initialization.
suite/e2e/tests/browser/firefox.test.ts — Confirms the web app loads and the unsupported-browser gate behaves correctly in Firefox, exercising the web entry point and composition root initialization.
suite/e2e/tests/browser/safari.test.ts — Confirms the unsupported-browser gate renders and can be bypassed in Safari, directly testing the web app shell and entry-point initialization.
suite/e2e/tests/browser/ios.test.ts — Verifies the iOS unsupported-platform page renders without a bypass option, exercising platform detection and the web composition root entry path.
suite/e2e/tests/bridge-tor/spawn-bridge.test.ts — Tests that the desktop app spawns and stops its bundled bridge process correctly across launch, reload, and quit, directly covering the desktop initialization and composition root.
suite/e2e/tests/bridge-tor/spawn-bridge-daemon.test.ts — Exercises the desktop daemon-mode startup path where the bridge is spawned separately from the UI, directly testing the desktop init and composition root changes.
suite/e2e/tests/trezor-connect/getAddress.test.ts — Validates TrezorConnect address export in suite-desktop core mode after onboarding, covering Connect initialization, permissions modal, and desktop app integration affected by the composition root and Connect logger setup.
suite/e2e/tests/trezor-connect/connectPopupWeb.test.ts — Covers the web-based TrezorConnect popup lifecycle and permission/address flows, exercising the web composition root, init, and Connect integration.

⚠️ Changes with no test coverage (7)

packages/suite/src/support/createGetBinFilesBaseUrl.ts
suite/test-utils/package.json
suite/test-utils/src/StoreProviderForTests.tsx
suite/test-utils/src/index.ts
suite/test-utils/src/initStoreForTests.ts
suite/test-utils/src/renderWithStore.tsx
suite/test-utils/tsconfig.json

Updated: 2026-09-08T10:56:08.980Z

🌐 Preview deployments

🌐 Suite Web preview: https://dev.suite.sldev.cz/suite-web/refactor-suite-app-initialization/web/

🔍 Currents Test Results

🔍 Suite desktop test results: View in Currents

🔍 Suite web test results: View in Currents

🔒 Quarantined E2E Tests

Trezor Suite (web) — 4 test(s)

Test	Type
Staking - Cardano > Stake Cardano	🤖 auto
Suite Sync - Update and Remove Labels > Update and remove labels syncs correctly to relay	🤖 auto
Trading - Swap > Swap SOL USDT token to ETH	🤖 auto
Trading - Swap coin to token > Swap Solana to USDC	🤖 auto

Updated: 2026-09-08T13:19:20.274Z • 4 test(s) total

Trezor Suite (desktop) — 5 test(s)

Test	Type
Trading - Sell BTC > Sell Bitcoin for best offer	🤖 auto
Staking - Cardano > Stake Cardano	🤖 auto
Suite Sync - Update and Remove Labels > Update and remove labels syncs correctly to relay	🤖 auto
Quarantine test: "Suite Sync - Quota Manager top-up,Exceeded wallet limit is topped up from the device pool"	🙋 manual
Trading - Swap > Swap SOL USDT token to ETH	🤖 auto

Updated: 2026-09-08T13:18:34.930Z • 5 test(s) total

github-actions · 2026-09-07T19:02:09Z

⚡️ Performance report (alpha)

Note

Alpha release: The limits are still being calibrated, and each measurement is
usually a single run on a shared CI runner, so expect noise between runs. Treat a
number as a hint to look closer, not as a verdict.

desktop / group 2

🟢 Within limits.

_{Each metric reads measured/limit (% of limit).}

Scenario	TBT	Long tasks	Longest task	React commits	Interaction
`multi-account-discovery [T3W1]`	1298/7000 ms (19 %)	44/100 (44 %)	295/800 ms (37 %)	230/400 (57 %)	7265 ms

desktop / group 5

🟢 Within limits.

_{Each metric reads measured/limit (% of limit).}

Scenario	TBT	Long tasks	Longest task	React commits	Interaction
`account-switch [T3W1]`	165/700 ms (24 %)	1/5 (20 %)	215/700 ms (31 %)	36/80 (45 %)	567 ms
`wallet-discovery [T3W1]`	390/2000 ms (20 %)	16/45 (36 %)	230/400 ms (57 %)	193/410 (47 %)	6071 ms

desktop / group 3

🟢 Within limits.

_{Each metric reads measured/limit (% of limit).}

Scenario	TBT	Long tasks	Longest task	React commits	Interaction
`account-switch [T3W1]`	150/700 ms (21 %)	1/5 (20 %)	200/700 ms (29 %)	29/80 (36 %)	500 ms
`wallet-discovery [T3W1]`	182/2000 ms (9 %)	10/45 (22 %)	130/400 ms (33 %)	212/410 (52 %)	5327 ms

desktop / group 1

🟢 Within limits.

_{Each metric reads measured/limit (% of limit).}

Scenario	TBT	Long tasks	Longest task	React commits	Interaction
`account-switch [T3T1]`	315/700 ms (45 %)	1/5 (20 %)	365/700 ms (52 %)	35/80 (44 %)	671 ms
`wallet-discovery [T3T1]`	818/2000 ms (41 %)	23/45 (51 %)	227/400 ms (57 %)	155/410 (38 %)	5329 ms

desktop / group 4

🟢 Within limits.

_{Each metric reads measured/limit (% of limit).}

Scenario	TBT	Long tasks	Longest task	React commits	Interaction
`account-switch [T3W1]`	244/700 ms (35 %)	2/5 (40 %)	290/700 ms (41 %)	35/80 (44 %)	742 ms
`multi-account-discovery [T3W1]`	1293/7000 ms (18 %)	47/100 (47 %)	245/800 ms (31 %)	227/400 (57 %)	6900 ms
`wallet-discovery [T3W1]`	608/2000 ms (30 %)	20/45 (44 %)	188/400 ms (47 %)	154/410 (38 %)	5231 ms

desktop / group 6

🟢 Within limits.

_{Each metric reads measured/limit (% of limit).}

Scenario	TBT	Long tasks	Longest task	React commits	Interaction
`multi-account-discovery [T3W1]`	1154/7000 ms (16 %)	46/100 (46 %)	252/800 ms (32 %)	229/400 (57 %)	6168 ms

desktop / group 7

🟢 Within limits.

_{Each metric reads measured/limit (% of limit).}

Scenario	TBT	Long tasks	Longest task	React commits	Interaction
`account-switch [T3W1]`	210/700 ms (30 %)	1/5 (20 %)	260/700 ms (37 %)	31/80 (39 %)	727 ms
`wallet-discovery [T3W1]`	691/2000 ms (35 %)	21/45 (47 %)	164/400 ms (41 %)	191/410 (47 %)	6482 ms

peter-sanderson · 2026-09-08T10:51:51Z

 export * from '@testing-library/user-event';
-
-// Re-export store types from @trezor/suite
-export type { PreloadedState } from '@trezor/suite';


This seems to be completely unused. 😱

vojtatranta · 2026-09-08T10:53:23Z

+        // When we run this first time `shouldRunTor` will tell if Tor should run according to previous settings,
+        // when it runs because of renderer (e.g. Ctrl+R) it will always be false.
+        if (shouldRunTor) {
+            await new Promise(resolve => {


like, wat? It runs root.render() twice? Here and line:84?

yup, masterpiece :D

Vere-Grey

Good from test POV

peter-sanderson added the no-project This label is used to specify that PR doesn't need to be added to a project label Sep 7, 2026