Posted by
ryuzaki0
on from the where-do-we-go-from-here dept.
Ashcrow writes "EWeek has posted an article on Microsoft's .NET initiative. It's been three years since we were first introduced to .NET and virtually none of the promised advantages have come true. Is it time for Microsoft to move on?"
I agree, as a Penn State Student I have worked with both.NET and Unix/PHP/Perl/Apache environments. Without a doubt, the latter of the two was far superior in every aspect, INCLUDING EASE OF USE. PHP has got to be the easiest freakin language ever, and Apache trumps IIS with the ability to do the majority of configuring with one file, instead of having to browse through a maze of tabbed windows with options, checkboxes, pop-up boxes, etc.
Without a doubt, the only reasons to use.NET would be if (a), you already have a Microsoft solution and for some reason you want to keep it, or (b), you fall to marketing hype.
Oh yeah, did I forget to mention STABILITY and SECURITY...
-- Let's get one thing perfectly clear, I did not vote for George W Bush, and I do not endorse what he does or says.
"
Re:So much...
by
PhysicsExpert
·
· Score: 5, Interesting
The main problem with.Net is that it ties you to a specific OS which makes it a pain from a business economics point of view
Here at the lab for example we run a lot of mission critical syatems written in Java. Although these systems are ultra reliable they are slow and as such we are severely hampered by the hardware we can afford.
A few months ago we got a.Net system to trial and we migrated some of the apps over to it for evaluation. The results showed that.Net was so much faster than java and the support for multi threaded processes far superior. From a technical point of view we wanted to switch but the university wouldn't let us. Switching to.NET would mean swapping from NT to XP and they just wouldn't meet that level of cost.
If someone would port.NET to linux it wuld become a viable option but until then I think will only ever be a niche product.
-- All that glitters has a high refractive index.
Re:.Net was never clearly defined
by
zero_offset
·
· Score: 5, Interesting
You are exactly correct in that most people don't know what.NET really is, and that includes people using it, and Microsoft itself. Once again, Microsoft marketing has screwed the pooch. They were so hot and bothered to tie.NET to the buzzword of the day (Web Services) that they overlooked a great deal of important features and capabilities.
If you ignore the marketing noise, though, it is itself a cohesive strategy, but it's quite a wide-ranging thing and it's hard to get the right perspective on it. The problem is that you probably started looking too early. The first round of books were all written based on the betas (I reviewed many of them for various publishers), and they were all targeted at teaching the world the basics of.NET.
There are now many books that explain the guts in great detail.
To continue with your specific example, there are MANY projects which support or are working to implement CORBA remoting for.NET. A simple Google search for ".NET CORBA remoting" yielded tons of results.
Microsoft marketing is Microsoft's own worst enemy...
--
Slashdot quality declines as the number of hot grits posts decreases. - Provolt's Law, Apr-09-2005
Re:Reality is quite nice though
by
CynicTheHedgehog
·
· Score: 5, Interesting
What I like about.NET:
- The way codebehind is implemented, and the ASP.NET page lifecycle - Custom controls - Properties and indexers - Collection and foreach - Events and delegates - app.config and web.config - XCopy deployment - Newsgroup support
What I don't like about.NET:
- Buggy implementation - Crappy file I/O package - DLL Versioning (Pain in the ass. Just deprecate!) - Crappy API documentation - A lot of default behaviors, little of which is intuitive, predictable, or documented - The inability to use classes effectively for things they weren't designed to work for, even though they would be perfect for the job. This is largely due to shortsighted design and access constraints (private methods, un-settable properties, etc)
In other words, I love the CLR design and syntactical shortcuts and hate the class libraries and implementation. The feature set is very wide but not very deep. It's painfully obvious where they've set their focus (ASP.NET, ADO.NET) and where they haven't (file I/O, date/time manipulation, string formatting, etc). You develop like lightening until you reach a point where you want to refine it a bit and make it do something very specific, then you spend weeks trying to figure out what it's doing, why it's doing it that way, and how to work around the default behavior.
It's a good product for small projects, but if you're doing enterprise applications, you're better off implementing a lot of this stuff yourself. A good example are typed DataSets...they manage rowstate and updates and such, which saves a lot of time in the short term, but a lot of the time you want much finer control and a looser coupling between business objects and the data schema. Unfortunately, you can't touch the rowstate directly, which leads to some pretty interesting (and ugly) solutions.
Heck no they shouldn't be moving on....
by
Asprin
·
· Score: 4, Interesting
There are some compelling advantages to.NET -- REAL compelling advantages. The thing is that it's takes a boatload of time for a new development platform to get to the mainstream: You're looking at two or three years to get the developers comfortable enough to start working with it, then another two or three years to get their apps ported over and another year or two to roll those out to customers.
I figure we should start seeing real concrete examples of the advantages of.NET in, like 2005-06.
