comparing to #embed about embed

[MKlimenko]

Hi Johannes,

Personally, I'd use the standard way, that way you'll get rid of an extra step in the build procedure.
However, looks like no compiler supports #embed so far, but I haven't been following that lately. Some sort of custom trunk was once live on Compiler Explorer, but it might be down now.

My library was inspired by std::embed proposal of the same author, but (long story short) this functionality landed in plain C first. Therefore, I designed it to be a resourse- rather than an array of bytes-oriented to be more C++ish.

It worked pretty well on my previous job for storing the binary images of the heterogeneous device programs, we integrated it into CMake as a pre-build step, so as a conclusion I'd say that it's pretty straightforward tool to use until we have a standard way.

[johannes-graeter]

Oh yeah you are right no compiler support for it currently. Although there is an active PR in Clang so perhaps we'll get it soon. llvm/llvm-project#68620
In the mean time l will happily employ yours :)

[MKlimenko]

It's a good idea to compare the performance against the native embed, and probably even create a PR with some performance improvement.
I'll do it once this support lands in Clang, but let me close this issue for now.