Comment by gpderetta
1 year ago
> I very much hope that the attribute will catch on, spreading to GCC, Visual C++, and other popular compilers,
AFAIK, attribute musttail is in the process of being added to GCC (the patch is under review) with semantics compatible with clang.
What about the preserve_most attribute? Is there any chance something like that will get into GCC? Without it, the non-tail calls ruin the interpreter.
Maybe, but not there yet: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110899
In the meantime some inline assembly macro trickery might help.
edit: code duplication can also be obviated by templating your op function with a fast/slow parameter, with the fast variant tail-calling the slow variant when it cannot perform the fast path, while guarding the slow code via the compile time parameter. The downside is yet more code obfuscation of course.
It's a hard problem because many ABIs cannot do tail calls even for very basic stuff, like calls to extern functions with matching argument and return types. It looks like Clang has some heuristics for changing call sequences for musttail calls. For example, it switches to noplt calls on i686. None of this is mentioned in the Clang documentation: https://clang.llvm.org/docs/AttributeReference.html#musttail
What's realistic here is that you get a diagnostic if the compiler cannot generate a tail call. For many users, that will likely be good enough. Guaranteeing a tail call as in Scheme is unlikely to happen.
Given some of the other scheme-like features GNUC has it's surprising they're lagging on this one.