Don't believe me?
USB.
Or even better, how about Win32? We *still* have at least two industry-specific Win16 apps that are under a current maintenance contract. Hell, most of the non-MSOffice Win16 crap was just replaced around four years ago with the Y2K upgrades, so we're still in the process of depreciating it!
All of MS's apps will be.NET in November, but contrary to what the open source community believes, MS Office will only get you so far -- it is by far not the most important piece of software we run. The developers are the key, and MS understands this. You need to get **THEM** interested in developing on a new platform (.NET, MONO, Java, LAMP, ELF or whatever) about five years before you want anything to happen.
-- "Lawyers are for sucks." - Doug McKenzie
Re:NO tolerance for standards wars
by
zero_offset
·
· Score: 4, Interesting
And again, the part everyone fails to understand about.NET (mostly due to Microsoft's crappy marketing) is that remoting in.NET is a fully pluggable artchitecture. So whatever standard emerges, you can still use.NET. Just handle your remoting in a reasonably abstract way, then switch the damned thing on the fly.
Hell, some of the basic tutorials that came with the.NET beta (and probably with the release version, I never got around to looking at them again) showed you how to do this. A local binary component communications channel was transparently switchable to an HTTP-based protocol using policies which were controllable by an administrator... re-programming and re-compiling not required.
Fight all the standards wars you want, then just plug in the winner and get back to work.
--
Slashdot quality declines as the number of hot grits posts decreases. - Provolt's Law, Apr-09-2005
Development good, marketing bad
by
boatboy
·
· Score: 5, Interesting
When.NET first came out, our development team took the plunge, and it has greatly improved development time and the quality of our code. Where scripts and hacks dominated our development before, it's now run off compiled, modular code..NET from a programming standpoint is a great tool.
The only problem I see is MS's marketing strategy of attaching ".NET" to everything. This just confused the term. There really was no reason to call "Windows 2003 Server" "Windows.NET Server", and they finally realized that. My guess is that their marketing geeks saw the success of the "development phase" and went overboard.
Whatever the case,.NET development is good, is here, and will stick around. Slashdotters should welcome it too- There's alot of open source momentum building behind.NET related tech. Take a look at the surge of C# projects in SourceForge, and the push to implement it in linux (Mono and Portable.NET).
From what I've read here, most of the objections fall into two categories:
I don't know what.NET is.
I don't like Microsoft as a company
On the first, if you limit the scope to.NET Framework and associated languages, it's pretty easy to grasp what it is, and see why it's good.
On the second, if this is your sole reason, you're being illogical. That would be like brushing off a good idea from a fellow developer because you didn't like his office.
Re:Reality is quite nice though
by
Glock27
·
· Score: 4, Interesting
See, you linux junkies don't really know crap about MS, do you?
Far too much, in most cases.
In most cases, the CLR out-performs native Win32 because of better heap management, caching, and other little things here and there.
Said heap management, caching, etc. couldn't have been implemented in a pre-compiled language?!? Sure.
And there will be cross-platform compatibility once linux developers finish Mono.
So long as Microsoft sees fit not to exercise it's massive patent portfolio. I'd sure bet my business on Microsoft playing nice...not.
If anything that runs on a VM is slow - it's Java. It has to JIT everything before running it while the CLR JITs on demand and it even does that faster!
That would depend on which Java implementation you're talking about. There are fully pre-compiled Java systems available, however the VM based versions are very competitive. They are certainly neck and neck with the CLR...and are available on many platforms, now. Even enterprise class platforms.:-)
Java has tremendous momentum - which.Not has largely failed to affect.
-- Galileo: "The Earth revolves around the Sun!"
Score: -1 100% Flamebait
.NET is hurting development
by
wandazulu
·
· Score: 5, Interesting
I'm a Windows developer who in the year 2003 is using a product that came out in 1998. The venerable Visual Studio 6. The first version of VS.net gave absolutely nothing to straight C/C++ developers who were not interested in C# or windows forms or what-have-you, but instead wanted to write good solid code using an ISO-standards compliant compiler for backend work. VS.net gave us nothing new.
VS.net 2003, that's a different story. It does all the things I want to do in a C++ compiler, but apart from the cost, what do you suppose is keeping the bosses from approving it? That's right:.NET. I have told everyone that it actually has a decent C++ compiler, but everybody thinks that it can only be used for.NET work.
So here I am, about to go back to a compiler that has no partial template specialization, a version of STL that I have to patch *by* *hand*, and if I want to look something up? Well, I've got my msdn help files from October 2001 to explain it to me, because that was the last version that integrated with VS6.
By pushing.NET they've done a good job of alienating the core base of people who write the back end code where too-fast-is-not-fast-enough. Maybe it'll come to the point where if you want to write services or databases or anything where speed and size are most important, you'll use a totally different compiler, say, Borland or Metrowerks. But if you're going to do that, why not also look at other platforms, say, Linux?
