Hacktoberfest 2026:メンテナが10月に向けて印を付けた、オープンで初心者向けの issue。 Hacktoberfest の issue を見る

query-db: a Query refetch that joins a loader's refetch acquisition does not report shifted pages

オープン
#2,185 コメント 1 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 1 日以内に返信

まだ誰も着手していません。

評価

難易度
5/5
見積もり時間
1週間以上
初心者へのやさしさ
8/100
issue の種類
バグ
明瞭さ
説明が足りない
活発さ
活発
技術スタック
typescript

調査の方向性

The gate is in packages/query-db-collection/src/query.ts in recordOrderedPageResult, where acquisitionRefetches suppresses results for a hash. Read how the guard counts overlapping acquisitions and how a joined Query fetch differs from the acquisition's own result. The issue asks for a named replay in the query-db refetch oracle with two live queries; done means that replay fails before a fix and passes after it. Expect the fix to be a design change, not a one-line edit.

索引モデルが issue の本文から書いたものです。

説明

Summary

Query DB does not report invalidateLoadedSubsets() for a Query refetch that joins a running refetch: true loader acquisition on the same Query hash. If that refetch shifts the page, other ordered live queries that rely on the page can keep a #2136-style gap.

Where

packages/query-db-collection/src/query.ts, recordOrderedPageResult. The gate skips a result while acquisitionRefetches holds its hash. This loop guard stops a repair's own result from starting another repair. Since #2154 (commit 0216c6026), the guard counts overlapping acquisitions, but it still cannot tell a joined user refetch from the acquisition's own result.

Shape of a trace (not yet reproduced)

  1. Loader A starts a bounded repair. Its refetch: true prefix acquisition on Query hash H is in flight.
  2. utils.refetch() or invalidateQueries runs. Query joins the in-flight fetch for H, after the backend changed.
  3. The result for H has different keys and a later page follows it. Without the guard it would report.
  4. The guard suppresses the report. Loader A repairs its own window from the result, but loader B, whose window also used H's page, gets no report.

Status

No history in the query-db refetch oracle reaches this. The oracle's limits section and the coverage map record it.

Suggested witness

A named replay in the query-db refetch oracle with two live queries over the same pages. Hold one loader's repair acquisition, change the backend, run a Query refetch, then release. The coverage law should fail for the second live query if the defect is real. A possible fix: record the fetch that the acquisition started, not the hash, and report other fetches' results.

This issue was written by Isaac.

主要言語
TypeScript
スター
3.9k
フォーク
272
平均マージ
22時間 54分
マージ済み PR(30日)
236

環境構築

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

TanStack/db のほかの issue

TanStack/db の issue をすべて見る

似ている issue

TypeScript の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。