@@marcgregoire I took the Cisco entry level c++ course and finished it, do you think I'm a good candidate for the "Professional C++, 5th edition" book? I'm also a fullstack developer with Javascript/Typescript. Thanks!!
Handful of new features in C++20 and eager to get my hands dirty and see how sweet/painful is it! Overall it was a long journey if we consider the whole transitions from late 90s.
first, view::ints(10) was wrong, it has to be "views::iota(10)". But, you better limit your sequence when using something like reverse, for example using views::take(...)
Good presentation of C++20. It has indeed many attractive new features (e.g. ranges; format; default comparison). Overall the C++ language still suffers from lack of standard libraries compared to e.g. .NET or JAVA; I hope they work on that in the future.
I took the Cisco entry level c++ course and finished it, do you think I'm a good candidate for the "Professional C++, 5th edition" book? I'm also a fullstack developer with Javascript/Typescript. Thanks!!
hi, please i start to leaern c++ for the first time, so i start with c++20, i got a book for it, but am having a big problem, i can't have a compiler for it, most of the IDE i use doesn't support the c-++20 features like (import), please i need your help because now i stopped the studying. i use these VSC extension codeblock embarcadero visual studio 2019 PLEASE HELP
So essentially C++ as some concepts from functional languages like F# and the pipeline processes of R? Nice really handy for Data Science task in larger datasets or streams where python or R are too slow and SQL is too limited 👍🏽
Usually. There's generally several draft standards and item-specific documents that get published as part of the standardization process, and compiler vendors are often involved in the process anyway. Sometimes features are just standardizing common extensions the compiler vendors already offer, but I don't know how much of that applies to C++20.
No, the code is ok. It uses co_yield to return something to the caller. co_return is optional, if you don't need to return something at the end, you can just leave it out. Just like you don't need to write "return;" in normal function.
How to install C++ 20 on Red Hat Linux 8 ? I have installed the development tools on RHEL,. GCC and G++ also have been installed. How I can know if the version is C++ 20 ?
No, if you make mm2 constexpr instead of const, then you'll get a compilation error. In my example I wanted to show that even though InchToMm() is marked as constexpr, that you are not guaranteed compile-time execution.
Right, in new code, I would recommend to switch to using modules. It'll benefit your build throughput. But not all compilers are supporting it at the moment ...
Did they fix std::set so we can have non const iterators again?! It was deliberately broken in the past in case people change the key, preventing changes to any other part of the nodes data!!!🤪
Some good and some bad ideas here. My biggest issue is the growing complexity of C++ and having people understand your code when there are so many constructs in the language. Ranges make perfect sense for C++. Views make sense but using | for pipe-filter in addition to OR in other contexts is yet another opportunity for people to see code who hadn't seen that feature just get lost at what they are looking at. It's not necessary to have such features. We can already do filters as a sequence of function calls.
You create an object named x of type T and use "uniform initialization" with the result depending on T. If T is a built in type then it will be uninitialized. If on the other hand it is a class, then the object is created using the default constructor.
re ranges: Finally! We have cast off the shackles of C (or at least one of many), where the "stylish" way to iterate over something was to use a begin and end pointers. Yuck!
the problem with source_location is that in a real logger you would expect to forward arguments to std::format.. but having arbitrary argument count is not possible with the last source_location param. nice summary tho, i enjoyed. Thanks!
many of these ... topics.. ...ideas... ...concepts... so to say were already discussed in the circles of programming languages creators and of programmers in the last years... in fact I find the D programming language has already all the benefits of those and has ALREADY today for example a better implementation of the module mechanisms... only sad that it is not as visible as C++ ....
It's a great lecture. Tho I don't really like computer science. It's all about keeping making new concepts and names for people to learn and make simple stuff more and more confusing to keep their job secured.
So you would like to keep developing computer science without making new concepts, or you accept making new concepts but you would like them to be left unnamed? I guess you have similar criticism for, say, medicine, chemistry, physics, etc.?
C++ 20 just clusters the c++ landscape. I just don't get it why the committee is constantly adding language features that have no impact on performance. I mean, no one comes to C++ for its "cool" language features.
