Comment 5 for bug 754583

We are doing user testing of the existing code, and this is one of the
key issues. So we are getting this bug discussed and some code written
in anticipation of a final report on that testing.

The challenge is that prior testing showed worse problems in the long
term with edge activation. I am reluctant to make this change so late,
especially since we don't actually have the opportunity to do long term
testing on the edge option either. I thought we could work up the
change, test it, and drop it in later BUT seb correctly points out we
have translation issues etc.

We know that long term muscle memory for corner activation is just as
good as the edge approach.

I think we can move forward based on previous testing, and address this
in Oneiric if needed.

Mark