11:42:04scymtymi have two changes for sb-sprof. which should go in during the freeze? 1) fix for the regression reported by dougk__ 2) a tertiary sorting criterion that amounts to x before y <=> y is a transitive caller of x (ignoring mutual recursion). i originally added 2) to fix unstable sorting in the report while comparing the old and new behavior. now i think 2) may be generally useful
11:42:54stassatssb-sprof is not as mission critical to fear the freeze too much
11:44:55scymtymok, i'm pushing both changes then. the previous code changes combined with unstable sorting would have caused user-visible ordering changes anyway. might as well produce a more predictable ordering instead
11:46:58scymtymjust to be clear: i'm only talking about the ordering of nodes whose sample count is identical. otherwise, there has already been a predictable ordering