Is D really that bad?
data pulverizer
data.pulverizer at gmail.com
Wed Nov 2 05:46:09 UTC 2022
On Tuesday, 1 November 2022 at 21:48:35 UTC, Bioinfornatics wrote:
> 4. Promote some killer libraries (Gsoc is a good way to see
> those libraries as it is cuurently done). But they need to
> survive to this events, be maintained or added to the the std
> library.
> What is the state of d dataframe? Mir ? D ai ? D web framework
> back and front included through wasm ?
>
> To me, the community have to create those libraries instead to
> complain in order to transform an application language to a
> system language.
My suspicion is that if we want inroads to scientific computing,
D needs to create packages that provide seamless utility with
current solutions. People are not going to learn D just to use
yet another AI, dataframe, or linear algebra library or framework.
If an analyst is working in R or Python, you have to go to them,
meaning that at first you speak their language. So a high quality
alternative to a current tool that does something their's does
not, perhaps it performs better or is easier to use and so forth
with a backend in D.
In terms of getting "enthusiast" programmers to write D code,
you'll need to persuade someone using Rcpp/cpp11 or even the new
extendr (for Rust) that they are better off writing D code using
something like embedr or whatever. This is where people will need
to be persuaded why they should switch to D. Things like:
1. Performance - super-important in scientific computing. Can D
stand up to C++ in those terms? Easy and well documented access
to popular HPC frameworks.
2. Memory safety - if they already have Rust why should they use
D? How memory safe is D?
3. Ecosystem - what does D's language ecosystem have to offer?
This one is not a deal breaker but is still very important.
4. What's so special about D and why should they use it?
If this was 5-6 years ago, I'd have said that the barrier was
much easier to penetrate, but now things have become much more
competitive, and there are fewer low-hanging fruit. There are
still opportunities out there but efforts need to be properly
focused. I suspect it needs to be an ongoing affair rather than a
summer project. It's not necessarily about expending a huge
amount of resource, just finding the right focus to wedge the
door open and then progressively capitalising.
More information about the Digitalmars-d
mailing list