Entries:
Comments:
Posts:

Loading User Information from Channel 9

Something went wrong getting user information from Channel 9

Latest Achievement:

Loading User Information from MSDN

Something went wrong getting user information from MSDN

Visual Studio Achievements

Latest Achievement:

Loading Visual Studio Achievements

Something went wrong getting the Visual Studio Achievements

Discussions

William Kempf wkempf
  • Question about using Linux.

    uriel wrote:
    Admittably that has happen to me with Fiesty Fawn Herd 4 (an Alpha) release, and a package known as beryl-settings-manager. Because of this, I could not properly change the settings of Beryl, and uninstallation of the package or reinstallation conventional did not work as expected. There will always be this problem with the way the Debian package manager works currently, but it is a managable problem, and it is rare. Indeed, I have yet to find a perfect operating system (and I have used quite a few), but Linux seems to be the best one out there.


    My experience (which I'm not claiming to be universal) is that it's not so rare.  I hit this wall within 6 months on every distro I've tried and with every package manager I've used.  Gentoo was by far the best here, where though it took technical knowledge you could do some amazing things to work around issues such as this, but you're not going to ask your "mainstream desktop" user to wait on ebuild.

    But I'm not trying to put Linux down here.  I'm just pointing out that Linux is the most popular *nix, and not necessarily the best *nix.  Depends on what you want a *nix for, for one thing.  Enterprise servers are much better off running Solaris or one of the other commercial *nix systems.  Desktop use is definately better off with a Linux variant, even if Linux still isn't ready for mainstream desktop usage.  Web servers can probably be run succesfully off of any *nix, so picking the right one comes down to other considerations.

    Etc, etc, etc.

  • Question about using Linux.

    uriel wrote:
    
    wkempf wrote:
    
    uriel wrote:
    Perhaps, since driver issues do plague Linux installs, but you should try installing the latest version of Fedora or Ubuntu if you are to believe it is difficult to install, at least try their installers. FreeBSD's installer is quite cryptic and requires some knowledge of UNIX systems, as well is being completely written in console. Big difference.


    1.  I never mentioned anything about installers.

    2.  I've used most of the top distros within the last few years.

    A good installer (which I can argue about with Linux... the initial install off CD is usually quite nice, but package management is a nightmare even with wonderful tools like apt/yum/etc) does not a "mainstream desktop" make.


    Can you give specific examples of how package management is a nightmare?


    I've done so in the past, and better than I'm going to make the attempt to today.  But I have experience with many of the package managers leading to broken packages caused by version conflicts in dependent packages.  It's usually possible to get out of these jams, but mom and pop won't figure it out.  It's basically the same as DLL hell on Windows, which has not been as big of an issue for many years because of SxS installations.

  • Question about using Linux.

    uriel wrote:
    Perhaps, since driver issues do plague Linux installs, but you should try installing the latest version of Fedora or Ubuntu if you are to believe it is difficult to install, at least try their installers. FreeBSD's installer is quite cryptic and requires some knowledge of UNIX systems, as well is being completely written in console. Big difference.


    1.  I never mentioned anything about installers.

    2.  I've used most of the top distros within the last few years.

    A good installer (which I can argue about with Linux... the initial install off CD is usually quite nice, but package management is a nightmare even with wonderful tools like apt/yum/etc) does not a "mainstream desktop" make.

  • Question about using Linux.

    uriel wrote:
    
    wkempf wrote:
    
    phreaks wrote:
    Not trying to hijack this thread, but it got me wondering...

    Since Solaris is now free, why aren't people using it in-place of various linux distros?

    Is *nix generally thought of as being *better* than Solaris 10?


    No, it's not better, just more popular.  BSD has also been available for a very long time, and is arguably a better *nix (BTW, that term means "Unix variant" not "Linux variant"), but Linux is still more popular.


    FreeBSD is really good for servers but I can't see it becoming a mainstream desktop OS unless they get better drivers and a much much better installation system.

    Can you explain why *BSD is better? Which one and why?


    All *nix variants are "really good for servers but I can't see it becoming a mainstream desktop OS unless they get better drivers and a much much better installation system".  Linux is no different here.  Some Linux distros are doing a better job then other *nix variants, but anyone who really thinks they are ready for the mainstream desktop is deluded.

    BSD is only "arguably" better.  In theory, it's more secure and has a better kernel (though in the past couple of years Linux has made great strides in both areas).

  • Question about using Linux.

    phreaks wrote:
    Not trying to hijack this thread, but it got me wondering...

    Since Solaris is now free, why aren't people using it in-place of various linux distros?

    Is *nix generally thought of as being *better* than Solaris 10?


    No, it's not better, just more popular.  BSD has also been available for a very long time, and is arguably a better *nix (BTW, that term means "Unix variant" not "Linux variant"), but Linux is still more popular.

  • Visual Studio 2008: Impressions.

    littleguru wrote:
    You won't see major updates in the WinForms controls... they are just wrappers for the native controls and offer only the build in functionality...


    We may not see major updates, but not for the reason you give.  Not all current controls are wrappers for native controls.  Panels, PropertyGrid, BackgroundWorker, Strip controls, ObjectDataSource, etc. are examples of controls that are not simply wrappers of native controls.  There's also new native controls that we don't have WinForm equivalents (or better yet, WPF equivalents) yet.

