Check variants with some @unboxed constructors like untagged ones by cknitt · Pull Request #9021 · rescript-lang/rescript

cknitt · 2026-10-24T12:33:35Z

A variant whose type isn't @unboxed, but which has @unboxed on some of its constructors, skipped the checks that keep untagged cases apart. check_invariant ran them only for @unboxed types. Three wrong declarations compiled, and their matches picked the wrong case:

two number cases;
an unknown payload beside another payload;
an untagged object beside a tagged block.

Meanwhile, the @as of an untagged case was compared with the tags of tagged cases, although it doesn't set one.

Now:

the payload kinds are checked as soon as one case is untagged;
tagged blocks count together as one object case;
only tagged cases' tags are compared for duplicates.

New tests:

fixtures untagged_mixed_two_numbers.res, untagged_mixed_unknown_payload.res and untagged_mixed_object_and_tagged.res (each now an error);
untagged_mixed_test.res: an untagged case's @as equal to a tagged case's tag compiles and matches correctly.

Found by a review of the untagged variant work (#8950).

🤖 Generated with Claude Code

https://claude.ai/code/session_01CMkrYaktctdKc4M41WF36t

@as

Why: The checks that keep the cases of an untagged variant apart ran only when the type was `@unboxed`. A variant with `@unboxed` on some constructors could have two number cases, an unknown payload beside another, or an untagged object beside a tagged block, and its matches then picked the wrong case (#9020). The `@as` of an untagged case was meanwhile compared with tags, which it does not set. The payload kinds are checked as soon as one case is untagged, tagged blocks count as one object case, and only tagged cases' tags are compared for duplicates. Diagnostic added (two number cases in a mixed variant): Example: type t = | @unboxed A(int) | @unboxed B(int) | C Before (diagnostic): none; `B(5)` matches `A` After (diagnostic): This untagged variant definition is invalid: At most one case can be a number type (int or float). Diagnostic added (untagged object beside a tagged block): Example: type t = | @unboxed A({x: int}) | B({y: int}) | C Before (diagnostic): none; `A({x: 1})` matches `B` After (diagnostic): This untagged variant definition is invalid: At most one case can be an object type. Diagnostic removed (an untagged case's @as): Example: type t = | @unboxed @as("x") A(int) | @as("x") B({y: int}) | C Before (diagnostic): This untagged variant definition is invalid: Duplicate literal x. After (diagnostic): none Fixes #9020 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CMkrYaktctdKc4M41WF36t Signed-off-by: Christoph Knittel <ck@cca.io>

github-actions · 2026-10-24T12:33:58Z

Changelog

🐛 Bug fix

Check variants with some @unboxed constructors like untagged variants, so that cases a match cannot tell apart are rejected, and an untagged case's @as no longer clashes with a tagged case's tag.

Semantic Diff

1 commit · Full page

Check variants with some @unboxed constructors like untagged ones

_{ef7c8c1 · cknitt + Claude · Oct 24, 2026, 12:33 UTC}

Why

The checks that keep the cases of an untagged variant apart ran only when the type was @unboxed. A variant with @unboxed on some constructors could have two number cases, an unknown payload beside another, or an untagged object beside a tagged block, and its matches then picked the wrong case (#9020). The @as of an untagged case was meanwhile compared with tags, which it does not set.

The payload kinds are checked as soon as one case is untagged, tagged blocks count as one object case, and only tagged cases' tags are compared for duplicates.

Diagnostic added (two number cases in a mixed variant)

type t = | @unboxed A(int) | @unboxed B(int) | C

diagnostic

-none; `B(5)` matches `A`
+This untagged variant definition is invalid: At most one case can be a number type (int or float).

Diagnostic added (untagged object beside a tagged block)

type t = | @unboxed A({x: int}) | B({y: int}) | C

diagnostic

-none; `A({x: 1})` matches `B`
+This untagged variant definition is invalid: At most one case can be an object type.

Diagnostic removed (an untagged case's @as)

type t = | @unboxed @as("x") A(int) | @as("x") B({y: int}) | C

diagnostic

-This untagged variant definition is invalid: Duplicate literal x.
+none

Fixes #9020

codecov · 2026-10-24T12:40:55Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.31%. Comparing base (d53b8ae) to head (ef7c8c1).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #9021   +/-   ##
=======================================
  Coverage   81.31%   81.31%           
=======================================
  Files         456      456           
  Lines       64003    64010    +7     
=======================================
+ Hits        52044    52052    +8     
+ Misses      11959    11958    -1

Files with missing lines	Coverage Δ
compiler/ml/ast_untagged_variants.ml	`89.03% <100.00%> (+0.60%)`	⬆️

