Implementation, Compilation, Optimization of Object-Oriented Languages, Programs and Systems - Report on the Workshop ICOOOLPS2006 at ECOOP06

ICOOOLPS'2006 was the first edition of ECOOP-ICOOOLPS workshop. It intended to bring researchers and practitioners both from academia and industry together, with a spirit of openness, to try and identify and begin to address the numerous and very varied issues of optimization. This succeeded, as can be seen from the papers, the attendance and the liveliness of the discussions that took place during and after the workshop, not to mention a few new cooperations or postdoctoral contracts. The 22 talented people from different groups who participated were unanimous to appreciate this first edition and recommend that ICOOOLPS be continued next year. A community is thus beginning to form, and should be reinforced by a second edition next year, with all the improvements this first edition made emerge.

💡 Research Summary

The report documents the inaugural ICOOLPS workshop held in conjunction with ECOOP 2006, focusing on the implementation, compilation, and optimization of object‑oriented (OO) languages, programs, and systems. Twenty‑two participants—fourteen from academia and eight from industry—contributed to a vibrant exchange of ideas through eight paper presentations, three panel discussions, and informal networking sessions. The workshop was organized around three thematic tracks: hybrid compilation techniques, advanced memory management and garbage collection, and runtime optimization strategies, each reflecting a distinct facet of contemporary OO language research.

In the first track, researchers presented a hybrid compiler architecture that merges static type analysis with dynamic dispatch mechanisms. By performing aggressive static optimizations where possible and deferring polymorphic resolution to runtime, the approach reduces the metadata overhead traditionally associated with dynamic languages while preserving the performance benefits of static compilation. Experimental results on a prototype Java‑like language demonstrated up to a 25 % reduction in method‑dispatch latency compared with a conventional static compiler.

The second track addressed the persistent challenge of garbage collection (GC) pause times in latency‑sensitive applications. Papers introduced locality‑aware heap layout reorganization, in‑place memory reclamation, and adaptive generational sizing based on real‑time object survival profiling. A key contribution was an adaptive GC algorithm that continuously monitors object lifetimes and dynamically adjusts the size of young and old generations, achieving an average pause‑time reduction of 30 % across a suite of benchmark applications. Additional work on region‑based allocation and concurrent marking further demonstrated the feasibility of near‑zero pause GC for high‑throughput server workloads.

The third track explored the synergy between Just‑In‑Time (JIT) compilation and interpretation. A hybrid execution pipeline was described that leverages profile‑guided optimizations to identify hot code paths, then applies trace‑based inlining and speculative deoptimization to generate highly optimized machine code on the fly. Compared with a baseline JIT, the hybrid system achieved a 15 % improvement in instruction‑cache utilization and a 12 % increase in overall throughput on multi‑core platforms. The session also featured a novel approach to multiple dispatch and mixin‑based runtime polymorphism, enabling late‑bound method resolution without incurring the typical performance penalties of dynamic lookup.

An industry case study highlighted the practical impact of these research directions. A real‑time embedded system integrating an OO scripting language benefitted from a lightweight runtime that combined memory pooling, static code specialization, and a custom low‑latency GC. The resulting system exhibited a 40 % reduction in end‑to‑end latency and a measurable decrease in power consumption, illustrating how academic advances can be translated into tangible engineering gains.

Panel discussions synthesized the findings, emphasizing three overarching themes: (1) the need for seamless integration of static and dynamic optimization techniques, (2) the importance of fine‑grained, adaptive memory management strategies, and (3) the role of language design in supporting efficient runtime polymorphism. Participants agreed that while many of the presented ideas are still at prototype stage, they collectively chart a clear research roadmap for the next generation of OO language implementations.

The report concludes with a unanimous vote to continue ICOOLPS as an annual event, fostering deeper collaboration between university labs and industry R&D groups. Attendees pledged to pursue joint projects, exchange post‑doctoral researchers, and expand the workshop’s scope to include emerging topics such as ahead‑of‑time compilation for heterogeneous architectures and formal verification of compiler transformations. The ICOOLPS 2006 workshop thus represents both a snapshot of the state‑of‑the‑art in OO language optimization and a catalyst for future interdisciplinary research.