Re: Vista and .NET (Win32 life may be limted)

"Atle Smelvær" <atle@xxxxxxxxxxxxxx> wrote:

It think this is bull***. Even Microsoft have replaced a lot of .NET code
they made in early Vista. Those are now win32. Actually, .NET is just
another Java. The only thing different is that it is more tied up to Windows
(ofcourse, because it's made by... Microsoft).

I think a lot of the code (apart from WPF and WCF) in .NET is in fact
library code (think compression, HTTP, WinForms, ASP.NET, they are all
just libraries), and doesn't represent "OS" functionality as such at all.

Win32/64 will always be around. All important sections in windows will be
available through standard API call's.

I think, especially WRT WPF & WCF, you will have to work a lot harder to
get the stuff .NET will be getting "for free", but you can get it if you
put in the work.

Still, you can get some premade
classes that help you on your way to use the different parts of Windows. But
then you have VCL and all other frameworks that does just the same.

We are thinking along the same lines.

I really don't like how bad the GC is today,

I can't agree with that. .NET lives mostly on the server, which ISTM is
the current focus of commercial development, and in my experience the .NET
server GC is pretty excellent (as in <2% of CPU usage under maximum
throughput in the server application I was deeply involved in), as long as
you are aware of its constraints.

For example, being aware that objects >80K bytes are allocated on the
large object heap (LOH) and won't be collected until the next Gen 2 GC,
and thus using pooling for such large objects helps hugely.

and I don't like the fact that I
have to pay remotesoft $1899 if I don't want to release my code to everybody
when compiling .NET code (http://www.remotesoft.com/salamander/index.html).

Open source has taught me how little people value source code that isn't
being maintained and updated, and business experience has taught me that
the risks of plagiarism, piracy and copyright violation are seriously
unworthy.

I actually think Borland/Delphi did a very stupid move by jumping on the
.NET wagon. They should concentrate on making the VCL and compiler better
than .NET. Especially RTTI, form persistance features, optimizations,
unicode support, win64.

Delphi's history and maturity is its strength, so it's only to be expected
that it's quite weak when trying to be a native platform for .NET.

A big problem with .NET's goal of language interoperability is that
(runtime and common) libraries for different languages usually use
different idioms, techniques and spelling conventions. As such, languages
interacting with .NET work best when they are eclectic in what they use,
and rely on what they inherit from their history, much like what VCL.NET
does.

A side effect of that is that when they try to be native .NET languages,
they often end up with not much more power than C#, yet saddled with a
runtime library requirement and a few things that look odd to other .NET
languages (I'm thinking of Delphi's metaclasses in particular).

.NET is in development stage and currently changing very fluid.

I strongly disagree. .NET's core libraries are increasing in capability,
but they are not changing in usage model or convention in any significant
way. Sure, C# 3.0 with LINQ is picking up lots of extra stuff, but it's
all in a complementary way to existing code.

Microsoft do not care about backward compatibility, so don't think
about moving .NET 1.1 projects to 2.0 easily.

I strongly disagree with this sentiment. I've ported two man years worth
of C# work to .NET 2.0 in less than 4 hours, and that's including the
build environment and project files. I tracked the changes from .NET 2.0
Beta 2 to each subsequent CTP to the final RTM, and nothing took more than
an hour to patch up. And this was in a project deeply integrated to the
CLR, doing in-memory codegen with Reflection.Emit, etc.

It's harder for Borland, because the debugging APIs (for example) have
changed quite a bit.

Delphi will also always be behind Microsoft on the .NET stuff.. it takes
time to implement all the new changes..

Speaking honestly, I cannot think of a reason to write a new .NET Delphi
application. Delphi's forte is Win32 client applications. .NET's forte
(especially currently) is managed server applications. Mixing the two is
always a stretch.

-- Barry
.

Loading