Consider a Rust tarball. It's a huge archive that contains multiple projects (rustc, cargo, other Rust-related tools, llvm, vendored libraries...). This alone is source of a very long scan step, plus the stone.yml outcome is probably going to be garbage due to the fact multiple projects are vendored in the tarball, whilst we are only interested in rust dependencies and build steps.
Implementing heuristics to circumvent this situation is probably hard; a relatively easy workaround is to hit CTRL+C to abort the scan and let boulder finish without pre-filling stone.yml.
The license scaninng algorithm isn't super efficient, however since we're in pre-alpha stage we're building tooling in debug mode. In truth, building in release mode improves scanning time by a factor of 10.
Consider a Rust tarball. It's a huge archive that contains multiple projects (rustc, cargo, other Rust-related tools, llvm, vendored libraries...). This alone is source of a very long scan step, plus the
stone.yml
outcome is probably going to be garbage due to the fact multiple projects are vendored in the tarball, whilst we are only interested in rust dependencies and build steps.Implementing heuristics to circumvent this situation is probably hard; a relatively easy workaround is to hit CTRL+C to abort the scan and let boulder finish without pre-filling
stone.yml
.Example log (partial):