Personal tools

Lucid

From HaskellWiki

(Difference between revisions)
Jump to: navigation, search
m (Fast-food: small rephrasings at the explanatory texts of the ``moving average'' example)
(Fast-food: Using terms ``syncronous'' vs ``asynchronous'')
Line 60: Line 60:
 
In fact, <code>first</code>, <code>next</code>, <code>fby</code> are not necessarily primitives in Lucid, but this fast-food implementation treats them as primitives.
 
In fact, <code>first</code>, <code>next</code>, <code>fby</code> are not necessarily primitives in Lucid, but this fast-food implementation treats them as primitives.
   
The above example may suggest that Lucid reacts on inputs in a one-output-for-each-input way. This is not the case: Lucid has general, unrestricted stream processing abilities.
+
The above example may suggest that Lucid be necessarily synchronous -- it reacts on inputs in a one-output-for-each-input way. This is not the case: Lucid has general, unrestricted stream processing abilities (thus able to implement also asynchronous processes).
   
 
=== Arrow ===
 
=== Arrow ===

Revision as of 21:01, 16 June 2006

Contents


1 Introduction

Weblogs Forum: Fluid Programming in Lucid -- good examples, and very good links.

Lucid page written by Bill Wadge, including an excerpt from Lucid Primer book.

Wikipedia article (yet a stub, but useful links).

2 Details

The most detailed online material on the details of this language I could find: Raganswamy Jagannathan, Chris Dodd. GLU programmer's guide (downloadable as the 4th paper of GIPSY Publications). The first pages seem to discuss other problems, but it is worth of reading further, because a detailed description of Lucid's syntax and semantics is hiding inside this paper (section 3.1.2 on pages 22--38). This paper is part of the Gipsy Project Home Page (GIPSY: A General Intensional Programming System).

3 Implementation

3.1 Fast-food

A fast-food implementation:

written to mimick and test the following Lucid example (I have made it a little Haskell-like -- it is not original Lucid syntax, but it is not an embedding into Haskell language, either -- in fact, it is a didactic intermediate syntax):

average n = sum n / card
sum n = first n `fby` (sum n + next n)
card = 1 `fby` succ card

This program computes the moving average of the input stream:

Snapshot Input Output
0  
1 10 10
2 10, 20 10, 15
3 10, 20, 30 10, 15, 20
4 10, 20, 30, 40 10, 15, 20, 25
5 10, 20, 30, 40, 50 10, 15, 20, 25, 30

In fact, first, next, fby are not necessarily primitives in Lucid, but this fast-food implementation treats them as primitives.

The above example may suggest that Lucid be necessarily synchronous -- it reacts on inputs in a one-output-for-each-input way. This is not the case: Lucid has general, unrestricted stream processing abilities (thus able to implement also asynchronous processes).

3.2 Arrow

A better implementation could be written using the arrow library available at the Downloads page of Arrows: A General Interface to Computation written by Antony Courtney, Henrik Nilsson and Ross Paterson.

3.3 Comonad

A professional approach based on comonads: see The Essence of Dataflow Programming paper written by Tarmo Uustalu. Comments on this paper can be found on Lambda the Ultimate (but the link to the mentioned given there seems to be broken, I have given an updated one here). The links to pages introducing the concept of comonad are Comonads and Haskell written by Einar Karttunen.