Update module github.com/grpc-ecosystem/go-grpc-middleware to v2 - #6
Merged
bootjp merged 1 commit intoDec 1, 2023
Merged
Conversation
bootjp
approved these changes
Dec 1, 2023
bootjp
deleted the
renovate/github.com-grpc-ecosystem-go-grpc-middleware-2.x
branch
December 6, 2023 14:32
5 tasks
bootjp added a commit
that referenced
this pull request
Apr 26, 2026
Three of the six items from Claude bot's round-1 review were already shipped in 9e3e44d (RowsInput stuck-when-cleared bug, fillRect comment, time-axis label overlap). Round-2 covers the remaining medium/minor items: - design doc 4.3: drop the stale "single-pass ImageData.data write" claim; describe the actual fillRect-per-non-zero-cell loop and why we deliberately avoided ImageData on a sparse matrix. - design doc 3.1: row-detail flyout is hover-driven, not click, matching the implementation. - design doc 4.2: document the time-axis stride floor that landed in code at 9e3e44d, so the doc explains why ceil(56 / cellW) is part of the formula. - KeyViz.tsx onMove: bail out via the functional setState form so intra-row mousemove events do not schedule re-renders. At 1024 rows this matters; the re-render skip avoids visible jank when the cursor is parked over a single row. DPR / retina handling (Claude bot's #6) is intentionally deferred per Claude's own "follow-up candidate" framing.
This was referenced Apr 26, 2026
bootjp added a commit
that referenced
this pull request
Apr 30, 2026
…e[bot] Must Fix x3 + Should/Minor) gemini-code-assist High #1+#2: Pre-cache cipher.AEAD inside Keystore. AES key expansion + GCM init was happening on every Encrypt/Decrypt -- a measurable hot-path regression. Keystore internal map switches from map[uint32][]byte to map[uint32]*keyEntry holding both [KeySize]byte dek and the pre-init cipher.AEAD. New AEAD(keyID) (cipher.AEAD, bool) lookup runs on the hot path; aes.NewCipher / cipher.NewGCM happen once at Set time. claude[bot] Must Fix#2: Keystore.Get returned a slice into the live map; a caller violating the "MUST NOT mutate" contract would corrupt the active DEK silently. Replaced Get with DEK(keyID) ([KeySize]byte, bool) returning a value copy by virtue of the array type. The signature itself enforces immutability of the live entry. claude[bot] Must Fix#1: Cipher.Encrypt doc gains a CRITICAL nonce-reuse warning. Under AES-GCM, reusing (key, nonce) leaks plaintext XOR and enables authentication-key recovery; the §4.1 deterministic node_id || local_epoch || write_count construction is the supported uniqueness scheme. claude[bot] Must Fix#3: Added TestCipher_Decrypt_RejectsReservedKeyID and TestCipher_Decrypt_RejectsBadNonceSize so both Encrypt and Decrypt entry points have explicit coverage of ReservedKeyID and bad-nonce-size rejection. claude[bot] Should Fix#4: EnvelopeVersionV1 doc clarified -- the §11.3 reserved range 0x02..0x0F is for FUTURE decoders; the current build accepts only 0x01 and rejects everything else (including the reserved range) with ErrEnvelopeVersion. claude[bot] Should Fix#5: Envelope.Encode() now validates Version and Body length and returns (bytes, error). Bad inputs fail fast at build time (ErrEnvelopeVersion / ErrEnvelopeShort) instead of producing undecodable bytes that surface later as a confusing read-side failure. gemini-code-assist Medium #3: New AppendHeaderAADBytes(dst, version, flag, key_id) lets storage callers compose AAD onto a pooled buffer without allocating. HeaderAADBytes is now a thin wrapper over AppendHeaderAADBytes. claude[bot] Minor #6: sort.Slice -> slices.Sort in Keystore.IDs (Go 1.25 idiomatic, no closure allocation). claude[bot] Minor #7: FileWrapper.Name() now returns "file:<path>" so log lines and the EncryptionAdmin status RPC can distinguish multiple configured KEK files. claude[bot] Minor #8: Removed unreachable post-Open length check in FileWrapper.Unwrap; the input-length check above already guarantees Open returns exactly fileKEKSize bytes on success. Replaced with an explanatory comment. Tests: 28 unit + 2 property tests, -race clean, 0 lint issues against the project .golangci.yaml.
2 tasks
bootjp added a commit
that referenced
this pull request
May 21, 2026
Six items called out across Codex P2, Gemini medium x2, and Claude bot on PR #798: 1. Codex P2 / Gemini medium / Bot bug #3 - clipboard write_text fires in non-secure contexts. Add capability check (navigator.clipboard?.writeText) and a try/catch around the await; on failure fall back to window.prompt with the JSON pre-selected so the operator can Ctrl/Cmd+C it manually. Surface the failure in a text-danger line so the action no longer silently fails. 2. Gemini medium - formatBytes used base 1024 but labelled kB/MB. Use KiB/MiB. (kB/MB are decimal SI units; the code multiplies by 1024.) 3. Bot accuracy #4 - the truncation display used message.body.length which is JavaScript's UTF-16 code-unit count, not bytes. The server applies BodyMaxBytes against raw UTF-8 so non-ASCII bodies (CJK, emoji) under-reported. Add utf8ByteLength(s) using TextEncoder and use it in the display. 4. Bot bug #2 - purgeRetryAfter was hardcoded to 60 regardless of the actual remaining cooldown. formatApiError already includes the server's "only one PurgeQueue per 60 seconds" message, so removed the state entirely. 5. Bot bug #1 - setLoading(false) in `finally` ran even when `catch` returned early on AbortError. Move setLoading out of finally into the explicit success and non-abort error branches so unmounted components don't get the state update. 6. Bot minor #6 - distilled the multi-line comment blocks on writePurgeInProgress, SqsPeekedAttribute, SqsPeekedMessage, SqsPeekResult, peekQueue, and purgeQueue to single-line summaries per CLAUDE.md conventions. Caller audit (semantic-change rule): no signature changes; the clipboard / formatBytes / utf8ByteLength helpers are private to SqsDetail.tsx; purgeRetryAfter removal touches only the same file. Verification: - cd web/admin && npm run lint passes - go test -race ./internal/admin/... passes - golangci-lint run ./internal/admin/... 0 issues Rebased onto main (Phase 4 PR #797 merged).
