Slashdot Mirror


First Look At Visual Studio 2010 Beta 1

snydeq writes "InfoWorld's Martin Heller takes VS2010 Beta 1 for a test drive and finds the upgrade promising, particularly with regard to improved thread debugging and a revamped UI. But the biggest enhancements have to do with parallel programming, Heller writes. 'I'm not sure that I've completely grasped the power of the new .Net Framework and native C++ support for task and data parallelism in VS2010, but what I've seen so far is impressive.' Heller points to intriguing parallel programming samples posted to CodePlex and offers numerous screenshots of VS2010 Beta 1 functionality. He also notes that the beta still lacks support for ASP.Net MVC, smart devices, and the .Net Micro Framework."

27 of 236 comments (clear)

  1. Re:oh by Anpheus · · Score: 5, Funny

    No, like this:

    developers
    developers
    developers
    developers
    deverlopee
    elsdeverlpr
    opesdeveos

    Or something like that. Threading is hard! :)

  2. Anyone even using VS 2008 yet? by tylersoze · · Score: 3, Informative

    Heck no one I've worked with has even upgraded to 2008 yet, it's been either VS 2005 or 2003.

    1. Re:Anyone even using VS 2008 yet? by sys.stdout.write · · Score: 4, Informative

      We run it at work. It is pretty much the same as VS 2005. The only relevant advantage that I see is access to the .NET 3.5 Framework, which may or may not matter to you depending on what you guys program.

    2. Re:Anyone even using VS 2008 yet? by shutdown+-p+now · · Score: 3, Informative

      We run it at work. It is pretty much the same as VS 2005.

      If you use it for C++ development, then the C++ compiler in 2008 has better standard compliance, and some nasty bugs in the libraries are fixed. Also, VS2008 SP1 adds C++ TR1 stuff.

    3. Re:Anyone even using VS 2008 yet? by Amazing+Quantum+Man · · Score: 3, Funny

      No, but you do get an add-in for FireFox... no questions asked!

      --
      Fascism starts when the efficiency of the government becomes more important than the rights of the people.
    4. Re:Anyone even using VS 2008 yet? by shutdown+-p+now · · Score: 4, Informative

      Visual C++ hasn't changed much since VS 2002.

      Eh? It's got C++/CLI since then, for starters. It has become much closer to ISO C++ (before 2003 it was a total joke, it didn't even get the scope of the for loop right - and 2003 was only so-so). It's got checked STL containers & iterators in 2005, and C++TR1 in 2008. And it is getting significantly improved code completion, and on-the-fly error checking in 2010. Doesn't sound "abandoned" to me.

      On the language front, Visual C++ in 2010 gets a bunch of C++0x features: lambdas, type inference (auto), static_assert , rvalue references (&&), and decltype. This is quite a lot, and lambdas are especially nice since they actually let you use STL algorithms as God intended without writing tons of boilerplate code for function objects.

      Then also there's Parallel Patterns Library, which provides STL-like algorithms with automatic parallelization.

      And no refactoring

      This one is interesting. I do not know of any C++ IDE or plugin that would provide working C++ refactoring, for very simple reason - it is extremely hard to properly parse C++, taking into account all templates and template specializations, and other context-dependent things. Heck, something like a<b>c can be parsed either as expression (a < b) > c, or as a variable declaration a<b> c, depending on the context - and that context, again, includes template instantiations, which form a Turing-complete language that has to be interpreted correctly to produce matching results. I once wrote a C++ program, for fun, which had in it a piece of code as described above, which was parsed and compiled either as expression or as variable declaration depending on whether char type was signed or unsigned was for a given compiler - so you could play with compiler options and get different results. How can IDE possibly handle this?

      You can say that it does it for code completion, but the truth is that a lot of it is guessing and heuristics. And there's the catch - when it guesses wrong, at worst, you get a wrong code completion list, or no list at all. But when you do a refactoring like, say, "rename class", and it fails to correctly determine that the class is referenced at some line of code, and doesn't rename it there, then your program no longer compiles...

      That said, VS2010 IDE C++ parser (used for code completion and "Go to definition") is EDG-based, so it should be much more accurate - so hopefully we'll get reliable C++ refactoring eventually. Just not in this release.

      ... decent GUI toolkit. Both MFC and Win32 API are incredibly difficult to code.

      I agree with that, but there are many good third-party libraries out there - most notably, Qt.

