Coffeehouse Thread

54 posts

Forum Read Only

This forum has been made read only by the site admins. No new threads or comments can be added.

Managed code (.NET) future...

Back to Forum: Coffeehouse
  • User profile image
    athan

    What do you really think about this?

    http://www.grimes.demon.co.uk/dotnet/vistaAndDotnet.htm

  • User profile image
    Lee_Dale

    the .Net framework may well be powerful but would you really integrate it deeply into a new OS ?

    Maybe when the framework is more advanced and proven we will see managed services and core functionality inside the OS.

  • User profile image
    rhm

    It's a crock. For a start Grimes is well known to be anti-NET so all his articles are written from the perspective of "and this is why .NET sucks".

    His conclusion that .NET is unsuitable because Vista doesn't use it extensively is what's known in the debating business as a "logical leap". It should really be called an illogical leap though. A lot of the .NET stuff that was in Longhorn was ripped out because there were too many integration issues due to Microsoft developing .NET2 and LH at the same time. That's all. A lot of what was ripped out will get added back in before Vista ships we are told, now that .NET2 is stable, so it's really no big deal and it certainly shouldn't put anyone off writing .NET code. He completely ignores the fact that ASP.NET2 is written in .NET and that is about as serious a bit of code there is.

  • User profile image
    AndyC

    Wow, imagine spending that much time researching something and yet still failing to take into account what actually went on in the real world.

    When the Longhorn "reset" occured, Microsoft made it very clear that there were simply too many interdependent modules that were causing all sorts of problems - similar issues were occuring in the VS2005 and SQL2005 projects.

    One of these core modules was .NET 2.0. The 1.1 runtimes simply didn't provide enough for most of the things developers wanted to do, yet the 2.0 runtime was still very buggy and changing on a daily basis. Code that worked yesterday would suddenly break, code that should work didn't and so on.

    It's a classic chicken and egg problem. The only way forward was to either halt Windows development till .NET 2.0 was finished or strip out the bits that were depending on .NET 2.0. Given that choice, it's obvious what has to be done.

    To extrapolate from that the idea that "Microsoft has lost its confidence in .NET." is plainly ridiculous.

  • User profile image
    Tonatiúh

    leeappdalecom wrote:
    the .Net framework may well be powerful but would you really integrate it deeply into a new OS ?

    Maybe when the framework is more advanced and proven we will see managed services and core functionality inside the OS.



    Perhaps, we may think in the .NET Framework as that very thing it is, and its name implies: A WAN connected frame to work... On what?... Over what?... Do you ask me?... Well at first glance, to work on advanced application developing projects for Windows Applications, Web Applications and Web Services... Note that from its by design point of view, such a framework have not been ever intended for serving as the platform where a System Application, like an operating system, should be developed... So how much is worth debating about how many .NET assemblies are inside every coming version of Windows Vista; or for the shake, inside any other Operating System?

    Sure... we will see the .NET Framework in the future... But how?... When the .NET Framework will become more advanced and totally proven we will see managed CODE running UNDER managed core functionality inside the OS... Do you ask me for Windows Vista as the OS?... Well... As far as I have learned about recent Microsoft Research investigation on Operating Systems, I give my bet to Singularity or to whichever OS comes out of the researchers hands and ready for been taken by the developers coding-eager fingers... But as far as the Singularity Project prototype outcome have shown, you will never see an application, though it managed and fully proven... INSIDE the NUT (kernell) of any truly reliable Operating System... Neither an application process, will be allowed to share, use or change in any way, by any means and because any "running needs" situation the TIME and SPACE of any other process, though owned by the same application... This will be a happy new for serious developers... But a very bad new for type-as-you-think developers... And whorst yet, for hackers and their akin.

    Tonatiúh

  • User profile image
    blowdart

    Beer28 wrote:

    vb- realbasic = slower ( even though they are compiled to native instructions - they are bloated over the top )
    java - mono - .NET = medium
    java - gcc - gjc = fast

    gcc - g++ C++ runtime = faster

    gcc - libc = fastest - i386
    gcc - libc = fastester - i686 or x86_64 builds

    The i686 set saves at least 30% in my tests over the i386 sets for cetain programs. MUCH FASTER + page caching like microsoft does with their own application code = fastester-est


    I thought you didn't own Windows machines? So how could you benchmark .net?

    Or are you just pulling these numbers out of your head? Again?

  • User profile image
    Wil

    AndyC wrote:
    The 1.1 runtimes simply didn't provide enough for most of the things developers wanted to do


    Yet the Win32 API and/or MFC did "provide enough for most of the things developers wanted to do"?  So it was easier to build Vista's wonderful new features, not just without using a version of .NET that was still being constructed, but without using .NET at all?  That's a whole different issue than the problem of .NET 2.0's interdependencies.  Is that a guideline that developers should apply not only to writing Vista but to writing anything, at least if performance matters - forget about Net 1.1 and go back to the unmanaged heap?  And was the problem that managed code was too slow (certainly that was to be anticipated!), or was it that MFC offered the developers more capability than .NET 1.1 did?

