I use C++ every day and this feels like an entirely different language and philosophy. I know this is the reality of C++, but I never go searching for it so when it pops up randomly my jaw drops a little.
I see there's an issue in the tracker to get more accurate data, and since it's using an under dev feature in compilers, it's not going to be definitive, but any rough numbers?
I don't have any numbers, but it is pretty slow. You can try making some edits in Compiler Explorer to see for yourself (though that of course has some impact from network requests).
One reason is that, like you said, the feature is still new. Additionally I made pretty liberal use of the std::ranges library in my implementation which has notoriously bad compile times. That could be an area to improve upon.
Another may be a bit more structural. If you want to call myObj.foo() via reflection, you have to linearly search members_of(myObj) for the one named "foo", and then call that. Actual compilers I assume use some kind of hash table.
The hand-waving solution is "put it in a PCH", but I am hoping to put some more effort into optimizing build time here in the future.
This library tries its best to mitigate that, catching common errors and whatnot, but it can definitely still happen. C++ doesn't have full token injection yet, so it avoids some of the more common pitfalls, if incidentally.
As an aside, you may want to check out Jai's approach. I believe everything you generate statically gets turned into a file by the compiler for debugging purposes, which it provides references to in the output.
I get a heart attack whenever I have to view core file from a decently complicated C++ program. The amount of template-in-a-template-in-a-template...(and this continues for some time) is not so readable to me. Maybe it is just me.
An include with a HTTP URL is a scary abomination straight put of hell. Please tell me that this is a compiler explorer specialty (which would still be cursed, but in a cool way) and not a GCC feature (which would be an absolute nightmare).
Yes, it's a Compiler Explorer feature. URL includes get rewritten client-side to work as if they were additional files supplied by the user. PR adding the current behavior is at [0], while I think the downloading code itself is at [1].
Yes. duck takes ownership of the vector by moving it into its internal storage.
As a bonus, if you tried passing in an lvalue, it will reject the input unless you add the "copyable" trait, so it ends up mitigating some hidden copies.
> If you’ve ever tried using type erasure for something more complicated than std::any or std::function, you’ve either written 100+ lines of easy-to-mess-up code or reached for a boilerplate-heavy library like Boost.TypeErasure or Folly.Poly
You mean "used void* and wrote lots and lots of linter suppression comments" because it's fundamentally amateurish to write C++ without at least clang-tidy, cppcheck, and a couple others.
I don’t really like much about C++ anymore, but I still enjoy reading C++ articles and listening to C++ podcasts, and I would consider it beautiful. Oftentimes the things I dislike about it are also the beautiful things. The term “beautiful mess” seems appropriate.
It’s a bit like a well-kept Victorian home. The amount of work, money, and dealing with discomfort that goes into maintaining one isn’t something I really want to experience for myself. But the amount of skill and craftsmanship that it takes to preserve one is still impressive, and I have to appreciate the respect for history and the care that goes into balancing it with modern concerns.
And talking to people who do live the life is always a great learning experience.
The problem with that is best described by Antoine de Sain-Exupery's saying "perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away." I guess the same goes for elegance and grace...
C++ is in the transitional phase where major bad things are being taken away very satisfactorily, usually by providing a simpler and more general replacement (for example, auto instead of long and pointless type declarations or modern initialization protecting against implicit conversions and surprise constructor overloads), but most progress of elegance and grace come from new features that enable something traditionally terrible or impossible (for example the gradual generalization of templates, culminating with concepts, the gradual extension of constexpr, consteval etc, and the new reflection).
I find it hard to see a language as beautiful that’s grown too complex for a single person to hold a complete mental model of.
I used to think that was a personal limitation, until I saw an interview with Bjarne explaining that he used to understand all of it but at this point it’s too big, no one can anymore.
That is exactly what I was thinking. I was a seasoned C++ programmer and always loved reading articles like this. I can't imagine I will every write my own C++ code again -- or in any language. I now program with English specifications now and I am 10000% times more productive.
Some of us are professionals and like to understand our systems and how they work. I don't write assembly instructions by hand either, nor do I design CPUs much, but I want to - and likely need to - know how they work to make the best judgements.
I could imagine that in near future when token prices are as high as they really are, programmers that can't imagine or remember how to write code anymore will clean our streets, drive our taxis and water our plants.
I use C++ every day and this feels like an entirely different language and philosophy. I know this is the reality of C++, but I never go searching for it so when it pops up randomly my jaw drops a little.
What's compilation time like when using it?
I see there's an issue in the tracker to get more accurate data, and since it's using an under dev feature in compilers, it's not going to be definitive, but any rough numbers?
I don't have any numbers, but it is pretty slow. You can try making some edits in Compiler Explorer to see for yourself (though that of course has some impact from network requests).
One reason is that, like you said, the feature is still new. Additionally I made pretty liberal use of the std::ranges library in my implementation which has notoriously bad compile times. That could be an area to improve upon.
Another may be a bit more structural. If you want to call myObj.foo() via reflection, you have to linearly search members_of(myObj) for the one named "foo", and then call that. Actual compilers I assume use some kind of hash table.
The hand-waving solution is "put it in a PCH", but I am hoping to put some more effort into optimizing build time here in the future.
Reflections, especially static ones, are horrible for debugging.
This library tries its best to mitigate that, catching common errors and whatnot, but it can definitely still happen. C++ doesn't have full token injection yet, so it avoids some of the more common pitfalls, if incidentally.
As an aside, you may want to check out Jai's approach. I believe everything you generate statically gets turned into a file by the compiler for debugging purposes, which it provides references to in the output.
