From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,INVALID_MSGID
	autolearn=no autolearn_force=no version=3.4.4
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 103376,7f4d16c4ee371eb5
X-Google-Attributes: gid103376,public
From: Paul Graham <pgraham@cadence.com>
Subject: Re: Why is it Called a Package?
Date: 2000/04/07
Message-ID: <38EE12F4.7972A964@cadence.com>#1/1
X-Deja-AN: 608059361
Content-Transfer-Encoding: 7bit
References: <38DF7F38.8D656ABD@lmtas.lmco.com>
 <38DFB0BC.9FF72EFC@callnetuk.com> <wcc3dpcnkcd.fsf@world.std.com>
 <87u2hq857e.fsf@deneb.cygnus.argh.org> <wcc4s9pr8o6.fsf@world.std.com>
 <38E2A4A4.E59E997C@research.canon.com.au> <wccya70bg8q.fsf@world.std.com>
 <Pine.BSF.4.21.0004061305320.6588-100000@shell5.ba.best.com>
 <8ck2mt$jrn$1@clnews.edf.fr>
X-Accept-Language: en
Content-Type: text/plain; charset=us-ascii
Organization: Cadence Design Systems
Mime-Version: 1.0
Newsgroups: comp.lang.ada
Date: 2000-04-07T00:00:00+00:00
List-Id: <comp.lang.ada>

Pascal Obry wrote:
> I don't think we want that. A very nice consequence here is that you can
> easily change an abstraction from:
> 
>    package P is
>       Some_Value : array (1 .. 10) of Integer;
>       --  first quick implementation using basic array
>    end P;
> 
> to
> 
>    package P is
>       function Some_Value (N : in Positive) return Integer;
>       --  real implementation using a complex structure
>    end P;
> 
> (or the other way around) without modifying all client code.

What about:

    x := Some_Value(1 .. 3);

Changing the implementation from an array to a function is not so easy
here.

If for efficiency reasons you want to re-implement a function as an
array lookup, why not just write a public function as an interface to a
private array, then inline it?

I've always thought that using [] for indexing, slicing, and aggregates
would have made the language much simpler.  How many times have you Ada
experts had to explain to someone that (1) is not a legal array
aggregate?

Another thing that has always bothered me is the omission of parentheses
in a subprogram without arguments.  It results in some peculiar corner
cases in the language.  For instance, sometimes f.x means variable x
declared within function f, and sometimes it is a call to f returning a
record (or an access to record :-) with element x selected.  And I seem
to remember some ambiguities with attributes of sbprograms: is
f'some_attribute an attribute of the function f or of the value returned
by f.  Sure, the LRM has rules to resolve these ambiguities, but why
introduce the ambiguities in the first place?
 
Paul