On Saturday, 23 July 2022 at 16:42:44 UTC, Martin B wrote:

What do you mean? you do not like objective opinions?

On Saturday, 23 July 2022 at 16:50:28 UTC, ryuukk_ wrote:

I was referring to tagged union (including pattern matching) and nullable being std libraries in both D and C++ as opposed to new languages making them builtin types, as i kept mentioning in all of my posts (check history)

If one stops being objective and is emotional about opinions and features, there is no room for evolution no more

Understanding strength and weaknesses of projects helps make proper documented choice

I try to stay objective, no matter what

On Saturday, 23 July 2022 at 15:47:40 UTC, ryuukk_ wrote:

So… an important concept in language design is to keep the core language minimal and try to implement as much as possible as library constructs. This isn't specific to D or C++, this is a core language design principle.

Some languages are more explorative in nature… C++ and D falls into this category. So they accumulate some bloat by not adhering to the idea of having a minimal core language. Then you end up in a situation where you want to make a major revision after some years once you have figured out what the core language ought to be.

Until there is consensus on the minimal language then things will not change significantly.

Well, it is easy for the Carbon designers to say that they will break things with a light heart now, but that is not really an appealing property, so it remains to see what that means in practice.

On Saturday, 23 July 2022 at 16:50:28 UTC, ryuukk_ wrote:

An objective opinion is like an exact estimate.

See...

https://en.wiktionary.org/wiki/contradiction_in_terms

On Tuesday, 19 July 2022 at 16:27:25 UTC, Tejas wrote:

Thanks to LLVM new languages are popping up frequently now, which is a good thing. Carbon was not the first and will not be the last. In my opinion Carbon (why did they name it that) is dead on arrival because what I've read it offers zero novelties compared to for example Rust. Also the syntax isn't particulary nice.

There will be a language up ahead that will be smash hit when you least expect it. Carbon is not one of them.

On Sunday, 24 July 2022 at 19:38:53 UTC, IGotD- wrote:

It is too early to tell. Depends on how good they are at taking feedback and cater to seasoned C++ developers. Their first goal is to release a minimal-viable-product which is quite a big task since it involves all of C++17 and their own generics solution. It was stated that another team at Google is looking at interop between Rust and C++, and they are following efforts of adding lifetimes to C++…

Anyway, it all hangs on the semantics and how fast they deliver. Syntax issues are easy to fix later…

On Tuesday, 19 July 2022 at 21:25:28 UTC, IGotD- wrote:

Because I am at a time in my life where I can do things such as this, I have written about 10,000 lines of Rust. I have described it as programming while wearing a hair-shirt while on a bed of nails. What I learned from this exercise is that there is a very large cost, in the sense of a steep learning curve and continuing demands on the programmer, to providing memory safety without a garbage collector. The Rust Kool-aid drinkers, who appear to be mostly from the C++ community, don't seem to have much experience with garbage-collected languages and therefore don't understand the tradeoffs. It's the classic shiny new object to them.

If I were charged with developing an application where C or C++ would be an appropriate choice (a situation requiring absolutely predictable latency and/or small footprint), I'd certainly consider Rust as a safer, less bug-prone alternative. But choosing Rust in situations where Go or Nim or Haskell or D would get the job done is, to me, completely crazy. As a (now-retired) software manager, if I had a deadline to meet, I wouldn't even think about using Rust. And I say this having tested the waters.

On Sunday, 24 July 2022 at 19:38:53 UTC, IGotD- wrote:

All the disinterest/lack of belief regarding Carbon's (potential)success is really making me think about Go, where people said something about it not having used any of the research in type theory since the 1970s, coupled with their insistence on not having generics, a stupid error handling system all combined to make it stand no chance in the future.

But it's still popular today

Maybe there's a non-trivial chance Carbon will end up the same? As we have already seen, it's not always about the technical merit.

On Monday, 25 July 2022 at 02:56:53 UTC, Tejas wrote:

Thanks Docker, Kubernetes and the cloud native foundation ecosystem that sprung from them.

clang is already lagging in ISO C++ support because Apple focus on Swift, Google's loss of interest (we know now why, it is even mentioned on the talk) and everyone else not caring enough about upstream.

Google only needs to push Carbon on Android or something like that.

On Monday, 25 July 2022 at 02:56:53 UTC, Tejas wrote:

Well, Go has some key technical merits: solid GC, stable non-breaking language, fast spinup time (compared to Java), easy build process, web-centric standard library.

So it is ok for smaller services that are to be maintained for years and years. I don't think Go is a good language, but I am also not able to point to another language that is more suited for hosting micro-web-services.

How does Carbon fit into this? By and large, slightly higher-level than C++, yet potentially better performing.

How can Carbon achieve this? By doing the opposite of D: throwing out C.