good theory, doesn't align with my (limited) experience. writing new code to be both 2 and 3 compatible was quite possible and not too tricky. six helped, `from

good theory, doesn't align with my (limited) experience. writing new code to be both 2 and 3 compatible was quite possible and not too tricky. six helped, `from __future__` imports, etc. however, everywhere i've worked still had some python 2 only code running - even to this day unfortunately. the real issue is 2to3 didn't work so well, or exposed the exact issues why python 3's breaking changes were needed (80% sloppy bytes/str/unicode handling).