Skip to content

Add opt-in cross-module inlining for Option helpers - #8689

Open
cknitt wants to merge 7 commits into
masterfrom
codex/cross-module-inline
Open

cknitt wants to merge 7 commits into
masterfrom
codex/cross-module-inline

Conversation

@cknitt

@cknitt cknitt commented Sep 28, 2026 •

Copy link
Copy Markdown
Member

Add @inline(crossModule) to opt individual functions into cross-module inlining, and use it for selected Option helpers.

Inlining copies a function’s body into its callers. If that body changes but its type does not, the caller must still be rebuilt; checking only the .cmi type signature would leave the old body in the generated JavaScript.

The compiler stores exported inline bodies in the existing .cmj file, alongside other JavaScript compilation metadata. Rewatch now compares the digest already stored in that file’s header before and after compiling a module. If it changes, Rewatch rebuilds dependents. This keeps implementation details out of .cmi and requires no new artifact. A .cmj change unrelated to inlining may also cause a dependent rebuild.

@codecov

codecov Bot commented Sep 28, 2026 •

Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 50.00000% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 81.44%. Comparing base (1a49d91) to head (08056a9).

Files with missing lines Patch % Lines
compiler/core/lam_stats_export.ml 66.66% 2 Missing ⚠️
compiler/ml/translattribute.ml 33.33% 2 Missing ⚠️
compiler/core/lam_analysis.ml 0.00% 1 Missing ⚠️
compiler/ml/printlambda.ml 0.00% 1 Missing ⚠️
Additional details and impacted files
@@            Coverage Diff             @@
##           master    #8689      +/-   ##
==========================================
+ Coverage   81.42%   81.44%   +0.02%     
==========================================
  Files         459      459              
  Lines       64547    64554       +7     
==========================================
+ Hits        52559    52578      +19     
+ Misses      11988    11976      -12     
Files with missing lines Coverage Δ
compiler/core/js_cmj_format.ml 92.98% <100.00%> (+0.12%) ⬆️
compiler/ml/lambda.ml 70.86% <ø> (ø)
compiler/core/lam_analysis.ml 78.23% <0.00%> (-0.69%) ⬇️
compiler/ml/printlambda.ml 8.81% <0.00%> (-0.04%) ⬇️
compiler/core/lam_stats_export.ml 88.23% <66.66%> (-2.88%) ⬇️
compiler/ml/translattribute.ml 75.55% <33.33%> (+1.64%) ⬆️

... and 6 files with indirect coverage changes

🚀 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

pkg-pr-new Bot commented Sep 28, 2026 •

Copy link
Copy Markdown

Open in StackBlitz

rescript

npm i https://pkg.pr.new/rescript@8689

@rescript/belt

npm i https://pkg.pr.new/@rescript/belt@8689

@rescript/darwin-arm64

npm i https://pkg.pr.new/@rescript/darwin-arm64@8689

@rescript/linux-arm64

npm i https://pkg.pr.new/@rescript/linux-arm64@8689

@rescript/linux-x64

npm i https://pkg.pr.new/@rescript/linux-x64@8689

@rescript/runtime

npm i https://pkg.pr.new/@rescript/runtime@8689

@rescript/win32-x64

npm i https://pkg.pr.new/@rescript/win32-x64@8689

commit: 08056a9

@github-actions

Copy link
Copy Markdown

@cknitt
cknitt marked this pull request as ready for review September 28, 2026 10:04

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 24646c5cb8

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".

Comment on lines +68 to +71
else if not (Lam_closure.is_closed lambda) then
Location.raise_errorf ~loc
"@inline(crossModule) requires a function without captured local \
values"

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Treat catch binders as local during inline validation

For an exported @inline(crossModule) function containing a try ... catch whose handler inspects the exception, this check incorrectly rejects the function as capturing a local value. Lam_closure.is_closed delegates to free_variables, but its Ltrywith branch ignores _exn instead of adding that binder to the local set (compiler/core/lam_closure.ml:118-120), so ordinary references to the caught exception are reported as free. Register the catch binder during closure traversal or use validation that handles all Lambda binders.

AGENTS.md reference: AGENTS.md:L51-L54

Useful? React with 👍 / 👎.

cknitt and others added 7 commits October 27, 2026 06:42
Signed-off-by: Christoph Knittel <ck@cca.io>
Signed-off-by: Christoph Knittel <ck@cca.io>
Signed-off-by: Christoph Knittel <ck@cca.io>
Signed-off-by: Christoph Knittel <ck@cca.io>
Signed-off-by: Christoph Knittel <ck@cca.io>
Signed-off-by: Christoph Knittel <ck@cca.io>
Why:
    The OCaml build system, now the default `rescript`, recompiled
    dependents only when a module's CMI changed. An `@inline(crossModule)`
    body lives in the CMJ and can change while the CMI stays the same, so
    callers kept the old inlined code, during a build and on the next build
    after the dependency was compiled separately. Rust rewatch already
    compares CMJ digests since the previous commits.

Publishing a byte-different CMJ now marks dependents dirty like a changed
CMI, and the startup freshness check compares a dependent's CMT with the
later of its dependency's CMI and CMJ.

Review: Does every implementation change now recompile dependents? Only
those whose CMJ bytes change, matching the Rust implementation.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Kbf3gKsWykY6KH4QSJ9Wrs
Signed-off-by: Christoph Knittel <ck@cca.io>
@cknitt
cknitt force-pushed the codex/cross-module-inline branch from 24646c5 to 08056a9 Compare October 27, 2026 10:35
@github-actions

Copy link
Copy Markdown

Changelog

🚀 New Feature

  • Allow selected functions to opt into cross-module inlining with @inline(crossModule) and inline common Option helpers in application code.

Semantic Diff

7 commits · Full page

Enable selective cross-module inlining for Option helpers

00d6bff · cknitt · Sep 28, 2026, 06:04 UTC

Inline additional Option helpers across modules

3da43ad · cknitt · Sep 28, 2026, 07:46 UTC

Document cross-module inlining feature

cb4faed · cknitt · Sep 28, 2026, 07:49 UTC

Refresh Option inlining snapshots after clean build

0e010a0 · cknitt · Sep 28, 2026, 07:55 UTC

Track CMJ changes for cross-module inlining

a7681f8 · cknitt · Sep 28, 2026, 09:19 UTC

Read CMJ digest header for incremental rebuilds

7ce5284 · cknitt · Sep 28, 2026, 09:53 UTC

✨ Recompile dependents of a changed CMJ in the OCaml build system

08056a9 · cknitt + Claude · Oct 27, 2026, 07:15 UTC · new

Why

The OCaml build system, now the default rescript, recompiled dependents only when a module's CMI changed. An @inline(crossModule) body lives in the CMJ and can change while the CMI stays the same, so callers kept the old inlined code, during a build and on the next build after the dependency was compiled separately. Rust rewatch already compares CMJ digests since the previous commits.

Publishing a byte-different CMJ now marks dependents dirty like a changed CMI, and the startup freshness check compares a dependent's CMT with the later of its dependency's CMI and CMJ.

Review

  • Does every implementation change now recompile dependents? Only those whose CMJ bytes change, matching the Rust implementation.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant