Update next and react versions - #97
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Review the following changes in direct dependencies. Learn more about Socket for GitHub.
|
Greptile SummaryThis PR bumps
Confidence Score: 3/5The react/react-dom patch bump is safe, but the lock file contains malformed duplicate YAML keys that could cause inconsistent behavior in tools that parse it independently of pnpm. The lock file contains duplicate top-level entries for all @next/* packages and duplicate libc: keys in multiple native-binary stanzas. While pnpm's own parser may silently accept the last occurrence, external tooling (security scanners, dependency auditors, CI lock-file integrity checks) could read different entries, potentially masking vulnerabilities or breaking audit steps. The lockfile should be cleanly regenerated before merge. frontend/pnpm-lock.yaml requires attention due to duplicate package entries and duplicate YAML keys introduced during regeneration.
|
| Filename | Overview |
|---|---|
| frontend/package.json | Bumps react and react-dom from 19.2.4 to 19.2.6 (patch update); next remains at 16.2.6. |
| frontend/pnpm-workspace.yaml | Adds react@19.2.6 and react-dom@19.2.6 to minimumReleaseAgeExclude, replacing the old 19.2.4 entries to match the updated versions. |
| frontend/pnpm-lock.yaml | Lock file regenerated with updated react/react-dom peer resolutions, but introduces duplicate YAML map keys — full package entries for @next/env@16.2.6 and all @next/swc-* packages appear twice, and multiple @img/sharp-* and @tailwindcss/oxide-* packages have duplicate libc: fields within their stanzas. |
Flowchart
%%{init: {'theme': 'neutral'}}%%
flowchart TD
A[PR: Update next and react versions] --> B[package.json]
A --> C[pnpm-workspace.yaml]
A --> D[pnpm-lock.yaml]
B --> B1[react: 19.2.4 to 19.2.6]
B --> B2[react-dom: 19.2.4 to 19.2.6]
B --> B3[next: 16.2.6 unchanged]
C --> C1[Add react@19.2.6 to minimumReleaseAgeExclude]
C --> C2[Add react-dom@19.2.6 to minimumReleaseAgeExclude]
D --> D1[Update all react peer dependency resolutions]
D --> D2[Duplicate @next/* package entries added]
D --> D3[Duplicate libc: keys in @img/sharp-* stanzas]
Comments Outside Diff (1)
-
frontend/pnpm-lock.yaml, line 205-208 (link)Duplicate package entries in lock file
The regenerated lock file contains duplicate top-level YAML map keys for every
@next/*package at16.2.6(e.g.,@next/env, all@next/swc-*variants), as well as duplicatelibc:fields within stanzas for several@img/sharp-*and@tailwindcss/oxide-*packages. Duplicate keys are undefined behavior in the YAML spec — most parsers silently keep only the last occurrence, but security scanners, audit tools, and some CI dependency-resolution steps that parse the lockfile independently of pnpm could see inconsistent or missing entries. This suggests the lockfile was regenerated with a mismatched pnpm version or via a manualpnpm installthat merged two lock files. It should be regenerated cleanly (delete the lock file and runpnpm installfrom scratch, or ensure a single consistent pnpm version is used) before merging.Prompt To Fix With AI
This is a comment left during a code review. Path: frontend/pnpm-lock.yaml Line: 205-208 Comment: **Duplicate package entries in lock file** The regenerated lock file contains duplicate top-level YAML map keys for every `@next/*` package at `16.2.6` (e.g., `@next/env`, all `@next/swc-*` variants), as well as duplicate `libc:` fields within stanzas for several `@img/sharp-*` and `@tailwindcss/oxide-*` packages. Duplicate keys are undefined behavior in the YAML spec — most parsers silently keep only the last occurrence, but security scanners, audit tools, and some CI dependency-resolution steps that parse the lockfile independently of pnpm could see inconsistent or missing entries. This suggests the lockfile was regenerated with a mismatched pnpm version or via a manual `pnpm install` that merged two lock files. It should be regenerated cleanly (delete the lock file and run `pnpm install` from scratch, or ensure a single consistent pnpm version is used) before merging. How can I resolve this? If you propose a fix, please make it concise.
Prompt To Fix All With AI
Fix the following 1 code review issue. Work through them one at a time, proposing concise fixes.
---
### Issue 1 of 1
frontend/pnpm-lock.yaml:205-208
**Duplicate package entries in lock file**
The regenerated lock file contains duplicate top-level YAML map keys for every `@next/*` package at `16.2.6` (e.g., `@next/env`, all `@next/swc-*` variants), as well as duplicate `libc:` fields within stanzas for several `@img/sharp-*` and `@tailwindcss/oxide-*` packages. Duplicate keys are undefined behavior in the YAML spec — most parsers silently keep only the last occurrence, but security scanners, audit tools, and some CI dependency-resolution steps that parse the lockfile independently of pnpm could see inconsistent or missing entries. This suggests the lockfile was regenerated with a mismatched pnpm version or via a manual `pnpm install` that merged two lock files. It should be regenerated cleanly (delete the lock file and run `pnpm install` from scratch, or ensure a single consistent pnpm version is used) before merging.
