endian check from runtime to compile time

[neheb]

No description provided.

[kevinbackhouse]

We're not on C++20 yet, are we? It's a shame we can't use std::endian::native.

[ghost]

👇 Click on the image for a new way to code review

Legend

[codecov]

Codecov Report

Merging #2668 (870ea3a) into main (a89c02f) will decrease coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #2668      +/-   ##
==========================================
- Coverage   63.92%   63.92%   -0.01%     
==========================================
  Files         103      103              
  Lines       22306    22305       -1     
  Branches    10795    10795              
==========================================
- Hits        14259    14258       -1     
  Misses       5826     5826              
  Partials     2221     2221              

Impacted Files	Coverage Δ
src/image.cpp	70.19% <100.00%> (-0.06%)	⬇️

[neheb]

We're not on C++20 yet, are we? It's a shame we can't use std::endian::native.

std::endian requires the minimum compiler to be bumped. Exiv2 can still be compiled with gcc7.

[Ryanf55]

gcc7

FYI the oldest non-EOL ubuntu version ships with 9.3.0. https://packages.ubuntu.com/focal/devel/gcc

[kmilos]

Backport to 0.28.x?

[kmilos]

Also, might be better to use compiler defined __BYTE_ORDER__ and __ORDER_BIG_ENDIAN__ macros rather than some header defined macros that are not standard?

[neheb]

That would only help with MinGW AFAIK. ORDER_BIG_ENDIAN is a GCC macro.

[kmilos]

And Clang as well, at least.

Where does __ORDER_BIG_ENDIAN come from?

[neheb]

GCC and clang.

[kmilos]

My GCC and Clang have only the variants with two underscores at the end 🤷