# Just-in-Time Index Compilation

## TL;DR

This paper introduces a flexible, just-in-time index reorganization method that applies small updates dynamically, formalizing a composable grammar and rewrite rules to optimize index performance.

## Contribution

It formalizes a new organizational grammar and rewrite framework enabling dynamic, incremental index reorganization for improved adaptability and performance.

## Key findings

- Flexible index structure adapts to various performance goals
- Competitive with existing structures like C++ STL map
- Effective in dynamic, background reorganization scenarios

## Abstract

Creating or modifying a primary index is a time-consuming process, as the index typically needs to be rebuilt from scratch. In this paper, we explore a more graceful "just-in-time" approach to index reorganization, where small changes are dynamically applied in the background. To enable this type of reorganization, we formalize a composable organizational grammar, expressive enough to capture instances of not only existing index structures, but arbitrary hybrids as well. We introduce an algebra of rewrite rules for such structures, and a framework for defining and optimizing policies for just-in-time rewriting. Our experimental analysis shows that the resulting index structure is flexible enough to adapt to a variety of performance goals, while also remaining competitive with existing structures like the C++ standard template library map.