Tuesday, 21 July 2009

Windows 7 Upgrade Paths: Uh Ohhh


Microsoft updated it's documentation on the upgrade paths for Windows 7. The details are contained in a document located here:

http://www.microsoft.com/downloads/details.aspx?displaylang=en&FamilyID=e170eba1-5bab-401f-bbf5-00f0ee7fe0fb


This is a pretty key document as it explicitly details what upgrade scenarios are NOT supported including;


  • Windows 95, Windows 98, Windows Millennium Edition, Windows XP, Windows Vista® RTM, Windows Vista Starter, Windows 7 M3, Windows 7 Beta, Windows 7 RC, or Windows 7 IDS
  • Windows NT® Server 4.0, Windows 2000 Server, Windows Server® 2003, Windows Server 2008, or Windows Server 2008 R2
  • Cross-architecture in-place upgrades (for example, x86 to x64) are not supported.
  • Cross-language in-place upgrades (for example, en-us to de-de) are not supported.
  • Cross-SKU upgrades (for example, Windows 7 N to Windows 7 K) are not supported.
  • Upgrades from Windows Vista to Windows N, Windows K, Windows KN, or Windows E are not supported.
  • Cross-build type in-place upgrades are not supported.
  • Pre-release in-place upgrades across milestones (for example, Windows 7 RC to Windows 7 RTM) are not supported.

And, hidden right in the middle (where I try put things that I don't want people to find) is Windows XP. Ok folks, as part of the BETA briefing of Windows 7, we were told that there "might" not be an upgrade path for Windows 7 from XP. This was caused a bit of a stink - and, it looks like "might not" has become "definitely not".

I don't think that this will affect corporates/enterprises in a big way, but the 5000 or mid-sized enterprises that Microsoft targets are typically consumers of a "upgrade" methodology rather than a clean install (generally with a new machine). We will have to see what this means in the industry, but I could see that the lack of an upgrade path from Windows XP could be a big issue.
Oh, and you might have guessed it, Windows N is not a supported upgrade path as attempting to compute this result set would have generated a NULL data-set and a biological equivalent of a divide by zero error.

No comments: