Page-level fragment for /events/past: skip the fetch pipeline on cache hits
维护者通常 1 天内回复
还没有人认领这个 Issue。
评估
- 难度
- 5/5
- 预计耗时
- 一周以上
- 新手友好度
- 42/100
- Issue 类型
- 功能
- 描述清晰度
- 基本清楚
- 活跃度
- 活跃
- 领域
- backend, performance
调研方向
Start at the controller action for GET /events/past and the view's existing list, pagination, ETag, and card-fragment code; compare them with the prototype on the measurement branch. Review deletion staleness and html_safe handling before settling the cache key. Done means cold HTML remains byte-identical, warm requests use the page fragment with one MAX(updated_at) query, and the chosen invalidation semantics are explicit.
由索引模型根据 Issue 内容生成。
描述
A benchmarked prototype cut the steady-state (warm-fragment) cost of GET /events/past by 84% (43,426 -> 7,104 GC objects per request) and its queries from 6 to 1 by serving the whole list from one page-level fragment: on a hit, the controller skips the COUNT, the eager loads, the decoration and the render entirely; only the ETag's MAX(updated_at) query runs.
Today each card fragment is checked after the pipeline has already loaded and decorated all 20 records, so warm requests pay the full fetch cost to render from cache.
Prototype shape (verified: cold HTML byte-identical, warm fragment serves the full card set)
- Key:
['events/past-page', I18n.locale, params[:page].to_i, MAX(updated_at) etag source] - Controller:
read_fragmenton that key beforefetch_past_events; on a hit, the view outputs the stored HTML - View:
cache(..., skip_digest: true)around the existing list + pagination markup - Measured: warm 43,426 -> 7,104 allocations (−84%), cold 73,401 -> 73,803 (+0.5%, the extra page write), cold rendered HTML identical
The prototype diff exists on the measurement branch and can be revived on request.
Staleness semantics to review before shipping: the key inherits the same exposure as the existing ETag and card fragments — record deletions do not bump MAX(updated_at), so a deleted event stays visible until another tracked table row is touched. Whether that is acceptable (and whether the key should incorporate a tombstone counter or shorter horizon) is the main design question. The stored HTML is output with html_safe, standard for view fragments but worth an explicit look.
- 主要语言
- Ruby
- 星标
- 105
- 派生
- 206
- 平均合并
- 1 天 3 小时
- 30 天内合并 PR
- 74
环境准备
- 提供 Dockerfile 或 Docker Compose 文件
- 没有 Pull Request 模板
- 阅读贡献指南
从这里开始
- 先读完整个 Issue，再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库，在一个分支上完成修改。
- 提交 Pull Request，并在描述里引用这个 Issue 编号。
codebar/planner 的其他 Issue
-
DashboardQuery eager-loads workshop_host with :sponsors in one join (same host-loss trap as #2975)未关闭
难度 2/5 1-3 小时 新手友好度 78/100
维护者通常 1 天内回复
-
enhancement
难度 2/5 1-3 小时 新手友好度 76/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 68/100
维护者通常 1 天内回复
-
难度 4/5 3-5 天 新手友好度 30/100
维护者通常 1 天内回复
-
难度 3/5 1-2 天 新手友好度 45/100
维护者通常 1 天内回复
相似的 Issue
-
bug
难度 2/5 1-3 小时 新手友好度 62/100
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 65/100
slovensko-digital/autogram-portal#392 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 65/100
voxpupuli/puppet-quadlets#122 · 5 条评论 ·
维护者通常 1 天内回复
-
难度 1/5 1 小时以内 新手友好度 82/100
TheOdinProject/curriculum#31452 ·
维护者通常 1 天内回复
-
难度 2/5 1-3 小时 新手友好度 70/100