Well, that's not gonna happen for sure. C++ borrows the idea based features from Higher Level Languages, but using Dynamic Type System like Python, will make it slow like Python :)
Time Stamps
-----------
2:46 - New Keywords
3:56 - Modules
9:03 - Ranges
15:00 - Coroutines
18:57 - Concepts
23:54 - Lambda Expression Changes
26:47 - constexpr Changes
27:54 - Concurrency Changes
33:33 - C++20 Synchronization Library
36:01 - Many more new features...
42:06 - Calendars & Timezones
45:27 - std::span
47:30 - Feature Testing Macros
48:36 - Version
49:12 - Immediate Functions - consteval
50:24 - constinit
51:19 - Class Enums and using Directive
52:07 - Text Formatting (std::format)
54:11 - Math Constants
54:31 - std::source_location
55:31 - [[nodiscard(reason)]]
55:51 - Bit Operations
56:34 - Small Standard Library Additions
Thanks for that informative video!
Thank you very much, very informative! Now I have to go rewrite all the code I have ever written, wish me luck!
Nice presentation. IMHO Modules and concepts are the most helpful features.
Glad you think so!
Marc Gregoire, thank you very much. Your book is one of the best in C++.
Thanks, I'm glad you like the book :)
@@marcgregoire I took the Cisco entry level c++ course and finished it, do you think I'm a good candidate for the "Professional C++, 5th edition" book? I'm also a fullstack developer with Javascript/Typescript. Thanks!!
Excellent talk! It really helps people like me get caught up with the latest developments in C++.
C++20 is like a new language
no more headers and #include directives 🤯🤯
I personally never had issues before. int a = 5; has work just fine for DECADES.
seems like C++ is getting even more functional
Very clear introduction to the new features of C++20. Thanks
Excellent! Great overview towards c++20 finishing in one go!
the speech is very informative. c++20 fabulous. i love it.
Very useful overview about new features of c++20 with comparison c++17!! Great!!!
This is the language version of Christmas morning.
Nice summary, found a few things that will I start to use immediatly.
This was a fantastic overview of C++ 20. Thank you.
Thank you much. Your talk was very comprehensive to follow.
C++ 2O - C++ BORN Again ❤️
Thanks, an excellent overview of C++20!
Very nice walk-through. Thanks a lot.
Handful of new features in C++20 and eager to get my hands dirty and see how sweet/painful is it! Overall it was a long journey if we consider the whole transitions from late 90s.
Nice to see Rusts Iterators coming to C++.
Thank you for this informative session.
Excellent
How does view::reverse work with infinite sequences like view::ints(10)?
first, view::ints(10) was wrong, it has to be "views::iota(10)".
But, you better limit your sequence when using something like reverse, for example using views::take(...)
This could be one of the most important talks in this year's cppcon
Great overview, thanks!
Great overview!
awesome !
This is amazing, so much good stuff!
Good presentation of C++20. It has indeed many attractive new features (e.g. ranges; format; default comparison). Overall the C++ language still suffers from lack of standard libraries compared to e.g. .NET or JAVA; I hope they work on that in the future.
Quite comprehensive! Thanks!
I took the Cisco entry level c++ course and finished it, do you think I'm a good candidate for the "Professional C++, 5th edition" book? I'm also a fullstack developer with Javascript/Typescript. Thanks!!
Thanks, good talk.
Thank you for this!
good talk and good audio!!!!
hi, please i start to leaern c++ for the first time, so i start with c++20, i got a book for it, but am having a big problem, i can't have a compiler for it, most of the IDE i use doesn't support the c-++20 features like (import), please i need your help because now i stopped the studying.
i use these
VSC extension
codeblock
embarcadero
visual studio 2019
PLEASE HELP
please i need anyone help
How about not starting with the newest features that are not yet supported universally?
Just use C++17/14 - they are widely supported.
C++20 is now wastly supported (but not fully) by compilers.
The latest Visual Studio Preview has full support for c++20 modules.
So essentially C++ as some concepts from functional languages like F# and the pipeline processes of R?
Nice really handy for Data Science task in larger datasets or streams where python or R are too slow and SQL is too limited 👍🏽
Great talk, thank you!
Does work on compilers happen at the same time as the standard is being decided?
Usually. There's generally several draft standards and item-specific documents that get published as part of the standardization process, and compiler vendors are often involved in the process anyway. Sometimes features are just standardizing common extensions the compiler vendors already offer, but I don't know how much of that applies to C++20.
Often yes.
I wonder why spacship operator is a minor update.
Can you please make a cheat sheet with a summary of all the new features? Thanks!
