On Sunday, 15 June 2014 at 17:19:42 UTC, bearophile wrote:
So you are missing most of the point of using D, and you are perhaps even relying too much on the D GC, that is much worse than the JavaVM GC.
I am aware of the D GC issue and know that I need to take another route to get the most out of D, but to compare languages it's also interesting to see how they compare with similar implementation.
The GC might never be as performant as Java's but it's nice if you can estimate to what extent it is slower and maybe trace improvement of the GC implementation over time.
The usual way of microbenchmarking is another perspective which should be looked into too but this viewpoint might show weaknesses better and might lead to clues on what should/can be improved.