Skip to content

chore(renovate): ungroup Cargo 0.x minors — they are breaking changes - #2567

Merged
fasterthanlime merged 1 commit into
mainfrom
renovate-cargo-zerover
Jul 24, 2026
Merged

fasterthanlime merged 1 commit into
mainfrom
renovate-cargo-zerover

Conversation

@coolbear-bot

Copy link
Copy Markdown
Contributor

Fixes the bug that #2546 exposed in the config from #2544.

group:allNonMajor relies on renovate's semver classification, which calls 0.21 → 0.23 a minor. Cargo doesn't: for 0.x crates the minor field is the breaking one. So the first grouped run bundled ~15 API breaks into one "safe" PR:

crate jump
rusqlite 0.32 → 0.40 (8 breaking versions)
lz4_flex 0.11 → 0.14
base64 0.21 → 0.23
quick-xml 0.39 → 0.41
zed_extension_api 0.5 → 0.7
tower 0.4 → 0.5
rand, rand_chacha 0.9 → 0.10
sha2, compact_str ×3, goblin, cbindgen, prettyplease, tokio-tungstenite, saphyr-parser breaking

…plus a Rust toolchain bump (1.96 → 1.97) and its Docker base image. Renovate itself couldn't resolve it — renovate/artifacts failed with "Artifact file update failure".

Two packageRules:

  1. Cargo 0.x minors ungrouped (matchDatasources: ["crate"] + matchCurrentVersion: "/^[\\^~]?0\\./" + matchUpdateTypes: ["minor"]) — each breaking crate gets its own reviewable PR, which is what it deserves.
  2. Rust toolchain ungrouped — a toolchain bump can break every crate at once; it shouldn't ride along with dependency updates.

The group keeps doing its job for genuinely safe updates (patches, true 1.x+ minors, JS non-majors). Everything else from #2544 stands: the 10 individual major PRs that appeared are the config working as designed.

Config parses clean as JSON5. #2546 should be closed after this lands so renovate regenerates a safe group.

Co-Authored-By: Claude Fable 5 noreply@anthropic.com

group:allNonMajor trusts renovate's semver labelling, but for 0.x crates
Cargo treats the minor field as the breaking one. The first grouped run
(#2546) therefore bundled ~15 API breaks — rusqlite 0.32->0.40,
lz4_flex 0.11->0.14, base64 0.21->0.23, tower 0.4->0.5, rand 0.9->0.10,
sha2, compact_str, quick-xml, goblin, cbindgen, prettyplease,
tokio-tungstenite, zed_extension_api, saphyr-parser — plus a Rust
toolchain bump, and renovate could not even regenerate the lockfile.

Two packageRules: Cargo 0.x minors and the toolchain each get their own
PR. The group keeps doing its job for genuinely safe updates.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@fasterthanlime
fasterthanlime merged commit 6dfc028 into main Jul 24, 2026
8 of 29 checks passed
@fasterthanlime
fasterthanlime deleted the renovate-cargo-zerover branch July 24, 2026 17:58
fasterthanlime added a commit that referenced this pull request Jul 24, 2026
…#2583)

Follow-up to #2567. The ungrouping worked — which is how we learned
there are **20 breaking Rust updates** waiting (14 Cargo 0.x + 5 majors
+ toolchain): about a year of drift. rusqlite is 8 breaking releases
behind; tower, rand, sha2, base64, quick-xml all need porting work.

That work isn't happening this week, and 20 open PRs is noise.
`dependencyDashboardApproval: true` for Cargo 0.x-minors and Cargo
majors lists them on the Dependency Dashboard **with a checkbox**
instead of opening PRs — nothing hidden, nothing demanding attention.
Tick a box when there's time and that crate gets its PR.

Deliberately *not* re-grouping them: #2546 proved renovate can't even
resolve a lockfile with 15 simultaneous breaking bumps
(`renovate/artifacts` failed).

Unaffected: JS/action updates and the rolling safe non-major group.

After this lands the 20 open Rust PRs get closed; they reappear on the
dashboard as pending-approval entries.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.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