Skip to content

chore(deps): update dependency score_tooling to v3 - #378

Open
eclipse-score-bot wants to merge 1 commit into
mainfrom
renovate/score_tooling-3.x
Open

eclipse-score-bot wants to merge 1 commit into
mainfrom
renovate/score_tooling-3.x

Conversation

@eclipse-score-bot

@eclipse-score-bot eclipse-score-bot commented Oct 26, 2026 •

Copy link
Copy Markdown
Contributor

This PR contains the following updates:

Package Type Update Change
score_tooling bazel_dep major 2.2.0 → 3.0.2

Release Notes

eclipse-score/tooling (score_tooling)

v3.0.2

Highlights

  • rules_score: root causes are argued by RootCauseAssessment records. They replace Mitigation and SafetyMeasure.
  • rules_score: failure modes trace to concrete public API methods, generated from the public API class diagrams.
  • rules_score: formal RST test specifications, and Given/When/Then text derived from gtest test names.
  • rules_score: documentation is keyed by the full target label, so equal target names in different packages or repositories no longer collide.
  • rules_score: $TransferInGate links fault trees, requirement ASILs are checked, and the certified-scope check is repository-aware.
  • PlantUML and validation: class template parameters and specializations.
  • cr_checker: // and /* */ headers are interchangeable.

Breaking changes

These changes require edits to build files, TRLC files or PlantUML diagrams.

rules_score: root cause assessments replace mitigations and safety measures (#​534)
  • New TRLC type ScoreReq.RootCauseAssessment with root_causes, measures, safety, rationale and version. An assessment without measures argues that the root cause does not apply.

  • Removed: ScoreReq.Mitigation, ScoreReq.SafetyMeasure, AoU.root_causes (and :root_causes: in .rst aou_req directives) and root causes in CompReq.derived_from.

  • Every CompReq.derived_from entry needs a version: Package.Item@<version>.

  • safety_analysis takes the assessment files in the new attribute assessments. safety_measures may only contain assumptions_of_use and component_requirements targets.

  • assumptions_of_use and component_requirements targets no longer need the fault_trees target in deps.

  • The traceability report has the levels "Assumptions of Use" and "Root Cause Assessments" instead of "Safety Measures". Custom LOBSTER templates must use {OWN_AOU_BLOCK} and {ASSESSMENT_BLOCK} instead of {SAFETYMEASURES_BLOCK}.

    # before
    safety_analysis(
        ...
        safety_measures = [
            "//docs:sample_aous",
            "sample_safety_analysis_safetymeasures.trlc",
        ],
    )
    
    # now
    safety_analysis(
        ...
        assessments = ["sample_safety_analysis_assessments.trlc"],
        safety_measures = ["//docs:sample_aous"],
    )
    ScoreReq.RootCauseAssessment ShmemCreatedWrongNameAssessed {
        root_causes = [sample_safety_analysis_fta.ShmemCreatedWrongName]
        measures = [SampleType.SampleAoU@1]
        safety = ScoreReq.Asil.B
        rationale = "The user prevents the wrong name through the assumption of use."
        version = 1
    }
    
rules_score: failure modes reference public API methods (#​555)
  • FailureMode.interface is a list of ScoreReq.ApiMethod records instead of a free-text string. The records are generated from the public_api class diagrams of an architectural_design: one per public method name (overloads collapsed) and one per free function.

  • architectural_design has the new attribute package, the TRLC package of the generated records. It is required with public_api .puml diagrams and rejected without them.

  • failure_modes must list the architectural_design in deps, and the failure mode files import its package.

  • safety_analysis.arch_design is removed.

  • The traceability level "Architecture" is replaced by "Public API". Custom LOBSTER templates must use {PUBLIC_API_SOURCES} instead of {ARCH_SOURCES}.

