Year: 1996
Type: article
Abstract: We argue for the use of aggressive interprocedural flow analysis to guide optimizations for higher-order languages such as Scheme [Clinger and Rees 1991] and ML [Milner et al. 1990].Functional languages provide abstraction through firstclass procedures and abstract datatypes.Without sophisticated interprocedural analysis, compilers must make overly conservative assumptions about how abstractions are used in the program.These assumptions often lead to poor performance.We present a strategy for building high-performance implementations that relaxes these assumptions and provides a framework upon which useful optimizations can be built.We describe two experiments that provide some evidence that this approach is viable.
Source: ACM Computing Surveys
Authors: Suresh Jagannathan, Andrew Wright
Language: English
Subfield: Software
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W1998579185