Win32, Linux, .NET >> Where is this mess going to?

From: Danijel Tkalcec (dtkalcec_at_hotmail.com)
Date: 10/27/04 

Date: Wed, 27 Oct 2004 14:25:18 +0200

Hi, everyone!

I'm wondering ... how many Delphi developers are thinking about moving to M$
Visual Studio and C# rather then staying with Borland Delphi 2005? To me, it
looks like Borland is doing more work for M$ then for their own products.
And this co-operation between Borland and M$ doesn't do any good to Borland
(maybe a few $$ from M$, but that's nothing, compared to what they're
losing -> customers).

>From what I saw until now, Borland is going to lose this "battle" with M$
and most developers who now work with Borland tools will switch to M$ in a
couple of years (if not sooner). And here is why, from my point of view:

1. In the beginning, Delphi 1 was there and it was for Win16. Then they
shipped Delphi 2, but it could ONLY compile Win32 applications. You would
need Delphi 1 to compile an app in Win16 mode. **** Instead of that, they
could have put 2 compilers and make a compiler switch for Win16 or Win32.

2. Then, a few years ago, Borland announced the beginning of a new era,
starting with their brand-new product, which they called "Kylix". But, they
decided to make this a completely new product and design a completely new
component library for this, which will NOT be comppatible with Win32 VCL.
**** Instead of that, they could have modified the VCL components internally
to map either to Win or to the Linux version and keep it transparent for the
end-user, by adding a simple "compile for Linux" option in the compiler.
That way, developers could really write applications that compile under
Windows and Linux, without the need to design new forms or use special
"Linux DLLs" for Windows.

3. When they saw Kylix was a "mistake" (the general idea was good, but the
implementation was not quite what I would expect), Borland bragged about
being the first company to ship a .NET compiler in Delphi 7. But, what's a
compiler good without IDE?

4. Last year, they decided it's "about time" to drop Win32 and bring out a
new Delphi, which only supports dotNET. Hmm .... how many of you moved from
D7 to D8 and have applications running under .NET? I don't think it's more
then 5% of all Delphi developers. Now that Borland saw .NET isn't that
ready, they decided to put Delphi 7, C++ Builder and Delphi 8 into one "new"
IDE and sell it as "Delphi 2005". But, other then having the IDE you need,
glued together with 2 more compilers you won't be using (since you can
compile the application you write ONLY with the compiler you wrote it for),
what's the point?

What I'm trying to say is ... Borland took a wrong turn with Kylix and now,
rather then combining the old and new things to form a unity, Borland again
decided to throw new stuff out and "forget about the past". If you want to
move to .NET, you have to start from scratch, or rewrite your code so it is
unusable for Win32. There's no way you can design your application to be
compileable for both "worlds" (Win32 and NET). It looks to me that we will
have to modify our existing applications over and over again, with every new
version of Delphi comming out.

But, I'm not just complaining here. I think there is a solution to this
problem. A solution that should have come directly from Borland. It would be
a lot easier if Borland did the job right, but there is also the possibility
that we, as component developers, do the job and keep a good product alive.

My idea is to define a component base, which will encapsulate most VCL, .NET
and CLX (visual and non-visual) components with all properties that exist in
all three worlds, into one unified component library. There is no need to
rewrite all components, since they are there. The only job would be to
create a new component wrapper set, which will (in the back) map all
peoperties, methods and events from the unified set component set to a
specific component set. That mapping could be triggered by using compiler
directives in Delphi, which would enable the Developer to define 1 form,
write 1 code and then compile for whatever system they need (Win32, Linux,
.NET, or whatever comes next). In case things change in .NET or Win32 or
Linux, it is much easier to adapt the back-end of your components to the
changes then to correct the change on every single Form you designed.

.NET is a long way from being stable and who knows when it will become
stable enough for sensitive bussiness applications. Win32 is here, Linux is
here. If there was a possibility to write applications that seamlessly
compile for Win32, Linux or .NET, I'm sure that every developer who worked
with Delphi, would except that as a solution to the problems we got now with
unstable .NET and expensive M$ systems.

What do you, as developers, think about this?
Is there anyone from Borland here, who has something to say?
I'm frustrated and I'm mad. If nothing really important happens in the
Borland world, I will be forced to switch to M$ and their Visual Studio
solution.

Best Regards,
Danijel Tkalcec

Relevant Pages

  • Re: A case for Microsoft
    ... specifically Delphi. ... set (in the meeting to make the decision, the developers' comments were ... "Borland looks like it is back on track" and "if they would stop wasting ... If you've used D2005 or D2006 and VS Studio - you know that the IDE ...
    (borland.public.delphi.non-technical)
  • A case for Microsoft
    ... I have been a Borland fan since turbo Pascal and paradox. ... specifically Delphi. ... set (in the meeting to make the decision, the developers' comments were ... The decision to sell the IDE products (which ...
    (borland.public.delphi.non-technical)
  • Re: Delphi 2005 Update 4 IS needed !!!
    ... I use a lot of 3rd party tools and am not doing much that requires web ... version just to make sure Borland didn't go completely out of business, ... I am wondering how many Delphi developers are in the same situation as ...
    (borland.public.delphi.non-technical)
  • Re: Delphi 8 ... is it a worthy investment?
    ... the directions Borland has taken is effecting people personally. ... are solo developers - ... That puts a lot of Borland's customers in the same position as me. ... Up until Delphi 6 this was not really an issue - it was always easy to ...
    (borland.public.delphi.non-technical)
  • Re: Is anything spectacular going to happen around Delphi anymore?
    ... > IDE tweaks on new bugs (broken multilanguage charset capability ... I have to agree with you that delphi should have a better Unicode Support. ... problem might then be that project compiled with VCL would no compile with UVCL... ... Borland is offering you a path. ...
    (borland.public.delphi.non-technical)