You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit 456eb3d
Browse filesBrowse the repository at this point in the historyBrowse files
Copy file name to clipboardExpand all lines: FRICTION_LOG.md
+5-1Lines changed: 5 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -26,7 +26,11 @@ Contributors wanting special elements must work within these representations or
26
26
27
27
[`tsconfig.json`](tsconfig.json) covers app source and excludes tests and selected GramJS generators, while [`tsconfig.script.json`](tsconfig.script.json) covers tests, development scripts, plugins, and configuration files. However, [`eslint.config.js`](eslint.config.js) applies the same broad Teact rules and both TypeScript projects to JavaScript and TypeScript files across the repository. The script config also inherits DOM/Web Worker libraries and bundler module resolution from [`tsconfig.base.json`](tsconfig.base.json). The split therefore does not clearly separate the rules and environment assumptions for browser code from tooling running locally in Node.
28
28
29
-
[`check:ts`](package.json) runs `tsc` against the app config, then repository-wide ESLint; it does not explicitly type-check the script project with a separate `tsc` invocation. This makes the compiler and lint coverage harder to reason about and maintain. With lockfile dependencies installed, the compiler succeeds on Windows, but a repository-wide ESLint run exceeds 4 GB of working memory after more than six minutes and is stopped without results; linting the changed audio/video files separately succeeds. The source of this full-run memory cost is not profiled. The preference is a clear app configuration and a separate local tests/helpers configuration, retaining browser types where tests need them: [Vitest](vitest.config.ts) runs in Node with a jsdom environment, and tests import app code. Shared settings still need to stay consistent without blurring those boundaries.
29
+
[`check:ts`](package.json) runs `tsc` against the app config, then repository-wide ESLint; it does not explicitly type-check the script project with a separate `tsc` invocation. This makes the compiler and lint coverage harder to reason about and maintain. With lockfile dependencies installed, the compiler succeeds on Windows, but a repository-wide ESLint run exceeds 4 GB of working memory after more than six minutes and is stopped without results; linting the changed audio/video files separately succeeds. On macOS with Node 26.12.1, the same full ESLint run aborts the process (`Abort trap: 6`) after `tsc` passes, while linting the changed files separately succeeds. The source of this full-run memory cost is not profiled. The preference is a clear app configuration and a separate local tests/helpers configuration, retaining browser types where tests need them: [Vitest](vitest.config.ts) runs in Node with a jsdom environment, and tests import app code. Shared settings still need to stay consistent without blurring those boundaries.
30
+
31
+
## Currency symbol test depends on the Node ICU data
32
+
33
+
On Node 26.12.1, [`walletAmountEntry.test.ts`](tests/walletAmountEntry.test.ts) fails in `gives no currency symbol where the code is the only one`: `getCurrencySymbol('CHF', 'en')` returns `Fr.`, while the test expects `undefined`. The test covers only currency formatting, yet it keeps the whole wallet test run failing locally. Whether other Node versions return a different symbol is not checked; the ICU data bundled with Node is the suspected cause.
30
34
31
35
## Generated icon files change line endings on Windows
0 commit comments