C++20 design is finalized

reddit.com/r/cpp/comments/au0c4x/201902_kona_iso_c_committee_trip_report_c20/

Modules and coroutines made it in

Attached: C_plus_plus_325.png (325x325, 8K)

Other urls found in this thread:

reddit.com
en.cppreference.com/w/cpp/compiler_support#cpp17
cs.rit.edu/~ats/books/ooc.pdf
godbolt.org/z/f_46UC
godbolt.org/z/QPg6qi
open-std.org/jtc1/sc22/wg21/docs/papers/2018/p1103r2.pdf
twitter.com/SFWRedditImages

I still write C++ like ANSI C with classes at work.

>polishing a turd
C++ was and is a mistake

>Modules
large if huge

>TL;DR: C++20 may well be as big a release as C++11.
welp at least Rust proved to be useful for something

Attached: 1549690647452.png (644x800, 15K)

What do you recommend instead?
>inb4 Cnile
>inb4 js/python brainlet

Cool, languages from the 80s had these.

I still barely use any C++ '11 features :/

you need to go back

>reddit.com
Why not link to an official source? Why do you think Jow Forums cares about some Reddit forum?

Can we finally have a cpp file for a class that has a template?

Can I get a quick rundown on Modules and Coroutines?

I swear im not stupid I just like coding in C

OP of that post is a C++ committee member it says it right next to his name, you idiot.

Attached: firefox_2019-02-23_18-11-00.png (753x41, 7K)

C++ is the true patrician taste.

Dunno mate, I just know that I get paid to C++ daily and it's a mess of bad decisions and missing features.

why do I care?
I didn't even click your link.
Find a source that isn't an ideological wasteland.

lmao back to with you, monkey

based digits, SEPPLES BTFO

>Modules
Glorified headers.
>Coroutines
Glorified Duff's device.
Or are C++ coroutines asynchronous?
And why are they posting on Reddit instead of an official mailing list or something? This is one step away from promoting yourself on Discord, like Rust is doing now.

back to >>>/reddit/ with you, newfag.

>all this buttdevastation ITT over a fucking domain name
go cry on and stop shitting up this thread

N-nothing in particular...

Attached: smug wojak rust.png (767x639, 109K)

They are and OP is just a stupid faggot who thinks that it's perfectly fine to advertise his trash website around here.
There'd be no problems if he'd try to find an actual source.

>trying this hard to fit in

>lets go on Jow Forums and post a link to another site, that'll generate discussion!
This thread is a shitpost.

>posting news is shitposting
Show me on the doll where OP touched you.

Ok I posted a comment asking the devs why they use reddit. I will get back to you guys when they respond.

gettaloadofthisguycam.tiff

Attached: urgay.png (881x135, 11K)

ok here you guys go

Attached: firefox_2019-02-23_18-35-01.png (1840x606, 122K)

Jow Forumsormless twits on suicide watch

Besides the linked thread here, the forum is largely devoid of activity. I seriously doubt it's any "official" community channel and it's just that one committee member choosing to share on their own accord. So OP is in the wrong for linking to Reddit instead of an official channel.

>reddit users are stupid enough to enable referrals
Color me surprised.

Attached: 1422742193013.jpg (250x250, 8K)

So basically C++ is confirmed a reddit language?

>he thinks c++ committee uses plebbit for official news
did you seriously take the other user's bait?

Attached: 1550762943183.jpg (562x437, 39K)

Yes. I say we go back to Speedcoding, the one and only programming language.

Attached: that 30-year-old boomer.jpg (234x215, 9K)

>links to links
wow the internet is amazing

Attached: 1550573883112.jpg (600x360, 61K)

I write amd64-only C++17 at work.

>be OP
>entire concept of the internet revolves around a few centralized resources like Reddit
>OP logs in to "the front page of the internet", and it really is for him, and checks their favorite subforums
>a C++ community member in charge of PR makes an unofficial announcement to various social media, including Reddit
>without this announcement, OP would have no idea that C++20 was finalized, because they don't follow the language and its design or committee, only what gets posted to Jow Forumscpp
>OP posts this announcement on HN enthusiastically
>finally, OP posts this unofficial random announcement to Jow Forums
Actually, why does HN link to the same post? Am I just yelling at the cloud here?

>Modules
last i read the modules proposal was absolutely garbage, but this more had to do with backwards compatibility with the language

basically we can't expect build tools to ever implement this in a sane manner

The level of job security you'll get by implementing some range-based, coroutine-driven, generic abomination, will be insane.
Really look forward to it.

This is actually genius.
With the new (>=c++14) compile-time instructions and these you can absolutely make yourself necessary by obscurity.

