Suggestion: Adopt C++ name mangling conventions for better compatibility about carp HOT 5 CLOSED

Agreed. My idea with the current naming scheme is that it's fairly close to how people name their C libraries. In theory that means better usage of Carp libraries from C code and nicer wrapping of existing C libraries. For instance, the SDL library could be wrapped in a SDL module with shorter function names since all its functions begin with SDL_. I'll read through the C++ mangling doc for ideas, but won't change any of this for now.

from carp.

The C++ standard does not specify a mangling scheme, and we will have to implement it in a few different ways to be able to be compatible with different compilers. This is neither an argument for nor against it—I have no opinion on this—, just a remark.

from carp.

Another thing to consider: when we write C code that interfaces with Carp, we mangle by hand. This would mean we would have to provide different versions for the core functions (because of the names) or autogenerate core.h, as well as doing that for any C code that interfaces with Carp. This sounds tedious to me, and I’m not sure what we gain from it.

Still I have no informed opinion about this, but I’d like to say that right now it is fairly comfortable to write C that interfaces with Carp. It would be prudent to keep it that way, so to whatever system we change, this should be our primary concern.

from carp.

That seems like a pretty good idea. Do you have a link to some kind of understandable spec for it?

from carp.

Hello Carp people!
This is the de-facto standard documentation for calling conventions and name mangling.
http://www.agner.org/optimize/calling_conventions.pdf

from carp.