Fix formatter blank line and over-indent for union template argument by oha-4 · Pull Request #11010 · microsoft/typespec

oha-4 · 2026-06-17T12:21:28Z

Problem

When a union expression is used directly as one of multiple template arguments and the argument list is long enough to wrap, tsp format inserts a blank line before the union and indents its leading-| variants one level deeper than the sibling arguments:

model Picked
  is PickProperties<
    Sample,

      | "alpha"
      | "bravo"
      ...
  >;

Fix

printTemplateParameters already emits a softline + an indent level before each argument in a multi-argument list. printUnion was also emitting its own leading line + indent (+ align(2)), so the two stacked, producing the blank line and the extra indentation level.

printUnion now detects when the union is a direct argument of a multi-argument template reference (isInMultiTemplateArgumentList) and, in that case, lets the surrounding argument list control the line break and indentation. Result:

model Picked
  is PickProperties<
    Sample,
    | "alpha"
    | "bravo"
    ...
  >;

The standalone/value-position cases (e.g. alias X = "a" | "b" | ...) and the single-template-argument (shouldHug) case are unchanged, since there the surrounding context does not supply the line break/indent.

Tests

Added a regression test for the multi-argument case (Formatter adds a blank line and over-indents a union used as a template argument #11009) and a test pinning the single-argument behavior in formatter.test.ts.
Reformatted two committed http-client-java test fixtures that contained the buggy output (whitespace only; no semantic change).
pnpm prettier --check "**/*.tsp", formatter test suite, tsc, and eslint all pass.

When a `union` expression was used directly as one of multiple template arguments and the argument list wrapped, the formatter inserted a blank line before the union and indented its `|`-prefixed variants one level deeper than the sibling arguments. The argument list already supplies a line break and an indentation level before each argument, so `printUnion` no longer adds its own leading line and indent when the union is a direct argument of a multi-argument template reference. Fixes microsoft#11009 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

oha-4 · 2026-06-17T12:24:02Z

@microsoft-github-policy-service agree

pkg-pr-new · 2026-06-17T13:18:05Z

Open in StackBlitz

npm i https://pkg.pr.new/@typespec/compiler@11010

commit: 6e0d095

github-actions · 2026-06-17T13:19:46Z

All changed packages have been documented.

✅ @typespec/compiler
✅ @typespec/http-client-java

Show changes

`@typespec/compiler` - fix ✏️

Fix formatter inserting a blank line and over-indenting a union expression used directly as one of multiple template arguments (e.g. PickProperties<Source, "a" | "b">)

`@typespec/http-client-java` - internal ✏️

Reformat union template arguments in test files to match updated formatter style.

timotheeguerin

Thanks for the contribution, I think there is also a regression this introduce in the formatting, i'll also need to see the effect this has on all of azure specs though it does seem like a good thing to fix

Address review feedback: the previous fix dropped align(2) on union variants when the union is a multi-argument template argument, which de-indented a nested template inside a variant. Keep align(2) unconditionally (matching prettier's union printer) so a breaking variant stays aligned under its "| " prefix, and add a regression test. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

oha-4 · 2026-06-17T14:26:06Z

Thanks for the careful review — you're right, fixed in 4daeb1b6b. The previous change dropped align(2) from the variants, which de-indented the nested template.

And to your question: this does follow prettier's union printer — indent-or-not depends on the parent (shouldIndent, union-type.js#L23-L38), but the per-variant align(2) is kept regardless (#L49-L56). I'd wrongly tied them together; now align(2) is unconditional, so a breaking variant stays aligned under its | , matching your screenshot. Added a regression test for it too.

Reformatting arm.tsp/response.tsp (union template arguments) changed the tcgc crossLanguageVersion hash of the generated metadata. Update the two *_metadata.json files to the regenerated hashes so the Java RegenCheck passes, and add an internal changeset for @typespec/http-client-java so the changelog check passes. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

timotheeguerin · 2026-06-18T14:16:15Z

+  }
+  let count = 0;
+  let node: Node | null;
+  while ((node = path.getParentNode(count++))) {


to dig a little more on this loop, as you pointed out the indent in typescript printer depends on the parent, which we have above, but I don't fully see how the type reference is involved here. Do we really need that loop, is thre maybe then some test missing to showcase those scenarios?

Thanks, that's a great catch — you're right, the loop was unnecessary. A TemplateArgument only ever exists inside TypeReference.arguments (the sole TemplateArgumentNode[] in the AST), so the owning TypeReference is always the next ancestor. I've replaced the loop with a direct getParentNode(1) check in 02007f6.

We still need the TypeReference itself (not just the parent) to read the argument count: a single argument hugs (no list indent, so the union keeps its own), while multiple arguments indent the list — that's the case the fix targets. Both are covered by the existing single-/multi-arg tests, and all 205 formatter tests still pass.