    # before
    architectural_design(
        name = "sample_seooc_design",
        public_api = ["public_api.puml"],
    )
    failure_modes(
        name = "sample_failure_modes",
        srcs = ["failure_modes.trlc"],
    )
    safety_analysis(
        ...
        arch_design = ":sample_seooc_design",
    )
    
    # now
    architectural_design(
        name = "sample_seooc_design",
        package = "sample_seooc_api",
        public_api = ["public_api.puml"],
    )
    failure_modes(
        name = "sample_failure_modes",
        srcs = ["failure_modes.trlc"],
        deps = [":sample_seooc_design"],
    )
    safety_analysis(
        ...
    )
    // before
    interface = "safety_software_seooc_example.SampleLibraryAPI.GetNumber"
    
    // now
    import sample_seooc_api
    ...
    interface = [sample_seooc_api.safety_software_seooc_example_SampleLibraryAPI_GetNumber]
    
rules_score: documentation is keyed by the full target label (#​542)
  • Pages and sources of units, components, unit designs, dependency modules, architectural designs and test-case coverage are keyed by <repo>/<package>/<name> (_main for the main repository) instead of the bare target name.
    • units/<label path>/index.rst, units/<label path>/<diagram>.puml, units/<label path>/<design>.rst.inc
    • components/<label path>/index.rst, components/<label path>/requirements/...
    • Page titles are the full label.
    • The HTML of a dependency module is merged to <repo>/<package>/<dep>/ instead of <dep>/. Links to published dependency pages change.
  • Custom rules that consume the providers must adapt:
    • SphinxModuleInfo.transitive_modules and SphinxNeedsInfo.needs_modules are structs with path (the label path) instead of name.
    • UnitInfo.dependent_labels and ComponentInfo.dependent_labels are depsets of structs (unit_label, dependencies), one per unit, instead of depsets of labels (#​533).
PlantUML: the provided-interface binding )- is removed (#​544)
  • Component )- Interface is rejected. Write the provided binding as -, -- or --(). Layout and style hints (-up-, --[bold]--) stay allowed.

    ' before
    Gateway )- ResponseApi
    
    ' now
    Gateway - ResponseApi
  • A bare -- between a component and an interface is a provided binding and no longer an association.

Skills
  • The SKILL.md files of rules-score, score-architecture, score-requirements, score-safety-analysis and score-testing changed. Downstream repositories must regenerate their sync_skills goldens.

Stricter validations

These changes need no new syntax, but input that built before can now fail.

TRLC requirement ASILs (#​540)
  • A FeatReq or CompReq must not have a lower ASIL than any requirement it is derived from. An upstream ASIL_B forbids QM, an upstream ASIL_D requires ASIL_D.
  • trlc is raised from 3.0.1 to 3.1.1. Raise it if you pin an older version.
Root cause assessments (#​534)

Errors in the safety_analysis build:

  • A root cause is listed by more than one assessment.
  • An assessment lists a root cause that no fault_trees target generates, or a measure that is not an own AoU or a CompReq of safety_measures.
  • The safety of an assessment is below the ASIL of a failure mode of its root causes or above the ASIL of a listed measure.

Failures of the traceability report of the dependable_element:

  • A root cause has no assessment.
  • An own AoU is not listed as a measure of any assessment.
Certified scope (#​533)
  • unit.scope entries are resolved relative to the package of the unit and matched together with the repository: an exact target, :__pkg__ or :__subpackages__ in the same repository. A scope in another repository no longer matches by package path alone. Add the repository, for example @seooc//:sample_library.
  • A component no longer implicitly certifies nested components. A unit implicitly certifies its own implementation.
  • The message is Unit <unit label> depends on <dependency label>, but the dependency is not in any certified scope (<reason>).
  • Violations warn with maturity = "development" and fail otherwise, so release builds can fail on dependencies the old check accepted.
Documentation names (#​542)
  • Basenames of design diagrams and images must be unique within a unit.
  • design_name must be unique within its parent component (case-insensitive).
Public API diagrams (#​555)
  • The public_api diagrams of an architectural_design must be class diagrams.
  • Public API classes must not declare conversion operators, new, delete or co_await. The parser fails with UnsupportedOperator.
$TransferInGate (#​545)
  • $TransferInGate(alias, connection) connects to the fault tree of the failure mode connection. The build fails when:
    • alias is not the fully qualified name of a failure mode (for example "Transfer1").
    • The target is unknown or ambiguous.
    • The transfers form a cycle, including a transfer to the own tree.
    • The gate alias collides with another node of the same diagram.
  • Transfers across different fault_trees targets are not supported.
  • A root cause inherited through a transfer is assessed once. The safety of its assessment must be at least the ASIL of all failure modes of all trees it belongs to.

