Subj : Re: Do all programming languages use files?
To   : comp.programming
From : Chris Sonnack
Date : Fri Sep 09 2005 07:06 pm

Michael Wojcik writes:

>>> The Standard defines the language; the Standard includes the standard
>>> library (and no others); the standard library includes file I/O; thus
>>> the language includes file I/O (for hosted implementations).  Your
>>> argument is utterly without merit.
>> 
>> 0. Community begins with *trying* to see the other guy's point.
> 
> I've seen it.  It's wrong.

In YOUR opinion using YOUR definition.  As you say above, the standard
defines the language and **includes** the standard **library**, so the
standard **also** defines that library.

Repeat: defines the **library**.

It is separate--even in the standard--from the "language", per se.

>> I do see a difference between the syntactical definition of a
>> language and its, for lack of a better word, "semantic" definition.
> 
> That's a valid argument.

As I read Scott's argument, that was essentially his point.

> "All libraries are equal in their degree of inclusion in the C
> language" is not, and that's the argument I was referring to.

I never got that sense from it.  I just went back and read all of his
posts in this thread, and I STILL don't get that sense from them.

>> And if you good look at the reserved keywords for C, you won't find
>> fprintf, printf, fputs or any other I/O family members.  Nor will
>> you find stdout or stdin.
> 
> I'm well aware of that, since I have a copy of the standard right
> by my desk.  "fprintf" is not a keyword.  It is, however, a reserved
> identifier, if stdio.h is included, so I don't find the "not a
> keyword" argument compelling.

Fine, but--again--that's YOUR OPINION.  Mine differs.

> "main" is not a keyword, either; does that make it not part of the
> C language?

By *my* (and apparently Scott's) definition of "language", no.

If you re-phrase the question to ask if that makes it part of the
C standard, then yes.

> "include" is not a keyword (C90 6.1.1); does that mean "#include"
> is not part of C?

The *language* or the *standard*?

And in any event, no--it's a preprocessor instruction.

Maybe another way to illustrate this is to point out that you can write
a 100% standard-compliant C compiler that knows *nothing* about #include
or fprintf.  You cannot, however, produce a C *implementation* that is
ignorant of them.

Now, if I can write such a compiler, how can fprintf be a part of the
language?

-- 
|_ CJSonnack <Chris@Sonnack.com> _____________| How's my programming? |
|_ http://www.Sonnack.com/ ___________________| Call: 1-800-DEV-NULL  |
|_____________________________________________|_______________________|

.