gareth-nx

> I am aware of the thread which showed little to no performance penalty of `optval` if the right compiler flags were used. Still, for the reasons just mentioned by...

How should we recommend that users set `-DCMAKE_BUILD_TYPE` ? If they use `-DCMAKE_BUILD_TYPE=Release`, then I believe that will append compiler flags that include `-O3`. This suffers the same problem that...

That sounds good. So in the README, I imagine that, rather than explaining `FFLAGS`, we would say: * Compiler options can be controlled by modifying the file [cmake/UserOverride.cmake](). Separate options...

@zoziha For what it's worth I've previously run into compiler problems at this part of the code (when using inter-procedural optimization). Although it would be good to hear from @Jim-215-Fisher,...

To give a bit more context - if I compile stdlib with: ``` make -B -f Makefile.manual ADD_FYPPFLAGS=-DMAXRANK=4 FC=gfortran FFLAGS="-O3 -flto" ``` then I get a segfault on the test...

@milancurcic Yes, to me this sounds like an improvement.

Also somewhat related: https://github.com/fortran-lang/stdlib/pull/573 (the test code compares with existing hash functions written in C and C++, and we will discuss how to best manage this).

Agree this would be nice to have, when a sufficient number of compilers are happy with it.

@leonfoks I like the subroutine interfaces you propose. IMO it is good to make this a subroutine as in your interfaces, rather than a function as in the matlab translation...

@Beliavsky Although I haven't checked comprehensively, I think it is true that, following a call to `qselect(a, k)` above, `a` is modified such that `a(1:(k-1))