Update github actions - #669
Merged
Merged
Conversation
renovate
Bot
force-pushed
the
renovate/github-actions
branch
from
September 5, 2026 10:12
7cfac8b to
4fd8788
Compare
renovate
Bot
force-pushed
the
renovate/github-actions
branch
2 times, most recently
from
September 8, 2026 22:09
418ac66 to
e79c440
Compare
renovate
Bot
force-pushed
the
renovate/github-actions
branch
from
September 9, 2026 21:30
e79c440 to
b8e9cf3
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR contains the following updates:
82b1b08→b944cc61.4.0→1.4.2v3.22.0→v3.24.0v4.37.9→v4.38.0v6.0.10→v6.1.0Release Notes
oven-sh/bun (bun)
v1.4.2: Bun v1.4.2Compare Source
To install Bun v1.4.2
Windows:
powershell -c "irm bun.sh/install.ps1|iex"To upgrade to Bun v1.4.2:
Read Bun v1.4.2's release notes on Bun's blog
Thanks to 3 contributors!
v1.4.1Compare Source
fallow-rs/fallow (fallow-rs/fallow)
v3.24.0: : Findings that say when the run behind them was incompleteCompare Source
This release is about output honesty: a number that is a floor should not be presented as a total, a zero that means "not measured" should say so, and a mutation resting on an incomplete picture should not be one click away.
Findings say when the run behind them was incomplete
A file that fails to parse, one the size guard skipped, one that could not be read: each of those silently yields zero imports, which can make a live export read as dead. Nine dead-code arrays now carry an optional
reachability_caveats[]naming exactly that, so a reader who never scrolls back to the diagnostics list still sees it.fallow fixrefuses to apply a mutation under a caveat (skip_reason: "low_confidence_incomplete_analysis"), the actions contract reportsauto_fixable: falsewith the reason in itsnote, and the review formats withhold the one-click```suggestionblock rather than hedging in prose above it. Findings themselves are never withheld, filtered, reordered, or re-severitied, and no exit code changes. A run that read every file it discovered is byte-identical.New commands
fallow trace --path <FROM> <TO>walks the import graph and returns the shortest chain between two modules, with type-only hops reported rather than skipped. Also over MCP astrace_import_path.fallow trace-error [FILE|-]resolves a runtime stack trace against the project graph. It refuses to overclaim: a frame matching several definitions isambiguous, one matching none staysnot_found. Also over MCP astrace_error.fallow doctorgains three advisory checks for installed dependencies and cache reuse.get_cloud_runtime_context, which reads a repository's runtime facts from fallow cloud and returns the sameruntime_coverageblock the local tools return. The key comes fromFALLOW_API_KEYin the server environment and is never a call argument. This is the only fallow MCP tool that makes a network call.Reproducible churn
Recency weighting, ownership staleness, and the churn window read the system clock at three separate points, so numbers drifted between runs over the same commit. All three now resolve one instant from HEAD's committer timestamp,
FALLOW_CLOCK_EPOCHpins it explicitly, andhotspot_summaryreports which clock it used and whether that is reproducible.Smaller, honest envelopes
The review brief carried its impact closure twice, uncapped, which on a one-file change was more than half the envelope.
graph_facts.reachable_fromis gone;impact_closurenow reports an exactaffected_count, a capped sample, and anaffected_by_dirrollup that says whether a change stayed inside its module.fallow dupes --no-fragmentsand the MCPfind_dupesinclude_fragmentsparameter return location-only clone instances. The MCP tool list stops inlining detail every session pays for on connect.Fixes worth calling out
const { method } = instancecounts as used.run:block is no longer read as an entry glob.fallow fixprints forward slashes on Windows.dupes --top Nreported two different scopes in one stats object; all four numbers now describe the measured corpus, with explicit shown and omitted counts.fallow coverage analyze --cloudkeeps the functions the runtime actually ran.fallow license refreshcan recover from a stale stored license using an API key.Thanks @Ha1baraA11, @kristersd, and @carlrannaberg for reports and patches in this release.
Three analysis caches change format together, so the first run after upgrading is cold.
Full changelog: fallow-rs/fallow@v3.23.0...v3.24.0
v3.23.0: : a readiness check, and reports that survive a moved checkoutCompare Source
An empty Fallow report has always been ambiguous. It can mean the project is clean, or it can mean Fallow never resolved your config and analyzed almost nothing, and until now the output looked the same either way.
