Rust .gitignore template

target is the whole story. Every compiled dependency, every intermediate artefact, the incremental compilation database and the final binaries all live there, and on a project with a moderate dependency tree it runs to gigabytes. It is also machine specific: it contains absolute paths and objects built for one target triple. Nothing in it is worth carrying and all of it is reproducible from the sources plus the lockfile.

debug is a legacy sibling of that rule. Older project layouts wrote debug builds to a top level directory of that name, and modern cargo puts them under target/debug instead. Keeping the line costs nothing, but it will also match a source directory called debug if you ever create one.

Cargo.lock is the deliberate judgement call, and the template makes it visible by shipping it as a comment rather than a rule, so by default the lockfile stays tracked. For an application or any binary you distribute, that is what you want: the lockfile pins exact versions so a build today and a build in six months produce the same dependency set. For a library, the lockfile is ignored by anyone who depends on you, so committing it only pins what your own CI tests against.

That second case is why the comment exists rather than a decision being made for you. The modern advice has drifted towards committing it even for libraries, on the grounds that a reproducible CI run is worth more than the mild dishonesty of a lockfile downstream users never see. Either way it should be a decision someone made on purpose.

The remaining two rules are small. Files ending in .rs.bk are backup files from an old rustfmt, and *.pdb files are the separate debug symbol files produced on Windows, which are large, uninteresting and regenerated on every build. Neither is likely to appear in a project started recently, but both cost nothing to keep, and the symbol file rule in particular saves a Windows contributor from committing something several times the size of the executable it describes.