alexcrichton wrote: There's been a fair bit of work in the realm of StarlingMonkey (js-runtime-compiled-to-wasm) and Python (compiled to wasm) as well as some other interpreter-based languages. In these scenarios dynamic linking is used for the interpreter itself, libc, and any "native plugins" loaded by the interpreter (e.g. the native module behind pandas). AFAIK it's working well for everyone and there's no major complaints so far, but it's also not been a super-stressed part of the ecosystem. In that sense I'd say there's strong validation that it works, but not strong validation that the performance/conventions/tooling are in the right place. Not to say there's points against it either, just that those aspects haven't been stressed.
Personally I think this is reasonable to do, and I'd agree it'd still make sense to revisit this in the future for a possible future iteration, but for now the experimental warning isn't buying much. https://github.com/llvm/llvm-project/pull/196566 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
