[aaronvg]

Fixes B-900.

Problem

baml fmt rejected any object literal containing a ...spread member:

error: while formatting: Expected token/node OBJECT_FIELD or R_BRACE, but found SPREAD_ELEMENT

The syntax parses and type-checks, so baml check and baml test passed on files baml fmt refused to format at all. Two-line repro:

class P { name: string, score: int }
function f(p: P) -> P { P { ...p, score: 1 } }

Cause

ObjectInitializer::from_cst in baml_fmt only admitted OBJECT_FIELD nodes, so a SPREAD_ELEMENT sibling fell through to the catch-all error arm.

Fix

Added an ObjectMember enum (Field | Spread) and a SpreadElement strong-AST node, and widened the object-literal member list to hold either.

Members stay in one ordered list rather than being split into "spreads" and "fields", because order is semantic — later members win at runtime, so P { score: 9, ...p } evaluates to 1 while P { ...p, score: 9 } evaluates to 9. Splitting them would silently change program behavior. ... prints tight against its operand.

The change is scoped to OBJECT_LITERAL, the only node where the parser emits SPREAD_ELEMENT (parse_object_literal_body is its sole caller). Map and array literals reject spread at parse time, so they keep using ObjectField directly.

Tests

Four regression tests in object_spread_tests, each asserting idempotency:

• single-line spread round-trips as P { ...p, score: 1 }
• order preserved across ...p, score, score, ...p, ...a, ...b, and ...a, score, ...b
• multi-line path with a comment above the spread member
• spread of a call expression (...base()), not just a bare name

Full suite: 124 passed, 0 failed — 120 pre-existing plus these 4, including the 1024-case formatter_scenario_matrix.

Verification

Formatting a messy input now produces stable output, and a second pass is a no-op:

function tight(p: Config) -> Config {
    Config { ...p, alpha: "x" }
}

function wide(base: Config) -> Config {
    Config {
        // inherit everything from base first
        ...base,
        alpha: "a much longer override value that forces the multi-line path here",
    }
}

Also confirmed against the case that surfaced this: Example 2 of plugins/baml/skills/core/SKILL.md in the baml-skill repo now passes baml fmt, baml check, and baml test. That example demonstrates spread while the same skill instructs agents to run baml fmt, so the guidance was self-contradicting until this landed.

🤖 Generated with Claude Code

https://claude.ai/code/session_017WeFQbEGpSJjzM79LHJ8du

---

Note

Low Risk
Formatter-only change scoped to object-literal AST/printing; no language semantics, auth, or data-handling paths are modified.

Overview
baml fmt now formats object literals that contain ...spread members, which previously failed with an unexpected SPREAD_ELEMENT error even though the syntax already parsed and type-checked.

ObjectInitializer members are modeled as an ordered ObjectMember list (Field | Spread) so field/spread order is preserved — later members win at runtime. Spreads print tight against their operand (...expr), including call expressions, on both single- and multi-line layouts.

Adds regression tests covering order preservation, multi-line comments, call operands, and formatter idempotency.

^{Reviewed by Cursor Bugbot for commit e9058c1. Bugbot is set up for automated code reviews on this repo. Configure here.}

Summary by CodeRabbit

• New Features

  • Added support for object spread syntax (...expression) in object initializers.
  • Preserved the original order of fields, spread elements, and comments when formatting.
  • Supported spreads across single-line and multi-line object formatting.
  • Enabled arbitrary spread expressions, including function calls.

• Bug Fixes

  • Improved validation messages for object initializer members.
  • Ensured formatting remains consistent when run repeatedly.

[linear]

B-900

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
beps	Ready	Preview	Aug 11, 2026 10:54pm
promptfiddle2	Ready	Preview	Aug 11, 2026 10:54pm

[github-actions]

⏭️ Performance benchmarks were skipped

Perf benchmarks (CodSpeed) are opt-in on pull requests — they no longer run on every push. They always run automatically after merge to canary/main.

To run them on this PR, do any of the following, then push a commit (or re-run CI):

• Add RUN_CODSPEED=1 to the PR description, or
• Include run-perf or /perf in the PR title or any commit message.

[coderabbitai]

📝 Walkthrough

Walkthrough

The formatter adds ObjectMember and SpreadElement support for object initializers. It preserves field and spread order, handles trivia and multiline output, and adds regression tests for formatting and idempotence.

Changes

Object spread formatting

Layer / File(s)	Summary
AST parsing and printing baml_language/crates/baml_fmt/src/ast/expressions.rs	Object initializers accept fields and spread elements through ObjectMember. SpreadElement parses, sizes, preserves trivia, and prints ...expr.
Formatting regression coverage baml_language/crates/baml_fmt/src/lib.rs	Tests cover spread syntax, member order, comments, multiline formatting, call expressions, and idempotence.

Estimated code review effort: 3 (Moderate) | ~25 minutes

Sequence Diagram(s)

sequenceDiagram
  participant ObjectInitializerParser
  participant ObjectMember
  participant SpreadElementPrinter
  ObjectInitializerParser->>ObjectMember: Parse field or spread element
  ObjectMember->>SpreadElementPrinter: Format spread expression
  SpreadElementPrinter-->>ObjectMember: Return formatted ...expr
  ObjectMember-->>ObjectInitializerParser: Preserve ordered members

Loading

Possibly related PRs

• BoundaryML/baml#4107: This PR also changes formatter AST and printing logic in expressions.rs for another expression syntax.
• BoundaryML/baml#4136: This PR also modifies ObjectField representation in the formatter AST.
• BoundaryML/baml#4308: This PR also changes formatter behavior and regression tests in expressions.rs and lib.rs.

Poem

A rabbit found three dots in line,
And made each spread element shine.
Fields stayed put, comments stayed near,
Repeated formatting brought no fear.
“Hop!” said the formatter, “all order is clear.”

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: support for struct-update spread in object literals.

✨ Finishing Touches 💡 1

🛠️ Fix failing CI checks 💡

• Create stacked PR
• Commit on current branch

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch aaron/b-900

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

Binary size checks passed

✅ 7 passed

	Artifact	Platform	File	Gzip	Gated on	Baseline	Delta	Status
✅	baml-cli	Linux	🔒 26.4 MB	11.2 MB	file	27.4 MB	-947.8 KB (-3.5%)	OK
✅	packed-program	Linux	🔒 17.0 MB	6.9 MB	file	18.6 MB	-1.6 MB (-8.6%)	OK
✅	baml-cli	macOS	🔒 20.6 MB	9.8 MB	file	21.3 MB	-698.5 KB (-3.3%)	OK
✅	packed-program	macOS	🔒 13.4 MB	6.1 MB	file	14.5 MB	-1.1 MB (-7.7%)	OK
✅	baml-cli	Windows	🔒 22.1 MB	10.0 MB	file	23.0 MB	-842.3 KB (-3.7%)	OK
✅	packed-program	Windows	🔒 14.2 MB	6.1 MB	file	15.5 MB	-1.4 MB (-8.8%)	OK
✅	bridge_wasm	WASM	15.8 MB	🔒 4.3 MB	gzip	4.6 MB	-347.7 KB (-7.5%)	OK

🔒 = the size this artifact is GATED on (ceiling + delta). Binaries gate on file size (installed binary); WASM gates on gzip (download size). The other size is shown for information only.

---

Generated by cargo size-gate · workflow run