I still think it shouldn't go in. This is the only feature in Engine 2 which might make non-OOP people convert. Once this isn't in Engine 2 we don't have a carrot for them.
Do you really think that people will convert en masse for this feature? There are multiple open-source extensions that provide this functionality (plus Thies' ZE backported patch) that enable debug_backtrace type functionality. Anyone who wants it as a feature can have it now. The point (IMHO) is that a) the work porting it is already done, b) it's extremely valuable (at least to a number of people who frequent this list), and c) its a feature that belongs in the engine (as is validated by your choice to write it into ZE2).
People will convert to ZE2 because it will eventually be the default in the stable release, and what vendors ship in their distros. And that will happen because of code-base stability and because it is the line of current development, not because it has a particular feature that php4/ZE1 doesn't have.
I'm looking forward to ZE2, but until it's stable I will use something I know and trust in production. And the ability to dump a backtrace is extremely valuable in large-scale application design. Why keep this away from people? Hack workarounds to support this sort of basic functionality shouldn't be necessary in a language with such a rich developer pool.
Just my 2 cents.
George