[johnnyreilly]

Speculative: converts ts-loader to the async function loader style, as far as webpack allows.

The catch: this.async() can't go away entirely

When a loader returns a promise, webpack's loader-runner does:

result.then(function(r) { callback(null, r); }, callback);

Only content is forwarded, so a loader that just returns its output loses its source map. ts-loader always produces one, so a "pure" async loader isn't an option without dropping source maps. You also can't mix the two: if an await happens before calling this.callback, loader-runner has already taken the returned promise as the result.

So loader is now an async function that calls this.async() once at the top and hands the result to it once at the end:

async function loader(this, contents, inputSourceMap) {
  this.cacheable && this.cacheable();
  const callback = this.async();
  let result;
  try {
    result = await compile(this, contents, inputSourceMap);
  } catch (error) {
    callback(error instanceof Error ? error : new Error(String(error)));
    return;
  }
  callback(null, result.output, result.sourceMap);
}

What changed

• successLoader + makeSourceMapAndFinish (which passed callback around) are replaced by one async compile() that returns { output, sourceMap } or throws.
• mapToInputSourceMap uses async/await with try/finally for the destroy() calls, instead of a hand-built new Promise.
• Output and behaviour are unchanged. The only snapshot changes are stack-frame names in three error tests (nodeModulesMeaningfulErrorWhenImportingTs, tsconfigInvalid, validateLoaderOptionNames): successLoader / makeSourceMapAndFinish → compile.
  • Note: the tsconfig-parse error used to be created in loader's own frame on purpose, to match classic ts-loader's single at Object.loader frame. It now shows at compile then at Object.loader. Easy to restore if we care.

Testing

npm run lint, npm run comparison-tests, npm run execution-tests all pass locally (macOS).

🤖 Generated with Claude Code

[github-actions]

Benchmark (Ubuntu)

Scenario	transpileOnly	PR branch median (ms)	base branch median (ms)	Δ vs base branch
Cold build	false	1267.9	1256.5	+0.9%
Incremental rebuild (leaf touch)	false	58.0	58.3	-0.4%
Incremental rebuild (hub touch)	false	476.9	485.6	-1.8%
Cold build	true	385.8	385.1	+0.2%
Incremental rebuild (leaf touch)	true	34.4	34.5	-0.2%
Incremental rebuild (hub touch)	true	33.4	34.3	-2.5%
Cold build (all orphan files)	false	1289.7	1288.7	+0.1%
Incremental rebuild (hub touch, all orphan files)	false	498.5	507.4	-1.8%

PR branch = /home/runner/work/ts-loader/ts-loader, base branch = /home/runner/work/ts-loader/ts-loader-main. 2 warmup + 10 measured iterations per scenario, median reported. Report-only - no threshold fails this check.

[github-actions]

Benchmark (macOS)

Scenario	transpileOnly	PR branch median (ms)	base branch median (ms)	Δ vs base branch
Cold build	false	786.2	772.4	+1.8%
Incremental rebuild (leaf touch)	false	50.3	50.3	-0.1%
Incremental rebuild (hub touch)	false	294.9	313.1	-5.8%
Cold build	true	252.3	251.7	+0.3%
Incremental rebuild (leaf touch)	true	47.9	48.5	-1.2%
Incremental rebuild (hub touch)	true	50.7	49.5	+2.4%
Cold build (all orphan files)	false	756.8	764.2	-1.0%
Incremental rebuild (hub touch, all orphan files)	false	300.4	296.3	+1.4%

PR branch = /Users/runner/work/ts-loader/ts-loader, base branch = /Users/runner/work/ts-loader/ts-loader-main. 2 warmup + 10 measured iterations per scenario, median reported. Report-only - no threshold fails this check.

[github-actions]

Benchmark (Windows)

Scenario	transpileOnly	PR branch median (ms)	base branch median (ms)	Δ vs base branch
Cold build	false	1640.9	1633.3	+0.5%
Incremental rebuild (leaf touch)	false	55.8	56.8	-1.8%
Incremental rebuild (hub touch)	false	677.7	691.2	-1.9%
Cold build	true	445.9	438.7	+1.6%
Incremental rebuild (leaf touch)	true	31.9	31.5	+1.4%
Incremental rebuild (hub touch)	true	32.6	32.1	+1.5%
Cold build (all orphan files)	false	1787.0	1798.6	-0.6%
Incremental rebuild (hub touch, all orphan files)	false	745.6	756.6	-1.4%

PR branch = C:\source\ts-loader, base branch = C:\source\ts-loader-main. 2 warmup + 10 measured iterations per scenario, median reported. Report-only - no threshold fails this check.