Wordlexpr: Compile-Time Wordle in C++20

Re:Wordlexpr: Compile-Time Wordle in C++20 by nneonneo on Monday February 28, 2022 @04:20AM (#30497958)

Neat hack! One minor improvement suggestion: if you seed the RNG with `__DATE__` (or a hash of that string), then you can replicate the one-puzzle-a-day experience with no need for an externally-supplied seed variable.

[ Parent | on HN ]

Re:Wordlexpr: Compile-Time Wordle in C++20 by SuperV1234 on Monday February 28, 2022 @10:14AM (#30500677)

Good point! I wanted to be able to modify SEED in order to test that everything was fine with multiple words, using __DATE__ didn't come to mind :)

[ Parent | on HN ]

Re:Wordlexpr: Compile-Time Wordle in C++20 by TheSoftwareGuy on Monday February 28, 2022 @11:37AM (#30501817)

    #ifndef SEED
    #define SEED __DATE__
    #endif

[ Parent | on HN ]

Re:Wordlexpr: Compile-Time Wordle in C++20 by fathyb on Monday February 28, 2022 @04:40AM (#30498087)

Wondering what happens if you allocate a chunk of uninitialized memory inside a constexpr? Could this be used for a quick constexpr RNG seed?

[ Parent | on HN ]

Re:Wordlexpr: Compile-Time Wordle in C++20 by jcelerier on Monday February 28, 2022 @05:59AM (#30498548)

A very neat thing of constexpr is that it does not have undefined behaviour ; any thing that would cause UB at run-time is flagged as a compile error. Every variable must be fully initialized.

[ Parent | on HN ]

Re:Wordlexpr: Compile-Time Wordle in C++20 by xroche on Monday February 28, 2022 @04:54AM (#30498178)

You simply can't. The constexpr environment won't let you have uninitialized variables/memory anywhere. It won't let you access dangling/out-of-bounds pointers either, overflow two-complement types, etc.

[ Parent | on HN ]

Re:Wordlexpr: Compile-Time Wordle in C++20 by flqn on Monday February 28, 2022 @05:59AM (#30498546)

Accessing uninitialized objects is undefined behaviour, which traps in the constexpr context.

[ Parent | on HN ]