Quantum computeras need to be invented before they are popular.
Quantum computeras need to be invented before they are popular.
Quantum computers (compilers?) must be understood before they're invented o_O
Originally Posted by Adak
Eh, no; try actually reading the thread. Neo1 has been defending his idea that Rust is an improvement over C++, rather than a waste of time as some others assert. There has been a hell of a lot more Rust hate in this thread than C++ hate.
Popularized? Quantum computing is not yet anywhere near useful, it's current state need a lot more than popularization.
If quantum computers need more than an abstracted and simple understanding of modern physics, I'm doomed if I live to see them!
But as one does not need to master electronics or VLSI design to be a good C++ (or..err..Rust) programmer, there is still some hope.
You're right. But there's a point in time where it's okay to let it go and just walk away. I guess it was a discussion about languages and possible things to fix them which I like but I feel like that ended a couple of pages ago. And I feel like a lot of this started because the OP mentioned Rust being a replacement over C and then it all turned into why Rust was superior and why it wasn't even though Neo never specifically said that they felt that way themselves. Though the adamant defending might indicate so and part of the fire was ignited when Neo mentioned people being skeptical of a new language as being old and dusty C++ programmers.
Ironically enough though, I did the same thing when posters here said they didn't like the Gnome 3 shell for Linux. But that is true though.
But I do agree, I don't understand people who use a C++ compiler and only write C code. You like constructors? That's not that hard to mimic in C!
And I don't get what's so hard about QC. It's just reading/writing binary from/to electron states and because heavier elements are so unstable you need to use multiple low-electron atoms and then you have a simple superposition which coheres upon quantum measurement into the appropriate binary string. I don't see how any of that would be hard to engineer at all. Nope. Not one bit. Or should I say... qubit?Popularized? Quantum computing is not yet anywhere near useful, it's current state need a lot more than popularization.
That was mentioned in the opening post.
True, Neo used the words "hyped as" in the opening post.
However .... in the next few posts, there was a question about the origin of "crate", an expression of scepticism about guaranteed memory safety, a comment that many hyped replacements for C have not replaced C, a query about whether this (Rust) was really needed, and an observation about how inventors of new languages often seek to replace other languages and historically rarely succeed.
That lead to ....
which Neo1 did in post 7, along with some claims about advantages of Rust over other languages, particularly C++.
That resulted in a number of posts with people pointing out that C++ was not as weak in those areas as Neo portrayed, and Neo responding to make points about Rust superiority, with comments that demonstrated narrow (and some incorrect) understanding of C++, all with the tone of one speaking to unenlightened bigots.
That has nothing to do with the virtues or otherwise of Rust, but Neo certainly took discussion down that path.
Last edited by grumpy; 06-14-2014 at 09:25 PM.
I don't personally consider quantum computers the next architecture. Despite the research being done, I find the whole notion too much into the realm of science fiction. We'll probably first experience a new generation of computers based on biochemistry, DNA and/or molecular biology.
Anyways, the merits of a programming language have really very little impact on that programming language ability to phase out a well established competitor. It really, for the most part, irrelevant how much better or worst Rust is compared to C and C++. People will just not flock to Rust because it is better. People will just not stop doing C or C++ because they are worse. The gains have to be significant for industry well established practices to die out. This is true of any industry, not just software engineering. We need more than a discussion on type safety. Many languages, including interpreted ones, offer much better solutions than c or C++ in particular areas of programming. Rust is just another one.
If Rust wants to establish itself in the industry, it better offer C and C++ interop. That's the simple truth of it in this industry of ours.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
But by virtue of being interpreted or just very high-level they cannot compete with C in the areas that have made C as popular as it is. Rust is one of a number of new languages (see Nimrod for example) that offers many of the features of high-level interpreted languages while still retaining the low-level control of C.
You're completely right. And it does, at least for C.If Rust wants to establish itself in the industry, it better offer C and C++ interop. That's the simple truth of it in this industry of ours.
The fact that you so readily offer such an analysis of the tone with which i'm communicating, but for some reason refrain from doing the same for the first few replies to this thread (or indeed many of the later ones), means i cannot take you seriously in the slightest.
How I need a drink, alcoholic in nature, after the heavy lectures involving quantum mechanics.
Not convenient to your wish to blame other members rather than yourself for this thread going off the rails, eh? The tone of the first few replies were either seeking background (the query about origin of "crate"), scepticism about one of the features, and observations based on previous experience (the others).
You're the one who made the first disparaging remark about another member. Yes, I've disparaged you since, but my first post in this thread was after you had posted several disparaging remarks.