New features

rules_score
  • Failure-mode-to-API traceability (#​555): the report traces failure modes to the public API methods in the "Public API" level. The safety analysis page shows one badge per guideword.
  • Formal test specifications (#​550): the new rule test_specification takes an RST file with a Test Goal, an optional Test Focus and at least one Test Step (optional Precondition, Action, Expected). A unit lists them in test_specs, a test refers to one with RecordProperty("test-spec", "//pkg:label"). The coverage lock gets a spec key per test case, which must not be mixed with Given/When/Then keys.
  • Given/When/Then from test names (#​505): a test named [<Prefix>_]Given<X>[_When<Y>]_(Expect|Then)<Z> does not need the given/when/then properties. This needs a lobster version with --derive-gwt-from-test-name. With lobster 1.0.6 only the properties are used. A test case in a coverage lock needs a Given and a Then. The When is optional.
PlantUML and validation
  • Class template parameters (type, non-type, template-template, parameter packs, optional names) are modeled in PlantUML and compared with the C++ implementation by position, count, kind and pack marker. An omitted design name is unconstrained, an explicit name must match (#​532).
  • Partial and explicit specializations are matched to the design by their fully qualified base name. Their template arguments are not compared. Explicit instantiations are ignored (#​532).
cr_checker
  • C, C++, Rust and TRLC headers may use either // or /* */. The other form is derived from the template, and --fix keeps the form a file already uses (#​554).

Bug fixes

  • cr_checker: --modified-only honors inputs and :(exclude) pathspecs. Before, it checked all modified files. A leading YAML frontmatter block of a Markdown file is kept in front of the header, and --fix no longer breaks it (#​546).
  • rules_score: the architecture key conflict check of a dependable element works for elements in non-root packages and external repositories (#​543).

Migration checklist

Breaking changes:

  1. Replace Mitigation/SafetyMeasure records by RootCauseAssessment records and pass them through safety_analysis.assessments. Remove root_causes from AoUs and root causes from CompReq.derived_from. Add @<version> to every derived_from entry.
  2. Add package to architectural_design targets with public_api, list them in the deps of failure_modes, remove arch_design from safety_analysis, and write FailureMode.interface as a list of ApiMethod records.
  3. Update custom LOBSTER templates: {OWN_AOU_BLOCK} and {ASSESSMENT_BLOCK} replace {SAFETYMEASURES_BLOCK}, and {PUBLIC_API_SOURCES} replaces {ARCH_SOURCES}.
  4. Adapt links to documentation paths (<repo>/<package>/<dep>/) and custom rules that read the changed provider fields.
  5. Replace Component )- Interface by Component - Interface in PlantUML component diagrams.
  6. Regenerate the sync_skills goldens.

Stricter validations:

  1. Check the ASIL of every FeatReq and CompReq against its upstream requirements.
  2. Give every root cause exactly one assessment and list every own AoU in an assessment.
  3. Add the repository to unit.scope entries of external targets and build with maturity = "release".
  4. Make design diagram and image basenames unique per unit and design_name unique per component.
  5. Make sure public_api diagrams are class diagrams without unsupported operators.
  6. Check $TransferInGate calls: use the fully qualified failure mode as alias and a failure mode of another tree as connection.

v3.0.1

Changelog

rules_score: unit documentation layout

  • Unit pages are rendered to units/<unit>/index.rst. Design files are staged in
    units/<unit>/design/. Previously: units/<unit>.rst and units/<unit>_design/.
  • Sphinx exclude pattern **/units/*_design replaced by units/*/design.
  • Two different units with the same name in one dependable_element now fail the build
    (previously the second was silently dropped).
  • File collisions while staging a unit's design files now fail the build.
  • PUML basenames no longer need to be unique across units.

Added

  • sync_skills(repo_skills_target = None): optional label of a filegroup with the repo's
    committed score-* skills. Required when .github/skills has its own BUILD file.
  • .devcontainer/run-tool: standard SCORE tool launcher.