I get a heart attack whenever I have to view core file from a decently complicated C++ program. The amount of template-in-a-template-in-a-template...(and this continues for some time) is not so readable to me. Maybe it is just me.
Depends pretty much on the language and IDE tooling being used.
Those against IDEs, well they already voted against good tooling in first place.
An include with a HTTP URL is a scary abomination straight put of hell. Please tell me that this is a compiler explorer specialty (which would still be cursed, but in a cool way) and not a GCC feature (which would be an absolute nightmare).
Yes, it's a Compiler Explorer feature. URL includes get rewritten client-side to work as if they were additional files supplied by the user. PR adding the current behavior is at [0], while I think the downloading code itself is at [1].
[0]: https://github.com/compiler-explorer/compiler-explorer/pull/...
[1]: https://github.com/compiler-explorer/compiler-explorer/blob/...
in the first example:
```
10: rjk::duck<Container> c{std::vector<int>{1, 2, 3}};
11: c.size(); // 3
12:
13: c = std::string{"hello"};
```
Does the assignment on line 13 call the destrucor for the vector of ints created on line 10?
Yes. duck takes ownership of the vector by moving it into its internal storage.
As a bonus, if you tried passing in an lvalue, it will reject the input unless you add the "copyable" trait, so it ends up mitigating some hidden copies.
Well, you’ve certainly convinced me to read your library.
Thanks for the blog post.
No problem! Hope you enjoy :)
Is there another option? Or were you asking if it leaks memory?
Maybe you were asking if it implements custom destructors? GC?
> Or were you asking if it leaks memory?
yeah,
If variable c where of type void* instead of duck<Container>, the assignment on line 13 would leak the memory used by the vector<int>.
> If you’ve ever tried using type erasure for something more complicated than std::any or std::function, you’ve either written 100+ lines of easy-to-mess-up code or reached for a boilerplate-heavy library like Boost.TypeErasure or Folly.Poly
Or you just used void* and went on with your life
You mean "used void* and wrote lots and lots of linter suppression comments" because it's fundamentally amateurish to write C++ without at least clang-tidy, cppcheck, and a couple others.
The things people describe as "beautiful" never cease to amaze me...
...but, as they say, beauty is in the eye of the beholder!
I don’t really like much about C++ anymore, but I still enjoy reading C++ articles and listening to C++ podcasts, and I would consider it beautiful. Oftentimes the things I dislike about it are also the beautiful things. The term “beautiful mess” seems appropriate.
It’s a bit like a well-kept Victorian home. The amount of work, money, and dealing with discomfort that goes into maintaining one isn’t something I really want to experience for myself. But the amount of skill and craftsmanship that it takes to preserve one is still impressive, and I have to appreciate the respect for history and the care that goes into balancing it with modern concerns.
And talking to people who do live the life is always a great learning experience.
> I would consider it beautiful.
If there's something that C++ actually lacks, that's the elegance, grace and beauty. The rest, it's all already there or will be there shortly :)
Think of it like a rainforest ecosystem. Very complex and tangled, even fragile. But it’s the tangled fragile complexity that makes it interesting.
I find C++ beautiful only when I come across simple, powerful things that use the minimal amount of advanced language features possible
The problem with that is best described by Antoine de Sain-Exupery's saying "perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away." I guess the same goes for elegance and grace...
C++ is in the transitional phase where major bad things are being taken away very satisfactorily, usually by providing a simpler and more general replacement (for example, auto instead of long and pointless type declarations or modern initialization protecting against implicit conversions and surprise constructor overloads), but most progress of elegance and grace come from new features that enable something traditionally terrible or impossible (for example the gradual generalization of templates, culminating with concepts, the gradual extension of constexpr, consteval etc, and the new reflection).
Oh, have they finally started to remove features?
> I would consider it beautiful
I had the same misunderstanding before I get to know CS. that was 30 years ago.
Beauty in C++ may be most similar to lipstick on a pig, but we try our best.
It's more like arranging pigs in beautiful patterns. If you don't look very closely, you don't see the pigs.
I admire you guy keep trying.
I'm also glad I do not write C++ on the daily anymore: luckily my software does not need that kind of performance characteristics.
I still reach for C++ on the backend. Honestly, with all its warts, I like it overall.
You can write pretty fast and reasonable code nowadays.
Yet I imagine your software depends on C++.
Either directly on top of a runtime/compiler written in C++, or as indirect dependency on a C++ compiler toolchain.
I find it hard to see a language as beautiful that’s grown too complex for a single person to hold a complete mental model of.
I used to think that was a personal limitation, until I saw an interview with Bjarne explaining that he used to understand all of it but at this point it’s too big, no one can anymore.
C++ is certainly a leader among the least feasible, but it's far from alone.
this is the most disgusting programming language ever invented!
I think that is perl, but I share what you mean, as more features they add, as uglyer it gets.
are we still hand writing code?
Yes
That is exactly what I was thinking. I was a seasoned C++ programmer and always loved reading articles like this. I can't imagine I will every write my own C++ code again -- or in any language. I now program with English specifications now and I am 10000% times more productive.
Good for you!
Some of us are professionals and like to understand our systems and how they work. I don't write assembly instructions by hand either, nor do I design CPUs much, but I want to - and likely need to - know how they work to make the best judgements.
I think you misunderstand -- I am still reading the article.
I could imagine that in near future when token prices are as high as they really are, programmers that can't imagine or remember how to write code anymore will clean our streets, drive our taxis and water our plants.
It will be easier than trying to learn to the new features of C++34.