This is a comment I wrote today in Mini-Microsoft's blog:
I think it is good that Microsoft shipped Visual Studio 2005, even in this state. At least RTM means that it has support, while CTP does not. I have been working intensively with the IDE for months and I know it is still buggy, but most problems have workarounds, and Microsoft can afford this situation better than the no-ship situation.
I don't need a service pack in six months. I prefer the 20% of the bugs that make for 80% of the annoyances are fixed much earlier. For me, that 20% means the "Visual Basic background compiler crash", and the "IDE vanishes while opening solution" bugs.
By the way, Microsoft right now has no means of estimating the actual number of occurrences of these crashes, because most developers click on "don't send" button when the Watson tool surfaces. You can read details here.
UPDATE: I must add to this that when the IDE simply vanishes, there is no Watson submission. Also, talking about workarounds: For the first, the best is usually to restart Visual Studio, clean the solution and rebuild. Besides, if you unload projects you are not updating, Visual Studio will be faster and more stable. For the second bug, my workaround is to create a blank Windows Forms VB project and exit the IDE. Next time you start it, your solution will probably load.
So, in the end, I agree with those that say VS 2005 had to be shipped. I love .NET 2.0 and I am happy my company took the risk some months ago to adopt it.
However, I think we are witnessing that Visual Studio code base and/or Microsoft's software process failed to scale to the level of complexity that the products required.
Of course, I am just guessing based on what I have read.
I have read about expert developers needing months to get familiar with Visual Studio code in order to fix bugs. I also read about developers not being able to fix bugs without triggering regressions. I have heard of Rascal, the lightweight IDE many developers choose internally over Visual Studio.
To some extent, this is all normal. But if things are getting worse at an exponential rate, as happened with Vista, then maybe it is time for a "reset".
This reset would hopefully imply both code base and process refactoring.
Is it time for Microsoft to go shared source with Orcas? Whatever helps streamlining the process and the code is good. Sometimes the sole process of preparing the code for being scrutinized and extended by an external community is worth the ticket. However, I expect 90% of future development would be by Microsoft personnel anyway.
I just hope they will think about this possibility.
But my most urgent need is that they fix VS 2005 steadily, and I need them to start right now.
They do that, or I will loose all my nails and hair!
UPDATE: Thinking more about it, I have the idea that the software development process have improved a lot at Microsoft over the last several months and few years. That is not to say that the process is good enough, but it is improving.
Then maybe the main problem is the code base, or perhaps the goals are simply set too high.
Regarding SQL Server 2005, I think it is a much more solid release, but I haven't really used it that heavily.