Add MANIFOLD_NO_IOSTREAM build-time option by zmerlynn · Pull Request #1690 · elalish/manifold

zmerlynn · 2026-05-03T14:33:29Z

Summary

Adds MANIFOLD_NO_IOSTREAM (default OFF) that strips iostream- and
filesystem-using bits from the public API and tests. Useful for
freestanding/embedded consumers — the immediate motivator is
wasm32-unknown-unknown via wasm-cxx-shim, which today carries
near-equivalent patches downstream that this PR consolidates upstream.
Idea floated and well-received in #1046.

When ON:

Defines MANIFOLD_NO_IOSTREAM and MANIFOLD_NO_FILESYSTEM as PUBLIC
compile defs on manifold, so consumers see the stripped API.
Triggers CLIPPER2_NO_IOSTREAM=ON for the bundled Clipper2 via a
carry-patch tracking Add CLIPPER2_NO_IOSTREAM build-time option AngusJohnson/Clipper2#1094 — drops once that lands.
Gates iostream-using TEST() blocks at compile time.
FATAL_ERRORs if combined with MANIFOLD_DEBUG / MANIFOLD_TIMING.

Always-on manifold_no_iostream_check / manifoldc_no_iostream_check
compile the public headers with the macro defined; build fails if
iostream use sneaks back in.

Test plan

Default build: unchanged.
MANIFOLD_NO_IOSTREAM=ON + MANIFOLD_TEST=ON: 270 tests register
(vs 383 default; tests requiring fixture-file reads are gated out at
compile time), all pass.
nm libmanifold.{a,dylib}: zero iostream/filesystem symbols.
MANIFOLD_DEBUG=ON + MANIFOLD_NO_IOSTREAM=ON fatals as expected.
New CI cell build_no_iostream exercises the option end-to-end
on Ubuntu/clang.
End-to-end with the consumer: wasm-cxx-shim builds against this
branch with its own iostream patches dropped — full integration
suite passes.

Notes

Windows: carry-patch uses GIT_CONFIG core.autocrlf=false, ships
with *.patch text eol=lf in .gitattributes, and applies with
--whitespace=nowarn. Without these, default Windows checkouts
EOL-convert and the patch fails. CI is Ubuntu-only; happy to add a
Windows lane on request.

codecov · 2026-05-03T14:53:39Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 95.22%. Comparing base (5f95a3a) to head (5011e65).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1690   +/-   ##
=======================================
  Coverage   95.22%   95.22%           
=======================================
  Files          36       36           
  Lines        7974     7974           
=======================================
  Hits         7593     7593           
  Misses        381      381

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 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.

