Hacker News new | past | comments | ask | show | jobs | submit
well they managed to get two's complement requirement into C++20. there is always hope.
Well then someone somewhere with some mainframe got so angry they decided to write a manifesto to condemn kids these days and announced a fork of Qt because Qt committed the cardinal sin of adopting C++20. So don’t say “a problem literally nobody has”, someone always has a use case; although at some point it’s okay to make a decision to ignore them.

https://lscs-software.com/LsCs-Manifesto.html

https://news.ycombinator.com/item?id=41614949

Edit: Fixed typo pointed out by child.

> because Qt committed the carnal sin of adopting C++20

I do believe you meant to write "cardinal sin," good sir. Unless Qt has not only become sentient but also corporeal when I wasn't looking and gotten close and personal with the C++ standard...

Wow.

https://theminimumyouneedtoknow.com/

https://lscs-software.com/LsCs-Roadmap.html

"Many of us got our first exposure to Qt on OS/2 in or around 1987."

Uh huh.

> someone always has a use case;

No he doesn't. He's just unhinged. The machines this dude bitches about don't even have a modern C++ compiler nor do they support any kind of display system relevant to Qt. They're never going to be a target for Qt. Further irony is this dude proudly proclaims this fork will support nothing but Wayland and Vulkan on Linux.

"the smaller processors like those in sensors, are 1's complement for a reason."

The "reason" is never explained.

"Why? Because nothing is faster when it comes to straight addition and subtraction of financial values in scaled integers. (Possibly packed decimal too, but uncertain on that.)"

Is this a justification for using Unisys mainframes, or is the implication that they are fastest because of 1's complement? (not that this is even close to being true - as any dinosaurs are decomissioned they're fucking replaced with capable but not TOL commodity Xeon CPU based hardware running emulation, I don't think Unisys makes any non x86 hardware anymore) Anyway, may need to refresh that CS education.

There's some rambling about the justification being data conversion, but what serialization protocols mandate 1's complement anyway, and if those exist someone has already implemented 2's complement supporting libraries for the past 50 years since that has been the overwhelming status quo. We somehow manage to deal with endianness and decimal conversions as well.

"Passing 2's complement data to backend systems or front end sensors expecting 1's complement causes catastrophes."

99.999% of every system MIPS, ARM, x86, Power, etc for the last 40 years uses 2's complement, so this has been the normal state of the world since forever.

Also the enterpriseist of languages, Java somehow has survived mandating 2's complement.

This is all very unhinged.

I'm not holding my breath to see this ancient Qt fork fully converted to "modified" Barr spec but that will be a hoot.

loading story #41876530
This person is unhinged.

> It's a desktop on a Linux distro meant to create devices to better/save lives.

If you are creating life critical medical devices you should not be using linux.

loading story #41876446