I don't find Neo1's intent other than friendly discussion. The internet is a closed book. No need to try to read others' feelings or expressions.
If you consider disparaging comments - such as characterising someone who questions you from a different perspective as "old and dusty" - to be friendly discussion, then fine. I do not.
I find no problem in it. It seems to be more like a casual comment you'd make jokingly in casual conversation.
Rust is still under development, but as of now it doesn't offer yet full interop with C. Structs aren't yet directly mapped to Rust type definitions and calls to C functions is still under development (iirc to avoid the stack switching that currently occurs).
Once completed, C interop will dictate Rust success as an usable application/systems programming language. Not because this will mean Rust is ready to phase out C, but because it can wrap C code or be wrapped in C code. This has been the fate of all major programming languages targeting the leading ISA in the industry. i.e. to become usable by certain niches and ignored by everyone else.
As for C++, it is hard to imagine Rust phasing this one out too. Particularly because C++11 addresses most of Rust improvements in one way or another. Lambdas and templates can implement Rust actor model (which I find, like Erlang, to be Rust biggest contribution considering how awful the idea behind threads always was). C++ meta programming can handle Rust type reasoning and auto and decltype can help with code branch analysis.
Certainly this increases C++ boilerplate code. It may even be seen as counterproductive. But were I to want to create something similar to Rust in C++, I would have the means. And this is exactly the question many will pose: Would I want to do it? The answer is almost always a resounding no. There's no point. Rust style brings nothing major that warrants the effort. And also because of this, it doesn't justify the tremendous effort of coding and maintaining a new language. Some specialized outfits will move on to Rust, most just won't see the gains in doing so.
Originally Posted by brewbuck:
Reimplementing a large system in another language to get a 25% performance boost is nonsense. It would be cheaper to just get a computer which is 25% faster.
The Rust Foreign Function Interface Guide
The Rust FFI no longer uses stack-switching when calling external C functions. As far as the C interface itself, it is still under development like the rest of the language, but it is nonetheless fully functional. Plenty of bindings have been written to C libraries at this point.
I don't agree that C++11 addresses the most significant improvements that Rust brings to the table, which is also why i don't agree that the creation of the language lacks justification. Regardless, your point about adoption/migration still stands. For a language which is not yet finished and for which adoption is being actively discouraged by the creators, Rust has gained significant traction. In 2013, the Rust repo closed the second-highest number of issues on all of Github and Mozilla has already begun work on Servo. The language is being hyped immensely, which i hope is a sign that we might see a change of attitude. Conversely, the hype might just be the result of me isolating myself in a sort of echo-chamber, i won't rule that out. I do hope that you are wrong, since i think it is high time we faced the fact we will always be dealing with shoddy software as long as we base everything on a language where program correctness is as much of an afterthought as it is in C.Once completed, C interop will dictate Rust success as an usable application/systems programming language. Not because this will mean Rust is ready to phase out C, but because it can wrap C code or be wrapped in C code. This has been the fate of all major programming languages targeting the leading ISA in the industry. i.e. to become usable by certain niches and ignored by everyone else.
As for C++, it is hard to imagine Rust phasing this one out too. Particularly because C++11 addresses most of Rust improvements in one way or another. Lambdas and templates can implement Rust actor model (which I find, like Erlang, to be Rust biggest contribution considering how awful the idea behind threads always was). C++ meta programming can handle Rust type reasoning and auto and decltype can help with code branch analysis.
Certainly this increases C++ boilerplate code. It may even be seen as counterproductive. But were I to want to create something similar to Rust in C++, I would have the means. And this is exactly the question many will pose: Would I want to do it? The answer is almost always a resounding no. There's no point. Rust style brings nothing major that warrants the effort. And also because of this, it doesn't justify the tremendous effort of coding and maintaining a new language. Some specialized outfits will move on to Rust, most just won't see the gains in doing so.
How I need a drink, alcoholic in nature, after the heavy lectures involving quantum mechanics.
Name a few of these improvements that Rust brings which C++11 cannot imitate.
I hope you aren't being serious. Correctness is not an afterthought in C++. And I seriously hope you are wrong. We don't need a new language that brings but a few minor improvements. We need more resources on improving existing languages. Put some research into looking if it is possible to bring a language add-on of smart pointers to C++, how it might be implemented such that it brings lots of improvements at little or no cost to compiler vendors instead.I do hope that you are wrong, since i think it is high time we faced the fact we will always be dealing with shoddy software as long as we base everything on a language where program correctness is as much of an afterthought as it is in C.
