I left Dyalog in June and have since been taking a break from employment to work on BQN roughly full time. Eventually I'll get another job, but I hope to keep working on BQN and won't seek to commercialize it.
I hope to have BQN running directly on at least Wasm and x86 eventually. Currently I'm focusing on the self-hosted parts of the implementation, so the VM is written in whatever's convenient, portable, and fast. Because the VM is very small it's easy to port between languages—the Go port I did in about three days without knowing any Go beforehand. But manual memory management adds a lot of difficulty, particularly because closures can form reference loops and have to be garbage collected. I'd rather stay in a memory-managed environment for as long as I can to work on the unknown aspects of BQN before attacking the known but hard memory management problems, and when I make that jump I might skip directly to assembly/machine code because I have more control that way.
Since I can't edit my current comment, I'll add that I wish your current project the best of luck and hope it morphs into something that people want to use and can use for development.
Comments
I left Dyalog in June and have since been taking a break from employment to work on BQN roughly full time. Eventually I'll get another job, but I hope to keep working on BQN and won't seek to commercialize it.
I hope to have BQN running directly on at least Wasm and x86 eventually. Currently I'm focusing on the self-hosted parts of the implementation, so the VM is written in whatever's convenient, portable, and fast. Because the VM is very small it's easy to port between languages—the Go port I did in about three days without knowing any Go beforehand. But manual memory management adds a lot of difficulty, particularly because closures can form reference loops and have to be garbage collected. I'd rather stay in a memory-managed environment for as long as I can to work on the unknown aspects of BQN before attacking the known but hard memory management problems, and when I make that jump I might skip directly to assembly/machine code because I have more control that way.
Since I can't edit my current comment, I'll add that I wish your current project the best of luck and hope it morphs into something that people want to use and can use for development.
Ah that makes sense and sounds good!