A third-generation SELF implementation: reconciling responsiveness with performance

  • Authors:
  • Urs Hölzle;David Ungar

  • Affiliations:
  • Computer Science Department, University of California, Santa Barbara, CA;Sun Microsystems Laboratories, 2550 Garcia Avenue, Building 29, Mountain View, CA

  • Venue:
  • OOPSLA '94 Proceedings of the ninth annual conference on Object-oriented programming systems, language, and applications
  • Year:
  • 1994

Quantified Score

Hi-index 0.00

Visualization

Abstract

Programming systems should be both responsive (to support rapid development) and efficient (to complete computations quickly). Pure object-oriented languages are harder to implement efficiently since they need optimization to achieve good performance. Unfortunately, optimization conflicts with interactive responsiveness because it tends to produce long compilation pauses, leading to unresponsive programming environments. Therefore, to achieve good responsiveness, existing exploratory programming environments such as the Smalltalk-80 environment rely on interpretation or non-optimizing dynamic compilation. But such systems pay a price for their interactiveness, since they may execute programs several times slower than an optimizing system.SELF-93 reconciles high performance with responsiveness by combining a fast, non-optimizing compiler with a slower, optimizing compiler. The resulting system achieves both excellent performance (two or three times faster than existing Smalltalk systems) and good responsiveness. Except for situations requiring large applications to be (re)compiled from scratch, the system allows for pleasant interactive use with few perceptible compilation pauses. To our knowledge, SELF-93 is the first implementation of a pure object-oriented language achieving both good performance and good responsiveness.When measuring interactive pauses, it is imperative to treat multiple short pauses as one longer pause if the pauses occur in short succession, since they are perceived as one pause by the user. We propose a definition of pause clustering and show that clustering can make an order-of-magnitude difference in the pause time distribution.