6 tasks
bootjp added a commit
that referenced
this pull request
May 26, 2026
…ad timing, barrier semantics, resolved open questions) Folds the PR #835 round-1 design review into the 7a proposed doc: - Critical #1 (timing): the registration-intent decision must run post-buildShardGroups (same phase as chainEncryptionStartupGuard), not in buildEncryptionWriteWiring — the default-group registry store isn't open until buildShardGroups returns. The epoch bump stays pre-buildShardGroups; the registry intent runs after. - #2 (registry handle): commit to threading the default-group WriterRegistryStore out of buildShardGroups (it already builds one per shard) rather than a new runtime/applier accessor. - #3 (async shutdown): the propose goroutine selects on both the apply signal and the process run-context, never closing the barrier on an uncommitted registration — no leak. - #4 (Dispatch seam): gate condition gated on hasMutatingElems (PUT/DEL/DEL_PREFIX), so read-only txns with only ReadKeys stay ungated; reads are structurally excluded. - #5 (barrier three-state): explicit nil (ungated) / open (pending, gated) / closed (committed) semantic; channel armed before gRPC starts; nil checked first to avoid a forever-block receive. - #6 (multi-shard): node-global barrier; a group-2 write waits on the group-1 registration commit — correct, DEK-wide not shard-scoped. - #7/#8 (wording + snapshot): closed-channel fast path is a select branch (not an atomic load); snapshot/compaction interaction noted. The three original open questions are resolved per the reviewer's recommendations (Dispatch / chan struct{} / async). Implementation follows in a separate 7a PR.
bootjp added a commit
that referenced
this pull request
May 26, 2026
…ut-committed registration releases barrier (codex P2) codex P2 #6: a Propose attempt can time out (the waiter is removed) yet Raft still applies the entry, so under consistently >registrationAttemptTimeout commit latency every attempt would return deadline-exceeded → retry forever → barrier never closes despite durable registration. runWriterRegistration now verifies-before-propose each iteration: re-reads the local registry last_seen via a verifyRegistered closure (built in buildProcessStartRegistrationGate from the default-group store) and closes the barrier when last_seen >= our epoch — covering both a prior timed-out-but-committed attempt and a registration that landed via another path. The first iteration sees last_seen < epoch (why we're proposing); a subsequent iteration after a committed-but- timed-out attempt observes the advance and releases. Test: TestRunWriterRegistration_VerifyCommittedClosesBarrier (verify true → barrier closes without reaching propose).
4 tasks
bootjp added a commit
that referenced
this pull request
May 28, 2026
Five findings from coderabbit + claude bot on commit 96223e9. - persist-credentials: false on the checkout step (zizmor `artipacked`, coderabbit Major, claude #3). The job does not push anything and the subsequent github-script step uses secrets.GITHUB_TOKEN explicitly, so the persisted credential is dead weight that a compromised later step could otherwise reuse. - MD040 language tag on the §7.3 chain-diagram fence (coderabbit Minor, claude #1). Tagged as `text`. - Re-draw the chain diagram so it visually shows tla-check.yml and tla-spec-ai-review.yml fire in parallel rather than implying a sequential chain (claude #6). - Sync-point count corrected from "three" to "four": (1) tla-check.yml paths, (2) tla-spec-ai-review.yml paths, (3) the ANCHORS regex in tla-spec-ai-review.yml, (4) the §7.3 list (claude #5). Both the doc prose and the inline workflow comment now enumerate explicitly. - Inline comment in the workflow explaining why `ready_for_review` is in the trigger types here but not in tla-check.yml (claude #4): an AI review request is cheap on draft→ready transitions; a redundant TLC run is not. Skipped (consistent with repo-wide convention, separate hardening pass): pinning actions/checkout@v6 and actions/github-script@v7 to commit SHAs. tla-check.yml and every other workflow in this repo floats on major version tags; introducing SHA pins here alone would be inconsistent. actionlint clean.
3 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for freeto join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR contains the following updates:
v1.4.0->v2.0.1Release Notes
grpc-ecosystem/go-grpc-middleware (github.com/grpc-ecosystem/go-grpc-middleware)
v2.0.1Compare Source
What's Changed
New Contributors
Full Changelog: grpc-ecosystem/go-grpc-middleware@v2.0.0...v2.0.1
v2.0.0Compare Source
This is the first stable release of the new v2 release branch 🎉
Many of the interceptors have been rewritten from scratch and the project has been upgraded to use the Go Protobuf v2 API.
See the project README for details and migration guide. Thanks to all contributors who made this possible! 💪🏽
What's Changed
logging.InjectFieldsusage description by @aimuz in https://github.com/grpc-ecosystem/go-grpc-middleware/pull/541New Contributors
Full Changelog: grpc-ecosystem/go-grpc-middleware@v1.4.0...v2.0.0
Configuration
📅 Schedule: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).
🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
🔕 Ignore: Close this PR and you won't be reminded about this update again.
This PR has been generated by Mend Renovate. View repository job log here.