Brad Larson

@lefticus A big monitor. 27" minimum.

OK C++ Twitter! I've asked you about your free software, and your commercial software choices. Now I have a really weird question! What piece of *hardware* is very important to you as a C++ developer?

@lefticus Drop the cruft! No more typedef, redo initializer lists, etc.

What do you most want from a C++ successor language?

@Booms_JustBooms 1. Yes they do, just not as clean. But I'm mostly wanting this for the appendAll() type API 2. I considered span<t>, wasn't sure about possible performance hits with non-vector backends? But span<> does have a size, so yeah I think it'd be fine.

@b_k_larson For the first case, do push_back or emplace_back not have the desired behavior? For both, would an "append(T)" and "append(span<t>)" be appropriate? Seems like the operator+= would boil down to those calls in the naive case.

std::vector<T> should provide += T to append T and += std::vector<T> to append a full vector. So much cleaner than the 3-parameter insert()

I'm planning the first ever "C++ Best Practices Game Jam"! It will begin April 1st, 2022! github.com/cpp-best-pract… And I need your help to make sure it goes smoothly! (plz retweet for exposure) 👇👇👇

@nicreations Same.

I felt this in KC

@lefticus @mattgodbolt Good point, that makes sense. I'm surprised the assembly is the same in arm - I didn't think about how the execution must be in x64. Using llabs in arm does produce different assembly

@b_k_larson @mattgodbolt You are comparing arm to arm, but executing on x64 In x64 the assembly is different: godbolt.org/z/3z7ndhaav

@mattgodbolt @lefticus same assembly but different output? What am I missing here? Feel like I'm going crazy: godbolt.org/z/h3rsq3hdh

@lefticus @craigw1701 They need to deprecate and remove much faster :)

@craigw1701 Every C++ standard release deprecates and removes something, that's already a given anyhow

Do you think the C++ standards committee should intentionally break ABI in C++23?

@skippytx @SwiftOnSecurity In the exact same boat, about to give up on the 3400g

@SwiftOnSecurity I can’t even find lower end kit in stock. Was going to build 3400g PCs for my kids. Gave up and ordered i3-10100 instead.

LinusTechTips gives up recommending high-end PC builds this holiday because there are no parts to buy period youtu.be/hW6_nX0g3to

@smarmyions Any tips on where to find a 3400g??

@mrmoneymustache reddit.com/r/IAmA/comment… might be worth a blog post?

@lefticus Template Meta Programming!

I think I now have a C++ Weekly episode for almost every C++ topic. So let's play a game! You respond with a topic/question, and I'll respond with an episode. The response with the most ❤️'s tomorrow at this same time will get a free digital C++ Best Practices book copy! 👇👇

This is by far my favourite "historical plaque" ever. It is truly spectacular.

@BramStolk @grujicbr @MrMadbrain ( Respectfully :) ) This sounds terrible. I want to include bar.h, so I need to go look up its dependencies to include foo.h first. But foo.h also has dependencies, so let me go look them up as well... I much prefer stand-alone headers.

@grujicbr @MrMadbrain No it does not. You can embed foo in bar as composition, not pointer. In main.c you just need to include foo.h first, then bar.h Being explicit in deps for every .c you write is a good thing. Dependencies should be front and centre, not hidden. Explicit dep > Implicit dep.

@chandlerc1024 Does this include language-breaking changes? E.g. removing std::auto_ptr?

2) Both software and language evolution "Change is the essential progress of all existence." -- Spock. Wait, you're still debating? ;] Seriously, things which do not change are replaced. C++ must enable both software and itself to change and grow.

So I wrote up (with a *huge* amount of help from colleagues both at G and elsewhere) what I think the goals and priorities for C++ should be: wg21.link/p2137 Finally published. I'm not sure the committee agrees, but it is definitely what I and my users need. Highlights:

5) Fast and scalable development Currently, we measure compile time in *minutes*. Seriously, MINUTES. I don't even know how that can be a thing. And the scaling problems are worse somehow. C++ Modules help, but are just the start.

@mattgodbolt @TitusWinters @hyrumwright Was just about to buy a couple copies for my team!

Yay! It's here!!! @TitusWinters @hyrumwright

@dascandy42 @normed_space I had a tablesaw accident back in November. I'm still finding blood splatters around my shop. I've since refused to use my tablesaw until I upgrade it to a sawstop.