apache/lucene

Create a simple "real world" regexp benchmark [LUCENE-9986]

Open

#11,025 opened on Jun 2, 2021

View on GitHub
 (6 comments) (0 reactions) (1 assignee)Java (2,179 stars) (879 forks)batch import
good first issuelegacy-jira-priority:Majortype:enhancement

Description

For issues like #11022, where we are struggling to decide which low-level optimizations to make for our (complicated!) determinize method, it would really help to have a large, real-world corpus of regexps to evaluate performance metrics of our automata operations, like CPU and HEAP required to parse the regexp and determinize.

Does anyone know of such an existing, hopefully compatibly licensed, corpus?

Probably we would add these benchmarks to luceneutil.

Migrated from LUCENE-9986 by Michael McCandless (@mikemccand)

Contributor guide

Tech stack
java
Domain
performance
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
blocked
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
clear
Prerequisites
JavaLucene internalsRegex
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
20
Research direction
First, review the existing discussion in issue #11022 to understand the performance concerns. Then, search for a real world regex corpus with a compatible license (e.g., regex corpora from programming language tests or open source projects). The benchmarks should be added to the `luceneutil` project, likely as a new module, measuring CPU and heap usage of regex parsing and determinization. Consult with maintainers, especially Michael McCandless, to align on the corpus selection and benchmark design.