>constexpr std::vector
neat

other than microsoft do any compilers even have c++17 implemented fully?
what's the point if every compiler is 2 standards behind?

imagine if the standards committee applied this line of reasoning 10 years ago when they were drafting C++11

en.cppreference.com/w/cpp/compiler_support#cpp17
Seems like both clang and G++ are doing okay.

gcc and clang have both been C++17 complete for months now, while icc is still struggling to get C++11 right.
If this isn't enough to convince you how much of a joke Intel is, nothing will.

What kind of C++ role allows you to shit out anything you want? Any job worth having or one that requires competent programmers is going to restrict you to a subset of C++.

>no fun allowed
We use static analysis, unit testing, code reviews and common sense. That works for us, at least.

intel doesnt give a shit about cpp, only C because its compatible
see MPI, OpenMP, intel intrinsics...

>If everyone cuts their dick off I'll cut my dick off too. -- redditfags
Take that pol!!!!!!!!! fucking dumpf.

And in reverse, faggots.

I use C++ as a convenient C, where I don't have to implement strings and containers myself.

lol

Attached: Screenshot_20190223_235202.png (600x380, 42K)

Modules as in systemc modules? I'm not too familiar with c++

based

go back to red-dit and kys

This is why I'm moving to C#...that and Unity.

Attached: 2uh4of.jpg (500x500, 56K)

what kind of shitty fucking meme is this?

reddittoddlers BTFO

Attached: 151.gif (498x278, 1.69M)

Is there a way to ban people with a reddit account on Jow Forums?

GOOD

>learning c++, how do i print a vector?
std::vector path;
// ...
for (std::vector::const_iterator i = path.begin(); i != path.end(); ++i)
std::cout

such retard post

Attached: 1550794774181.jpg (900x657, 207K)

I've implemented garbage collection with a simple 50 line C++ header

>No metaclasses
>No static reflection
Who cares?

>reddit
As i expect from a C++ programmer. Faggotry.

std::vector path;

for(int i : path)
{
std::cout

D

>t. tranny rust shill

I prefer the chad traditional C.

>C++ has finally made it into 2005
i'll stick with nim and kotlin, thanks

how do you not have a reddit account in 2k19? there are so many small specific tech subs filled with people with way more knowledge than on Jow Forums. for a tech enthusiast it's basically a must-have. also good for when you get sick of the Jow Forumskids over here spamming every board with their shit

Then join your fellow brainlet Pajeets

Attached: 1549755547828.png (500x750, 347K)

Kill yourself

cs.rit.edu/~ats/books/ooc.pdf

>int i
>vector path

retard

using C++ the right way
also daily reminder C++ is not an oop language.

this is awful very unbased and cringe pilled.

it also compiles differently as the path.end() causes to make a lookup on every iteration while for each loop saves the length into register at the beginning
although dunno how much of a hit is that, could just be register vs L1 cache lookup
anyway linters warn on using iterators in this case

What are the practical advantages though?

Good C++ looks nothing like C.
Although I agree that using C++ as an OOP language only is suboptimal.

Too bad C++ modules, the way they are currently defined, are useless and broken.

You're wrong. path.end doesn't create any function call as vector is a template and is thus inlined. The .end function definition just returns a pointer. I'd be surprised if any decent compiler generated different code for these two loops.

Why aren't they announcing it on Jow Forums, the smart board?

Attached: 1534551491812.jpg (500x500, 100K)

>other than Microsoft
they still can't get expression SFINAE exactly right

Just ban all phoneposting, the overlap should be big enough and there are no negative side effects.

then be surprised and check compiler output

godbolt.org/z/f_46UC
to make it easier for you

>modules
why? what's wrong with the headers?

Can people stop blaming Jow Forums for everything.

What would you do different?

The absolute state of Jow Forums.
These two functions aren't even equivalents.

Use a const reference in the second loop instead, the function size are the same.

Better: remove the iostream noise and the assembly is absolutely identical

godbolt.org/z/QPg6qi

>C++20 design is finalized
I already saw a job posting for 5 years experience coding in C++20

Attached: 1550033408339.jpg (227x222, 6K)

>entry-level

How do "modules" differ from standard includes? Is it just a preprocessor directive/hint?

>open-std.org/jtc1/sc22/wg21/docs/papers/2018/p1103r2.pdf

Attached: dd.png (1037x611, 74K)

Oh no no no
>the utter state of sepples
HAHAHAHA

Attached: キャプチャ.png (1000x771, 45K)

>Swedish
>filenames in Japanese
>weeb

>the utter state of "swedistan"
HAHAHAHA

I don't think you can call asians weebs.

nice

what's your problem? they differ precisely in what I described

report1
# init
cmp rbx, QWORD PTR [rdi+8]
# loop
cmp QWORD PTR [rbp+8], rbx


report2
# init
mov rbp, QWORD PTR [rdi+8]
cmp rbx, rbp
# loop
cmp rbx, rbp

>Use a const reference in the second loop instead.
results in same code