Why O(1) does not guarantee high speed: four data structures for the signal-effect graph

In this article, the author analyzes the efficiency of dependency graph implementations in reactive systems, where it is necessary to quickly find and remove connections between signals and effects. Despite the theoretical O(1) complexity for search and deletion operations, performance can vary significantly in practice due to the specifics of the V8 engine and memory management. The author conducts a comparative analysis of four different data structures used to store the signal-effect graph. The study evaluates not only the time cost of operations but also memory consumption. Local benchmarks demonstrate that identical asymptotic complexity does not guarantee identical execution time. This article is useful for developers working on optimizing reactive libraries and state management systems, as it clearly illustrates the impact of low-level factors on the real-world performance of JavaScript applications.