    5. Re:Anyone even using VS 2008 yet? by zlogic · · Score: 3, Interesting

      Eclipse does C++ refactoring, I think Netbeans can do it too. I've used Eclipse for renaming values, implementing methods and generating getters/setters, it didn't ever break anything and showed all code that was about to be changed before doing something irreversible. Even if it breaks something, there's Local History which acts as a simple version control server, committing code on every save operation.
      If a compiler can parse the code, the IDE should be capable of doing that.

    6. Re:Anyone even using VS 2008 yet? by oiron · · Score: 3, Insightful

      And it is getting significantly improved code completion, and on-the-fly error checking in 2010.

      If you read the comments in that post, it looks like C++/CLI isn't supported by the new, improved intellisense.

      Honestly, we took to calling it "Intellinonsense" at work, given the number of times it fails to complete; you can rate it by failures per second...

      Doesn't sound "abandoned" to me.

      On the language front, Visual C++ in 2010 gets a bunch of C++0x features: lambdas, type inference (auto), static_assert , rvalue references (&&), and decltype. This is quite a lot, and lambdas are especially nice since they actually let you use STL algorithms as God intended without writing tons of boilerplate code for function objects.

      Then also there's Parallel Patterns Library, which provides STL-like algorithms with automatic parallelization.

      Yes, the actual c++ compiler and library support has definitely improved. But there seems to be no corresponding improvement in the IDE's functionality. When Visual Assist X becomes a requirement for working with any kind of productivity, it's a rather sad situation. You've got to spend on VS, and then on VA-X for every developer, when the alternative is eclipse with mingw, giving all those above-mentioned features (aka, parts of the standard) for free, and a real usable IDE along with that!

      This one is interesting. I do not know of any C++ IDE or plugin that would provide working C++ refactoring, for very simple reason - it is extremely hard to properly parse C++, taking into account all templates and template specializations, and other context-dependent things. Heck, something like a<b>c can be parsed either as expression (a < b) > c, or as a variable declaration a<b> c, depending on the context - and that context, again, includes template instantiations, which form a Turing-complete language that has to be interpreted correctly to produce matching results. I once wrote a C++ program, for fun, which had in it a piece of code as described above, which was parsed and compiled either as expression or as variable declaration depending on whether char type was signed or unsigned was for a given compiler - so you could play with compiler options and get different results. How can IDE possibly handle this?

      You can say that it does it for code completion, but the truth is that a lot of it is guessing and heuristics. And there's the catch - when it guesses wrong, at worst, you get a wrong code completion list, or no list at all. But when you do a refactoring like, say, "rename class", and it fails to correctly determine that the class is referenced at some line of code, and doesn't rename it there, then your program no longer compiles...

      That said, VS2010 IDE C++ parser (used for code completion and "Go to definition") is EDG-based, so it should be much more accurate - so hopefully we'll get reliable C++ refactoring eventually. Just not in this release.

      Eclipse and the still-in-beta KDevelop 4.x give you at least some basic refactoring support. Eclipse has done so for a few years now...

      I agree with that, but there are many good thir

  3. Re:Yay! by sys.stdout.write · · Score: 5, Insightful

    I love how you have to post anonymously in order to support Microsoft products on Slashdot.

    Screw it, I second this. Visual Studio has the best code completion implementation ever written. I can type lines like obj.GetSomething().Append(item) in about four keystrokes.

    This makes me warm and fuzzy inside.

  4. Re:More security? by Mr2001 · · Score: 4, Informative

    Look at that fourth screenshot. What possible harm could loading a project do, I wonder? Does it already (partially?) execute even when it is just sitting there in the development environment? Is this an attempt to banish evil compilers from accidentally compiling source?

    As it says right there in the screenshot, the possible harm is from custom build steps.

    Unix developers are already used to this, because makefiles have the same risks: if you untar an untrusted project and type "make", you might find that one of the build steps erases your home directory.

    --
    Visual IRC: Fast. Powerful. Free.
  5. Re:More security? by wzinc · · Score: 5, Informative

    Does it already (partially?) execute even when it is just sitting there in the development environment

    Actually, it kind-of does execute. Most controls, even user-created ones, have "design mode." That's a special view that gets rendered while you're designing pages or forms. I never thought about it, but it is just code that executes. I don't know if there's anything that prevents you from opening up an FTP connection or calling "del /f /s /q C:\*" from a control in design mode.

