Publicação
Design, implementation and calculation of circular programs
| Resumo: | Circular programming is a powerful technique to express multiple traversal algorithms as a single traversal function in a lazy setting. Such a (virtual) circular program may contain circular definitions, that is, arguments of function calls that are also results of that same calls. Although circular definitions always induce non-termination under a strict evaluation mechanism, they can sometimes be immediately evaluated using a lazy evaluation strategy. The lazy engine is able to compute the right evaluation order, if that order exists. Indeed, using this style of circular programming, the programmer does not have to concern him/herself with the definition and the scheduling of the different traversal functions, since a single (traversal) function has to be defined. Moreover, because there is a single traversal function, the programmer does not have to define intermediate gluing data structures to convey values computed in one traversal and needed in following ones, either. In this Thesis, we present our studies on the design, implementation and calculation of circular programs. We start by developing techniques to transform circular programs into strict ones. Then, we introduce calculation rules to obtain circular programs from strict equivalents, both in the context of pure and monadic programming. Because we use calculation techniques we guarantee that the resulting circular programs are equivalent to the strict ones we start with. In this Thesis, we also perform a series of benchmarks comparing the running performances of circular programs and the programs we are able to derive from circular programs. |
|---|---|
| Autores principais: | Fernandes, João Paulo Sousa Ferreira |
| Ano: | 2009 |
| País: | Portugal |
| Tipo de documento: | tese de doutoramento |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | Circular programming is a powerful technique to express multiple traversal algorithms as a single traversal function in a lazy setting. Such a (virtual) circular program may contain circular definitions, that is, arguments of function calls that are also results of that same calls. Although circular definitions always induce non-termination under a strict evaluation mechanism, they can sometimes be immediately evaluated using a lazy evaluation strategy. The lazy engine is able to compute the right evaluation order, if that order exists. Indeed, using this style of circular programming, the programmer does not have to concern him/herself with the definition and the scheduling of the different traversal functions, since a single (traversal) function has to be defined. Moreover, because there is a single traversal function, the programmer does not have to define intermediate gluing data structures to convey values computed in one traversal and needed in following ones, either. In this Thesis, we present our studies on the design, implementation and calculation of circular programs. We start by developing techniques to transform circular programs into strict ones. Then, we introduce calculation rules to obtain circular programs from strict equivalents, both in the context of pure and monadic programming. Because we use calculation techniques we guarantee that the resulting circular programs are equivalent to the strict ones we start with. In this Thesis, we also perform a series of benchmarks comparing the running performances of circular programs and the programs we are able to derive from circular programs. |
|---|