From: jcoffin{at}taeus.com
To: c_echo{at}yahoogroups.com

At 05:48 PM 9/21/2003 +0100, you wrote:

[ ... ]

>  >>Why would you get .NET working on other platforms like Linux or other
>  >>UNIX'es.
>
>  JC> Because somebody wants it to run there, obviously.
>
>I mean do you actually believe that Microsoft would make a VM for Linux?

That depends on whether they think it would be profitable -- but it's more
or less irrelevant whether MS does it.  One of the benefits of it being an
open standard is that it allows others to implement it on Linux (or
Solaris, HP-UX, OS/400, etc.) if they want to.  MS is a large and
influential company, but they're hardly the sole source of all software, or
anything like that.

>What generation of programming language would you say .NET is then?

.NET isn't a language, it's a platform.  Of the languages that run on it,
most are solidly in the 3rd generation camp.  Offhand, I don't claim
intimate knowledge of all the languages available on .NET -- it's possible
somebody's implemented a 4GL on it, but if so I'm not really aware of it.

Since it may not be obvious, I'll point out that 4GL is really fairly
independent of the age of the language -- C (sticking to topicality) is
clearly and definitely a 3GL.  APL has been around since 1964 or so, but
it's clearly a 4GL.

To a minimal extent, the standard library for C++ attempts to add some
4GL-like features, but they're clearly parts of the library, not the language.

>  JC> Offhand, I'm hard put to think of much of .NET that's _not_ an
>  JC> impelementation of some open standard.
>
>Is there specifications how the platform in implemented?

Not really of _how_ it's implemented, but of what it has to
implement.  I.e. they're pretty careful to stick to specifying the
externally observable behavior, not things like the data structures you
need to use to implement it.  This is common to nearly all standards --
e.g. the C standard mostly specifies the language a compiler has to accept,
but contains essentially nothing about how to implement a lexer, parser,
code generator, optimizer, etc.

>Sorry, I'm just sick of hearing Microsoft are _allways_ doing the right
>thing, and Microsoft's products is _allways_ the best...

I didn't say (or intend to imply) that. IMO, MS has designed a somewhat
better VM for .NET than Sun did for Java -- but that's only saying that MS
avoided some of the more egregious and obvious errors Sun made.  The fact
is, VM technology is nothing new, but for reasons known only to themselves,
Sun seems to have ignored nearly everything that's been learned from all
the other VMs, and instead started over from the beginning.  If the Java VM
had come out just about the time the USCD P-system was running out of steam
(i.e. about 20 years ago) I'd have said it was a nice incremental
improvement.  Unfortunately, it ignores the massive amount that's been
learned about VMs since then; because of that, it was basically obsolescent
before it was even finished.

IOW, in this case I think MS has done a better design, but that's not
saying MS did a particularly wonderful job, only that Sun did a rather poor
one.
         Later,
         Jerry.

--- SoupGate-Win32 v1.05