OMG that's insane.

Well, our Windows builds may just have to move from 2005 to 2010 and plan to be stuck there for the forseeable future.

I was hoping I could use a C++11 compiler for Windows development someday. But it's probably more likely that I'll stop developing for Windows before my company's customers stop running XP.

2010 builds them OK. You can also look at mingw with g++ if you want to try some of the C++11 features on Windows XP. I was surprised to hear that MS did this. XP is still widely used in industry and is a good baseline platform to target.

2010-built executables only work on XP SP2+ btw, so they've been creeping it forwards every release for a while.

> XP is three years past the end of official support and only two years until the end of extended support

I don't think I've ever met anyone who managed fewer than 25 servers who has ever looked that that calendar.

I know you mean to point out how old and crufty XP is (and that is true), but if you stop and think about it... it's really a testament to C++ and Microsoft engineering that it has done so well for so long.

Well that, and

* the long gap between XP and Vista

* that it was still selling on new netbooks until quite recently, and

* the leveling off of CPU performance.

Customers don't care about 'generations'.

If it works for them, they will not break it.

That said: XP will eventually fade off, but not because of MS official support or extended or whatever. It will fade because end users will eventually stop installing and using it.

Our company only recently started the full roll out of Windows 7, and some computers still require Windows XP (for accounting applications).

These few computers will use Windows XP for the foreseeable future, official support or not.

And they can still use previous versions of VS to develop with XP. It won't stop working, it will gradually fade to governments and big businesses with inertia and businesses who don't want to pay to upgrade.

Then to smaller and smaller niches.

I'm not saying it should die, or it is bad, but that it is old and new tools are not 'insane' to not support it. IE9 came out almost a year ago and doesn't support XP, for example. DirectX vNext in Vista isn't XP compatible.

I'll go out on a limb here and predict that XP extended support will itself be extended. There are simply too many desktops out there using it to be ignored.

I'm afraid XP won't be ignorable for another 3 or 4 years at least.

According to: http://gs.statcounter.com/#os-ww-monthly-201101-201201 XP dropped 14% in the last year, and is now on 34%. So a really dumb calculation says about 2 and a half years if that keeps up, but I guess there will be a longer tail of die-hard users... hard to tell, since it's pretty unprecedented for a particular version of a major OS to be so popular for so long.

So I guess in 2 or 3 years not supporting XP will only cost you 50 to 100 million potential users.

Microsoft stops providing XP security patches in April 2014. Any organization that wants to use XP after that will need to pay Microsoft to maintain support. Any organization that continues to use XP without support is irresponsible.

Yea, it is called a Custom Support agreement. Costs $50,000 per quarter for first year and increases every year, BTW.

As "Mike" commented there:

"One big thing that works against us having any sort of nice workaround for this, is the fact that as of Visual C++ 2010, the CRT and MFC rebuild makefiles are no longer included with the source code. In 2008 and earlier, it was so easy to rebuild CRT and MFC DLLs because there was a clear set of makefiles included. Now all there is, is a silly "these are the compiler and linker switches we use, go find your own solution to rebuild these" web page."

It's also not something that would take them work to do, as the older library versions already implemented the code paths needed for Windows 2000 and XP.

I suspect MS wants the support for older OS-es banned to make new apps unrunable on Wine too.

> Status: Closed as "By Design"

Some people might consider that a feature.