ai slop

[robobun]

This PR has been marked as AI slop and the description has been updated to avoid confusion or misleading reviewers.

Many AI PRs are fine, but sometimes they submit a PR too early, fail to test if the problem is real, fail to reproduce the problem, or fail to test that the problem is fixed. If you think this PR is not AI slop, please leave a comment.

[coderabbitai]

Walkthrough

Changes

uWS termination diagnostics

Layer / File(s)	Summary
Runtime termination diagnostics packages/bun-uws/src/*.h	uWS validation and contract-violation paths write descriptive messages to stderr before calling std::terminate().
Source-lint and CI enforcement test/internal/source-lints/uws-terminate-diagnostic.test.ts, .github/workflows/source-lints.yml	A source-lint test checks termination sites for preceding stderr writes, and workflow path filters include bun-uws and bun-usockets changes.

Possibly related PRs

• oven-sh/bun#34682: Adds the Source lints workflow and related test coverage.
• oven-sh/bun#35256: Modifies the same bun-uws termination paths.

Suggested reviewers: cirospaciari

🚥 Pre-merge checks | ✅ 4

✅ Passed checks (4 passed)

Check name	Status	Explanation
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title is concise and accurately summarizes the main change: restoring stderr diagnostics before std::terminate() in bun-uws.
Description check	✅ Passed	The description covers what changed and how it was verified, though it uses different section headings than the template.

---

_{Comment @coderabbitai help to get the list of available commands.}

[robobun]

^{Updated 4:05 AM PT - Jul 24th, 2026}

@robobun, your commit abf490d is building: #79523

[robobun]

Status: restores the five std::cerr messages that #35256 dropped ahead of std::terminate() in bun-uws, as fputs(stderr). Source-lint test at test/internal/source-lints/uws-terminate-diagnostic.test.ts fails without the fix and passes with it; no-iostream-include.test.ts still passes. Review feedback addressed in 8df9fa7 (workflow paths: filter + corrected terminate-handler wording). Waiting on CI.

[github-actions]

Found 1 issue this PR may fix:

1. Bun.serve expects idleTimeout to be 255 or less #15589 - Directly related to idleTimeout bounds validation; the PR restores the diagnostic message for invalid idleTimeout values in App.h

If this is helpful, copy the block below into the PR description to auto-close this issue on merge.

Fixes #15589

🤖 Generated with Claude Code

[robobun]

This does not fix #15589. That issue is about the HTTP-level idleTimeout cap at 255 in ServerConfig.rs; this PR only restores the diagnostic string for the WebSocket idleTimeout guard in App.h, which Bun already clamps before it reaches uWS.

[claude]

LGTM — mechanical restore of the five stderr diagnostics that #35256 dropped, plus a source-lint guard mirroring no-iostream-include.test.ts. The inline nit about source-lints.yml paths: is worth doing but is a pre-existing gap (the sibling no-iostream-include test scans packages/bun-uws under the same filter) and merge_group: backstops it, so not blocking.

Extended reasoning...

Overview

Five fputs("Error: ...\n", stderr) lines added immediately before each std::terminate() in packages/bun-uws/src/{App.h,HttpContext.h,TopicTree.h}, plus #include <cstdio> in each, plus a new source-lint test that scans packages/bun-uws/src/**/*.{h,hpp,cpp} and asserts every std::terminate() is preceded by an fputs|fprintf ... stderr write. Grep confirms exactly five std::terminate() sites in bun-uws/src and all five are covered.

Security risks

None. These paths are internal invariant guards the PR description shows are unreachable from user JS today; the only runtime change is a fixed-string fputs to stderr immediately before a process-abort that already existed.

Level of scrutiny

Low. The C++ change is a mechanical restore of upstream-uWebSockets diagnostic strings that #35256 deleted rather than translated to fputs. No control-flow, allocation, or state changes. The lint test copies the structure of the adjacent no-iostream-include.test.ts (same glob root, same expect(count).toBeGreaterThan(0) vacuous-pass guard, same expect(violations).toEqual([]) shape) and its look-back loop matches the actual shape at every site (fputs on the line directly above std::terminate(), block-open the line before that).

Other factors

The one flagged nit — source-lints.yml paths: not listing packages/bun-uws/** — is real but (a) equally applies to the pre-existing no-iostream-include.test.ts which scans the same directories, (b) is backstopped by the unfiltered merge_group: trigger, and (c) is a two-line CI-config follow-up rather than a defect in this change. The PR description shows the test failing on the pre-fix tree and passing after, and no-iostream-include.test.ts still passing, which is the right USE_SYSTEM_BUN-style verification for a source-lint.

[claude]

LGTM — the workflow paths: addition addresses the earlier note; the rest is unchanged.

What was reviewed:

• Five fputs(..., stderr) restorations before std::terminate() in App.h/HttpContext.h/TopicTree.h with <cstdio> added — matches upstream uWS wording, no behavior change on non-terminate paths.
• Confirmed these are the only std::terminate() sites under packages/bun-uws/src, so the new lint's terminates > 0 guard and lookback both hold; no-iostream-include still passes since only <cstdio> was added.
• source-lints.yml paths: now covers packages/bun-uws/** and packages/bun-usockets/** in both push and pull_request.

Extended reasoning...

