I'm just wondering if we could do a better fix... There might be other code (other regressions) depending on BufferID to be initialized to invalid. So if we don't fix BufferID itself, then we can't be sure we've caught all the regressions of this kind.
Verified the fix works.
I'm just wondering if we could do a better fix... There might be other code (other regressions) depending on BufferID to be initialized to invalid. So if we don't fix BufferID itself, then we can't be sure we've caught all the regressions of this kind.