I've been giving D a shot at making my usual learn new language project, making a JRPG styled game. So far it's been looking pretty good while porting an old engine someone wrote for their senior thesis from D1 to D2. The type system is fairly well featured and flexible to the point where lots of bugs while updating code and porting it to the new standard library and language features were caught by the compiler itself. It's at the point now where while it builds I'm having issues with the OpenGL code since there isn't a good binding that supports OpenGL 2.1 the way the engine was originally using it.

https://bitbucket.org/simcop2387/yage

The other side of that is that D has had a long time to percolate and hasn't found the success of Go, and soon Rust. So, we come back to the common startup ailment, which is having an awesome product but it just inexplicably doesn't get traction, while the inferior (depending on perspective) product does.

I know there are pockets of D throughout the industry. I know it's catching on at Facebook (and, because I know Facebook folks, I know what an uphill battle those folks are having too). I just don't hear about D nearly as much as I'd expect to over a decade in.

Nothing against D, mind.

In my opinion the fact that there were two competing standard libraries (Tango and Phobos) certainly brought confusion and needlessly fragmented the community.

IIRC, at the beginning the whole thing was not being developed in the open and the compiler was proprietary (although the front-end became open source at one point).

Those two things (especially the former) pretty much killed it for me.

D finally matured at a time where I was already more interested and invested in the Lisp/ML family of languages, thus I ended up never considering it for a project.

Those were definitely the two things that kept me from really considering it in the past. Both of those have now changed for the better and there's a more active community building up again. I think the facebook use of D has also brought some much needed attention to it to the point where the language itself is starting to accelerate changes to move forward. The DMD reference compiler still has some restrictions on the backend of it I think but because of the frontend being available there's now LDC and GDC that are LLVM and GCC versions of the compiler that do much better at optimizing and are fully open source anyway. It's also seemed to help find bugs in both the language spec and the DMD backend from what i've read on the news groups because they've got three very different implementations of the language now that are all mostly compatible (the DMD version is always a little ahead since the other two are essentially ports of it to LLVM/GCC).

The D front end was just relicensed from GPL to Boost. (The front end is shared by DMD, GDC, and LDC.)

I knew something had changed but was unsure of the details. Thanks for the correction.