Compile character range patterns in linear time

[cknitt]

Fixes #8866.

A character range pattern is expanded into one or-pattern alternative per character. Three steps were quadratic in the number of alternatives:

• Typing (Typecore, Ppat_or): each level of the nested or-pattern ran alpha_pat over the entire right subtree, copying it even when there are no variables to rename. This caused the memory blowup. When alpha_env is empty the copy is now skipped. enter_orpat_variables has already checked that both sides bind the same (here: no) variables, so it would be a no-op.
• Exhaustiveness and unused-case checks (Parmatch.filter_all): every row did a linear search for its group. For int, char and string constants and for polymorphic variants, simple_match is equality, so their groups are now looked up by hashing. Group order and row order stay the same.
• Default matrices (Parmatch.get_mins le_pats, used by Matching.as_matrix, check_unused, check_partial): compared all pairs of rows. The new get_mins_pats only compares a row starting with an int/char/string constant with rows starting with the same constant or with a different kind of pattern. No other row can cover it, so the result is identical.
• Switch compilation (Matching.divide for constants): the cell for each key was found with List.find; it now uses a hash table for int/char/string keys. Floats and bigints keep the linear search, since they compare semantically (1.0 = 1.).

The hash tables are created lazily, because most calls are on small matrices.

	before	after
'\u{D7FE}' .. '\u{E001}' (~2k chars)	0.64 s, 327 MB	0.01 s, 25 MB
'\u{100}' .. '\u{1100}' (~4k chars)	2.64 s, 1.2 GB	0.03 s, 28 MB
'\u{4E00}' .. '\u{9FFF}' (CJK, ~21k chars)	killed after 39 s at 15 GB	0.16 s, 72 MB
'\u{0}' .. '\u{FFFF}' (65k chars)	effectively hangs	0.52 s, 138 MB
switch with 4,000 int cases	1.09 s	0.23 s
switch with 4,000 string cases	0.88 s	0.16 s

Generated JavaScript and warnings are byte-identical to master across tests/tests/src, the super_errors fixtures, the runtime, and a stress file. That file has overlapping ranges, duplicate int/string/float/bigint/polyvariant cases, tuples with mixed heads, guards, constructors and arrays, and exercises warnings 8, 11 and 12. On ordinary code (tests/tests + Belt), CPU time is unchanged within noise and allocation is up about 0.5%.

Big int/string switches are not fully linear yet. The remaining cost is check_unused checking each row against the previous ones.

New test: tests/tests/src/char_range_test.res covers the CJK block range and several adjacent ranges, checking boundaries at runtime. Before this change the compiler ran out of memory on it (killed after 39 s at 15 GB).

🤖 Generated with Claude Code

https://claude.ai/code/session_01KDBKAdTcAxA3HsjtVESRCr

[cknitt]

@codex review

[chatgpt-codex-connector]

Codex Review Summary

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

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-15T17:10:15.505228Z	d61ce0b	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.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. More of your lovely PRs please.

Reviewed commit: d61ce0bade

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".

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.02%. Comparing base (bc2e0e1) to head (4bb9898).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8868      +/-   ##
==========================================
+ Coverage   80.99%   81.02%   +0.02%     
==========================================
  Files         453      453              
  Lines       62989    63030      +41     
==========================================
+ Hits        51020    51070      +50     
+ Misses      11969    11960       -9     

Files with missing lines	Coverage Δ
compiler/ml/matching.ml	77.38% <100.00%> (+0.17%)	⬆️
compiler/ml/parmatch.ml	77.31% <100.00%> (+0.65%)	⬆️
compiler/ml/typecore.ml	86.72% <100.00%> (ø)

... and 4 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]

Open in StackBlitz

rescript

npm i https://pkg.pr.new/rescript@8868

@rescript/belt

npm i https://pkg.pr.new/@rescript/belt@8868

@rescript/darwin-arm64

npm i https://pkg.pr.new/@rescript/darwin-arm64@8868

@rescript/darwin-x64

npm i https://pkg.pr.new/@rescript/darwin-x64@8868

@rescript/linux-arm64

npm i https://pkg.pr.new/@rescript/linux-arm64@8868

@rescript/linux-x64

npm i https://pkg.pr.new/@rescript/linux-x64@8868

@rescript/runtime

npm i https://pkg.pr.new/@rescript/runtime@8868

@rescript/win32-x64

npm i https://pkg.pr.new/@rescript/win32-x64@8868

commit: 4bb9898

[github-actions]

Developer playground preview: https://rescript-lang.github.io/rescript/dev-playground/?version=pr-8868

[cristianoc]

TIL this was not done in linear time earlier