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

pkg-pr-new · 2026-10-24T12:45:46Z

Open in StackBlitz

rescript

npm i https://pkg.pr.new/rescript@9021

@rescript/belt

npm i https://pkg.pr.new/@rescript/belt@9021

@rescript/darwin-arm64

npm i https://pkg.pr.new/@rescript/darwin-arm64@9021

@rescript/linux-arm64

npm i https://pkg.pr.new/@rescript/linux-arm64@9021

@rescript/linux-x64

npm i https://pkg.pr.new/@rescript/linux-x64@9021

@rescript/runtime

npm i https://pkg.pr.new/@rescript/runtime@9021

@rescript/win32-x64

npm i https://pkg.pr.new/@rescript/win32-x64@9021

commit: ef7c8c1

github-actions · 2026-10-24T12:49:21Z

Developer playground preview: https://rescript-lang.github.io/rescript/dev-playground/?version=pr-9021

cknitt · 2026-10-24T13:28:26Z

@codex review

chatgpt-codex-connector · 2026-10-24T13:28:39Z

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-24T16:22:36.933581Z	`ef7c8c1`	Manual request

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

cknitt · 2026-10-24T16:20:21Z

@codex review

chatgpt-codex-connector · 2026-10-24T16:22:35Z

Codex Review: Didn't find any major issues. Delightful!

Reviewed commit: ef7c8c1bc6

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

@as

Why: The checks that keep the cases of an untagged variant apart ran only when the type was `@unboxed`. A variant with `@unboxed` on some constructors could have two number cases, an unknown payload beside another, or an untagged object beside a tagged block, and its matches then picked the wrong case (#9020). The `@as` of an untagged case was meanwhile compared with tags, which it does not set. The payload kinds are checked as soon as one case is untagged, tagged blocks count as one object case, and only tagged cases' tags are compared for duplicates. Diagnostic added (two number cases in a mixed variant): Example: type t = | @unboxed A(int) | @unboxed B(int) | C Before (diagnostic): none; `B(5)` matches `A` After (diagnostic): This untagged variant definition is invalid: At most one case can be a number type (int or float). Diagnostic added (untagged object beside a tagged block): Example: type t = | @unboxed A({x: int}) | B({y: int}) | C Before (diagnostic): none; `A({x: 1})` matches `B` After (diagnostic): This untagged variant definition is invalid: At most one case can be an object type. Diagnostic removed (an untagged case's @as): Example: type t = | @unboxed @as("x") A(int) | @as("x") B({y: int}) | C Before (diagnostic): This untagged variant definition is invalid: Duplicate literal x. After (diagnostic): none The same fix as #9021 on master, for the stack's check_invariant; rule 6 says so. Fixes #9020 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CMkrYaktctdKc4M41WF36t Signed-off-by: Christoph Knittel <ck@cca.io>

cristianoc

I guess @unboxed on single cases should go.
Now seems redundant. It was needed when we did not have untagged variants.

#9021 checks a variant with some `@unboxed` constructors like an untagged one: tagged blocks count as one object case, and an untagged case's `@as` is not a tag. Rule 6 now says so and names its tests. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CMkrYaktctdKc4M41WF36t Signed-off-by: Christoph Knittel <ck@cca.io>

cknitt marked this pull request as ready for review October 24, 2026 13:28

cknitt requested a review from cristianoc October 24, 2026 16:20

cknitt mentioned this pull request Oct 24, 2026

Fix and simplify the untagged variant stack after review #9023

Open

cknitt enabled auto-merge (squash) October 24, 2026 16:32

cristianoc approved these changes Oct 24, 2026

View reviewed changes

cknitt merged commit 41a2860 into master Oct 24, 2026
26 checks passed

cknitt deleted the fix-mixed-variant-checks branch October 24, 2026 17:55

Navigation Menu

Repository navigation

Check variants with some @unboxed constructors like untagged ones - #9021

Check variants with some @unboxed constructors like untagged ones#9021
cknitt merged 1 commit into
masterfrom
fix-mixed-variant-checks

cknitt commented Oct 24, 2026

Uh oh!

github-actions Bot commented Oct 24, 2026

Uh oh!

codecov Bot commented Oct 24, 2026

Uh oh!

pkg-pr-new Bot commented Oct 24, 2026

Uh oh!

github-actions Bot commented Oct 24, 2026

Uh oh!

cknitt commented Oct 24, 2026

Uh oh!

chatgpt-codex-connector Bot commented Oct 24, 2026 •

edited

Loading

Uh oh!

cknitt commented Oct 24, 2026

Uh oh!

chatgpt-codex-connector Bot commented Oct 24, 2026

Uh oh!

cristianoc left a comment

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

cknitt commented Oct 24, 2026

Uh oh!

github-actions Bot commented Oct 24, 2026

Changelog

🐛 Bug fix

Semantic Diff

Uh oh!

codecov Bot commented Oct 24, 2026

Codecov Report

Uh oh!

pkg-pr-new Bot commented Oct 24, 2026

Uh oh!

github-actions Bot commented Oct 24, 2026

Uh oh!

cknitt commented Oct 24, 2026

Uh oh!

chatgpt-codex-connector Bot commented Oct 24, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codex Review Summary

Uh oh!

cknitt commented Oct 24, 2026

Uh oh!

chatgpt-codex-connector Bot commented Oct 24, 2026

Uh oh!

cristianoc left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

chatgpt-codex-connector Bot commented Oct 24, 2026 •

edited

Loading