On 3 Jul 2003 at 8:24, Lester Caine wrote:

> I am seeing comments elsewhere that I will HAVE to support
> .NET at some point because it will replace everything else
> when the next versions of Windows come out.
>
> I don't subscribe to that view, but would appreciate a
> little support :)
>
> With Borland 'messing up' Builder by making it compile C#
> and support .NET I am being hit from one side, and the lack
> of support for Kylix seems to be 'assisting'.
>
> Anybody got any 'insight' it what is going to happen in the
> next few years ;)

Well, I follow this theme for about a year now (researching, give
training, speaking at events or "pub meetings" with developers etc.), so
here you are my 0.02 EUR.

MS future depends on .NET, at least until they do not invent another big
thing. So what you hear is true, WIN32 API + ActiveX + COM+ etc. will be
replaced by .NET framework + other .NET facitilties built up on it. We
really have another major DOC->WIN32 switch before us, but don't expect
it tomorrow or as quick mass transition. It will really start with
release of Longhorn (scheduled to 2005 by recent MS statements), despite
MS hard push with VS.NET + .NET Framework for W98/ME and W2000/XP. It's
still not fully developed (including C# itself), and we will see at least
one major "redesign" in near future. It's like WIN32s (do you remember it
?).

So, WIN32 IS still better for client GUI apps, and will be for some time.
You don't need to wory at least until Longhorn arrive, and since that
we'll have around five years transition phase. It's a lot of time when
anything may happen, including demise of .NET, fast evolution of Java
platform toward desktop and rise of Linux on desktop :-)

But .NET core (C# + CLI/CRE + base classes) is interesting alternative to
Java more suitable for desktop (now, but that may change) applications
with great potential in cross-platform, language neutral development. If
MS do not stops Mono and Dot.GNU in some way, we would get something
really good from MS :-)

About Borland: They bet the future of whole RAD division on .NET (Delphi,
Kylix, C++Builder and C#Builder, but not J Builder!). Their main selling
point is integration with CaliberRM, Together, StarTeam and OptimizeIT!
While it's interesting and definitelly good strategy, there is nothing in
it for small and medium sw shops. Borland definitelly left this market
and aimed the enterprise. They are still Inprise in strategy, just
renamed. Unfortunatelly for all, C#Builder is not ready for prime time,
despite their try to convince all about the opposite (it's worse than
Kylix 1 or Kylix 3 for C++). But Borland thrown all R&D resources from
RAD division to it, so they're making a lot of progress. I would suggest
to take another look on C#Builder in six months from now.

The hard part for Borland is that MS did a big leap in VS development, so
Visual Studio .NET is really very good, low priced (try to find your
local MS promotion discounts offering) and ready for prime time. Borland
has still very good engineers, but lost his head visionary (Anders) to
MS, so now they play catch-up. MS inovated more in RAD tools recently
than Borland. It's pity to say that, but it's true.

What's really pity for Delphi and Kylix developers is that all
development on D8 and K4 stoped, as C#B has the top priority, and RAD
division R&D is small. Kylix may not survive this lack of development as
people may drift out to other tools to develop for Linux (C#+Mono, gcc,
Python etc.). D7 still haven't a patch, D8 is scheduled to Q1/2004 and
you should not expect much from it (just integration with CaliberRM etc.,
VCL.NET which is a horrible mistake to use it and bastardized OP for
.NET). So, if one wants to develop for WIN32, there will be no reason to
pay the price for D8, and those who want to develop for .NET will make
better with VS.NET or even C#Builder. All in all, Borland is doomed for
small and medium WIN32/Linux developers. I'll predict a decline in sales
of Delphi, C++Builder and Kylix in future (it would be only worse, and
never get better), and no much
success with C#Builder in next year or two. But Borland will not fade
away, as they will make money
from big companies and from Java.

Summary: If you're happy with D5-7 or K3 for Delphi, you don't need to
wory about next five years, as you'll have a lot of contracts for
WIN32/Linux. But you should look around for new tools (C#, gcc, whatever)
for WIN32/.NET/Linux development that you'll use in five years from now,
and don't hope much that they will come from Borland.

Best regards
Pavel Cisar
http://www.ibphoenix.com
For all your upto date Firebird and
InterBase information