[hypertele-Xii]

D makes a compelling promise, but it's still ongoing metamorphosis. Changes too rapidly for libraries to take root. Every time I look at it there's a "new, better way" of doing everything so all the tools and libraries are out of date, at all times. And the whole thing doesn't sound so clean and elegant anymore.

[WalterBright]

All languages (even C to some extent) that are in wide use evolve towards new and better ways, and obsolete old ways. I don't even recognize C++ anymore, and I wrote a frakkin compiler for it.

[raphlinus]

Speaking of people who have written C++ compilers, what do you think of Circle? Have you given any thought to targeting GPU shaders from D? I think "C-like low level control but with strong compile time metaprogramming" might be something of a sweet spot, though there is a lot of complexity to the space (mostly the oddball storage spaces).

[WebFreak001]

there is DCompute for writing D programs on the GPU: https://github.com/libmir/dcompute

it's pretty awesome

[raphlinus]

Very interesting, thanks. I'll add that to my collection of alternative languages targeting GPU compute.

[WalterBright]

I don't know anything about Circle. Sorry.

[7thaccount]

BSG fan Walter?

[destructionator]

> Changes too rapidly for libraries to take root.

I have code that has worked unmodified for many years. The language doesn't change that often.