Re: [lmi] PATCH: fix math_functions_test build with clang

[Greg Chicares]

On 5/21/22 13:40, Vadim Zeitlin wrote:
[...]
> [2]: https://github.com/let-me-illustrate/lmi/pull/210

I think the changes I just pushed take care of that entire PR.

>  BTW, on a somewhat related note, we currently use __BYTE_ORDER (note the
> lack of the trailing underscores) in md5.cpp. I wonder if we shouldn't
> switch to using __BYTE_ORDER__ there too, as this symbol is predefined by
> the compiler, and so would avoid the need for including <endian.h>.

Done. Thanks.

If "#pragma GCC' should be guarded with a test for LMI_GCC rather
than for __GNUC__, then was it right to fix only a couple of cases,
when many more exist? For example:

  $git grep -B3 'pragma GCC'
  bourn_cast.hpp-#if defined __GNUC__
  bourn_cast.hpp:#   pragma GCC diagnostic push
  bourn_cast.hpp:#   pragma GCC diagnostic ignored "-Wsign-compare"
  bourn_cast.hpp:#   pragma GCC diagnostic ignored "-Wsign-conversion"

Why isn't that a problem for clang, which AIUI defines __GNUC__ ?