Lambda the Ultimate

One (probably unobvious) lesson is that choices taken by PL designers are guided by their envisioned applications - and the target audience.

While it may be ok for Bigloo not to use trampolining to implement TCO on JVM because of potential performance loss, there are applications for which TCO is much more important than difference in performance by factor of two or even more. One such application I witnessed recently is persistent business processes (to a less degree the same effect can be observed in continuation-based web frameworks).

This may sound trivial to many of the readers, but how often we see people discussing which PL is better without qualifying - for what.