[r6rs-discuss] [Formal] Allow compilers to reject obvious violations

[Anton van Straaten]

Joe Marshall wrote:
> On 2/24/07, John Cowan <cowan at ccil.org> wrote:
> 
>> Matthias Felleisen scripsit:
>>
>> >  1. When an implementation is allowed to reject a program without
>> > running it, it comes with a filter also known as a type system.
>>
>> I think that rather stretches the definition of "type system".
> 
> 
> It doesn't, actually, if you use the terminology of the static type
> community.  A `type system' would encompass any system that
> performs analysis on the program without actually attempting to
> execute it.

This assumes that you actually have "a program".  Type theorists don't 
usually refer to syntax errors as type errors.

John Cowan's proposed text (if (foo)) can be rejected as not being a 
valid program without involving a type system, so even without using a 
stretchy definition of "type system", it doesn't contradict Matthias' 
statement.

Anton