Well, all that effort and now my manager and his manager have convinced me to wait 'til I return from the PDC to checkin.  Normally I'd checkin now, but Chris will be on vacation next week and I'll be at the PDC so it's better to wait 'til one or both of us is around just in case there are problems.

Not that I expect problems, of course.  :-)

Seriously, the pervasive nature of this checkin does warrant extra care, which is I ran those 300 tests.  Turns out that only 1 out of all 300 needs to be investigated.  Here's how I determined this:

  1. First I ran all 300 with my changes.
  2. Then I ran all 300 without my changes.
  3. I compared the results looking for those that failed with my changes, but succeeded without.  (All others were ignored.) 

It turns out there are 4 in this category.  3 of them passed when I re-ran them.  I'm debugging the remaining one tomorrow since the VC team doesn't want me to checkin until it's issues are resolved/understood.