Allegedly HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\CurrentVersion is now locked at 6.3 (i.e. win8.1) and there are new, undocumented values HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\CurrentMajorVersionNumber and HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\CurrentMinorVersionNumber that need to be checked. cf http://stackoverflow.com/questions/31072543/reliable-way-to-get-windows-version-from-registry. Workaround is to use ProductName instead.
Oh that is nuts. Thanks for altering me to this change.
Related question: I only care about this because virt-install isn't autodetecting --os-variant on an existing disk image (or at least it's not documented as doing so). Should I file a bug report with them on this? Or are virt-manager dependencies on libguestfs a problem?
(In reply to Richard Tollerton from comment #2) > Related question: I only care about this because virt-install isn't > autodetecting --os-variant on an existing disk image (or at least it's not > documented as doing so). Should I file a bug report with them on this? Or > are virt-manager dependencies on libguestfs a problem? Yes, file separate bugs for anything that affects a different component.
Untested patch posted: https://www.redhat.com/archives/libguestfs/2015-November/thread.html#00137
I have validated the fix in a build from patched master.
For reference, the correct fix was: https://www.redhat.com/archives/libguestfs/2015-November/msg00141.html
Fixed in 1.31.26.