The Most Important API Design Guideline - No, It's Not That One - Jody Hagins - C++Now 2024

[@Heater-v1.0.0]

I was involved in test ing the Fly By Wire software of the Boeing 777 back in 1993. I observed what looked like a failure in the rudder control under certain circumstances to me and filed a Problem Report (We were not allowed to say "bug report"). My report was investigated by the developers at the subcontractor I was working at (This was all happening in England) and they determined it performed according to the specification. I protested that it seemed obvious the spec was wrong. My PR was forwarded to Boeing in Seattle and they eventually replied that I was seeing expected behaviour. I'm no avionics expert so that had to be good enough for me. Then, many months later, on the second or third test flight the test pilot did some stall testing which resulted in him permanently loosing control of the rudder. Just as I had predicted. Not good. The test pilot was furious and that bug got fixed. Or I assume that it got fixed. You see Boeing let the twenty or so guys on the PFC testing team go when we were still only half way through the thousands of tests that had to created. About 20 years later I finally got a flight on a 777 over to the USA. It worked fine....

[@peregrin71]

No naked types, and here is something I really really miss natively in C++. A strong type alias (one that says Price is a double but also a unique type)

[@Arwahanoth]

I do believe writing a struct Price will always be more valuable than using strong_typedef. (also the performance cost of a struct with only a double and a naked double should be the same).

[@douggale5962]

I believe in mean, nasty, hideous, evil tests that are crafted specifically to attack clever stuff like lock free skiplists and threading. This talk is the first time I have ever heard another developer mention this sort of nastiness. Everyone I meet in tests wants to do every microscopic unit test they can dream up, in pristine conditions. If you can honestly describe some tests as an attack, you have good tests. Tests should test everything you thought was going to work correctly. You shouldn't bother testing things that you think won't work. They will. If you think the code is easy, it's a buggy mess. If you thought it was scary and you were afraid it was going to screw up, you would be meticulous about everything in the code.

[@Roibarkan]

1:31:27 Jody’s talk from cppcon 2022: kzbin.info/www/bejne/ipLJZpinir90d5I

[@Roibarkan]

1:32:53 Jody’s talk from cpponsea: kzbin.info/www/bejne/qamulqeHi7WcrKs

[@Roibarkan]

49:18 Mark Rendle’s talk: kzbin.info/www/bejne/j2rGnXV-erOsarM

[@Roibarkan]

1:02:34 Stephen Dewhurst’s talk: kzbin.info/www/bejne/hnfHiKSXa51_o6M

[@TNothingFree]

No offense to this man particular or any other speaker but CPP NOW have to do some guidelines on presentation. I cannot withstand these panels with ton of text in them, 200-500 slide presentation is worth of 4 hours, not 1.5. Minimize concepts and minimize information output, it's just...too much.

[@Roibarkan]

1:23:21 Alan Kay’s talk: kzbin.info/www/bejne/pXzKYpuKhLaLpbs

[@perghosh8135]

WTF, reusable code is easy to test and very flexible. I think that Jody are mixing domain code where someone tries to do reusable domain code, thats not possible. Look at stl, isn't that reusable code?

[@tialaramex]

Jody is telling you what Knuth says about reusability.

[@Roibarkan]

9:35 kzbin.info/www/bejne/l2iZi5KVoLZ9mtE

[@mo3k]

I really can't comprehend why that one person kept asking "buy why?" questions, and "what's the penalty for stopping trading?...like dude, he said 1000x times, this presentation is about the programming aspect, do you not understand how the world works? Do you think the programmers at JP Morgan get to decide whether or not JP Morgan trade or not? Do you think a programmer needs to understand all the SEC rules and fees for the market? Wasted so much time interrupting with off-topic questions.