Fixed

  • component(requirements = ...) accepts assumed_system_requirements targets, so component
    requirements can derive directly from assumed system requirements without a feature layer.

v3.0.0

Highlights

  • rules_score: the safety analysis is split into failure_modes, fault_trees and safety_analysis. A dependable_element now derives its own AoUs from the safety analysis measures.
  • PlantUML tooling: an element's identifier now comes from its written name, not its alias. Qualified and rooted declarations and free functions in class diagrams are supported.
  • validation: free-function consistency check between PlantUML class design and C++ implementation.

Breaking changes

rules_score: safety analysis and AoUs
  • safety_analysis is split into three rules to avoid a dependency cycle between the fault trees, the measures and the analysis:

    • failure_modes(srcs, deps): the FailureMode records.
    • fault_trees(srcs, deps, fta_package): the fault-tree diagrams and the generated RootCause records.
    • safety_analysis(failure_modes, fault_trees, safety_measures, arch_design): the FMEA page.
  • safety_analysis attributes are renamed:

    Before Now
    failuremodes failure_modes (labels of failure_modes targets)
    root_causes fault_trees (label of a fault_trees target, required)
    safetymeasures safety_measures
    arch_design (optional) arch_design (required)
  • dependable_element.assumptions_of_use is removed. The element's own AoUs are the assumptions_of_use targets listed in the safety_measures of its safety_analysis targets.

    • An assumptions_of_use target that no safety_analysis lists is neither rendered nor forwarded.
    • A dependable_element without a safety analysis has no own AoUs.
  • ScoreReq.AoU.root_causes is now mandatory.

    • .rst aou_req directives need :root_causes: <fta_package>.<RootCause>, ....
    • assumptions_of_use.deps must list the fault_trees target. Other assumptions_of_use targets are rejected there.
    • :root_causes: on non-AoU directives is rejected.
  • aou_forwarding.yaml entries must carry the AoU version: aou_id: "Package.AOU_ID@1". Unknown IDs, version mismatches and duplicate entries fail the build.

  • The default fta_package is now the package and name joined by __ plus _fta (e.g. pkg_sub__name_fta). Set fta_package explicitly if you reference <fta_package>.<RootCause> from elsewhere.

  • Raw .trlc files in safety_measures may only contain Mitigation records. A root cause whose failure mode is missing from failure_modes fails the build.

  • List one safety_analysis per dependability_analysis. Use several dependability_analysis targets for several analyses.

  • The .github/skills/*/SKILL.md files changed. Downstream repos must regenerate their sync_skills goldens.

PlantUML: identifiers are built from the name, not the alias
  • The written name of an element is its identity. The alias (as X) is only a local reference key and no longer takes part in the identifier.

  • Names must be identifier paths: segments of letters, digits and _, separated by . or ::. Free-text names are errors, for every element kind.

  • Sequence participants:

    • The "instance : pkg::Type" label form is removed.
    • Referencing an aliased participant by its name instead of its alias is an error.
  • Diagrams must be migrated.

    ' before
    component "Unit 1" as unit_1 <<unit>>
    participant "Order Service" as OrderService
    
    ' now
    component unit_1 <<unit>>
    participant "orders::OrderService" as OrderService
  • validation matches by this identifier. Component and sequence checks match participant uids, and the Bazel check matches the last segment of the entity id (case-insensitive) against the Bazel target name.

  • See plantuml/parser/docs/element-identifiers.md for the full rules.

New features

PlantUML
  • Qualified names in nested declarations are appended to the enclosing scope (like relative names in C++). A leading . or :: roots the name:
    package outer {
        class core::X     ' outer.core.X
        class ::core::X   ' core.X
    }
    :: is accepted wherever a name is written: declarations, package/namespace names, ports, extends/implements targets and relationship endpoints.
  • Free functions in class diagrams: entity Name <<function>> { + f(...) } groups free functions, which resolve in the enclosing namespace. They are serialized to FlatBuffers. See plantuml/parser/docs/class-diagram.md. (#​511)
  • Class-diagram parser and AST simplified. (#​517)
  • Clearer sequence-diagram diagnostics for aliased participants and lifecycle commands.
Sphinx clickable_plantuml
  • Links are resolved by uid. A last-segment name index lets a bare unit_1 in a sequence or component diagram find the diagram that defines pkg.comp.unit_1.
  • *.idmap.json gains match_by_name (default false). It is true for component, class and sequence diagrams and false for FTA diagrams.
  • Aliases that are not plain identifiers are quoted in url of "...". Aliases that cannot be written (double quote, line break) are reported with a warning.
validation
  • New ClassDesignImplementationFreeFunctionConsistency check: a free function in the unit design must exist in the C++ implementation with a matching return type, parameters and template parameters. (#​518)
  • Clearer errors for bazel_component, sequence and class checks.

