On Thursday, 9 March 2017 at 07:24:12 UTC, aberba wrote:
So technically and from experience, the current state of D is not the primary issue?
I don't have enough experience with D yet, hopefully someone else can tell you better. But my two cents.
"Current state" is a very general thing. Going back to your original post, I would not really think memory management performance to be an issue at all (in the hands or a competent programmer who avoids 'new' inside loops etc); specially for the kind of software you have in mind, or even for the most demanding real-time applications, where D provides some possibilities -- maybe not as idiomatic as the GC 'new', but much better than getting void pointers from malloc().
Even in C# for example one's able to make things like this,[1] granted it's not quite Photoshop but quite nice; and there must be more impressive examples. So I don't think application developers should worry so much about GC pauses of death.
Also regarding the "current state", another issue is the availability of libraries that you can use in your project. Of course D can't yet compare to C++, Java, Python, or .NET in this regard, even though there are lots of nice things around (plus you can call C and C++ and the bindings are often already available[2]). However I think that providing the demand from a killer app project could get a lot of support from the existing community in completing useful library projects.
I get the impression that most people in the D community are computer scientists or engineers who use it for backends, or more often for pet weekend projects.[3] At work they are constrained to work in another, more commercially successful language. Just like Andrei at Facebook.
[1] https://en.wikipedia.org/wiki/Paint.NET [2] https://github.com/filcuc/dqml [3] http://code.dlang.org