Just my $0.02
It's all about the Pentiums
by
Joe+U
·
· Score: 4, Interesting
We're all missing the real point of.net
The true reason behind the.net push is to create a bunch of easy to use high level languages to compile down to basically the same code, then let that code run on Win32 platforms and Win64 platforms without making changes.
When the 32 to 64 bit switch starts, the.net apps will be ready to go. The win32 apps will require a translation layer.
Combine that with the fact that the Windows (NT/XP) kernel already supports multiple architectures, win32, posix and os/2 are the 3 common ones. I'm willing to bet that.net will show up in the kernel in the next version of Windows.
Re:From "Great" to old ideas
by
goodviking
·
· Score: 4, Interesting
When Java was first released, umpty squat years ago, it introduced a lot of good concepts to the wider programming community (yeah yeah, smalltalk blah blah blah). The good news is, the language is adapting and evolving based on a community input process, and real world feedback. There are some things that maybe should or could have been done in different ways, but all in all, I keep comming back.
Re:.Net was never clearly defined
by
alext
·
· Score: 4, Interesting
.Net has is superior native execution
I'd be interested in some benchmarks. My experience in fiddling with some numerically-intensive code is that Sun JVM 1.4.1 is about 4 times faster than a Dotnet release of 18 months ago. I haven't tried a more recent version.
Slashdot Mirror
I agree, as a Penn State Student I have worked with both .NET and Unix/PHP/Perl/Apache environments. Without a doubt, the latter of the two was far superior in every aspect, INCLUDING EASE OF USE. PHP has got to be the easiest freakin language ever, and Apache trumps IIS with the ability to do the majority of configuring with one file, instead of having to browse through a maze of tabbed windows with options, checkboxes, pop-up boxes, etc.
Without a doubt, the only reasons to use .NET would be if (a), you already have a Microsoft solution and for some reason you want to keep it, or (b), you fall to marketing hype.
Oh yeah, did I forget to mention STABILITY and SECURITY...
Let's get one thing perfectly clear, I did not vote for George W Bush, and I do not endorse what he does or says.
"
The main problem with .Net is that it ties you to a specific OS which makes it a pain from a business economics point of view
.Net system to trial and we migrated some of the apps over to it for evaluation. The results showed that .Net was so much faster than java and the support for multi threaded processes far superior. From a technical point of view we wanted to switch but the university wouldn't let us. Switching to .NET would mean swapping from NT to XP and they just wouldn't meet that level of cost.
.NET to linux it wuld become a viable option but until then I think will only ever be a niche product.
Here at the lab for example we run a lot of mission critical syatems written in Java. Although these systems are ultra reliable they are slow and as such we are severely hampered by the hardware we can afford.
A few months ago we got a
If someone would port
All that glitters has a high refractive index.
If you ignore the marketing noise, though, it is itself a cohesive strategy, but it's quite a wide-ranging thing and it's hard to get the right perspective on it. The problem is that you probably started looking too early. The first round of books were all written based on the betas (I reviewed many of them for various publishers), and they were all targeted at teaching the world the basics of .NET.
There are now many books that explain the guts in great detail.
To continue with your specific example, there are MANY projects which support or are working to implement CORBA remoting for .NET. A simple Google search for ".NET CORBA remoting" yielded tons of results.
Microsoft marketing is Microsoft's own worst enemy...
Slashdot quality declines as the number of hot grits posts decreases. - Provolt's Law, Apr-09-2005
What I like about .NET:
.NET:
- The way codebehind is implemented, and the ASP.NET page lifecycle
- Custom controls
- Properties and indexers
- Collection and foreach
- Events and delegates
- app.config and web.config
- XCopy deployment
- Newsgroup support
What I don't like about
- Buggy implementation
- Crappy file I/O package
- DLL Versioning (Pain in the ass. Just deprecate!)
- Crappy API documentation
- A lot of default behaviors, little of which is intuitive, predictable, or documented
- The inability to use classes effectively for things they weren't designed to work for, even though they would be perfect for the job. This is largely due to shortsighted design and access constraints (private methods, un-settable properties, etc)
In other words, I love the CLR design and syntactical shortcuts and hate the class libraries and implementation. The feature set is very wide but not very deep. It's painfully obvious where they've set their focus (ASP.NET, ADO.NET) and where they haven't (file I/O, date/time manipulation, string formatting, etc). You develop like lightening until you reach a point where you want to refine it a bit and make it do something very specific, then you spend weeks trying to figure out what it's doing, why it's doing it that way, and how to work around the default behavior.
