On Tuesday, 24 June 2014 at 16:37:06 UTC, Chris Cain wrote:
Of course, that's yet more playing around with restrictions. Plus your proposed relaxed stable sort is also the same as relaxed unstable sort so your restriction just made them identically fast.
By "relaxed" here, I was thinking in terms of not having to worry about the restriction of memory bounds (since you're required to create new memory anyway). Technically it's viewable from both the perspective of a restriction and a relaxation of a normally natural restriction.
