Can/should we open up thread to thread state allocator in IndexWriter? [LUCENE-5657]
- Dominant language
- Java
- Stars
- 3.6k
- Forks
- 1.4k
- Avg merge
- 2d 11h
- Merged PRs (30d)
- 88
Description
In #6706 we removed the ability to provide a custom DocumentsWriterPerThreadPool, when we switched to a simpler LIFO scheduler ... I think this (a custom thread pool) is a super-expert thing to do... should we open it up again?
If so I think we need better testing around it; I added two separate tests but neither would fail either the old or the new thread pool ...
---
Migrated from [LUCENE-5657](https://issues.apache.org/jira/browse/LUCENE-5657) by Michael McCandless (@mikemccand)
Contributor guide
Research direction
Start by reviewing issue #6706 and the current IndexWriter scheduling around DocumentsWriterPerThreadPool. Examine the two tests mentioned in the issue and determine whether they distinguish the old and new thread pools. Done means reaching and documenting a decision on reopening custom thread-pool support, with tests that meaningfully cover the chosen behavior.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- java
- Domain
- search
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100