Visual C++ Team Blog

Why did it take over three months to fix the fix, especially if it was just an error with the packaging/installer?

"Long test cycles" don't seem like a good reason considering the bugs which still slip through (e.g. the first fix not working).

I wish MS would react to these problems more rapidly. By all means take time to do a good amount of testing, but testing for three months gives you diminishing returns and still does not (and cannot) find all the problems. It's inevitable that problems will slip through occasionally and when they do we shouldn't have to wait *another* three months for them to be fixed.

If this was the only example of a fix-fix taking three months (or more) then I'd assume something just went wrong and wouldn't complain but it seems to be a common occurrence across multiple Microsoft products.

I'm not having a go at the VC team in particular as it's not just them. MS as a whole really need to improve on how they do testing and releases. And that may mean testing *less* and being prepared to deal with problems quickly.

Other examples which spring to mind: The amount of time it took for the data corruption bug in Windows Home Server -- which was known about when it was still in beta -- to be solved (one year). Or the way the mid-2008 Vista Media Center roll-up regressed back to XP's behaviour of unpausing/restarting video when the screen saver kicked in, which was not fixed for three months, and when a fix finally arrived it didn't actually fix all occurrences of the problem (we're still waiting!).

Hi dvy,

this KB article might help:

http://support.microsoft.com/kb/925336

--

SvenC

Thanks for the re-release, it certainly helps with a number of pieces of code I have.

There still seems to be a problem with the tr1::result_of machinery. The following code fails to compile, but it should (it does with boost). It seems to add a reference to all the arguments its passed.

#include <functional>

template <class T>

struct some_nested_result {

 template <class>

 struct result;

 template <class Func>

 struct result<some_nested_result(Func)> {

   typedef typename

     std::tr1::result_of<Func(T)>::type

    type;

 };

};

int main()

{

 typedef std::tr1::function<int (int)> f;

 typedef std::tr1::result_of<some_nested_result<int>(f)>::type

   type;

 return 0;

}

I've posted a connect issue on this.

https://connect.microsoft.com/VisualStudio/feedback/ViewFeedback.aspx?FeedbackID=425054

Should have some feedback shortly. Thanks again for all the effort getting TR1 in to MS.

[njoubert]

> There still seems to be a problem with the tr1::result_of machinery.

(This wasn't affected by the hotfix.) Thanks for the bug report. It's Dev10#641210 in our internal database, currently assigned to me. It won't be fixed in VC10 Beta 1, but (no promises) I believe that we'll be able to fix it in Beta 2.

[anon]

> This hotfix seems to make intellisense always freeze the editor for a

> few seconds when typing an identifier. Anyone know how to fix it?

I talked to one of our Intellisense testers, and we're not sure what's going on here. (This hotfix affects the libraries only, not Intellisense.) Our best guess is that your NCB file has gotten corrupted. Try closing the IDE, deleting your NCB file, and seeing if the problem goes away.

If it doesn't, please E-mail me at stl@microsoft.com , and I'll put you in touch with the right people.

Stephan T. Lavavej

Visual C++ Libraries Developer

Is there a re-distributable available for the newer runtimes?