Hacker News Reader

>> Just try writing (without hacks/cuts) a run-length encoder that produces only one answer ([9,9,9,8,8,7] -> [[9,3], [8,2], [7,1]]) and works forwards and backwards and then tell me the execution model is not a limitation.

But, why not use the cut? It's a feature of the language and it's there exactly so you don't have to bust your head trying to find a way to do things the hard way.

I honestly don't understand what's the fuss with the poor cut. If you want an efficient language that runs on a computer then you have to be pragmatic and recognise that concessions have to be made to accommodate the limitations of the hardware. If you want something pure and ideal, then you should probably not be trying to do that on a computer. You should probably not be trying to do it at all. There are no pure formal systems, or if there are, we can't explain or understand them because they can't communicate with the external world (hey, no side effects, OK?). See Haskell, for instance. The article above says something really strange about Haskell, that it:

>> ... finally broke through the wall of pure-functional programs being seen as ivory-tower languages that were impractical, or at least really awkward, for any system that needed to do I/O or interact with users

Which to me is insane. I think it's talking about monads. Monads, as an escape from the ivory tower, impractical, or at least awkward formalisms of... what, LISP? I mean for the love of dogs.

Prolog is a balanced language. People complain about it because it's not unbalanced, that's the problem. If it were 100% declarative and nobody could program anything in it, it would have legions of dotting admirers praising it for its perfection, and whining because they have to do all their work in languages that are hardly 10% declarative, like Rust, or Python or Ruby (and I'm being generous). People have to get their head around this: our computer science is primitive and it is full of holes, dark pits of despair and undecidability, intractability, and terminal inefficiency. To be practical, a language has to give up a little part of its true, pure self, that it could achieve in a perfect world. So Prolog has the cut, and you can use it to control its backtracking, which you need because it's implemented as a depth-first search, because that's efficient and convenient. So use the cut, it's a feature.

View on HN

%! run_length_encoding(+List, -Run_length_encoding) is det. % % Run_length_encoding is a list of all key-value pairs where each % key is an element in List and value the number of consecutive % repetitions of that element up to the first differing element. % % For example: % == % ?- run_length_encoding([a,a,a,b,b,b,b,b,b,c,c,c,a,a,f], RLE). % RLE = [a-3, b-6, c-3, a-2, f-1]. % == % run_length_encoding([], []-0):- !. run_length_encoding([H|List], Run_length_encoding):- run_length_encoding(List, [H-1], Run_length_encoding). % Last element in input list or single-element input list. run_length_encoding([], [C-F|Fs], Run_length_encoding):- reverse([C-F|Fs], Run_length_encoding). % Run of N consecutive identical elements run_length_encoding([C|Cs],[C-F|Fs], Acc):- ! % Orange ish; backtracking will produce successive % counts of repetitions of C from different indices in the list % I think. ,F_ is F + 1 ,run_length_encoding(Cs, [C-F_| Fs], Acc). % End of run of N consecutive identical elements. run_length_encoding([C|Cs], Fs, Acc):- run_length_encoding(Cs,[C-1|Fs], Acc).

rle2([], []). rle2([H],[[H, add1(zero)]]). rle2([H | T], [[H, CountPlus1] | More]) :- Count = add1(_), CountPlus1 = add1(Count), rle2(T, [[H, Count] | More]). rle2([H | T], [[H, add1(zero)], [Element, Count] | More]) :- Count = add1(_), dif(Element, H), rle2(T, [[Element, Count] | More]). ?- rle2([1,1,1,2,2,3],X), rle2(Y,X). X = [[1, add1(add1(add1(zero)))], [2, add1(add1(zero))], [3, add1(zero)]], Y = [1, 1, 1, 2, 2, 3] ; false.

%! peano(?Arabic,?Peano) is nondet. % % Convert between Arabic and Peano number notations. % % Use this program to quickly convert between Peano notation and % Arabic notation. % % Accepted modes are (+,-) and (-,+). In mode (?,?) this predicate % returns only 0 deterministically. % % Examples: % == % ?- peano(1,P). % P = s(0). % % ?- peano(I,s(0)). % I = 1. % % ?- peano(N,P). % N = P, P = 0. % == % peano(N,P):- peano(P,0,N). peano(0,N,N):- !. peano(s(P),N,Acc):- succ(N,N_) ,peano(P,N_,Acc).