comp.lang.ada
 help / color / mirror / Atom feed
From: "Marin David Condic" <marin.condic.auntie.spam@pacemicro.com>
Subject: Re: Ada Generic vs. C++ Templates
Date: Thu, 5 Apr 2001 17:04:08 -0400
Date: 2001-04-05T21:04:09+00:00	[thread overview]
Message-ID: <9aimk9$h99$1@nh.pace.co.uk> (raw)
In-Reply-To: wccsnjnjdev.fsf@world.std.com

I don't know why the actual hardware would have to be limited at all. The
implementation is pretty much free to define what parts of the actual
hardware it wants to utilize, isn't it? I recall Ada implementations that
did not take advantage of all the address space that may have been available
to them, so why couldn't an implementation limit its address space to one
byte? (Or just enough bytes to write "Program Error!" onto the screen?)
Worse comes to worse, I go off and invent "Marin's Virtual Machine" that
executes "Marin's Byte Code" - which has the quality of a very, very,
limited instruction set & memory.

Of course, there are other ways of building an unusable compiler by properly
defining implementation defined limits. How about a line length of 1? Or
zero?

Its a silly notion, but only brought up to illustrate that a legal
implementation of Ada can be totally useless.

Why wouldn't it (legally speaking) pass validation? It would produce correct
results, would it not? (In the sense that all programs would result in an
outcome that is legal within the definition of the language.)

Practically speaking, I'm sure that the keepers of the validation suite
would have better things to do than waste their time getting into the
Guinness Book Of World Records for validating the world's most useless
compiler. But that's not the same thing as saying there is a technical,
legal reason why such a compiler wouldn't pass.

MDC
--
Marin David Condic
Senior Software Engineer
Pace Micro Technology Americas    www.pacemicro.com
Enabling the digital revolution
e-Mail:    marin.condic@pacemicro.com
Web:      http://www.mcondic.com/


"Robert A Duff" <bobduff@world.std.com> wrote in message
news:wccsnjnjdev.fsf@world.std.com...

> Yes, it's perfectly legal (presuming the target machine really is that
> limited).
>
> But no, it won't get validated.
>





  reply	other threads:[~2001-04-05 21:04 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2001-04-02 14:28 Ada Generic vs. C++ Templates Josef Widder
2001-04-02 14:38 ` Ted Dennison
2001-04-02 20:57   ` Francois Godme
2001-04-02 21:26     ` Ted Dennison
2001-04-03  0:53       ` David Starner
2001-04-04  7:12         ` Pascal Obry
2001-04-04 12:37         ` Stephen Leake
2001-04-04 14:16         ` Ted Dennison
2001-04-03 22:09       ` Francois Godme
2001-04-04 16:17         ` Brian Rogoff
2001-04-04 16:21         ` Jeffrey Carter
2001-04-04 16:49           ` Ayende Rahien
2001-04-05 22:31             ` Colin Paul Gloster
2001-04-04 16:56           ` Ted Dennison
2001-04-04 17:02             ` Ayende Rahien
2001-04-05  0:35         ` James Rogers
2001-04-05  3:38           ` DuckE
2001-04-05 14:25             ` Marin David Condic
2001-04-05 20:32               ` Robert A Duff
2001-04-05 21:04                 ` Marin David Condic [this message]
2001-04-06 16:19                   ` Robert A Duff
2001-04-06  0:37             ` James Rogers
2001-04-06 10:38               ` Colin Paul Gloster
2001-04-11  3:33                 ` Stephen Howe
2001-04-11 14:33                   ` Colin Paul Gloster
2001-04-04 13:24 ` Georg Bauhaus
2001-04-05  8:46   ` Jean-Marc Bourguet
2001-04-04 17:30 ` Ehud Lamm
2001-04-05 22:04   ` Colin Paul Gloster
2001-04-10  7:03     ` Simon Wright
2001-04-13 14:11       ` Pat Rogers
2001-04-17  8:27         ` Colin Paul Gloster
2001-04-08  5:00 ` Lao Xiao Hai
replies disabled

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox