I had never heard of the idea of keeping a ring of rings for a limited stack trace (the "history"). It's my experience that seeing the last few calls in a backtrace is usually enough for me to see where I went wrong.

There comes a point when you're just circling the wagons, be careful of that.

My gut feeling is that this guy is right. This kind of stuff is above my pay grade, so if I'm wrong I'm wrong, that's the risk with intuition.

I think that important decisions need to be made on firmer grounds than 'pythonicity'. Reminds me of when PG talks about 'yellowism'.

The deal with what's "Pythonic" is that Guido believes it's better to have Python be clear upfront about what it does and doesn't do well, and to have a relatively straightforward way to express those things. This is in reaction to languages such as Perl that have several ways to do the same thing, sometimes with counter-intuitive differences in performance. It's good to ask whether the benefit of adding feature X to the language is worth the cost of making it that much larger.

While I would choose a different subset of features, I think he's got a good point.