quickwit-oss/tantivy

Tantivy should work without a store/fieldnorm/position file

Open

#124 建立於 2017年4月24日

在 GitHub 查看
 (0 留言) (0 反應) (0 負責人)Rust (8,354 star) (499 fork)batch import
enhancementgood first issue

描述

Currently tantivy creates empty file even if the schema does not requires them (e.g. no field in the schema is stored for instance). The segment reader also needlessly opens and mmaps these files. jwolfe lost a bit of time when writing a converter from a different search index to tantivy.

貢獻者指南

技術棧
rust
領域
backend
議題類型
feature
難度面向新貢獻者的預計實作難度，1 表示很小改動，5 表示專家級工作。
3
預計時間有經驗貢獻者完成調查、實作、測試並準備 pull request 的粗略時間範圍。
half day
活動狀態議題目前的可參與程度：新鮮、活躍、陳舊、阻塞或等待維護者輸入。
stale
清晰度議題是否清楚說明預期改動、驗收標準和下一步。
clear
前置要求
Rust programmingfile I/O concepts
新手友善度1-100 的估計分數，表示該議題對首次貢獻者的友善程度。
60
研究方向
Examine the segment reader code in Tantivy that creates and opens store, fieldnorm, and position files. The issue suggests that these files are created even when the schema doesn't require them. Start by locating where these files are initialized (likely in the segment or index writer). Then, modify the logic to conditionally create and open them based on the schema's stored fields, field norms, and positions. Also review the reader side to avoid mmap'ing unnecessary files. Check the repository's git history for any partial attempts or discussions around this issue. Update existing tests to verify the new behavior.