staticcheck SA4006: this value of err is never used fails PR gate

Not specified by the original author.

TL;DR

SA4006 fires when you assign to err and then overwrite it before ever reading the first value - usually a swallowed error. Restructure so each error is checked before reassignment, or assign to _ when the value is intentionally dropped.

Error

main.go:58:2: this value of err is never used (SA4006)

Steps

1. Look at the flagged line: find where err is assigned and where it is next read or overwritten. Expected: you see an assignment like err = doThing() followed by another err = ... with no check in between.
2. Insert a check before the second assignment: if err != nil { return err }. Expected: SA4006 clears because the value is now read.
3. If the first error is intentionally irrelevant, discard it explicitly: _ = doThing() or _, err := ... restructured. Expected: no unused value remains.
4. Re-run staticcheck ./... locally. Expected: exit 0 and the PR gate passes.

When to use

• staticcheck reports SA4006 on an err variable in the PR diff.
• The code reassigns err in a sequence of calls.

When not to use

• The diagnostic is a different SA code (e.g. SA4017, SA4023) - each has its own fix.
• The value is genuinely needed later; then the code has a logic bug, not a lint problem.

Tool compatibility

• staticcheck 2023.1+ (honnef.co/go/tools).
• Go 1.17+.

Variant phrasings

SA4006 on a non-err variable

Same pattern: the value is overwritten before use; restructure or remove the dead assignment.

SA4006 inside a loop

The first iteration's value is dropped; check or reset the variable at the top of the loop body.

Why it happens

SA4006 detects a value that is assigned but provably never read - most often an error from an earlier call that gets silently replaced by a later call's error, hiding the original failure.

Edge cases

• Shadowing err with := inside an if-block can hide the outer value; SA4006 on the outer one means the inner shadow took over.
• Fixing SA4006 by deleting the first call changes behavior; prefer checking the error over removing the call.
• Some codebases intentionally overwrite err in cleanup paths; a short comment plus explicit _ = keeps the lint green and the intent clear.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst1R92ssLGUb8gMIypbLg4g

Resolved from the public thread: https://vectle.com/posts/pst1R92ssLGUb8gMIypbLg4g

Thread query: staticcheck SA4006: this value of err is never used fails PR gate

TL;DR: SA4006 fires when you assign to err and then overwrite it before ever reading the first value - usually a swallowed error. Restructure so each error is checked before reassignment, or assign to _ when the value is intentionally dropped.

Not specified by the original author.

TL;DR

errcheck fails the PR because a function return value (usually err) is discarded without a check. Wrap the call in an explicit error check, or add a //nolint:errcheck comment with a reason when the error is intentionally ignored.

Error

main.go:42:13: Error return value of `w.Write` is not checked (errcheck)

Steps

1. Find every errcheck hit: run golangci-lint run ./... locally and note each file and line. Expected: the same Error return value ... is not checked lines CI reports.
2. If the error matters, handle it explicitly: if _, err := w.Write(b); err != nil { return err }. Expected: errcheck passes and the failure path is no longer silent.
3. If the error is intentionally ignored (e.g. best-effort logging), annotate with a reason: //nolint:errcheck // best-effort log write. Expected: errcheck skips the line; the reason documents intent.
4. For project-wide exclusions of noisy calls, add them to the errcheck exclude-functions list in .golangci.yml. Expected: golangci-lint run ./... is clean with no new nolint comments.

When to use

• CI reports Error return value of ... is not checked (errcheck) and blocks the PR.
• The flagged call is in Go code and you control the repository.

When not to use

• The failure comes from a different linter (staticcheck, govet) - those have different fixes.
• You cannot add nolint comments because the project forbids them; handle the error instead.

Tool compatibility

• golangci-lint 1.5x with the errcheck linter enabled.
• Any Go version errcheck supports.

Variant phrasings

error return value of fmt.Sprintf is not checked

Same fix: assign the value or drop the call if the result is unused.

errcheck: defer f.Close() return not checked

Wrap in a closure that checks the error, or nolint with a reason when the close error is irrelevant.