  6. Not just parallel by shutdown+-p+now · · Score: 5, Interesting

    ParallelFX is definitely interesting, but I'd say that another very major addition is Visual F# - to the best of my knowledge, this is the first time a primarily functional language goes mainstream, and gets documentation, tooling (IDE/debugging/profiling), and general support on par with the likes of C# and VB. It's not Haskell (read: no typeclasses), and it's not quite OCaml either (no functors), even though the core language is recognizably ML. But it's got most of the nice FP bits OCaml has to offer, some syntactic sugar on top of that (e.g. ability to declare locals as mutable when needed, and arithmetic operators overloaded for all numeric types), and it's got direct and full access to one of the largest class libraries on the market today.

    (I'm sure someone will remind me of Scala, which is in many ways similar to F#. It's definitely comparable, but its tooling support is lagging behind, and, most importantly, it's not backed by any of the "big players" in Java land - not Sun, not Google, not IBM - or indeed, any other company.)

    The second, smaller, but still interesting bit is improved language interop. It seems that, as new core (i.e. MS-supported) .NET languages are added to the batch, the framework itself is extended as needed to provide primitives for them where more than one language uses them. For example, both F# and IronPython work with tuples, but they have previously each defined their own type for that - and so .NET 4 introduces the standard System.Tuple type, and all languages are changed to use that. So now you can actually make a tuple in IronPython, and pattern-match it in F# - nice.

    Another bit along the same lines is C# 4 dynamic type - which is nothing but opt-in duck typing - and the associated DLR framework for exposing runtime dynamic type information in a common way. This means that static/dynamic language interop on .NET is now two-way - previously, you could easily call C# class methods from IronPython/IronRuby, but there was no easy way to call methods on IronPython/IronRuby objects in C# - but now you can do the latter just as easily.

    1. Re:Not just parallel by shutdown+-p+now · · Score: 3, Insightful

      Again, we'll see... to be quite frank, I'm firmly convinced that functional programming is just far too confusing for most people, and so it'll never truly go mainstream (I can definitely see niche penetration, but that's not mainstream is it? :), regardless of the tooling and vendor support available. But, hey, I could be wrong.

      You're probably right regarding full-fledged FP, but FP techniques have certainly been making inroads in the last 10 years or so. First-class functions have gone from an arcane concept to something every mainstream language today has in one way or another (even Java has a castrated version of it in form of anonymous inner classes). Python, Ruby, and now C# programmers happily use map/filter/fold (or whatever it's called in their libraries) as opposed to hand-coding loops. And so on.

      To that extent, it should be noted that F# is not a pure functional language a la Haskell - it has mutable locals and fields, it has strictly imperative constructs such as loops or if-without-else, it is eager and not lazy, and it has well-defined evaluation order. In practice, it can just as easily be used as "advanced C#" - exploiting its better type inference, using pattern matching as a glorified typeswitch, and simply enjoying the little bits like inline array/list/sequence generators or anonymous objects - and ignore the traditional FP patterns and approaches. When used that way, it gets really close to Scala. And, in fact, I suspect that it's how most people on .NET will use it in the end.

      As an aside, F# does have one huge advantage going for it: the ability to very easily incorporate it as a component in a larger project that *is* implemented in a mainstream language. So you can write most of your app in nice, friendly C#, and use F# where its advantages really shine.

      That is very true, and the F# team has stated many times that it is indeed the intended goal. It's why VS2010 doesn't come with visual UI designers for F#, for example - it can be done, but what's the point? But if I need to e.g. write a parser for my .NET project, I would very strongly consider using F# + FParsec for it - it's just so much concise, easier, and more powerful than all available C# solutions.

  7. Re:Themes by WarwickRyan · · Score: 4, Informative

    Yes, it is slower. However, being able to put form designer on one monitor and code-view on another makes it all worth while.

  8. Regarding C++ by xquark · · Score: 4, Informative

    Compatibility and conformance with standards (TR1), also going that extra step forward and implementing some of the upcoming 0x features I can truly say that since VS05 MS has gone a long way. WRT Language/IDE/Debug integration nothing comes close in the OSS world for the C++ language (and please don't say CDT, I've tried using 5 and it can't even do the simple C++ syntax properly let alone templates or even simple metaprograms).

    Disappointing/sad thing with VS10 is that a lot of the interesting source code metric/analysis stuff is only available for C++\CLI. For pure C++ code metrics I've been pinning my hopes for the past 5 years on someone getting around to implementing to-do #6 of doxygen.

    --
    Arash Partow's Philosophy: Be a person who knows what they don't know, and not a person who doesn't know.
  9. Re:Yay! by doti · · Score: 4, Insightful

    No, what you need is a sane programming language/technique.

