P. Sestoft. ML pattern match compilation and partial evaluation. In O. Danvy, R. Gluck, P. Thiemann (eds.), Partial Evaluation. Proceedings, LNCS 1110, 446--  Home/Search   Document Details and Download   Summary   Related Articles   Check

....decision trees, with a primary focus on reducing code size. It is unclear to us how to combine column and row optimization in practice and whether this would yield noticeable improvements or not. There also exists a partial evaluation based approach to pattern matching optimization. 8] and later [18] specialize an ultra naive pattern matching interpreter to create an e cient pattern matching compiler. Both authors use context information as we do. By contrast, their target is decision trees. In the end, the automatic process of partial evaluation does not nd as many optimizations as we do. ....

Sestoft, P. ML pattern match compilation and partial evaluation, 1996.

....a binding time monovariant o#ine specializer to the modified program. The overall e#ect of the transformation is that of a binding time polyvariant o#ine specializer, even though only a binding time monovariant o#ine specializer was used to produce the residual program. Similarly, it is known [21, 27, 30, 31] that optimizing translators can be generated from suitable interpreters. Such techniques can also be used in self interpreters to improve the specialization of programs. Theorems 1 and 2 carry over to the interpretive case, provided we replace in the rhs of their implication the quantification ....

P. Sestoft. ML pattern match compilation and partial evaluation. In O. Danvy, R. Gluck, P. Thiemann (eds.), Partial Evaluation. Proceedings, LNCS 1110, 446--464. Springer-Verlag, 1996.

.... handle Fail = Tp [ pn ; v) en) handle Fail = raise Match The simple translation given here does not make use of the many sophisticated techniques for improving the efficiency of pattern matching, such as jump tables, or Sestoft s technique for using information known from previous matches [13]. Adapting these techniques to handle recursive pattern definitions is one interesting area for further research. Tp : p Theta Vars) Gamma Exp Gamma Exp Tp [ e = e Tp [ v) r] e = Tp r e Tp [ x; v) r] e = Tp r [ let x = v in e] Tp [ fail; v) r] e = raise Fail] Tp ....

P. Sestoft. ML pattern match compilation and partial evaluation. In O. Danvy, R. Gluck, and P. Thiemann, editors, Partial Evaluation. International Seminar. Selected Papers, pages 446--464. Springer-Verlag, Berlin, February 1996.

....explicit type annotations. Channel types and channel kinds (input, output, local, etc. are checked, pattern matches are checked for exhaustiveness, redundancy, and so on. The program is simpli ed and compiled into the intermediate language Core: pattern matches are compiled into decision trees [18], record expressions a = 26, b = true and record patterns are replaced by tuples (26, true) and datatype constructors SOME( are replaced by integer tags (e.g. 1) The Core language is untyped, so that the typed constructed value SOME (true, 26) from PMC may be represented by the untyped ....

Peter Sestoft. ML pattern match compilation and partial evaluation. In Danvy, Gl#ck, and Thiemann, editors, Partial Evaluation. Dagstuhl Castle, Germany, February 1996, volume 1110 of LNCS, pages 446464. Springer-Verlag, 1996.

No context found.

P. Sestoft. ML pattern match compilation and partial evaluation. In O. Danvy, R. Gluck, P. Thiemann (eds.), Partial Evaluation. Proceedings, LNCS 1110, 446--

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC