Re: Clause "with and use"

[18k11tm001@sneakemail.com (Russ)]

Vinzent 'Gadget' Hoefler <ada.rocks@jlfencey.com> wrote in message news:<borog2$1fpdhl$1@ID-175126.news.uni-berlin.de>...
> Gautier Write-only wrote:
> 
> >Vinzent 'Gadget' Hoefler:
> >
> >> For *sale*. Exactly. BTW, I still don't understand what is redundant
> >> on with and use. I don't use "use" (well, at least not globally).

> >In this case, effectively, you are not likely to see a redundance.
> 
> Hmm. True. Good point. :-)

I say it's redundant because it doesn't make sense to have "use"
without a corresponding "with". Let's say "with" is level 1, and "use"
is level 2. If you're at level 2, that implies you're already past
level 1, so specifying both level 1 and level 2 is redundant. It's
like saying that the Florida Marlins won the World Series, then also
saying that they won the NL Championship. The former implies the
latter and is therefore redundant. It belongs in the Department of
Redundancy Department.

I appreciate your discipline in shunning "use," but I assume the
language provides it for a good reason. For basic I/O and math, for
example, I think "use" is perfectly appropriate. I'm sorry, but "sin"
and "cos" will always be just "sin" and "cos" to me, and I prefer to
see them that way. (I gave an example on this forum a few months ago
of great-circle equations involving no less than 10 math function
calls in a single expression.)

> >In your place I wouldn't worry at all about this proposal.
> 
> Hey, it's still *my* language! Or so. ;)
> 
> Ok, seriously, being able to use "use" without its corresponding with
> in one case and differently in another looks inconsistent for me.

I think the cleanest approach is to let "use" imply "with". What does
that mean? If the compiler sees "use x", but x is not yet "in scope",
let the compiler simply pretend that the line "use x" had been
preceded by "with x".

Now, admittedly, that involves some semantic subtleties because the
"withed" package spec need be identical to the "used" spec (due to the
conventions for child packages). I suspect that any potential
ambiguity is easily resolvable with simple conventions, but I could be
wrong. OK, suppose I am wrong. Then the fallback position is to allow
"with and use x" or "with/use x", as previously suggested. That
removes all semantic ambiguity, and is logically equivalent to
separate "with" and "use" clauses, but without the repetitious
clutter.

> To bring it to the point and the way I see it: "with" is for
> thecompiler/linker to state dependencies, use is for the programmer.
> For me that's two different things and I wouldn't like them to get
> mixed up.

Well yes, but ... "with" tells the compiler which packages will be
used (ironically), and "use" tells the compiler how the elements of a
package will be referred to, so it's for the compiler too.

For those who insist that this whole issue is "minor" or even
"trivial", here's what I have to say to you. You're all idiots and
dumbsh*ts. Relax, I'm just kidding. You are completely right. It *is*
trivial. But as I said before, it's like the clutter in the front yard
of a house for sale: it can do harm to the market value of Ada, and
it's trivial to clean up too. For crying out loud, just do it and be
done with it.