Reviews (1): Last reviewed commit: "merge main" | Re-trigger Greptile
Replaces frontend/pnpm-lock.yaml with main's lockfile minus the duplicate keys introduced by PR #97 (so pnpm 10 can parse it). No package versions change vs main; pnpm install --frozen-lockfile passes. The previous commit on this branch ran a non-frozen pnpm install, which silently re-resolved every floating semver in package.json and produced ~2000 lines of incidental lockfile churn. This commit reverses that.
* fix(tps-chart): smooth Transaction volume line Switch the TPS Area interpolation from `linear` to `monotone` so the Transaction volume chart renders as a smooth curve instead of connected straight segments between data points. * fix(tps-chart): animate new data points instead of curving the line Revert line interpolation to linear and enable Recharts' built-in animation so newly arriving data points ease into the chart instead of snapping in. This is the smoothing that was actually requested: the visible motion of new points appearing, not the line shape. * feat(tps-chart): smooth streaming via sliding time-window X axis Switch the X axis to a continuous time scale with a [now - 5min, now] domain advanced each animation frame, so new points slide in smoothly from the right edge instead of popping into discrete category slots. Keeps isAnimationActive disabled to avoid Recharts re-animation loops on each TPS event. Clamp the domain's left edge to the oldest point so the chart fills immediately instead of showing an empty 5-minute lead-in. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore: update pnpm lockfile and allow sharp/unrs-resolver builds Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore: minimize lockfile diff vs main Replaces frontend/pnpm-lock.yaml with main's lockfile minus the duplicate keys introduced by PR #97 (so pnpm 10 can parse it). No package versions change vs main; pnpm install --frozen-lockfile passes. The previous commit on this branch ran a non-frozen pnpm install, which silently re-resolved every floating semver in package.json and produced ~2000 lines of incidental lockfile churn. This commit reverses that. * ci: unified frontend workflow (lint + typecheck + build) * ci: add rust job (fmt + clippy + build) * ci: run rust job in rust:1.91-slim container for libclang/bindgen parity * perf(frontend): reduce GC pressure and re-renders in the live event pipeline (#98) **Motivation:** After ~1h on node.monad.xyz the renderer process is killed for OOM. There is no single growing array — every component bounds its state — but at ~200 TPS the per-event fan-out from EventsContext drives enough short-lived allocations to fragment the heap. Backgrounded tabs are worse because the browser does not throttle WebSocket onmessage. **Modifications:** 1. Memoize EventsContext value so consumers don't re-render every time TopAccesses updates. 2. Drop WebSocket messages while document.visibilityState === 'hidden' to avoid running the dispatch + decode pipeline for an invisible tab. 3. Move the in-flight block's transactions out of React state into a ref-backed Map<txnIndex, Transaction>. Per-event updates are now O(1) instead of O(N), and the block is materialized into React state once at BlockFinalized/BlockVerified. 4. Allow subscribers to register an eventTypes filter on subscribe, and apply it at dispatch time so swap/transfer/totals hooks no longer run on BlockStart/TxnHeaderStart/etc. **Result:** Live event dispatch does far less work per WebSocket message, in-flight blocks no longer reallocate their transactions array on every txn event, and a backgrounded tab is effectively idle until it becomes visible again. Co-authored-by: dak-agent[bot] <284037069+dak-agent[bot]@users.noreply.github.com> * Smooth transaction volume chart clean (#105) * feat(tps-chart): smoothly extend the line to each new point Switch the X axis to a continuous time scale whose right edge eases toward the newest point's timestamp each animation frame. A freshly appended point sits just past the edge (clipped by allowDataOverflow) and is revealed sliding in from the right instead of snapping into a discrete category slot; the animation halts once the edge catches up, so the chart is still between points. isAnimationActive stays disabled to avoid Recharts re-animation loops on every TPS event. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * chore: update pnpm lockfile and allow sharp/unrs-resolver builds Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * drawing lines smoothly between the points creation * fix the x absis to have only 1 now --------- Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * fix(backend): resolve clippy lints (#101) * fix(backend): resolve auto-fixable clippy lints Addresses 6 of the 8 clippy findings surfaced by the new rust CI job: - needless_return (server.rs) - match_like_matches_macro (server.rs) - ptr_arg: &Vec<T> -> &[T] (event_filter.rs) - needless_borrow (event_filter.rs) - manual_is_multiple_of (event_listener.rs) - clone_on_copy (serializable_event.rs) Remaining (require manual judgment, deferred): large_enum_variant (server.rs), should_implement_trait for from_str (event_listener.rs). * fix(backend): resolve non-auto-fixable clippy lints - large_enum_variant: box the 640-byte Event variant of EventDataOrMetrics (Event(Box<EventData>)). Since From::from infers its parameter type from the argument (deref coercion does not apply), the match-site call passes &*event_data to select From<&EventData>; construction uses Box::new(..). - should_implement_trait: rename inherent EventName::from_str -> from_name (and its single caller) so it no longer shadows std::str::FromStr::from_str. * fix(backend): remove redundant u64 casts in client (unnecessary_cast) Surfaced once the lib compiled clean: timestamp_ns is u64, so the (u64 - u64) as u64 casts are redundant; drop the cast and parens. --------- Co-authored-by: dak-agent[bot] <284037069+dak-agent[bot]@users.noreply.github.com> Co-authored-by: Camillebzd <bcamille99@gmail.com> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Co-authored-by: Camillebzd <48495021+Camillebzd@users.noreply.github.com>
No description provided.