Course Notes on Formal Languages and Compilers

10. PROGRAMMING LANGUAGE IMPLEMENTATION ISSUES
51
5
10
15
20
25
30
35
A high level rogramming language has a hierarchic structure for statements
expressions and definitions of functions, procedures, data and their types.
Assembly language has a sequential structure (except for expressions) and
does not support a large variety of data types. Machine language is
completely sequential in structure and has no concept at all of data type.
There is therefore a large gap between a high level language and machine
or and the main problem to be handled is this large gap. It is for this reason
that there are many phases in compilation and interpretation, where each
phase handles a part of the problem. The multiphase design makes also it
easier to transfer the compiler to another machine. This is because the
code generation phase and the interpretation phase are the only really
machine dependent phases in the compiler.
Let us now give examples of different programming language features,
associated problems and some solutions.
Code generation or interpretation - Associating properties with names
Recall first the distinction between identifier and name, that is names are
unique but identifiers can have multiple definitions. Certain languages (such
as Fortran IV) allow determining all the properties of a name at compile
turns e.g. type of data, its size, and its absolute run time address. Other
languages such a Pascal, need the absolute address to be determined at
run time even though sizes are all known. At compile time, relative
addresses are known e.g. relative to top of a stack.
Other languages such as ADA, the size of data is not known and even this
needs to be determined at run time e.g. on block entry. In languages such
as LISP and PROLOG the type may only be known at run time. In view of
these approaches to languages, different implementations are used. Code
generation can be used when the types of the data are known at compile
time but is much harder or impossible to use when the data types are
known at run time, and interpretation is a simpler solution in this case.
Type declarations and checking
Here are three possibilities for programming languages.
40
45
Strongly typed language: Type declarations and compile time type
checking. e.g. Pascal, ADA
Weakly typed language: Type declarations but no type checking.
e.g. assembly language, C.
Utyped language: No type declarations but run time type checking.
e.g. LISP, PROLOG