I think that would be a rather big sheet, as the C++20 update is quite massive.
Is the coroutine example incorrect because it doesn't use `co_return` after the loop finishes?
No, the code is ok. It uses co_yield to return something to the caller. co_return is optional, if you don't need to return something at the end, you can just leave it out. Just like you don't need to write "return;" in normal function.
Think of python iterators you just need the yield in the example
How to install C++ 20 on Red Hat Linux 8 ? I have installed the development tools on RHEL,. GCC and G++ also have been installed. How I can know if the version is C++ 20 ?
Excellent Talk!
49:33 - did you miss constexpr keyword instead of "const" on both slides?
No, if you make mm2 constexpr instead of const, then you'll get a compilation error. In my example I wanted to show that even though InchToMm() is marked as constexpr, that you are not guaranteed compile-time execution.
constexprt strings, YES.
Ranges looks somewhat similar to C# linq
can I start c++20 without knowing c++???
So header Files are no longer needed to use anywhere or anytime?
Right, in new code, I would recommend to switch to using modules. It'll benefit your build throughput. But not all compilers are supporting it at the moment ...
Who will use all of these???
very comprehensive
It's really annoying when they use the expression " Now with CXX you don't have to this, you can do this instead... " But why !😢
Haven't done much c++20 yet but I do love concepts. Quite similar to typeclasses in haskell
Did they fix std::set so we can have non const iterators again?!
It was deliberately broken in the past in case people change the key, preventing changes to any other part of the nodes data!!!🤪
If somebody knows about memory fences guarantees w.r.t. to synchronisation c++ primitives please let me know.
Some good and some bad ideas here. My biggest issue is the growing complexity of C++ and having people understand your code when there are so many constructs in the language. Ranges make perfect sense for C++. Views make sense but using | for pipe-filter in addition to OR in other contexts is yet another opportunity for people to see code who hadn't seen that feature just get lost at what they are looking at. It's not necessary to have such features. We can already do filters as a sequence of function calls.
Is it only me who thinks that declaring variables like this makes it barely readable?
auto var { int { 10 } };
54:55, how about a standard way of getting __PRETTY_FUNCTION__-like info ?
I'm not sure I understand what you want to do...
at 25:00, what is mean by T x { } ?
You create an object named x of type T and use "uniform initialization" with the result depending on T. If T is a built in type then it will be uninitialized. If on the other hand it is a class, then the object is created using the default constructor.
That just declares a variable x of type T and zero-initializes it.
Default value for T (default ctor or default 0 for int, float, double...)
Not a word about operator | for vector 15:00
With every release C++ is Approaching python semantics and syntax
It seems like the C++ standardization committee is trying to win over Javascript developers 😂
re ranges: Finally! We have cast off the shackles of C (or at least one of many), where the "stylish" way to iterate over something was to use a begin and end pointers. Yuck!
One day all code will be constexpr
Reminds me of Fortran.
Thank Typescript for giving me better TypeC++20.
the problem with source_location is that in a real logger you would expect to forward arguments to std::format.. but having arbitrary argument count is not possible with the last source_location param.
nice summary tho, i enjoyed. Thanks!
You still can forward it to std::format as it ignores unused args. Or you can provide a specialization (I think) that will do nothing.
many of these ... topics.. ...ideas... ...concepts... so to say were already discussed in the circles of programming languages creators and of programmers in the last years...
in fact I find the D programming language has already all the benefits of those and has ALREADY today for example a better implementation of the module mechanisms...
only sad that it is not as visible as C++ ....
It's a great lecture.
Tho I don't really like computer science. It's all about keeping making new concepts and names for people to learn and make simple stuff more and more confusing to keep their job secured.
So you would like to keep developing computer science without making new concepts, or you accept making new concepts but you would like them to be left unnamed? I guess you have similar criticism for, say, medicine, chemistry, physics, etc.?
C++ 20 just clusters the c++ landscape. I just don't get it why the committee is constantly adding language features that have no impact on performance. I mean, no one comes to C++ for its "cool" language features.
C++ 23: pointers are removed, no data types- it's basically Python! That's my guess.🙄
Well, that's not gonna happen for sure.
C++ borrows the idea based features from Higher Level Languages, but using Dynamic Type System like Python, will make it slow like Python :)
Is it going the other direction; i.e Python and JavaScript getting type annotations?.