Overview

This PR restores five diagnostic fputs("Error: ...\n", stderr) writes immediately before std::terminate() in packages/bun-uws/src/{App.h,HttpContext.h,TopicTree.h}, which #35256 dropped when it banned <iostream>. It adds <cstdio> to each header, adds test/internal/source-lints/uws-terminate-diagnostic.test.ts to keep every std::terminate() in bun-uws paired with a preceding stderr write, and (in response to the earlier review) adds packages/bun-uws/** / packages/bun-usockets/** to both paths: filters in .github/workflows/source-lints.yml.

Security risks

None. The only runtime effect is a fixed-string fputs to stderr on paths that immediately std::terminate() afterward — internal-invariant guards that (per the PR description and code) are not user-reachable today. No user-controlled data is formatted; no new inputs are parsed. The workflow change only widens a read-only lint job's trigger.

Level of scrutiny

Low. The C++ delta is purely additive diagnostic output on already-fatal paths and cannot affect any surviving control flow. The new test is a hermetic source-tree scan that mirrors the neighboring no-iostream-include.test.ts (same Glob/read pattern, same violations.sort(); expect(violations).toEqual([]) shape, same expect(count).toBeGreaterThan(0) vacuous-pass guard). The workflow edit is a two-line paths-filter widening.

Other factors

I grepped packages/bun-uws/src for std::terminate and confirmed exactly the five sites this PR touches, all now preceded by an fputs(..., stderr) on the immediately preceding non-blank line inside the same block — so the lint's lookback matches every current site and terminates > 0 cannot pass vacuously. <cstdio> is not covered by the <iostream> ban (the shim explicitly recommends fputs/fprintf). My prior review's only note (workflow paths filter) was addressed in 8df9fa7 and the thread is resolved. No CODEOWNERS apply to the touched paths.

[claude]

LGTM — the workflow paths: filter feedback from the earlier pass is addressed in 8df9fa7.

What was reviewed:

• Confirmed all five std::terminate() sites in packages/bun-uws/src now have a preceding fputs(..., stderr); <cstdio> is not in src/banned-includes/ and no-iostream-include.test.ts only bans <iostream>.
• The new lint follows the no-iostream-include pattern and guards against a vacuous pass with expect(terminates).toBeGreaterThan(0).
• The fputs calls sit only on crash-imminent paths, so there's no hot-path or behavioral impact.

Extended reasoning...

Overview

Restores five diagnostic error strings that #35256 dropped when it stripped std::cerr << ahead of each std::terminate() in bun-uws, replacing them with the fputs(..., stderr) form the iostream-ban shim itself prescribes. Adds <cstdio> to App.h, HttpContext.h, and TopicTree.h; adds a source-lint test enforcing the pairing; and (per the earlier review comment) adds packages/bun-uws/** and packages/bun-usockets/** to the source-lints workflow paths: filters.

Security risks

None. The only runtime change is writing a fixed string to stderr immediately before std::terminate() on internal-invariant guards that are not reachable from user JS today.

Level of scrutiny

Low. This is a mechanical restoration of upstream uWebSockets diagnostic strings on crash-only paths, plus a source-tree lint. There are no control-flow changes, no allocations, and no new codepaths reachable in normal operation. <cstdio> is lightweight and not on the banned-includes list; only iostream is shimmed.

Other factors

• Verified via grep that the five std::terminate() sites in packages/bun-uws/src are exactly the five the PR touches — none missed.
• The new lint mirrors no-iostream-include.test.ts (same Glob/readFileSync/sorted-violations shape) and includes a non-vacuous-scan guard. The look-back heuristic is simple but adequate for the current code shape and will name the exact file:line if a future site drops its diagnostic.
• The one review comment from the previous pass (workflow paths: gap) was addressed and the thread is resolved.
• PR evidence shows the lint fails on the pre-fix headers and passes post-fix, and no-iostream-include.test.ts still passes.

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In @.github/workflows/source-lints.yml:
- Around line 17-18: Update the source-lint test command in the workflow’s
affected job to run through the repository-required debug-build wrapper, using
“bun bd ...” rather than invoking “bun test test/internal/source-lints/”
directly. Apply this consistently to both newly added bun-uws and bun-usockets
trigger paths while preserving the existing lint target.

In `@packages/bun-uws/src/App.h`:
- Around line 422-434: Update the idleTimeout and maxLifetime validation
diagnostics before std::terminate() to include each rejected behavior value and
its valid range. Replace the current fixed fputs messages in the idleTimeout and
maxLifetime checks with formatted stderr output, preserving the existing
validation thresholds and termination behavior.

🪄 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: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: 21344c9e-4f98-46d7-8617-4a89697341a2

📥 Commits

Reviewing files that changed from the base of the PR and between df84f8d and abf490d.

📒 Files selected for processing (5)

• .github/workflows/source-lints.yml
• packages/bun-uws/src/App.h
• packages/bun-uws/src/HttpContext.h
• packages/bun-uws/src/TopicTree.h
• test/internal/source-lints/uws-terminate-diagnostic.test.ts

[github-actions]

This PR has been closed because it was flagged as AI slop.

Many AI-generated PRs are fine, but this one was identified as having one or more of the following issues:

• Fails to verify the problem actually exists
• Fails to test that the fix works
• Makes incorrect assumptions about the codebase
• Submits changes that are incomplete or misleading

If you believe this was done in error, please leave a comment explaining why.