    --
    factor 966971: 966971
  10. Re:I tried it, not impressed. by Photo_Nut · · Score: 4, Informative

    I heavily use both Visual Studio 2005 and 2008, so I was excited to use 2010. The thing I found most obnoxious about it was the the text in the code editor was blurry at normal font settings (Consolas 10pt). Let me restate that. The text in the primary window of the software that you will be staring at for hours daily... is blurry. How on earth did that get past QA?

    Section 2.4.2.2 of the Readme describes why the code can appear blurry:

    2.4.2.2 Text may appear slightly blurry
    Text may appear slightly blurry. This can occur with any font, although different users may see different fonts as more or less blurred. MSGothic, the default font for Japanese system locales, is known to be more blurred than other fonts.

    To resolve this issue:

    We recommend that you keep the default fonts for Visual Studio, for example, Consolas for English SKUs of Visual Studio 2010. There is no alternative that is consistently sharper than MSGothic for Japanese characters, although some users find Meiryo more readable.

    TrueType fonts are the least blurred of the non-default fonts. These fonts appear in bold in the font list on the Fonts and Colors page of the Options dialog box (Tools menu, Options, Environment, Fonts and Colors). The WPF and Visual Studio teams are working to improve font rendering in the text editor before the release of Visual Studio 2010.

    You can also take a look at this white paper for more information on the issue.
    The ClearType Tuner PowerToy can also help. If you are running Windows 7, it's built into the control panel.

  11. Re:More security? by shutdown+-p+now · · Score: 5, Informative

    As it says right there in the screenshot, the possible harm is from custom build steps.

    To clarify (since it may not be obvious to those who haven't used VS, and, in fact, even to many who did) - Visual Studio projects are nothing more but MSBuild makefiles, which has roughly the same expressive power and extensibility as, say, Apache Ant. In particular, the build steps can include file system operations, and execution of arbitrary shell commands. By default, VS-created projects have nothing like this, and so the verifier lets them load without asking. But if the file was hand-edited to include any such things, you'll see the dialog such as one on the screenshot.

  12. Re:I tried it, not impressed. by Anonymous Coward · · Score: 4, Informative

    Sadly the real problem is that it uses Windows Presentation Foundation, which renders text in an idealized grid rather than snapping glyphs to a pixel grid. In other words it renders the font exactly as it is supposed to look, and then blends this into the pixel grid, which makes it appear blurry. GDI, on the other hand, will deform the glyphs to fit them into the pixel grid thus offering better readability at the expense of accuracy.

    WPF can't render aliased text either, which some people prefer, and also has a silly text animation system that will wait a second before doing the blending meaning (to prevent scrolling artifacts) so that if you scroll fast you'll initially get really blurry text that gradually turns somewhat clearer.

    These are well-known problems with WPF, and has always been pushed as "by design" even though it results in horrible text. Microsoft seems to have changed their tune lately and will apparently be adressing the problem in .NET 4. I suspect the reason is because VS 2010 now uses it and the poor text quality was just not acceptable for an application like that.

    Messing with the ClearType settings in Windows doesn't really help much, as WPF's renderer is completely separate.

    WPF is just a big mess really, and far too complex. I'm all for a new GUI toolkit (even though it's .NET-based), but WPF just isn't it.

  13. Top 3 features by alphabetsoup · · Score: 5, Informative

    Quite a bit actually. Personally for me, the top 3 features are:

    • F# - Finally a functional programming language with a real chance of becoming mainstream. I personally would have liked Haskell though :(
    • Parallel Patterns Library - An STL like library for doing parallel computation. For example, instead of the STL for_each you can use the PPL parallel_for_each. Combine this with lambda functions for best results.
    • C++0x goodies - These includes lambdas, auto, rvalue references, etc.

    Apart from the above it includes a completely new intellisense for C++, using the EDG frontend. All this in addition to the usual .Net stuff.

    1. Re:Top 3 features by shutdown+-p+now · · Score: 3, Informative

      I personally would have liked Haskell though :(

      I'm sure it's possible to get Haskell running on .NET (in fact, rumor it is that it's what they were trying to do before switching to F#). The problem is that if you implement it using the existing CLR, you'll have to build your own framework for Haskell genericity and typeclasses, since CLR is not powerful enough on its own - and then your Haskell code will only be usable from other Haskell code. On the other hand, F# has the advantage that all its constructs are reasonably accessible from other .NET languages.