A TemplateArgument node only ever exists as an element of TypeReference.arguments (the sole TemplateArgumentNode[] in the AST), so when the parent is a TemplateArgument the owning TypeReference is always the next node ancestor. Replace the variable-depth loop with a direct getParentNode(1) check; behavior is unchanged and all formatter tests pass. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

weidongxu-microsoft

approve on java lib to unblock.

but unsure why metadata file changes, they should not be affected by typespec source formatting. though I think Alan merged another unblocker.

timotheeguerin · 2026-06-22T13:36:40Z

approve on java lib to unblock.

but unsure why metadata file changes, they should not be affected by typespec source formatting. though I think Alan merged another unblocker.

yeah we investigated with Allen, I believe the issue is that the crossVersionId hashes the file content which the formatting affect. Should be good now

…osoft#11011) Fixes the `is`-breaking part of microsoft#11009 (the blank-line/over-indent part is addressed separately in microsoft#11010). ## Problem When a model/scalar heritage clause (`is X` / `extends X`) references a template with multiple arguments, the formatter wrapped the keyword and the template reference in a single `group(indent(...))`. Prettier measures that group against the full flat width of the template argument list, so the group broke and pushed the keyword onto its own indented line — even when the declaration line itself was short: ```tsp model Picked is PickProperties< Sample, ... >; ``` ## Fix `printHeritageClause` now keeps the keyword on the declaration line when the base is a template reference with multiple (breakable) arguments, and lets the template argument list control the line break: ```tsp model Picked is PickProperties< Sample, ... >; ``` Single-argument (hugged) and non-template bases keep the previous behavior of breaking the keyword onto its own line when the declaration is too long. ## Tests - Added regression tests for the multi-argument case on `model is`, `model extends`, and `scalar extends` in `formatter.test.ts`. - Reformatted one committed `http-client-java` test spec (`arm-customization.tsp`) that contained the old output (whitespace only, no semantic change). - Formatter test suite, eslint, tsc, and `prettier --check "**/*.tsp"` all pass locally. > Note: the `crossLanguageVersion` hash in the corresponding `*_metadata.json` is whitespace-sensitive; it will be updated from the RegenCheck output once CI runs. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

oha-4 requested review from XiaofeiCao, anuchandy, bterlson, catalinaperalta, haolingdong-msft, iscai-msft, markcowl, srnagar, timotheeguerin, weidongxu-microsoft and witemple-msft as code owners June 17, 2026 12:21

microsoft-github-policy-service Bot added compiler:core Issues for @typespec/compiler emitter:client:java Issue for the Java client emitter: @typespec/http-client-java labels Jun 17, 2026

timotheeguerin reviewed Jun 17, 2026

View reviewed changes

Comment thread packages/http-client-java/generator/http-client-generator-test/tsp/arm.tsp Outdated

oha-4 requested a review from timotheeguerin June 17, 2026 15:37

oha-4 mentioned this pull request Jun 17, 2026

Keep is/extends inline for multi-argument template heritage #11011

Merged

timotheeguerin reviewed Jun 18, 2026

View reviewed changes

weidongxu-microsoft reviewed Jun 22, 2026

View reviewed changes

weidongxu-microsoft approved these changes Jun 22, 2026

View reviewed changes

Merge branch 'main' into fix/formatter-union-template-arg

6e0d095

oha-4 requested a review from alzimmermsft as a code owner June 22, 2026 05:24

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix formatter blank line and over-indent for union template argument#11010

Fix formatter blank line and over-indent for union template argument#11010
oha-4 wants to merge 5 commits into
microsoft:mainfrom
oha-4:fix/formatter-union-template-arg

oha-4 commented Jun 17, 2026

Uh oh!

oha-4 commented Jun 17, 2026

Uh oh!

pkg-pr-new Bot commented Jun 17, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Jun 17, 2026 •

edited

Loading

`@typespec/compiler` - fix ✏️

`@typespec/http-client-java` - internal ✏️

Uh oh!

timotheeguerin left a comment •

edited

Loading

Uh oh!

Uh oh!

oha-4 commented Jun 17, 2026

Uh oh!

timotheeguerin Jun 18, 2026

Uh oh!

oha-4 Jun 19, 2026

Uh oh!

weidongxu-microsoft left a comment •

edited

Loading

Uh oh!

timotheeguerin commented Jun 22, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

oha-4 commented Jun 17, 2026

Problem

Fix

Tests

Uh oh!

oha-4 commented Jun 17, 2026

Uh oh!

pkg-pr-new Bot commented Jun 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented Jun 17, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

@typespec/compiler - fix ✏️

@typespec/http-client-java - internal ✏️

Uh oh!

timotheeguerin left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

oha-4 commented Jun 17, 2026

Uh oh!

timotheeguerin Jun 18, 2026

Choose a reason for hiding this comment

Uh oh!

oha-4 Jun 19, 2026

Choose a reason for hiding this comment

Uh oh!

weidongxu-microsoft left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

timotheeguerin commented Jun 22, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

pkg-pr-new Bot commented Jun 17, 2026 •

edited

Loading

github-actions Bot commented Jun 17, 2026 •

edited

Loading

`@typespec/compiler` - fix ✏️

`@typespec/http-client-java` - internal ✏️

timotheeguerin left a comment •

edited

Loading

weidongxu-microsoft left a comment •

edited

Loading