[V4] Explicit scheduling

[ConorWilliams]

Summary by CodeRabbit

• New Features

  • Added context-switching awaitable support for tasks across multiple scheduler pools.
  • Enhanced benchmark suite with pool switching and random pool selection measurements.

• Bug Fixes

  • Corrected spelling errors in code comments.

• Tests

  • Added comprehensive explicit scheduling test coverage.
  • Expanded awaitable concept validation tests.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 7e56f764-19da-4306-8640-b3cce4e04aae

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch v4-explicit-sched

---

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 and usage tips.}

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@benchmark/src/libfork/switch_random.cpp`:
- Around line 89-95: The split logic for threads_total uses threads_a and
threads_b but forcibly sets threads_b = 1 when zero, which turns a 1-thread run
into two workers; change the behavior so threads_b is not silently increased:
compute threads_a = (threads_total + 1) / 2 and threads_b = threads_total -
threads_a and if threads_b == 0, leave it zero (or clamp to max(0,...)) and
adjust downstream code that constructs pool_b to skip creating pool_b when
threads_b == 0; update any code that assumes pool_b exists to handle the
zero-worker case (references: variables threads_total, threads_a, threads_b and
the pool creation logic in switch_random.cpp).

In `@test/src/explicit_scheduling.cpp`:
- Around line 757-759: The split currently over-allocates by one because pool_b
is constructed with thr - half + 1; change the construction of TestType pool_b
to use thr - half so the two pools sum to thr (i.e., replace TestType pool_b{thr
- half + 1} with TestType pool_b{thr - half}), keeping the computation of half
(std::size_t const half = thr / 2 + (thr % 2)) and the existing TestType
pool_a{half} unchanged.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: de3c1a9b-eed7-4aac-adc4-d4418adab1c1

📥 Commits

Reviewing files that changed from the base of the PR and between 586bd4c and 0e80b01.

📒 Files selected for processing (16)

• CMakeLists.txt
• benchmark/src/libfork/CMakeLists.txt
• benchmark/src/libfork/switch_io_pool.cpp
• benchmark/src/libfork/switch_random.cpp
• src/batteries/deque.cxx
• src/batteries/slab_stack.cxx
• src/core/awaitables.cxx
• src/core/concepts/awaitable.cxx
• src/core/core.cxx
• src/core/execute.cxx
• src/core/final_suspend.cxx
• src/core/handles.cxx
• src/core/promise.cxx
• test/src/adaptors.cpp
• test/src/concepts.cpp
• test/src/explicit_scheduling.cpp