Could D have fit Microsoft's needs?

[Exil]

On Tuesday, 23 July 2019 at 10:15:52 UTC, Russel Winder wrote:
> On Tue, 2019-07-23 at 03:01 +0000, Exil via Digitalmars-d wrote:
>> […]
>> 
>> The problem is who has control of these things. I've made pull 
>> requests to dub, they just sit around and nothing gets done. 
>> Very few people have access to actually enact change. Don't 
>> say stuff like "you have to do action yourself" cause it's not 
>> on me or anyone else. There's hundreds of pull requests for 
>> DMD and DUB combined. There's only so much people can do when 
>> pull requests sit idle for months/years, especially worse when 
>> there's no responding. When something somewhat default 
>> appears, no one on the D team is willing to do anything about 
>> it. They all look to Walter, and Walter is usually to busy 
>> doing something else so he never gets around to it. But yah, a 
>> very naive thing to say considering the current state of how 
>> things are managed with the project. God bless Seb for having 
>> enhanced permission status ^TM.
>
> Go and Rust got this right from the very outset – make the 
> standard library as small as possible and have everything else 
> provided through user packages/crates. Go only got this partly 
> right in that there was no central repository, just DVCS 
> packages. Rust got this very right in that it has a central 
> repository and Cargo can use DVCS crates and even local file 
> crates. This creates a vibrant community with some competition 
> (which has bad as well as good sides to it).
>
> I am sure Dub can do for D most of the things Cargo does for 
> Rust, but the crucial difference is Phobos in D vs std in Rust. 
> Phobos is driven by the "batteries included" philosophy which 
> is a philosophy past it's time. Python and C++ are also failing 
> to come to terms with this.
>
> What D needs is to rip everything out of Phobos that is not 
> actually needed
> for the compiler to compile code and put it into the Dub 
> repository as
> separate packages – this is what the Rust team did, and it was 
> the right thing
> to do.
>
> If this means creating D v3, then that seems like a really good 
> thing to do.

Dub will probably never be as good as Cargo. I actively avoid Dub 
as it is. Even its documentation is atrocious, I loath having to 
look through it when I am forced to. There's no one on the D team 
with a vision for it, to make sure it gets to where it needs to 
go. No one actively developing it. This isn't something you can 
just thrust on someone else like someone in this thread was 
proposing to do with their "just do it yourself" mentality.