Results 1 
3 of
3
The Calculation of a Polytypic Parser
, 1996
"... In this paper it is shown how inverses can be used to calculate a parser. A polytypic unparser is given and by using rules for calculating inverses a polytypic parser is calculated from it. It can be instantiated automatically for all data types that can be described by a regular functor. The idea t ..."
Abstract

Cited by 5 (0 self)
 Add to MetaCart
In this paper it is shown how inverses can be used to calculate a parser. A polytypic unparser is given and by using rules for calculating inverses a polytypic parser is calculated from it. It can be instantiated automatically for all data types that can be described by a regular functor. The idea that a parser can be calculated as the inverse of an unparser is not new, but because polytypical functions are used here the calculated parser is very general. Inverses are defined in a general way and rules are given to calculate them. The set monad has a strong connection with inverses and for many monadic concepts the instantiation with this monad gives rise to rules about inverses. In this way the inverses of catamorphisms and anamorphisms can be characterized. As we know that the unparser and the rules that were used in the calculation are correct, the calculated parser is known to be correct too. In general the parser that results from such a calculation is not very efficient and it is possible to construct much more efficient parsers by hand. Because it is possible to prove the equality of these two parsers, this parser is correct too. An implementation of parsers for a small subset of html and latex is given as an illustration of how the polytypic functions are instantiated for a particular datatype. Contents 1
Relational Programming, Program Inversion and the Derivation of Parsing Algorithms
, 1993
"... The spec calculus is a programming algebra that supports and even encourages the derivation of relational programs in a clear and calculational style. After a brief introduction of the spec calculus two particular uses are investigated. First, the notion of program inversion is introduced. This noti ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
The spec calculus is a programming algebra that supports and even encourages the derivation of relational programs in a clear and calculational style. After a brief introduction of the spec calculus two particular uses are investigated. First, the notion of program inversion is introduced. This notion from imperative programming allows one to construct from a given program R that satisfies fPg R fQg a program R 1 that satis es fQg R 1 fPg . It is shown that, as may be expected, a relational framework allows easy definition of and calculation with inverses of arbitrary programs. Several examples are discussed and used to introduce different methods to calculate implementable inverses. The second topic is the derivation of precedence parsing algorithms. A simple algorithm for the calculation of the inorder traversal of a tree is transformed in two steps into an algorithm for precedence parsing. The first transformation is the use of program inversion to obtain an algorithm that constru...
Using underspecification in the derivation of some optimal partition algorithms
, 1990
"... Indeterminacy is inherent in the specification of optimal partition (and many more) algorithms, even though the algorithms themselves may be fully determinate. Indeterminacy is a notoriously hard phenomenon to deal with in a purely functional setting. In the paper “A Calculus Of Functions for Progra ..."
Abstract

Cited by 1 (0 self)
 Add to MetaCart
Indeterminacy is inherent in the specification of optimal partition (and many more) algorithms, even though the algorithms themselves may be fully determinate. Indeterminacy is a notoriously hard phenomenon to deal with in a purely functional setting. In the paper “A Calculus Of Functions for Program Derivation ” R. S. Bird tries to handle it by using underspecified functions. (Other authors have proposed to use ‘indeterminate ’ functions, and to use relations instead of functions.) In this paper we redo Bird’s derivation of the Leery and Greedy algorithm while being very precise about underspecification, and still staying in the functional framework. It turns out that Bird’s theorems are not exactly what one would like to have, and what one might understand from his wording of the theorems. We also give a derivation in the BirdMeertens style of a (linear time) optimal partition algorithm that was originally found by J. C. S. P. van der Woude. 1