LLVM9.0 treats writes to const as "dead code"

What about several nested consts or const passed through several layers of functions? Will it know the source memory is writeable or treat it as "dead code"(which it eliminates)?

the fuck should I know. go test it anus.

https://news.ycombinator.com/item?id=21021344

On the plus side kernel compiles cleanly. They finally added asm goto.

>>1
it would only work for const values
references and pointers to const can obviously have the values they refer to be changed elsewhere

boringcc, djb was right again

rurban is a crankpot, he said that siphash is broken but did not show any evidence for that. He instead started Mentifex-style shitposting.

LLVM should be Not-C. They're telling me it's Esperanto but it's ebonics with a funny name. White man's languages (RPG, COBOL, BASIC, FORTRAN, SNOBOL, PL1, PASCAL, ALGOL, ASSEMBLY) give me the freedom to store my data however I want and use whatever types I want. These niggers can't handle freedom like white people can so they want to take more freedom away from C. C compilers have no meaning because they're written in C and C has no meaning. That bald man must have gold inside his head, so let's decapitate him. That's how niggers think.
https://melmagazine.com/en-us/story/bald-men-in-mozambique-are-being-hunted-down-for-the-gold-in-their-heads
https://www.thetimes.co.uk/article/bald-man-decapitated-in-mozambique-for-gold-in-his-head-7bkph7z7f

>>3

How about you fix your code to not rely on undefined behavior specific to a compiler version? Poor coding standards is not an excuse.

Whoever thinks coding standards are the solution to this is divorced from reality. How about the Clang developers make the effort to emit a warning when processing such code, if it is so obviously broken, and add an optional sanitizer if it is not.

No amount of benchmark improvement is worth generating incorrect results, especially if this is hard to detect.

It's about time LLVM stops treating whites as dead niggers.