Add prototype to generate COPYRIGHT from REUSE metadata
#104439
Conversation
rustbot
added
T-bootstrap
This comment was marked as resolved.
This comment was marked as resolved.
pietroalbini
force-pushed
the
pa-generate-copyright
branch
from
f9dc676 to
4af7de1
Compare
|
Fixed the tidy failure and revamped the Markdown output to actually be readable. This is ready for review. |
|
|
||
| let mut result = Vec::new(); | ||
| for file in document.file_information { | ||
| let license = interner.intern(License { |
There was a problem hiding this comment.
as someone unfamiliar with SPDX and with the overall plans here: What is the expected interning behavior when the file has differing copyright_text but the same concluded_license ? Are those supposed to be interned to the same LicenseId, or to distinct LicenseId's?
There was a problem hiding this comment.
Same license but different copyright holder would result in different LicenseIds, as that's effectively a different license (since most licenses require attribution, and the copyright statement is the attribution).
|
@pietroalbini do you have any thoughts on how to encode some tests for this prototype? E.g. some sample files and maybe a mocked up I won't block landing a prototype here on the presence of such tests, but it is something worth a little bit of thought. (e.g. Cargo already uses some solution to automatically generate directory/file structures to use to drive its tests; is there work there that could be leveraged here, rather than making something from scratch?) |
Co-authored-by: Felix S Klock II <[email protected]>
|
Yeah I'll add tests to a followup PR (added it in my todo list). |
|
Addressed all feedback. |
| ) -> Node<&License> { | ||
| match node { | ||
| Node::Root { childs } => Node::Root { | ||
| childs: childs.into_iter().map(|child| strip_interning(child, interner)).collect(), |
There was a problem hiding this comment.
... how is this code still building, given that there are references to the old name in this recursive call here that were not updated as part of the rename to expand_interned_licenses?
There was a problem hiding this comment.
Well, that's because it's not building anymore. Whoops, good catch.
|
( r=me once @pietroalbini addresses my concern about my hypothesized-to-be-broken alpha renaming. ) |
|
@bors r=pnkfelix |
bors
added
S-waiting-on-bors
|
@bors rollup |
…iaskrgr Rollup of 11 pull requests Successful merges: - rust-lang#104439 (Add prototype to generate `COPYRIGHT` from REUSE metadata) - rust-lang#105005 (On E0195 point at where clause lifetime bounds) - rust-lang#105098 (propagate the error from parsing enum variant to the parser and emit out) - rust-lang#105243 (remove no-op 'let _ = ') - rust-lang#105254 (Recurse into nested impl-trait when computing variance.) - rust-lang#105287 (Synthesize substitutions for bad auto traits in dyn types) - rust-lang#105310 (Be more careful about unresolved exprs in suggestion) - rust-lang#105318 (Make `get_impl_future_output_ty` work with AFIT) - rust-lang#105339 (support `ConstKind::Expr` in `is_const_evaluatable` and `WfPredicates::compute`) - rust-lang#105340 (Avoid ICE by accounting for missing type) - rust-lang#105342 (Make `note_obligation_cause_code` take a `impl ToPredicate` for predicate) Failed merges: r? `@ghost` `@rustbot` modify labels: rollup
This PR adds a prototype to generate the
COPYRIGHTfile from the metadata gathered with REUSE. There are two new tools:src/tools/collect-license-metadatainvokes REUSE, parses its output and stores a concise JSON representation of the metadata insrc/etc/license-metadata.json.src/tools/generate-copyrightparses the metadata generated above, (in the future will) gather crate dependencies metadata, and renders theCOPYRIGHT.mdfile.Note that since the contents of those files are currently incorrect, rather than outputting in the paths above, the files will be stored in
build/and not committed. This will be changed once we're confident about the metadata.Eventually,
src/etc/license-metadata.jsonwill be committed into the repository and verified to be up to date by CI (similar to our GitHub Actions configuration), to avoid having people install REUSE on their local machine in most cases.You can see the (incorrect) generated files in https://gist.github.com/pietroalbini/3f3f22b6f9cc8533abf7494b6a50cf97.
r? @pnkfelix