    If managed code were a great idea for OS development, then Solaris would be written in Java.  And as for me, I've never trusted code where you couldn't actually see the bits by holding it up to the light, as my avatar enabled you to do...Wink

  • User profile image
    Tonatiúh

    Beer28 wrote:
    [...] but if you make system software your goal is "fastester"`[...]


    I will make my try on translating English-Beer dialect into just common English... Taking that word highlited in yellow by my own, I translate it as: Fast-est-er that is, faster than fastest... If so is what you have intended to denotate which that term, though it could appear as a funny word, it is a linguistically correct construct by adding two modifiers which denotate, respectively, the most fast, and more fast...

    But Beer, are you sure that for those Operating Systems far from now on the road ahead (that is, in the future) speed will be important; when taking the Moore's Law for granted, it will be the chips over which every Operating System runs, the ones which will be fastester than fastester?

    Tonatiúh

  • User profile image
    dotnetjunkie

    Well, I talked to a Microsoft employee on the Windows Client team the other day, so this is straight from the horse's mouth:

    The reality is in fact just the complete opposite from this article's conclusion:
    Microsoft is going 100% for .NET, is completely committed to it and will phase out the Win32 API.  Vista is just a starting point for a movement towards a completely new platform, he said it was the same as going from 16-bit (DOS) to 32-bit (Windows 95) ten years ago, and that WinFX will be the platform for everything in the next 10 years.

  • User profile image
    Tonatiúh

    Wil wrote:
    [...] And as for me, I've never trusted code where you couldn't actually see the bits by holding it up to the light, as my avatar enabled you to do...


    Well... mh, mh... As for me, I've never trusted bits which you couldn't actually see where are attempting to go... [6]

    Tonatiúh

  • User profile image
    Charles

    Future of Managed Code (.NET)?

    Bright. Very bright. Get yourself some shades. You'll need 'em.

    C

  • User profile image
    Tonatiúh

    Beer28 wrote:
    [...] If you walk from where you are to 10 feet in front of you, it may take you 3 or 4 seconds, if you walk to that same point, first taking a detour around the block it will take you alot longer. [...]


    That is what I have said... Accordingly to the Moore's Law, the size of every transistor in a chip, will become the half almost every two years... That is, as transistors size goes down, number of components in a chip goes up...

    But that is as well... An increment in speed... From the moment that, for walking a block half longer than now-a-days blocks, will take half the time you need to take a detour around the block where your home is.

    Tron world is going smaller!... Makes it sense?

    Tonatiúh

  • User profile image
    Tonatiúh

    Beer28 wrote:
    Moore's law isn't scalable across the real world where people have machines of different speeds.

    Moore's law isn't an actual law of science, or business for that matter.

    If somebody running an old PIII has to run your app, oh will they feel the pain of managed code runtimes.



    So... You will stay just where you are here and now... Let us walk the road ahead... Even if some of us *olderest* dinosaurs, are near to our personal road ahead end.

    Tonatiúh

  • User profile image
    Tonatiúh

    Beer28 wrote:
    [...] Moore's law isn't an actual law of science, or business for that matter.


    Ask Intel.

    Tonatiúh

  • User profile image
    amotif

    dotnetjunkie wrote:

    Microsoft is going 100% for .NET, is completely committed to it and will phase out the Win32 API.


    And in the early '90s Microsoft told us we'd be moving to an Object File System, replacing the clunky hierarchical file systems we know and love.

    I'm just saying, don't hold your breath. Smiley



    Seriously, though, there are a great many Win32 apps. Think of how long it would take to get ISVs and IT shops to phase those out in favor of a managed code approach even if Microsoft had a complete and compelling solution today. Win32 won't be going away anytime soon.

  • User profile image
    Tonatiúh

    amotif wrote:
    dotnetjunkie wrote:
    Microsoft is going 100% for .NET, is completely committed to it and will phase out the Win32 API.


    And in the early '90s Microsoft told us we'd be moving to an Object File System, replacing the clunky hierarchical file systems we know and love.

    I'm just saying, don't hold your breath.



    Seriously, though, there are a great many Win32 apps. Think of how long it would take to get ISVs and IT shops to phase those out in favor of a managed code approach even if Microsoft had a complete and compelling solution today. Win32 won't be going away anytime soon.



    I just have let my breath go... Cetrainly <quote>Win32 won't be going away anytime soon.</quote>... But it is as certain that such will be only while this very transition stage reliefs.

    Tonatiúh

    EDIT: Releases changed to reliefs.

  • User profile image
    footballism

    Charles wrote:
    Future of Managed Code (.NET)?

    Bright. Very bright. Get yourself some shades. You'll need 'em.

    C

    Hype alert:p

    Sheva

  • User profile image
    footballism

    dotnetjunkie wrote:

    Microsoft is going 100% for .NET, is completely committed to it and will phase out the Win32 API.

    This statement is groundless

    Sheva

Conversation locked

This conversation has been locked by the site admins. No new comments can be made.