Slicing and Functional Programming Simon Thompson University of - PowerPoint PPT Presentation

Slicing and Functional Programming Simon Thompson University of Kent

Haskell and Erlang • Pure language … • Impure language … • … but has monads • … but single assignment • Strongly typed • Weakly typed • Lazy evaluation • Strict evaluation • Sequential core • Concurrent

Concurrency in Erlang -module(echo). -export([go/0, loop/0]). go() -> loop() -> Pid = spawn(echo, loop, []), receive Pid ! {self(), hello}, {From, Msg} -> receive From ! {self(), Msg}, {Pid, Msg} -> loop(); io:format("~w~n",[Msg]) stop -> end, true Pid ! stop. end.

Laziness in Haskell if True t f = t • Arguments are only if False t f = f evaluated when needed. if (2>3) ⊥ 4 ↝ 4 • Arguments are only sft (a:b:xs) = a+b evaluated to the extent ones = 1:ones that is needed for sft ones ↝ 2 evaluation to proceed.

Slicing parseMessage :: MessageList -> (Message, MessageList) parseMessage [] = ([], []) parseMessage xs = (takeWhile (/= ’&’) (tail ys), dropWhile (/= ’&’) (tail ys) ) where ys = dropWhile (/= ’&’) xs parseMsgL :: MessageList -> Message parseMsgL [] = [] parseMsgL xs = takeWhile (/= ’&’) (tail (dropWhile (/= ’&’) xs))

Clone detection • Search for common generalisation. • What about insertion, deletion or permutation of statements? • Can slicing help?

Ideas • Slicing for debugging functional programs. • Slicing components of complex structures. • … and others … ?