Talk on D at DORS/CLUC

[Lance Bachmeier]

On Monday, 20 May 2024 at 08:03:12 UTC, RazvanN wrote:

> Templates are quite cool and they do appeal the inner nerd, but 
> the reality is that not so many people use templates. If you're 
> not a library writer, chances are you'll stick to using normal 
> functions.

The complexity of templates is probably more important (though I 
agree with your point). Maybe after years of using the language 
they will have uses here and there, but templates are not a 
positive when showing off the language. Same for ranges.

> After my talk, some people that were using C++ came to talk to 
> me. They're biggest problem was they were having buffer 
> overflows which corrupted memory that were not caught by their 
> linters. They liked the idea of automatic bounds checks and the 
> fact that you can incrementally transition to D, however, they 
> were wondering how safe is D when you are not using the GC. 
> They also mentioned that they considered using Rust, however, 
> it's pretty much an overkill for their use case. I told them 
> that for their use case they can simply manually allocate 
> memory and then create an array which contains the size 
> information. The compiler will then check every access to the 
> array for out of bounds indexes, however, they will still need 
> to manage the memory manually. All in all, they seemed willing 
> to try D, however, we cannot since if they will actually do it.

D has a GC that is the source of many complaints, but it also has 
SafeRefCounted. Maybe I don't keep up with the C++ community, but 
I thought they were fans of RC, based on the repeated claims that 
you should seldom need to do manual memory management in C++. 
SafeRefCounted is part of Phobos so it's not like you're relying 
on a third-party package author to maintain it.