Skip to content

fix: stale unlicensed key in deny.toml and rgbuilder LFS path in .gitattributes - #89

Merged
sshaaf merged 1 commit into
sshaaf:mainfrom
nerdalytics:fix/deny-config-and-lfs-path
Sep 21, 2026
Merged

sshaaf merged 1 commit into
sshaaf:mainfrom
nerdalytics:fix/deny-config-and-lfs-path

Conversation

@nerdalytics

@nerdalytics nerdalytics commented Sep 20, 2026 •

Copy link
Copy Markdown
Contributor

Description

cargo deny check fails on every run, and the Git LFS filter never applies to the ONNX weight file. One reference is out of date with cargo-deny, the other with the rgbuilder to rgctl rename.

Type of Change

  • Bug fix (non-breaking change which fixes an issue)

Changes Made

deny.toml. cargo-deny 0.20.2 removed the unlicensed key in EmbarkStudios/cargo-deny#611, so validation fails before cargo-deny reads any crate. That same PR made deny the new default for an unlicensed crate, so removing the line changes nothing about the outcome.

.gitattributes. The LFS pattern still names crates/rgbuilder-analysis, which the rename to rgctl removed, so git check-attr filter returns unspecified for the real asset. The committed pointer still resolves because git-lfs reads blobs directly, but the next write to that file would commit 205.75 MiB of raw weights, over GitHub's 100 MB limit.

Testing

Neither file is Rust, so there is nothing here for cargo test or clippy to cover. I checked both fixes by hand:

  • cargo deny check moved from error[deprecated] ... failed to validate configuration file at deny.toml:17 to reaching the crates.io index fetch. I could not finish the run, because my network blocks crates.io.

  • git check-attr filter -- crates/rgctl-analysis/assets/code-daemon-embed-v1/model.onnx.data moved from unspecified to lfs.

  • Unit tests added/updated

  • Integration tests added/updated

  • Tested with cargo test

  • Manual testing performed

Test environment:

  • OS: macOS 26.6.2 (Darwin 25.6.0)
  • Rust version: rustc 1.98.1 (48a229cea 2026-09-01)
  • cargo-deny 0.20.2

Checklist

  • I have performed a self-review of my code
  • I have commented my code where necessary
  • Documentation updated (if applicable). Neither file is referenced in the docs.
  • CONTRIBUTING.md guidelines followed
  • All commits are signed and include DCO sign-off

Breaking Changes

None. No workflow under .github/workflows/ calls cargo-deny today, so CI behavior does not change, and the .gitattributes fix rewrites no committed blob.

Additional Notes

I found the deny.toml failure while adding this repo's gates to a local task runner. The .gitattributes path turned up in the same pass.

cargo-deny 0.20.2 removed the `unlicensed` key (EmbarkStudios/cargo-deny#611),
so `cargo deny check` aborts on config validation before it inspects anything.
The `allow` list already denies a crate whose license cannot be determined.

.gitattributes still names `crates/rgbuilder-analysis`, a path that no longer
exists after the rgbuilder to rgctl rename, so `git check-attr filter` reports
`unspecified` for the real asset.

Signed-off-by: nerdalytics <97166791+nerdalytics@users.noreply.github.com>
@sshaaf
sshaaf merged commit ef4faba into sshaaf:main Sep 21, 2026
@nerdalytics
nerdalytics deleted the fix/deny-config-and-lfs-path branch September 21, 2026 20:08
sshaaf pushed a commit that referenced this pull request Sep 22, 2026
The comment I added in #89 said the `allow` list rejects a crate whose
license cannot be determined. It does not. EmbarkStudios/cargo-deny#611 made
`deny` the new default for `unlicensed`, and that default is what the removed
key was configuring. The outcome was right, the stated mechanism was not.

Also restores the space in `allow = [`, which #89 dropped by accident.

Signed-off-by: nerdalytics <97166791+nerdalytics@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants