Lucid (programming language)
Paradigm | Dataflow |
---|---|
Designed by | Edward A. Ashcroft, William W. Wadge |
Developer | Wadge |
First appeared | 1976 |
Typing discipline | Typeless |
Major implementations | |
pLucid | |
Dialects | |
GIPSY, Granular Lucid | |
Influenced by | |
ISWIM | |
Influenced | |
SISAL, PureData, Lustre |
Lucid is a dataflow programming language. It is designed to experiment with non-von Neumann programming models. It was designed by Bill Wadge and Ed Ashcroft and described in the book Lucid, the Dataflow Programming Language.
Model
Lucid uses a demand-driven model for data computation. Each statement can be understood as an equation defining a network of processors and communication lines between them through which data flows. Each variable is an infinite stream of values and every function is a filter or a transformer. Iteration is simulated by 'current' values and 'fby' (read as 'followed by') operator allowing composition of streams.
Lucid is based on an algebra of histories, a history being an infinite sequence of data items. Operationally, a history can be thought of as a record of the changing values of a variable, history operations such as first and next can be understood in ways suggested by their names. Lucid was originally conceived as a kind of very disciplined, mathematically pure, single-assignment language, in which verification would be very much simplified. However, the dataflow interpretation has been a very important influence on the direction in which Lucid has evolved.
Details
In Lucid (and other dataflow languages) an expression that contains a variable that has not yet been bound waits until the variable has been bound, before proceeding. An expression like x + y
will wait until both x and y are bound before returning with the output of the expression. An important consequence of this is that explicit logic for updating related values is avoided, which results in substantial code reduction, compared to mainstream languages.
Each variable in Lucid is a stream of values. An expression n = 1 fby n + 1
defines a stream
using the operator 'fby' (a mnemonic for "followed by"). fby defines what comes after the previous
expression. (In this instance the stream produces 1,2,3,...).
The values in a stream can be addressed by these operators (assuming x is the variable being used):
'first x'
- fetches the first value in the stream x,
'x'
- the current value of the stream,
'next x'
- fetches the next value in the stream.
'asa'
- an operator that does some thing 'as soon as' the condition given becomes true.
'x upon p'
- upon is an operator that repeats the old value of the stream x, and updates to the new values only when the stream p makes a true
value available. (It serves to slow down the stream x)
i.e.: x upon p
is the stream x with new values appearing upon the truth of p.
The computation is carried out by defining filters or transformation functions that act on these time-varying streams of data.
pLucid was the first interpreter for Lucid.
Examples
Factorial
fac where n = 0 fby (n + 1); fac = 1 fby ( fac * (n + 1) ); end
Fibonacci sequence
fib where fib = 0 fby ( 1 fby fib + next fib ); end
Total of a Sequence
total where total = 0 fby total + x end;
Running Average
running_avg where sum = first(input) fby sum + next(input); n = 1 fby n + 1; running_avg = sum / n; end;
Prime numbers
prime where prime = 2 fby (n whenever isprime(n)); n = 3 fby n+2; isprime(n) = not(divs) asa divs or prime*prime > N where N is current n; divs = N mod prime eq 0; end; end
Dataflow diagram
Quick sort
qsort(a) = if eof(first a) then a else follow(qsort(b0),qsort(b1)) fi
where
p = first a < a;
b0 = a whenever p;
b1 = a whenever not p;
follow(x,y) = if xdone then y upon xdone else x fi
where
xdone = iseod x fby xdone or iseod x;
end
end
Data flow diagram
--------> whenever -----> qsort --------- | ^ | | | | | not | | ^ | |---> first | | | | | | | V | | |---> less --- | | | | | V V ---+--------> whenever -----> qsort -----> conc -------> ifthenelse -----> | ^ ^ | | | --------> next ----> first ------> iseod -------------- | | | -----------------------------------------------------------
Square root
sqroot(avg(square(a))) where square(x) = x*x; avg(y) = mean where n = 1 fby n+1; mean = first y fby mean + d; d = (next y - mean)/(n+1); end; sqroot(z) = approx asa err < 0.0001 where Z is current z; approx = Z/2 fby (approx + Z/approx)/2; err = abs(square(approx)-Z); end; end
Hamming problem
h where h = 1 fby merge(merge(2 * h, 3 * h), 5 * h); merge(x,y) = if xx <= yy then xx else yy fi where xx = x upon xx <= yy; yy = y upon yy <= xx; end; end;
Dataflow Diagram
References
- Wadge, William W.; Ashcroft, Edward A. (1985). Lucid, the Dataflow Programming Language (PDF). Academic Press. ISBN 0-12-729650-6. Retrieved 8 January 2015.
External links
- pLucid
- Archived May 14, 2011, at the Wayback Machine.
- Archived June 29, 2007, at the Wayback Machine.
- Programming in Lucid
- Archived October 5, 2000, at the Wayback Machine.