Bug fixes

  • validation: a class reachable through several implementation targets (e.g. a shared header) was reported as defined more than once. The duplicate-class error now says "design" or "implementation". (#​522)
  • rules_score: units listed directly in dependable_element.components were missing from the generated toctree. (#​519)

Build and CI

  • Rust crates are pinned to minor versions (~x.y). Cargo.lock and cargo-bazel-lock.json are added, and the crate repository is renamed from @crates to @crates_score_tooling. (#​515)

Migration checklist

  1. Split each safety_analysis into failure_modes, fault_trees and safety_analysis, and rename its attributes.
  2. Remove assumptions_of_use from dependable_element. List the AoU targets in safety_analysis.safety_measures instead.
  3. Add root_causes to every AoU and list the fault_trees target in the deps of each assumptions_of_use target.
  4. Add @<version> to the aou_id entries in aou_forwarding.yaml.
  5. Rewrite PlantUML declarations to identifier-path names. Use an alias only for long or qualified names.
  6. Regenerate the sync_skills goldens.

Full changelog: eclipse-score/tooling@v2.3.1...v3.0.0

v2.3.1

Release v2.3.x contains breaking changes:

Renames (Bazel)
fmea(...) is now safety_analysis(...).
dependability_analysis(fmea = [...]) is now dependability_analysis(safety_analysis = [...]).
The controlmeasures attribute is replaced by safetymeasures. These files hold Mitigation, AoU and CompReq records.
New optional attribute fta_package sets the package name of the generated RootCause records. It defaults to _fta.
unit(...) and component(...) get an optional design_name. Use it when the PlantUML <>/<> name differs from the Bazel target name.

FTA .puml diagrams
$TopEvent(...) is now $FailureMode(name, "Pkg.FailureMode", ...). It accepts up to 8 failure modes per node.
$BasicEvent(name, "Pkg.Alias", conn) is now $RootCause(name, "Alias", conn). The alias is a plain name without a package prefix.
Root causes are generated as TRLC records in fta_events.trlc (package <fta_package>). They are not written by hand.
Each root cause links to the failure modes of the tree it sits under.

TRLC model (ScoreReq)
Removed: ControlMeasure, PreventiveMeasure and Measure.
Removed: the mitigates field on AoU and Mitigation.
Added: RootCause, the abstract SafetyMeasure, and FailureMode-to-RootCause links.
Mitigation no longer extends AssumedSystemReq. It now requires root_causes = [...] and justification.
AoU can have optional root_causes = [...].
CompReq.derived_from can reference a RootCause, unversioned. This makes the CompReq a control measure for that root cause.
CompReq.derived_from may only reference AoUs received from other dependable elements. An element's own AoUs are never a source.

Wiring (BUILD)
A safety_analysis target that an AoU references via root_causes must be listed in the deps of assumptions_of_use.
The same target must also be passed via safetymeasures to safety_analysis(...).

Migration:

BUILD
fmea becomes safety_analysis.
controlmeasures becomes safetymeasures.
dependability_analysis(fmea=...) becomes dependability_analysis(safety_analysis=...).

FTA .puml
$TopEvent becomes $FailureMode.
$BasicEvent(n, "Pkg.Alias", c) becomes $RootCause(n, "Alias", c). The alias is a plain identifier with no package prefix.

TRLC measures
Add import _fta, the generated package. It defaults to the target name plus _fta.
Delete ControlMeasure and PreventiveMeasure records and all mitigates fields.
Convert each former measure:
Mitigation: add root_causes = [_fta.] and justification.
CompReq: add derived_from = [_fta.], with no @​version. Add the safety_analysis target to that component_requirements target's deps.
AoU: add root_causes = [_fta.]. Add the safety_analysis target to the assumptions_of_use deps, and pass the file in safetymeasures.

