How much has the complexity of the underlying computing platform grown in that time?
We’ve come a long way. Architectures are much easier to target by compilers. But at the same time, compilers have increased their complexity enormously. We’ve seen the microprocessor architectures get much more complex and we’ve seen compilers in general get much more complex. It used to be that an optimizer was somewhat of a novel feature in compilers — only the most exotic compilers had an optimizer, or really good ones. Nowadays it’s pretty much a prerequisite. One of the implications is that there’s a lot more reluctance than there used to be to try out something new. Or it’s much more difficult to introduce something radically new. I think 20 years ago people were more open to work with radical new languages like Forth and things like that, playing with microprocessor architectures that didn’t support the same instruction sets and so on. Now we’re in a position where the investment is so high, and [with] the infrastructure that exists, that any revolutionary idea needs to have an
