The objective of this lab is to write an interpreter for a small,
untyped functional programming language.
The interpreter should walk through programs
and print out the value of the
Before the lab can be submitted, the interpreter has to pass some tests, which are given on the course web page via links later in this document.
The recommended implementation is via a BNF grammar processed by the BNF Converter (BNFC) tool. No type checker is needed.
The approximate size of the grammar is 15 rules, and the interpreter code should be about 100 lines, depending on the programming language used for the implementation. You can use this grammar if you want.
All BNFC supported languages can be used, but guidance is guaranteed only for Haskell and Java 1.5.
Makefile similar to lab2. The interpreter should
be compilable via calling
Calling the interpreter should work by the command
lab4 (-n|-v) <File>
-n forces call-by-name evaluation, the flag
-v forces call-by-value. The default, i.e. when no flag
is present, is call-by-value.
The language is the same as in lecture notes, Chapter 7.
The main category is
Program. A program is a sequence of definitions,
which are terminated by semicolons. A definition is a function name followed by
a (possibly empty) list of variable names followed by the equality sign
followed by an expression:
fun x1 ... xn = exp
fun and the variables
xn are lexically identifiers.
fun is the function to be defined, and
are its arguments. These variables are considered bound in
exp. Notice that
the all such definitions can be converted to definitions of just
a lambda abstraction over its arguments.
Expressions are of the following forms
Applications and operations are left-associative. Abstractions are right-associative.
The available operations are
+, -, <.
Here is an example of a program in the language:
-- example mult x y = if (y < 1) then 0 else if (y < 2) then x else (x + (mult x (y-1))) ; fact = \x -> if (x < 3) then x else mult x (fact (x-1)) ; main = fact 6 ;
Comments are line tails starting with
There is just one type of basic values: integers. Closures or abstraction expressions are also possible values of expressions.
Evaluation is parametrized so that it can be performed in both call-by-value and call-by-name manners.
The function defined in a definition is in scope in the entire program, including the expression part of that definition (which results in recursive and mutually recursive functions).
The variables bound on the left-hand-side of a definition are in scope in the expression part of the definition.
x in an abstraction
\x -> exp is bound in
the body of the abstraction, i.e.
Bindings made inside a scope overshadow those made outside.
+, -, < have their usual integer semantics.
< has value 1 if it is true, 0 if false.
if c then a else b is evaluated "lazily"
so that if
c has value 0,
b is evaluated, otherwise
a is evaluated.
The output of a program is the value of the
main function, and
it must be an integer.
A program may also exit with an error, due to an unbound identifier.
It should then say what identifier is unbound. It is also an error
main function is missing or has wrong type.
Arithmetic operations on non-integers are also errors, e.g.
f x = x + x ; main = f + f ;
All these errors occur at run time, because there is no type checker.
-- file good.fun mult x y = if (y < 1) then 0 else if (y < 2) then x else (x + (mult x (y-1))) ; fact = \x -> if (x < 3) then x else mult x (fact (x-1)) ; main = fact 6 ;
Running the interpreter
./lab4 good.fun 720
-- file bad.fun mult x y = if (y < 1) then 0 else if (y < 2) then x else (x + (mult x (y-1))) ; fact = \x -> if (x < 3) then x else mul x (fact (x-1)) ; main = fact 6 ;
Running the interpreter
./lab4 bad.fun ERROR: unknown identifier mul
-- file infinite.fun grow x = 1 + grow x ; first x y = x ; main = first 5 (grow 4) ;
Running the interpreter
./lab4 infinite.fun <infinite loop> ./lab4 -n infinite.fun 5
The interpreter is submitted as source files that can be compiled by typing
Run the programs in the test suite before submitting the lab.
Include a log on the test run, showing the call of
lab4 for every program
in the testsuite.
The interpreter must give acceptable results for the test suite and meet the specification in this document in all respects.
All "good" programs must work with at least one of the evaluation strategies; need not work on both (because of loop or long time); see comments in test programs to see which one is expected to work.
The solution must be written in an easily readable and maintainable way.
Submit your lab by using Fire. Please include exactly all the files that are required for building your solution, including a Makefile. Do not however submit any generated files, and kindly avoid using archives (upload each file individually).