c-blake

Happy to help use it, if I can. The command line driver is pretty fancy. To just get estimated coefs you should be able to just set up the SVD...

!nim c ```Nim import parseUtils # works import std/strUtils # fails import seq_utils # fails ```

!nim c ```Nim import seq_utils ``` Repeated but simplified just to show the non-`std/`-qualified `import` with style variation also fails on all hot-loadable test Nim versions.

It seems the `std/` package qualification also changes things...

It's really more an issue that nimble never updates the `@#head` version after it is first installed. This surprised me. I would have thought it was "always updated". Instead they...

The 32-bit output is even more an issue than the speed, IMO, as it is surprising. There was originally some concern about having the string hash values from the C/C++...

Thanks for the contribution, @demotomohiro. I don't think anyone disputes that a limited range of output `Hash` limits the scalability of tables using it. You can define your own `hash(x:...

So, with the below Nim program (and [`cligen`](https://github.com/c-blake/cligen) installed, compiled with `-d:ubench` and PGO under both `clang-17` and `gcc-13` on A)lder-L)ake & S)ky-L)ake), I get the attached graph (in gnuplot...

I think all agree what is most a problem is the counterintuitive restriction of `Hash` to 32-bits. [This addresses](https://github.com/nim-lang/Nim/pull/23735) that as well as giving a speed boost at the same...

Well, for the performance curious, since this issue did start with performance complaint, a Nim built with the [referenced PR](https://github.com/nim-lang/Nim/pull/23735), and this little cligen utility program: ```Nim import std/[times, monotimes,...