It's a good product for small projects, but if you're doing enterprise applications, you're better off implementing a lot of this stuff yourself. A good example are typed DataSets...they manage rowstate and updates and such, which saves a lot of time in the short term, but a lot of the time you want much finer control and a looser coupling between business objects and the data schema. Unfortunately, you can't touch the rowstate directly, which leads to some pretty interesting (and ugly) solutions.
There are some compelling advantages to
I figure we should start seeing real concrete examples of the advantages of
Don't believe me?
USB.
Or even better, how about Win32? We *still* have at least two industry-specific Win16 apps that are under a current maintenance contract. Hell, most of the non-MSOffice Win16 crap was just replaced around four years ago with the Y2K upgrades, so we're still in the process of depreciating it!
All of MS's apps will be
"Lawyers are for sucks."
- Doug McKenzie
Hell, some of the basic tutorials that came with the .NET beta (and probably with the release version, I never got around to looking at them again) showed you how to do this. A local binary component communications channel was transparently switchable to an HTTP-based protocol using policies which were controllable by an administrator... re-programming and re-compiling not required.
Fight all the standards wars you want, then just plug in the winner and get back to work.
Slashdot quality declines as the number of hot grits posts decreases. - Provolt's Law, Apr-09-2005
The only problem I see is MS's marketing strategy of attaching ".NET" to everything. This just confused the term. There really was no reason to call "Windows 2003 Server" "Windows
Whatever the case,
From what I've read here, most of the objections fall into two categories:
- I don't know what
.NET is.
- I don't like Microsoft as a company
On the first, if you limit the scope toOn the second, if this is your sole reason, you're being illogical. That would be like brushing off a good idea from a fellow developer because you didn't like his office.
Far too much, in most cases.
In most cases, the CLR out-performs native Win32 because of better heap management, caching, and other little things here and there.
Said heap management, caching, etc. couldn't have been implemented in a pre-compiled language?!? Sure.
And there will be cross-platform compatibility once linux developers finish Mono.
So long as Microsoft sees fit not to exercise it's massive patent portfolio. I'd sure bet my business on Microsoft playing nice...not.
If anything that runs on a VM is slow - it's Java. It has to JIT everything before running it while the CLR JITs on demand and it even does that faster!
That would depend on which Java implementation you're talking about. There are fully pre-compiled Java systems available, however the VM based versions are very competitive. They are certainly neck and neck with the CLR...and are available on many platforms, now. Even enterprise class platforms. :-)
Java has tremendous momentum - which .Not has largely failed to affect.
Galileo: "The Earth revolves around the Sun!"
Score: -1 100% Flamebait
I'm a Windows developer who in the year 2003 is using a product that came out in 1998. The venerable Visual Studio 6. The first version of VS.net gave absolutely nothing to straight C/C++ developers who were not interested in C# or windows forms or what-have-you, but instead wanted to write good solid code using an ISO-standards compliant compiler for backend work. VS.net gave us nothing new.
.NET. I have told everyone that it actually has a decent C++ compiler, but everybody thinks that it can only be used for .NET work.
.NET they've done a good job of alienating the core base of people who write the back end code where too-fast-is-not-fast-enough. Maybe it'll come to the point where if you want to write services or databases or anything where speed and size are most important, you'll use a totally different compiler, say, Borland or Metrowerks. But if you're going to do that, why not also look at other platforms, say, Linux?
VS.net 2003, that's a different story. It does all the things I want to do in a C++ compiler, but apart from the cost, what do you suppose is keeping the bosses from approving it? That's right:
So here I am, about to go back to a compiler that has no partial template specialization, a version of STL that I have to patch *by* *hand*, and if I want to look something up? Well, I've got my msdn help files from October 2001 to explain it to me, because that was the last version that integrated with VS6.
By pushing
Just my $0.02
We're all missing the real point of .net
.net push is to create a bunch of easy to use high level languages to compile down to basically the same code, then let that code run on Win32 platforms and Win64 platforms without making changes.
.net apps will be ready to go. The win32 apps will require a translation layer.
.net will show up in the kernel in the next version of Windows.
The true reason behind the
When the 32 to 64 bit switch starts, the
Combine that with the fact that the Windows (NT/XP) kernel already supports multiple architectures, win32, posix and os/2 are the 3 common ones. I'm willing to bet that
Show me boxing in Java.
...etc. It's a fun read.
OK. Boxing, Typesafe Enums,
When Java was first released, umpty squat years ago, it introduced a lot of good concepts to the wider programming community (yeah yeah, smalltalk blah blah blah). The good news is, the language is adapting and evolving based on a community input process, and real world feedback. There are some things that maybe should or could have been done in different ways, but all in all, I keep comming back.
.Net has is superior native execution
I'd be interested in some benchmarks. My experience in fiddling with some numerically-intensive code is that Sun JVM 1.4.1 is about 4 times faster than a Dotnet release of 18 months ago. I haven't tried a more recent version.