C++ compilers are supposed to optimize the heck out of your code, right? Well, there are some rules. My latest blog post explains why consumeByValue(getANewObject()) will be smaller & faster than auto x = getANewObject(); consumeByValue(std::move(x));

You can’t go around making salutes like that and expect me to stick around on your app. Find me elsewhere

done --

@patio11 @LizardOrman @SRNissen So do people who walked away from underwater mortgages find themselves debanked, because they demonstrably defaulted on a big loan on purpose and accounts are credit products? (Thanks for the big debanking post, just read it)

Will anything go wrong if does a cmake build with a build directory that isn't under build_temp?

(std::sqrt(2) is easily constant-folded) Reference: code is from the fast Hadamard transform implementation in ExecuTorch:

@ESYudkowsky @patio11 To saturate bureaucratic inputs using letters, doesn’t each letter have to claim to be a real person? There are structural limits and USPS can read envelopes. Also, mailpieces come with additional metadata (notably, the physical origin) that could be used to filter spam.

__is_long(): it's UB: (search undefined behavior) memcpy loophole: OK fine my worldview is not shattered but I'm amused

@olson_dan Do you think this is actionable advice for C++ or just another wrong default to throw onto the pile?

@chofter Worked on iOS apps for years, haven’t installed a PWA since before 2011 and it was a PWA I wrote

@marksaroufim Never looked into factorio, doesn’t seem like my cup of tea from what I’ve seen incidentally

Thanks are also due to @KhuuJack for his work pushing the pin bump through to torchchat!

C++ pet peeve (not that I have a suggestion for how to fix it): The following are not the same. struct Outer { T1 x; T2 y; T3 z; }; struct Outer { struct Inner { T1 x; T2 y; } inner; T3 z; }; (For example, if T1 is int64_t and T2/T3 are both int8_t)

@NavinFS @patio11 Why can’t they do that?