Allen Bauer (Borland/DTG) writes:
Quote
There's always room for innovation and improvements. However, I also
want to try and preserve as much of the "spirit and intent" of the
language as possible. It makes for a more consistent and cohesive
story.
Yes, but I argue that .NET *is* very different from Win32 and the real
opportunity in .NET is innovation and improvement, not consistency and
compatibility.
Quote
So far it is been not been that bad. In fact there have been a lot of
folks that have lauded this compatibility.
You have only heard from those that care enough to complain. Many will
simply move quietly to VS.NET for their .NET work.
I'm also quite certain some of those that lauded the compatibility would
praise anything that comes forth from Borland. that is not to say that
this compatibility isn't a good thing, it is.
I just don't believe in hindsight, the sacrifices made for this
compatibility are worth it going forward. And as I have argued before,
there are other ways to solve the "preserve code investment" problem
than porting code to .NET that would have solved your problem of
customer retention/satisfaction.
Quote
OK. that is Nick's bias, which is cool. You have another bias, and
that's cool too.
I agree, I was just clarifying what my reference was. If I had a "with"
statement with a scoped pronoun I'd have used it so your
parser/de{*word*81} would have picked it up unambiguously. ;)
Quote
To me it is like arguing the notion of right-hand
drive versus left-hand drive between the States and the UK. it is a
matter of what you are exposed to and are comfortable with.
Personally, I think left-hand drive is better... is it, really? Tell
that to a British citizen.
I don't think it is like arguing that. We're not talking about
preferences like begin/end and {}.
IMO, it is about looking at the coupling of something logical to
something physical. Does Delphi force you to place each class in a
separate physical file? Java's prescribed directory structure is the
same simply because I think there is unnecessary coupling in the design.
Quote
>If it were better, I believe clean slate languages would have taken
>the same approach that Delphi for .NET did. ISTM, if Nick designed a
>clean slate language, he would take the same approach and I would
>promptly call him crazy.
Good point. However, there's really no way to quantify that.
There is a way to quantify it empirically and anecdotally. If you had a
clean slate what would you do?
Another question: when you guys were thinking about what Delphi was
going to be and pondered "Label.Caption := 'foo';" (or whatever the line
of code was), did you guys think a lot about backward compatibility with
Turbo Pascal? If not, why did you do so with .NET?
IMO, a revolutionary kind of mindset is what I'd have liked to see
Borland have towards .NET. "How can we bring elements of Delphi to .NET
in a rich and powerful way", not "how can we port Delphi to .NET."
The whole porting strategy will *never* work. It didn't work with Kylix
and it is not working with .NET either. And if you guys wanted to bring
.NET to Delphi (and I think you needed to), I contend a mixed mode
compiler is a more natural and powerful solution.
Quote
Hindsight is really nice... :-)
Indeed, and I fully acknowledge that my opinions are formed in
hindsight. Understand that I am not bashing the past at all; what I'm
calling for though is the courage to take some risks now and execute on
a .NET strategy that makes sense using the knowledge gained in hindsight
and through past mistakes.
Quote
>>I just like
>>to challenge folks' assumptions.
>So do I.
Then keep doing so.
As long as I still care, I will. Right now, I do care. In my perfect
world, R&D would advocate radical change and execute a .NET strategy
that has potential of being significantly better than what MS offers;
they would also reread the above paragraph about hindsight and realize
that the worst thing is not making mistakes but not learning or taking
actions to remedy them going forward.
--
Brian Moelk
Brain Endeavor LLC
XXXX@XXXXX.COM
