Informatics 10 Years Back , 10 Years Ahead

Interpretation is a theory of approximation of the be— havior of dynamic discrete systems such as the formal seman­ tics ofprograms. Since such behaviors can be characterized by fixpoints, the theory essentially provides constructive and ef— fective methods for fixpoint approximation and checking by abstraction. Non—conventionalapplications of abstract interpretation in— clude the design of hierarchies of semantics at different levels of observation of program execution, the generation of heuris­ tics for search problems in artificial intelligence, program watermarking, etc. The origin and most well-known application of abstract in­ terpretation is program analysis that is the automatic static determination of dynamic run-time properties of programs. Formally, this subsumes and includes data flow analysis, set or constraint based analysis, type inference, etc. Program analy— sis techniques are used for compile—timeprogram optimization, partial evaluation, program abstract model—checking, abstract program debugging and testing and more generally semantic­ based software verificationand manipulation. In the past ten years, program analysis by abstract interpre— tation moved from an intensive research area to its first indus­ trial applications leading to the emergence of innovative startups. The impressive evolution of hardware by a factor of 106over the past 25 years has lead to a similar explosion of the size of programs. The size, complexity and scope ofapplication of large programs is likely to continue expanding rapidly in the next decade. These big programs will have to be modified and main— tained during their whole lifetimewhich often exceeds 20 years. The size and efficiency of the programming and maintenance teams in charge of their design and follow-up cannot grow up in similar proportions. At a not so uncommon rate of one bug per thousand lines such huge programs might rapidly become hardly manageable in particular for safety critical systems.