Pre-release for v0.4.0. Pins manifold to a current upstream master commit (5f95a3ac) and replaces the three shim-side iostream patches with a single vendored diff of elalish/manifold#1690 — the upstream PR that adds MANIFOLD_NO_IOSTREAM natively. Once #1690 lands and the manifold pin moves past it, the carry-patch drops entirely and a v0.4.0 (non-alpha) release follows. Helper refactor: * Drops the Clipper2 pre-declaration. Manifold's manifoldDeps.cmake (post-#1690) owns the FetchContent_Declare(Clipper2 ...) and derives CLIPPER2_NO_IOSTREAM=ON from MANIFOLD_NO_IOSTREAM=ON automatically. * Sets MANIFOLD_NO_IOSTREAM=ON as a CMake cache var and lets manifold's option chain propagate the macros as PUBLIC compile defs on the manifold target. * API: removes CLIPPER2_GIT_TAG and EXTRA_CLIPPER2_PATCHES (manifold owns Clipper2). MANIFOLD_GIT_TAG / EXTRA_MANIFOLD_PATCHES / SKIP_BUILTIN_PATCHES continue to work. Test expansion: 71 -> 121 tests run on the shim. Adds boolean_complex, manifoldc, smooth (alongside existing boolean, cross_section, sdf). Skipped: hull / properties / samples (need samples helper library); manifold_test (uses std::set/<thread> without direct includes our libcxx subset doesn't ship); polygon (RegisterPolygonTests is no-op under MANIFOLD_NO_IOSTREAM, file would compile but contribute zero tests). Size budget: manifold_tests_size_budget bumped 1.10 -> 1.40 MB to accommodate the expanded test set (current ~1.07 MB; ~30% headroom). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Pre-release for v0.4.0. Pins manifold to a current upstream master commit (5f95a3ac) and replaces the three shim-side iostream patches with a single vendored diff of elalish/manifold#1690 — the upstream PR that adds MANIFOLD_NO_IOSTREAM natively. Once #1690 lands and the manifold pin moves past it, the carry-patch drops entirely and a v0.4.0 (non-alpha) release follows. Helper refactor: * Drops the Clipper2 pre-declaration. Manifold's manifoldDeps.cmake (post-#1690) owns the FetchContent_Declare(Clipper2 ...) and derives CLIPPER2_NO_IOSTREAM=ON from MANIFOLD_NO_IOSTREAM=ON automatically. * Sets MANIFOLD_NO_IOSTREAM=ON as a CMake cache var and lets manifold's option chain propagate the macros as PUBLIC compile defs on the manifold target. * API: removes CLIPPER2_GIT_TAG and EXTRA_CLIPPER2_PATCHES (manifold owns Clipper2). MANIFOLD_GIT_TAG / EXTRA_MANIFOLD_PATCHES / SKIP_BUILTIN_PATCHES continue to work. Test expansion: 71 -> 121 tests run on the shim. Adds boolean_complex, manifoldc, smooth (alongside existing boolean, cross_section, sdf). Skipped: hull / properties / samples (need samples helper library); manifold_test (uses std::set/<thread> without direct includes our libcxx subset doesn't ship); polygon (RegisterPolygonTests is no-op under MANIFOLD_NO_IOSTREAM, file would compile but contribute zero tests). Size budget: manifold_tests_size_budget bumped 1.10 -> 1.40 MB to accommodate the expanded test set (current ~1.07 MB; ~30% headroom). Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

pca006132 · 2026-05-04T12:04:23Z

Do we need this? Or can we do it by compiling examples?

Fair point — quick defense was "catches creep on every build," but the new build_no_iostream CI cell already builds manifold + manifoldc with MANIFOLD_NO_IOSTREAM=ON, so iostream creep in public headers fails there. Dropped the targets + .cpp files in 40586ef.

Strips iostream- and filesystem-using bits from the public API and tests, for freestanding/embedded consumers (e.g., wasm32-unknown-unknown via wasm-cxx-shim). Default OFF. When ON, defines MANIFOLD_NO_IOSTREAM and MANIFOLD_NO_FILESYSTEM as PUBLIC compile defs on manifold, gates iostream-using TESTs at compile time, and triggers CLIPPER2_NO_IOSTREAM=ON for the bundled Clipper2 via a carry-patch tracking AngusJohnson/Clipper2#1094 (drops once that lands and the SHA pin moves past it). Incompatible with MANIFOLD_DEBUG / MANIFOLD_TIMING (FATAL_ERROR). Discussion: elalish#1046 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

pca006132

Thanks. Btw, it will be better if you just commit the changes without force-push. We will squash the updates anyway.

zmerlynn · 2026-05-04T19:07:58Z

Thanks. Btw, it will be better if you just commit the changes without force-push. We will squash the updates anyway.

Noted - still getting used to local preferences. If I need to rebase, would you prefer I keep the commit stack so it's easy to re-review?

pca006132 · 2026-05-04T19:14:47Z

yes, it is easier if I can see the diff, especially if the change is large.