A readiness command
fallow doctorsettles it before you run anything. The command checks the project root, which config actually resolved, whether workspace discovery found what you expect, whether external plugins load, and whether the optional type-aware companion is available. Every check is a local read, so nothing is analyzed and nothing is written.A failed required check exits
2. There is no exit1, because that code means error-severity findings anddoctorlooks at your setup, not your code. For agents,--format json --quietgives you astatusto branch on and achecks[].messageper check to report back.Reports that stay put when the checkout moves
Duplication groups that share a full content hash used to take an ordinal derived from absolute-path digests. Move an unchanged checkout to a different directory and the same handle could land on a different group. Those groups now use report-scoped
dup:<16hex>-rNhandles ordered by canonical fragment, location, and metric. Health complexity ties break on project-relative path, line, column, and function name, so--topselects the same finding wherever the repository lives.If you suppress duplication findings, read this before upgrading. Legacy numeric collision handles (
dup:<16hex>-N) are still valid syntax and are deliberately not aliases for the corrected ones. An old collision suppression or baseline key will resurface its finding for review, rather than match some other group without telling you. Regenerate the report, look at the group, then refresh itsignoredCloneskey or baseline. Update every installation before you store-rNkeys, since older versions reject that syntax;minimumVersionpins the released version a shared config needs.Ordinary
dup:<8hex>and wideneddup:<16hex>handles are unaffected, which is most of them.Detection accuracy
Four people filed four bugs this cycle, and all four are fixed here.
A class member reached through an object container or an object alias, rather than directly on the instance, was never followed (#2546, thanks @Ericlm). Binaries invoked through
varlock run -- <bin>lost their argument boundaries in package-script parsing (#2551, reported and fixed by @PrinceD96 in #2552).fallow-ignore-next-line unused-catalog-entrywas suggested by the finding and then did nothing, because quoted YAML scalar content was dropped in catalog parsing (#2548, thanks @michalius). Scoped package names rendered as empty strings inreview --brief(#2553, thanks @Elia97).Oxlint configs that declare
jsPluginsnow count those as referenced dependencies, so Ultracite-style setups stop reporting their lint plugins as unused.Editors and CI
Zed reaches the parity its extension API allows. Exact diagnostic codes go in
initialization_options.mutedCategories, where a team can commit them alongside the project. Inline complexity Code Lens is opt-in throughhealth.inlineComplexity, and advisory security candidate diagnostics turn on through your project rules like anywhere else. The Zed guide separates those LSP features from fullfallow healthandfallow securityCLI runs, and says plainly that the current extension API offers no contribution point for a Fallow-owned sidebar or status-bar item.Inline review comments now carry clone evidence: the stable
dup:handle with line and instance counts, plus each peer copy as a repository-relative range, so a reviewer can open the other copies from the comment (thanks @Jerc92 for #2550).The VS Code integration frames LSP messages by bytes, which fixes stream desynchronization on workspace paths containing non-ASCII characters.
For agents
Every MCP tool parameter now carries a description taken from the CLI help, so a tool reads the same way in an agent as it does in a terminal.
guardnow spells out its own contract. It reads config and runs no analysis, and a path matching no configured zone comes back as an empty rule set instead of an error. The server instructions point attools/listand the resources instead of listing every tool inline, which keeps them from drifting each time the roster changes.The agent instruction surface got a prompt audit. Numbers that had rotted were deleted rather than refreshed, since a refreshed number rots again next release, and the CLI reference now names
fallow dead-codein the three places it still used the legacycheckalias.Install
Full Changelog: fallow-rs/fallow@v3.22.0...v3.23.0
github/codeql-action (github/codeql-action)
v4.38.0Compare Source
linux-arm64CodeQL bundle when available. #4072pnpm/action-setup (pnpm/action-setup)
v6.1.0Compare Source
What's Changed
Full Changelog: pnpm/action-setup@v6.0.10...v6.1.0
Configuration
📅 Schedule: (UTC)
🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
👻 Immortal: This PR will be recreated if closed unmerged. Get config help if that's undesired.
This PR was generated by Mend Renovate. View the repository job log.