  • Visual Studio 2008: Impressions.

    SecretSoftware wrote:
    
    Well, The backgroundWorker control is the exact reason why you would want a new progress indicator, because the PercentWorkDone, is really unkown. So you are forced to make guesses that are not accurate. A more accurate representation is to use what the SQL Server 2005 guys use, which is a circular thing that indicates that the server is working on task.


    <shrug>Not sure I care.  The control you're asking for is just one of a million different "answers" to this problem.  And since I'm moving to WPF instead of WinForms, I can easily use an actual progress bar for this.</shrug>

    SecretSoftware wrote:

    Another things is that. Yes BackgroundWorker has a callback that you can use to update say a progressbar or something, but that is only in the case of using that control. What if I want to update a control from a thread that I generate using my code? Why do I have to think about Invokability or not. I think its a compiler's job to do that not mine.


    1.  Why are you creating a thread instead of using the BackgroundWorker?  Doesn't buy you anything.

    2.  I'm uncertain the compiler could actually do what you want it to.  I do know that the obvious thoughts on how to make it do this (for any method calls on objects of types with thread affinity it would emit the normal code to test if invoke were needed and do that instead it it is) would add needless overhead that in many situations would be significant, and would lead to poor designs.  This would encourage coders to not segment their design so that thread context switches would be minimized.

    Don't get me wrong, I think it's extremely important that languages move to a better way of handling concurrency.  But this topic is extremely complex, and we don't have any answers to it today.  Microsoft is working on it, with various research including the threading APIs included in robotics studio, but I'm willing to bet you aren't going to see what you want in the next few releases.

    SecretSoftware wrote:

    I as a dev, should only think about what functionalities my program will do, the pluming should be done by the compiler. It should figure out how to update the control, from a different thread. It should generate a way to update it by setting up some sort of events and call backs automatically.

    Programming using threads can be painful experiance, and I think MS should change that for the better.

    Its not only the Progress Bar control. I just used that as an example, of something that is relic. The Sockets (TcpClient etc..) classes are really limited and are not robust so as to allow for rapid customizable app development.

     
    In what way?  You'll have to get a lot more specific in this area.

    SecretSoftware wrote:

    Devs should not be left with the task of making their own custom controls for simple things that MS should have thought of before releasing VS.


    Uhm... even though there's a lot of functionality that would be nice to have included, I could never go along with a statement like that.  First of all, it's simply not possible to have an all encompasing standard library.  No language does, nor should they really.  Second, the way you phrase this implies a lot of negativity towards Microsoft that is unfounded at best.

    [qoute user="SecretSoftware"]Why low priority? I seen people ask for this kind of support in VS, since 5 years ago or so.[/quote]

    I explained that.  It's the 80/20 rules.  20% of the people and projects will benefit, while 80% won't.  Other things have this ratio in reverse, and so they should have a higher priority.

    SecretSoftware wrote:

    True. But there are controls that are not really UI elements that are in existance today in VS.

    But anything that does this is good.


    Not sure I agree.  There's non-visual controls, yes, but they are very closely related to the UI.  I/O channels aren't.

    SecretSoftware wrote:
    Why not add both? At the end of the day, its about organization and cleanness. You can either have a messy experience, or a really great clean experience. I just prefer things happen in-situ.


    Like I said, I don't have issue with this debugging aid.  I won't call it testing, however, and I won't agree that the current testing framework and IDE support is clunky and unusable.

    SecretSoftware wrote:
    Its just about organization. I dont want to reference alot of DLLs They can add them all into one dll, and then with different namespaces. So in my using statement, I can say using System.LINQ.Xml , or System.LINQ.Entities, etc... its better that way.


    I already explained why they should not be in the same assembly.  This is the same design mistake as the "God Object" anti-pattern.  It's a waste of resources.

    SecretSoftware wrote:

    The reason is to be able to open one session to conserve resources. Believe you me, this will improve productivity a lot.


    You won't save much.  If you notice how quickly second (and later) instances of VS open, you'll know that many resources are already shared.  I will, however, agree with the point about settings being persisted.

  • Visual Studio 2008: Impressions.

    mstefan wrote:
    
    wkempf wrote:
    
    I have no issues with the performance of 2005...


    You must not be working on large projects in C++ with IntelliSense enabled then. Visual Studio 2005's performance is nothing short of  fantastically horrid. That's why virtually every C++ programmer (at least the one's I've talked with) ends up renaming VC\VCPackages\feacp.dll to effectively disable it.


    True, I'm using large projects in C# and (heaven forbid) VB.NET with ItelliSense enabled.  I can certainly see how things would be worse in C++, being a very difficult language to parse.  If that area is performing badly, by all means, it should be fixed (no idea if Orcas has done this).  But since everyone else is talking about .NET here, I doubt this is the performance issue other people are having, which brings us back to the "hmmm".

  • As much as I bash Microsoft Marketing..

    What are you talking about?  Details, so we can follow along, please.

  • Who is Jason Weber?

    1.  His commercial product includes the same license clause(s) that he thinks are wrong in the Microsoft license.

    2.  He didn't disclose the e-mails in which he tried to leverage "perks" out of Microsoft in exchange for removing the Express versions.

    He's using publicity to push his product.  It's wrong, regardless of whether or not Microsoft is in the wrong or not (and for full disclosure, I don't think Microsoft is, at least legally).