On Tuesday, 14 June 2022 at 07:09:23 UTC, Ola Fosheim Grøstad wrote:

D is honestly a mess compared to when I first started using it over a decade ago, back then it was simple and you could easily do something in it.

Now you have this huge amount of attribute soup that you either need to sprinkle everywhere or you have to completely ignore.

The community and libraries etc. also feel a lot more divided, even compared to when tango was a viable alternative to phobos; at least that's how it feels from my point of view.

Also it's like D chooses the most complex implementations for every new feature that has to be added.

I came to D because it was easier to use and get things done in than any other languages, but over the past couple years I've slowly stopped starting new projects in D because it's becoming a drain to use and all its selling points are no longer selling points because other languages are either on pair with D or does it better.

D was on the path to greatness, but no more. It's spiraling down a bad path and it's honestly sad. It had so much potential and I'm not sure it'll ever recover.

The only way for D to ever succeed would be to start D3 ASAP and start by figuring out how to get rid of the attribute soup etc. because all it does is adding clutter and clutter is what makes a language difficult to use. C++ wasn't hard because of the technical stuff, but because there's a lot of clutter in the language with templates etc. IMHO.

On Tuesday, 14 June 2022 at 07:34:32 UTC, bauss wrote:

It is my viewpoint that D would have been more successful if it had chosen to be simple system level language without GC/templates and instead focused on high quality builtins and high level optimizations… but that ship has sailed for sure!?

Rust is taking that space, except it isn't all that convenient. I don't have much belief in Zig and the other newcomers… C++ is just too far ahead at this point for me to view minor system level languages as interesting. And by not being interesting, I mean that I am not even willing to download and give them one spin…

There is also much less need for system level programming today than 10 years ago, so it is a shrinking market with more competition… Only 2 or 3 can succeed in building an eco system in that space and also reach an acceptable level of portability (full support for various hardware, iOS, Android, compilation to FPGA, etc).

Depends on SDC, if SDC implements the core language and don't sacrifice compiler internals to support other cruft then I guess you could attract compiler devs that would evolve it into a clean slate D3.

You could reimagine D3 as a high level language with the possibility of going system level. If you design from that angle then all the clutter will vanish as you cannot tolerate clutter in a high level language design.

Clutter in D has for the most part been added in the name of "system level".

(Which is a bit odd, because system level programming is even more in need of clean simplicity due to the drastic consequences of making a misstep.)

Well, there isn't all that much incomprehensible clutter in C++ anymore if you want to avoid it, but you still have to deal with "evolved surprises", so it isn't easy for beginners.

C++ cannot become acceptable for high level programming though, as it has a heavy focus on enabling compiler optimizations. As a programmer you have to focus heavily on the correctness of your code, rather than relying on wrongdoings being caught. That makes C++ unsuited for evolutionary programming. You basically need a design before you code in C++. So C++ is not a good alternative for D programmers who like to experiment and prototype.

On Tuesday, 14 June 2022 at 07:34:32 UTC, bauss wrote:

Totally agree with this, it puts me off using D. TBH I haven't started a project in D for a while now, I am simply reaching for Python or C++20. Both get the job done and while not great C++ is at least moving in the right direction.

I much prefer Phobos over Tango but I agree the core development community seem very divided. Instead of trying to grow the user base with a great experience, adding syntax sugar and frition reducing features like int[$] arr = [1, 2, 3]; etc. But no, a massive amnount of energy is spent on chasing C? Seriously?? C interop is hugely important but D already interop'd with C seamlessly enough....clearly not enough for a core few because that is the main focus for D development right now.

This is also my experience, unfortunately it is at the point where I do not start new projects in D.

Totally agree but I wouldn't say sad, it just is. I have moved back to C++ for hobby projects and I'm rather enjoying C++20. I also like the fact I'm staying up to date the latest C++. There are a lot more C++ jobs than D jobs out there.

Agreed, I like the improvements in C++20 but it has a long way to go to make things readable That said int func(auto value) {} is pretty easy to read template. D code now with attributes, I wouldn't know where to start and TBH would probably just ignore them.

On Tuesday, 14 June 2022 at 08:35:53 UTC, norm wrote:

Do you combine Python and C++, i.e. call C++ code from Python? Or do you "chase" higher level and lower level aspects in different projects?

It is being done for fun. It is not unreasonable that people do things they find interesting in their own spare time.

The problematic part is that incomplete features are being merged instead of being polished in an experimental branch. Even more problematic when it is stated that the feature perhaps will never become complete (macro expansion). Such experiments should never be done on the main branch and that pretty much defines almost all of D's challenges.

There is quite a difference between this practice and C++ compilers making features available for preview when they are following a formal spec!

It is a bit sad if you count all the hours put into attempts to build an eco system around it. The current approach guarantees a mode of perpetual instability, and that prevents an eco system from emerging.

It would be much easier to build an eco system if you had one big fat breaking change to set the record straight (D3) than what is perceived as random movements. If people feel you are moving about randomly they become passive.