But the halting problem for a subset of programs isn't The Halting Problem, which is the question whether a UTM can exist to calculate any function, over its entire domain.
The parent's wording was a little sloppy, but the point was that the grandparent was incorrectly using the Halting Problem to argue that you can't even do it for a meaningful subset of programs, which of course does not follow.
http://research.microsoft.com/apps/mobile/news.aspx?post=/en...