Coverage
Every $RootCause needs a Mitigation, AoU or CompReq. Otherwise release maturity fails.
Every FailureMode needs a $RootCause under it.

v2.3.0

What's Changed

Full Changelog: eclipse-score/tooling@v2.2.3...v2.3.0

v2.2.3

What's Changed

Full Changelog: eclipse-score/tooling@v2.2.2...v2.2.3

v2.2.2

What's Changed

New Contributors

Full Changelog: eclipse-score/tooling@v2.2.0...v2.2.2

v2.2.1

What's Changed

Full Changelog: eclipse-score/tooling@v2.1.1...v2.2.1


  • If you want to rebase/retry this PR, check this box

This PR was generated by #infrastructure automation. Contact us if you have any questions or feedback.

@github-actions

github-actions Bot commented Oct 26, 2026 •

Copy link
Copy Markdown
Contributor

License Check Results

🚀 The license check job ran with the Bazel command:

bazel run --lockfile_mode=error //:license-check

Status: ⚠️ Needs Review

Click to expand output
[License Check Output]
Extracting Bazel installation...
Starting local Bazel server (8.6.0) and connecting to it...
INFO: Invocation ID: 88dbdf53-5505-4d98-a965-9a7e3b3438d6
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
WARNING: For repository 'rules_cc', the root module requires module version rules_cc@0.2.17, but got rules_cc@0.2.25 in the resolved dependency graph. Please update the version in your MODULE.bazel or set --check_direct_dependencies=off
WARNING: For repository 'score_platform', the root module requires module version score_platform@0.7.2, but got score_platform@0.7.3 in the resolved dependency graph. Please update the version in your MODULE.bazel or set --check_direct_dependencies=off
WARNING: For repository 'score_crates', the root module requires module version score_crates@0.0.11, but got score_crates@0.0.13 in the resolved dependency graph. Please update the version in your MODULE.bazel or set --check_direct_dependencies=off
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
Computing main repo mapping: 
WARNING: /home/runner/.bazel/external/score_baselibs+/score/json/BUILD:24:12: target '@@score_baselibs+//score/json:base_library' is deprecated: base_library no longer has any effect, vaJson is the only parser backend. Remove this flag from your build configuration; it will be deleted in a future release.
Loading: 
Loading: 4 packages loaded
DEBUG: /home/runner/.bazel/external/score_docs_as_code+/docs.bzl:716:14: INFO: ⚠️ DEPRECATED: Passing a Needs inventory through docs(data = [...]) is deprecated!!! Move it to external_needs = [...] instead.
WARNING: Target pattern parsing failed.
ERROR: Skipping '//:license-check': no such target '//:license-check': target 'license-check' not declared in package '' defined by /home/runner/work/logging/logging/BUILD
ERROR: no such target '//:license-check': target 'license-check' not declared in package '' defined by /home/runner/work/logging/logging/BUILD
INFO: Elapsed time: 17.214s
INFO: 0 processes.
ERROR: Build did NOT complete successfully
ERROR: Build failed. Not running target

@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from 6a49f57 to b124e57 Compare October 27, 2026 13:21
@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from b124e57 to 2449a73 Compare October 28, 2026 09:15
@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from 2449a73 to 1078650 Compare October 28, 2026 10:10
@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from 1078650 to a9d880b Compare October 28, 2026 11:10
@eclipse-score-bot eclipse-score-bot changed the title Update dependency score_tooling to v3 chore(deps): update dependency score_tooling to v3 Oct 28, 2026
@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from a9d880b to 5eae1a3 Compare October 28, 2026 12:11
@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from 5eae1a3 to 8558b3d Compare October 28, 2026 14:09
@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from 8558b3d to 74ddc47 Compare October 28, 2026 15:10
@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from 74ddc47 to 7265694 Compare October 29, 2026 07:10
@eclipse-score-bot
eclipse-score-bot force-pushed the renovate/score_tooling-3.x branch from 7265694 to 02ac61e Compare October 29, 2026 08:11

This branch is waiting to be deployed

1 waiting deployment
workflow-approval — 2ce6a70c Waiting Oct 30, 2026 by eclipse-score-bot via run-tests / coverage approval gate #1586
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

Status: Backlog

Development

Successfully merging this pull request may close these issues.

1 participant