Manifold's `MANIFOLD_NO_IOSTREAM` build option is now upstream (merged in elalish/manifold#1690 on 2026-05-04). The single vendored carry-patch shipped at v0.4.0-alpha.1 is no longer needed — manifold owns the option, and its `manifoldDeps.cmake` carries a Clipper2 tracking patch for AngusJohnson/Clipper2#1094 internally. The shim drops the carry-patch entirely: * `cmake/manifold-patches/0001-manifold-no-iostream.patch` deleted (and the now-empty directory removed). * `wasm_cxx_shim_add_manifold()`'s SKIP_BUILTIN_PATCHES argument removed (no builtin patches to skip). MANIFOLD_GIT_TAG and EXTRA_MANIFOLD_PATCHES continue to work. * `install(DIRECTORY .../manifold-patches ...)` rule dropped. * Helper smoke test updated for the new file layout. Manifold pin bumped from `5f95a3ac` (master pre-#1690) to `3ce9622b` (master at release time, includes #1690 + one downstream commit). Build + test verified end-to-end on the new pin: 18/18 ctest, 121/121 inner manifold tests pass. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Manifold's `MANIFOLD_NO_IOSTREAM` build option is now upstream (merged in elalish/manifold#1690 on 2026-05-04). The single vendored carry-patch shipped at v0.4.0-alpha.1 is no longer needed — manifold owns the option, and its `manifoldDeps.cmake` carries a Clipper2 tracking patch for AngusJohnson/Clipper2#1094 internally. The shim drops the carry-patch entirely: * `cmake/manifold-patches/0001-manifold-no-iostream.patch` deleted (and the now-empty directory removed). * `wasm_cxx_shim_add_manifold()`'s SKIP_BUILTIN_PATCHES argument removed (no builtin patches to skip). MANIFOLD_GIT_TAG and EXTRA_MANIFOLD_PATCHES continue to work. * `install(DIRECTORY .../manifold-patches ...)` rule dropped. * Helper smoke test updated for the new file layout. Manifold pin bumped from `5f95a3ac` (master pre-#1690) to `3ce9622b` (master at release time, includes #1690 + one downstream commit). Build + test verified end-to-end on the new pin: 18/18 ctest, 121/121 inner manifold tests pass. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

zmerlynn force-pushed the add-no-iostream-option branch from 2c6a4c3 to 067b40b Compare May 3, 2026 14:57

zmerlynn mentioned this pull request May 3, 2026

Bump manifold pin + fold to single #1690 carry-patch (v0.4.0-alpha.1) zmerlynn/wasm-cxx-shim#13

Merged

7 tasks

zmerlynn mentioned this pull request May 3, 2026

Bump wasm-cxx-shim to v0.4.0-alpha.1, drop wasm-uu ray_cast cfg-gates zmerlynn/manifold-csg#40

Merged

8 tasks

pca006132 approved these changes May 4, 2026

View reviewed changes

zmerlynn force-pushed the add-no-iostream-option branch from 067b40b to 40586ef Compare May 4, 2026 13:18

zmerlynn force-pushed the add-no-iostream-option branch from 40586ef to 5011e65 Compare May 4, 2026 13:44

pca006132 approved these changes May 4, 2026

View reviewed changes

pca006132 merged commit e73c078 into elalish:master May 4, 2026
40 checks passed

zmerlynn mentioned this pull request May 4, 2026

Drop carry-patch, bump manifold pin past #1690 merge (v0.4.0) zmerlynn/wasm-cxx-shim#14

Merged

5 tasks

zmerlynn mentioned this pull request May 4, 2026

Bump wasm-cxx-shim to v0.4.0 + manifold pin to 3ce9622b zmerlynn/manifold-csg#43

Merged

5 tasks

elalish mentioned this pull request May 23, 2026

v3.5 #1730

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add MANIFOLD_NO_IOSTREAM build-time option#1690

Add MANIFOLD_NO_IOSTREAM build-time option#1690
pca006132 merged 1 commit into
elalish:masterfrom
zmerlynn:add-no-iostream-option

zmerlynn commented May 3, 2026 •

edited

Loading

Uh oh!

codecov Bot commented May 3, 2026 •

edited

Loading

Uh oh!

pca006132 May 4, 2026

Uh oh!

zmerlynn May 4, 2026

Uh oh!

pca006132 left a comment

Uh oh!

Uh oh!

zmerlynn commented May 4, 2026

Uh oh!

pca006132 commented May 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

zmerlynn commented May 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test plan

Notes

Uh oh!

codecov Bot commented May 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

pca006132 May 4, 2026

Choose a reason for hiding this comment

Uh oh!

zmerlynn May 4, 2026

Choose a reason for hiding this comment

Uh oh!

pca006132 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

zmerlynn commented May 4, 2026

Uh oh!

pca006132 commented May 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

zmerlynn commented May 3, 2026 •

edited

Loading

codecov Bot commented May 3, 2026 •

edited

Loading