C++26 Preview - Jeffrey Garland - C++Now 2024

[@zxxvcc]

I love that c++ is so complicated that when megabrained c++ experts get together even they scratch their heads about exactly what you can do in it.

[@adamrushford]

C++'s whole goal was to keep everything out, patterns were made, things were tested, and always c++ stayed behind to do it the right way, so now they struggle to add all the best parts of other languages, lambdas and modules from java for example, so there is a lot of discussion now of how it should move forward, this one set of ways, or the other, to simplify what's really driving this kind of behavior. Make no mistake though these people know programming better than the code knows itself.

[@lbgstzockt8493]

Yet they insist on making it even more complex. If C++ had package management that didn't drive everyone insane we would not be doing it like this.

[@elliott8175]

I'm only 15 minutes in, but really enjoying the presentation. I also learnt from this that you can delete any member function - and I already have a use-case for this at work. Cheers! One bit of feedback I have, though, is that it'd be nice if the presenter repeated the questions. Lots of them I can hear, but some are more muffled and I couldn't work out what was being asked. I always appreciate it when the question is repeated. Otherwise it seems like a great summary!

[@pauloperbone]

Great talk thanks!

[@shavesha]

Its sad that we might not get pattern matching in c++26. I don't understand why and how is was so easy for std::ranges to make it into the standard. Am jealous that everyone else always gets what they want

[@theintjengineer]

Sad to see Networking crossed out, but happy for the progress regarding the other stuff. Besides, asio is cool. If it can be fully integrated with std::execution, the new Coroutines stuff, Contracts, etc., then everything is good👌.

[@SeregaZinin]

Network was promised in C++17, std::experemental_network even appeared in Visual Studio 2017, but then all of this was removed and completely deleted! Very sad (

[@ciCCapROSTi]

It's a bit sad that Jeff is not an expert on EVERY one of these topics, but it's good that the audience can expand on the topics.

[@MindGameArcade]

Looks great! C++ is slowly becoming the definitive best language out there, we need to address dependency management though, it shouldn't be hard, Go did it well

[@sneeddeens9895]

i mean, as far as i understand with cmake and later with its full CPS (Common Package Specification) implementation there is not much about dependency management that needed to be addressed. Btw you can look recent talk by Bill Hoffman about cmake and CPS on this channel

[@aniketbisht2823]

vcpkg exists. So does conan.

[@keris3920]

​@@aniketbisht2823 the fact that you mentioned multiple approaches is a problem for me. I want a standard.

[@ciCCapROSTi]

Mate not knowing about middle click / opening on a new tab cost the presentation like 5 minutes.

[@Ariccio123]

37:00 I'll be very disappointed if the standardized contracts feature is mostly runtime and not the kind of static checking that could subsume SAL

[@llothar68]

If you ever worked in practice you will see that it is exactly what we need. Not even more static typesystem support

[@ciCCapROSTi]

How would you even static check runtime parameters?

[@foobar69]

24:00 is this Fedor Asking questions?

[@pawello87]

I don't understand this hype about contracts. What's the problem with using a strong type that will throw an exception/assertion? Instead of using the great type system that C++ has, they invent more keywords and clutter the language.

[@drinductor8150]

Contracts, as proposed, will be enforced at the call site allowing optimization across translation units, but more importantly, allows for better constexpr support.

[@CyberDork34]

Design by contract is a programming paradigm supported by other languages and some extensions to C and C++. We want to support it first-hand, we've wanted to support it for over a decade, and it makes other things, like the C++ lifetime profile, easier to implement

[@DDO-56]

- Exceptions are not for programming errors. - Assertions are used in the implementation, while contracts are specified in the interface. Moreover, tools may be implemented to test an interface based on its contract, while it is impossible to do the same with the assertions since they are part of the source code.

[@DeathSugar]

I wonder if reflection will upgrade typeinfo

[@bernhardmanfredgruber7156]

No, it won't. std::type_info is runtime information. Reflection is only compile-time information.

[@unixux]

Awesome video. But on substance, is it me or did WG21 lose 99% of real life C++ developers roughly half a decade to a decade ago ? (trick question : the answer is "all of the above") ?

[@spyofgame200]

Nice

[@douggale5962]

Audience dude was helpful.

[@Axacqk]

Excuses not to switch back to C++ are disappearing as we watch!

[@breezebanana]

how to make so complex

[@abbycin7280]

the wg21 never meets the needs of developers

[@arsalananwari4377]

Its 2024 and intellisense still does not work with C++20 modules in visual studio 😂. My lord at this rate i think its best to switch to Rust (jk).

[@SeregaZinin]

I still use C++11 and don't need these modules ))

[@llothar68]

More worry is that all support for c++ at microsoft was stopped last year. Zero support for cpp26 because all developers go AI now

[@WalderFrey]

Interesting, but very little there for app developers. Networking being kicked down the road, again. C++11 was awesome, but each release since feels like diminishing returns. Where I work we're rapidly moving away from C++, primarily towards C#, which just feels like a more intelligible language with decent library support. Nobody I work with is interested in C++ any more.

[@sanjaygatne1424]

कोड फाॅन्ट साईज बहोत छोटी है. पढने मे दिक्कत होती है.

[@destiny_02]

watch in 1080p, if still can't see get an eye checkup

[@kito4525]

Gonna have to give pieces of code to AI to understand what random feature it's using, the learning curve is not curving.