BibTeX
@MISC{Nielson96perspectiveson,
author = {Flemming Nielson},
title = {Perspectives on Program Analysis},
year = {1996}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
eing analysed. On the negative side, the semantic correctness of the analysis is seldom established and therefore there is often no formal justification for the program transformations for which the information is used. The semantics based approach [1; 5] is often based on domain theory in the form of abstract domains modelling sets of values, projections, or partial equivalence relations. The approach tends to focus more directly on discovering the extensional properties of interest: for constant propagation it might operate on sets of values with constancy corresponding to singletons, and for neededness analysis it might perform a strictness analysis and use the strictness information for neededness (or make use of the "absence" notion from projection analysis and attempt to discover the di#erence). On the positive side, this usually gives rise to provably correct analyses, although there are sometimes complications (due to deciding what information to stick onto the
Keyphrases
program analysis extensional property sometimes complication strictness information strictness analysis domain theory neededness analysis abstract domain semantic correctness projection analysis formal justification di erence partial equivalence relation program transformation constant propagation positive side absence notion negative side
