Maybe D is right about GC after all !

[Ola Fosheim Grøstad]

On Tuesday, 2 January 2018 at 04:43:42 UTC, codephantom wrote:
> Well, consider the silent 'minority' too, who still think that 
> increasing performance, and reducing demands on resources, 
> still matter, a lot, and that we shouldn't just surrender this 
> just to make programmers more 'productive' (i.e so they can 
> ship slower GC code, more quickly).

I think most of the people in this minority (which actually I 
think was a majority a few years back) has given up on D as a 
production language.  I am certainly in that group.  It is 
starting to be a bit late to change direction now IMO. I mean, it 
is still possible, but that would require a mentality shift, 
which has been surprisingly difficult to establish.

Given the increased availability of memory in computers I think 
an application language with built in compiler-supported arena 
allocator will be a big win, but the only mainstream language 
that is going for this seems to be Golang. (Go is an application 
language, not a systems language).


> What it really comes down to though, is language designers 
> ensuring that any language that defines itself as a 'modern 
> systems programming language', gives control 'to the 
> programmer', and not the other way around.

Right now, I think only C++ and Rust fits the "modern system 
programming" description… GC and refcounting is for application 
level programming, so it shouldn't even be on the table as a core 
solution for that domain.

But D seems to be content with application level programming and 
that's ok too, but a bit misleading if you also say that you aim 
to be the best language for low level programming… I don't really 
think it is possible to be good at both without a massive budget.

You have to pick what you want to be good at. And that is the 
main problem with the evolution of D; a lack of commitment to a 
specific niche.