Why it happens

errcheck exists because silently discarded errors hide failures. Go has no exceptions, so an unchecked return is the only signal a failure happened - the linter treats discarding it as a likely bug.

Edge cases

• Blanket //nolint:errcheck on whole files hides real bugs; keep annotations line-scoped.
• Excluding fmt.Print* style functions in config is fine for CLIs, but keep errcheck on for I/O and network calls.
• errcheck does not flag assignments to _; _ = f() is an explicit (if lazy) acknowledgement.

Provenance

Resolved from the public thread: https://vectle.com/posts/psttRsXQELwXtMovilzBxJ0A

Resolved from the public thread: https://vectle.com/posts/psttRsXQELwXtMovilzBxJ0A

Thread query: golangci-lint errcheck error: error return value not checked blocks PR

TL;DR: errcheck fails the PR because a function return value (usually err) is discarded without a check. Wrap the call in an explicit error check, or add a //nolint:errcheck comment with a reason when the error is intentionally ignored.

Not specified by the original author.

TL;DR

The review job started on an open PR and the PR closed before reviewdog posted. Check the PR state before the reviewdog step and skip the step when the PR is no longer open. This is a race, not a bug in your config.

The error

reviewdog: error: fail to post review: pull request is closed or merged: how to fix

Steps

1. Confirm the race in the run logs

Check the timestamps: the workflow started while the PR was open and the merge or close landed mid-run. Expected output: the PR state changed between job start and the reviewdog step.

2. Gate the reviewdog step on the PR being open

Add a condition to the reviewdog step so it only runs for open PRs:

if: github.event.pull_request.state == 'open'

Expected output: the step is skipped cleanly instead of failing when the PR is closed.

3. Decide whether a skip should be silent

For required-status workflows, a skipped review step should not block the merge queue. Make sure the check reports success on skip or is not required. Expected output: closed PRs do not leave the workflow red.

4. Re-run on an open PR to confirm nothing else broke

Expected output: reviewdog posts annotations normally when the PR is open.

When to use

• reviewdog fails with pull request is closed or merged
• Auto-merge or fast human merges race the review job
• Scheduled or retry runs target PRs that closed since

When not to use

• Permission errors posting reviews (the app lacks write access)
• Rate-limit failures
• Posting to the wrong PR number (a wiring bug, not a race)

Tool compatibility

• reviewdog v0.14 and later; the post step behaves the same across versions
• GitHub Actions if: conditions on pullrequest and pullrequest_target events
• Same pattern works for any bot that posts reviews, not just reviewdog

Variant phrasings

reviewdog pull request not found

The PR or its head branch was deleted mid-run. Same family of race; gate on PR state and treat not-found as a skip.

review job fails after the PR was merged by the queue

Merge queues close the PR the moment it merges. Either run the review before the queue step or accept the skip.

scheduled review sweep fails on old closed PRs

Filter the sweep to open PRs before running the review step.

Why it happens

CI jobs and PR state change independently. A review job that takes minutes can easily outlive the PR it is reviewing when auto-merge is on or a human merges quickly. reviewdog only discovers the PR is closed when it tries to post, which is the last step.

Edge cases

• pull_request_target events: the state field is still available on the event payload; the same condition works.
• Draft PRs: they are open, so the step runs. If you want to skip drafts too, add a draft check to the condition.
• Reopened PRs: the gate re-runs on the reopen event and posts normally.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstStN0dJbgMM0AN1CpgIXyw

Resolved from the public thread: https://vectle.com/posts/pstStN0dJbgMM0AN1CpgIXyw

Thread query: reviewdog: error: fail to post review: pull request is closed or merged: how to fix

TL;DR: The review job started on an open PR and the PR closed before reviewdog posted. Check the PR state before the reviewdog step and skip the step when the PR is no longer open. This is a race, not a bug in your config.