Civility

[Ola Fosheim Grøstad]

On Friday, 24 June 2022 at 09:14:04 UTC, forkit wrote:
> On Friday, 24 June 2022 at 09:03:48 UTC, Ola Fosheim Grøstad 
> wrote:
>>
>> ..
>> The only way to resolve this is for the core team to narrow 
>> down and announce clearly which application area the language 
>> is supposed to cover really well.
>
> I don't think that's enough.

Not enough, but it is generally difficult to keep a project 
focused without a clear target. How can you make a good 
evaluation of the design of a new feature if you don't understand 
the use context? You most likely can't…


> Someone else mentioned that D needs an experimental branch, 
> where the barrier is lowered to what can be put into it.

I said that D needs an experimental branch so that incomplete 
features like ImportC, @live etc won't be made available to 
casual users until it is known if it is at all possible to bring 
it up to being a 99.5% solution. A 90% solution is not marketable.

(The architecture of the existing code base is not suitable for 
evolution by outsiders, so something like SDC would be better for 
that. And that won't happen tomorrow or the day after.)


> Since most users come to D from major languages, with vastly 
> more features than D, they will not look nicely on a stagnant 
> language that is missing features that are very important to 
> them.

Actually, many C++ programmers seem to use Python as an add on. I 
would not bet on adding many additional high level features when 
the current ones are not completed.

I would bet on a higher abstraction level than C++, simplicity, 
cleaner syntax than today. Yet, clearly system level.

Being dragged into the feature set of C#, Rust and other 
«paradigms» is not going to end well. Those are distractions. 
Perhaps fun, but also very time consuming distractions that 
pretty much ensures that a percentage of the feature set will 
remain incomplete/broken/unsound/inefficient.

It will lead to a lack of focus and nothing will ever be brought 
to 99.9% that way.

It is quite clear already that staying focused is a problem, 
Walter now says he is looking at APL for inspiration and 
encourages the addition of async/await by volunteers.

It would be much more comforting if Walter just said flat out no 
to any new features before the existing feature set is fixed.

D needs to nail down a believable memory management strategy that 
is no more complex than Swift and that does not involve blocking 
"all" threads.

System programmers taking a break from their main language to do 
something in D are looking for something simple and clean with 
comparable performance, not something equally complex.

Start there, branch out to exploration of less critical features 
later.