Bee SmaRT

It's been quite some time since last post, we'be been really busy cooking Bee Smalltalk. You wouldn't believe how much work is needed in order to connect the dots! We continued trying to make libraries load in bee, but we felt that performance was abysmal, library loading was taking ages and tests were ran so slow that they were slowing down the whole development process. In order to measure performance, we ported a bunch of benchmarks to bee. Bee nchmarks The first benchmark we took was the simplest we know: tinyBenchmarks. It consists of a rather simple pair of metrics. The first one estimates bytecode speed and the second one dispatch speed. The bytecode test does a lot of arithmetic (which is executed directly without message sends) and sends only few messages: #new: , #at: , #at:put: . On the other hand, the dispatch test calculates Fibonacci recursively: ^self < 2 ifTrue: [1] ifFalse: [(self - 1) fibonacchi + (self - 2) fibonacchi +