      That said, typeclasses on CLR level sure would be cool. And I recall Don Syme said something about them being a possible addition to F# 2.0, so who knows...

  14. Re:I tried it, not impressed. by shutdown+-p+now · · Score: 4, Informative

    The ClearType Tuner PowerToy [microsoft.com] can also help.

    It will not, since it only tunes Windows's built-in ClearType renderer. Visual Studio uses WPF, which has its own renderer. There isn't really anything you can do to make it better.

    That said, the blurriness is recognized as a problem, and will be fixed for 2010:

    We are replacing WPF's text rendering stack in WPF 4.0, and this should allow you to render text with comparable sharpness to what you're used to with GDI. The reason the existing text stack in WPF looks blurrier than GDI's is that GDI text is typically rendered with Compatible Width Layout, whereas WPF's existing text stack always uses Ideal Width Layout. Compatible Width Layout snaps glyphs to pixel boundaries, Ideal Width does not, which is why WPF's text looks blurrier than GDI's. WPF's existing text stack also does not support use of the embedded bitmaps that are included in many fonts and are intended to be used when rendering at smaller sizes.

    The new text stack in WPF 4.0 will allow Compatible Width Layout, and it will also support embedded font bitmaps. We believe this will solve all of our text blurriness issues.

    Thanks!
    -The WPF Graphics Team

    It just didn't get into beta 1 yet.

  15. Re:Whatever. by genghisjahn · · Score: 5, Informative

    Allow me to introduce you too... http://www.microsoft.com/Express/ There will be VS2010 versions. It's not the high-end dev environment, but it does quite a bit.

    --
    Sorry about the mess.
  16. Re:Yay! by bertok · · Score: 4, Interesting

    Screw it, I second this. Visual Studio has the best code completion implementation ever written. I can type lines like obj.GetSomething().Append(item) in about four keystrokes.

    Have you tried IntelliJ IDEA? It's Java-only, but I found its code completion to be many, MANY times better than VS. For one, instead of showing you every symbol that matches a prefix, it narrows down to the appropriate type.

    For example, in VS, if you create two methods with similar names ("int Test1()", "string Test2()"), and try and tab complete something like string "foo = T", it'll show you Test1() first, even though Test2() is a far better match.

    Note that I use VS 2008 daily, and I've got 2010 installed as well. I just tested that in VS 2010, and it still shows you every single identifier available, including class names. I know that technically, the intention may have been to reference a static, but in practice, they could go to some lengths to select a "most likely" set and an "alphabetical" set, and show the most likely first, and only show the complete set if you try to complete twice, or something.

    It's a great IDE, but it could be a lot better. Microsoft really needs to get over their "not invented here" attitude, install a competing IDE at least ONCE, try it, and learn that other people sometimes do things better.

  17. Re:Let me get this out of the way... by Rycross · · Score: 3, Insightful

    If you think that writing software in Visual Studio consists of "point and click programming" then you haven't used it enough to have a valid opinion, or at all.

  18. Re:Whatever. by wjsteele · · Score: 3, Informative

    And just to reiterate... it's been free (as in beer) for years!!!

    Bill

    --
    It's my Sig and you can't have it. Mine! All Mine!
  19. native ANSI C/C++ support .... by NullProg · · Score: 4, Insightful

    Doesn't exist in Visual Studio anymore without some tweaks. If your program targets multiple platforms beyond Microsoft your in for a few headaches.

    I wonder if Martin Heller used the VS10 compiler for cross platform Wx/Gtk/Qt development (Check Audacity out). I (or someone) should do this in a future slasdot review.

    The OpenWatcom, g++, and Intel compilers are a much better solution if your targeting multiple platforms (ARM, Mac, Power5, Mainframes, cellphones etc.) I use VS6 and GCC, but your mileage may vary.

    I appreciate the fact that Microsoft is pushing for VS studio C#/.Net acceptance. As of today, that solution is just as slow and portable as Java is/was ten years ago. For some strange reason I refuse to write a program that takes twenty to thirty megabytes of RAM to run when it should only take two. Why? Because that RAM belongs to the user and the other programs they may be running, not me. Waste not, want not. If you can do it faster and for less RAM in a different language then you owe your users to do so.

    And no, I've never written a C/C++ program that was un-secure (yet), thanks for asking. And yes, I like C#/Java programming, I just have deployment issues that I've never recovered from.

    My opinion or experiences may not be yours.

    Enjoy,

    --
    It's just the normal noises in here.