🚀 go-pugleaf

lgm@cbnewsc.ATT.COM (lawrence.g.mayka) writes:

>In article <1990Jan5.084746.17836@cs.eur.nl> reino@cs.eur.nl (Reino de Boer) writes:
>>	all interpreters (CL, Smalltalk, etc.) by definition do only
>>	run-time checking (not being able to compile at all).

>Common Lisp programs are typically compiled into native machine
>instructions. 
>An interpreter is usually also available but most CL programmers
>use it primarily to execute top-level environment commands such as
>ED, COMPILE-FILE, LOAD, TRACE, DESCRIBE, INSPECT, and DISASSEMBLE.

The point I was trying to make, is that an interpreter can BY DEFINITION
do no compile-time checking.
Some `compilers' allow a form of run-time checking because they include
a run-time interpreter in the compiled code. I would like to classify
these compilers under interpreters.
Some `interpreters' allow a form of compile-time checking because they
interpret some kind of internal code. Depending on the semantics of the
internal code, these could be classified under compilers.

The only reasonable way for a compiler to allow run-time type checking
would be to include type-checking code in the compiled code.

However, there is more than one issue being discussed here:
1. What would be reasonable definitions for
   o run-time type checking;
   o compile-time type checking.
2. Given a solution to (1), what are the advantages and disadvantages of
   both kinds of type-checking.
3. Given an answer to (2),
   o what would be the best way to let a programmer influence the
	 checking mechanism (e.g., type declarations as in Pascal, or
	 polymorphic typing as in Miranda)
   o does type checking have to be `strong' or `weak' (and what would be
	 suitable definitions for `strong' and `weak').
	 Example:
	 Strong type checking in Pascal would give the programmer the
	 opportunity to get a compiler error on the following:
	 program wrong( output );
	 type distance = real;
		  money    = real;
	 var  d        : distance;
		  m        : money;
	 begin d := 5.5; m := 10.0; writeln( d + m : 10 : 2 ) end.

Any comments? -- Reino

-- 
Reino R. A. de Boer
Erasmus University Rotterdam ( Informatica )
e-mail: reino@cs.eur.nl