Should we reduce where we define the target runner machine?

[HChughtai]

Short Description

Prompted by https://github.com/UCL-MIRSG/MIRSG/issues/237

With composite actions, the calling workflow and job defines runs-on and what runner machine is targeted. This means that we have to define this machine in each calling workflows -e.g. every repository includes that includes a copy of https://github.com/UCL-MIRSG/mirsg-template/blob/main/.github/workflows/manage-projects.yaml#L11

This makes it difficult to update the targeted runner across the organisation. Reusable workflows however can define the type of machine used within them in a single location.

Could we use a combination of composite actions, reusable workflows and organisational variables to reduce duplication here?

Details

e.g. in a structure like this the machine would only need to be specified in two places (or as many reusable workflow that exist) and be independent of the number of repositories (and only defined once if a variable was used)

In contrast, if each repository called each composite action directly and defined the target runner there, it would scale linearly with the number of repositories.

graph TD;
    subgraph Composite Action 1
        direction TB
        A1[Checkout code]
        A2[Set up environment]
        A3[Run build script]
        A4[Run tests]
    end

    subgraph Composite Action 2
        direction TB
        B1[Checkout code]
        B2[Set up environment]
        B3[Run linting]
        B4[Deploy]
    end

    subgraph Composite Action 3
        direction TB
        C1[Checkout code]
        C2[Set up environment]
        C3[Run security scans]
        C4[Build and test]
    end

    subgraph Composite Action 4
        direction TB
        D1[Checkout code]
        D2[Set up environment]
        D3[Generate documentation]
        D4[Publish artifacts]
    end

    subgraph Reusable Workflow 1
        direction TB
        E1[Runs on: ubuntu-latest]
        E2[Call Composite Action 1]
        E3[Call Composite Action 2]
        E4[Call Composite Action 3]
        E1 --> E2
        E2 --> A1
        A1 --> A2
        A2 --> A3
        A3 --> A4
        E1 --> E3
        E3 --> B1
        B1 --> B2
        B2 --> B3
        B3 --> B4
        E1 --> E4
        E4 --> C1
    end

    subgraph Reusable Workflow 2
        direction TB
        F1[Runs on: ubuntu-latest]
        F2[Call Composite Action 3]
        F3[Call Composite Action 4]
        F1 --> F2
        F2 --> C1
        C1 --> C2
        C2 --> C3
        C3 --> C4
        F1 --> F3
        F3 --> D1
        D1 --> D2
        D2 --> D3
        D3 --> D4
    end

    subgraph Main Workflow 1
        direction TB
        G1[Trigger: Push to main]
        G2[Call Reusable Workflow 1]
        G3[Call Reusable Workflow 2]
        G1 --> G2
        G2 --> E1
        G1 --> G3
        G3 --> F1
    end

    subgraph Main Workflow 2
        direction TB
        H1[Trigger: Pull request]
        H2[Call Reusable Workflow 1]
        H3[Call Reusable Workflow 2]
        H1 --> H2
        H2 --> E1
        H1 --> H3
        H3 --> F1
    end

    subgraph Main Workflow 3
        direction TB
        I1[Trigger: Schedule]
        I2[Call Reusable Workflow 1]
        I3[Call Reusable Workflow 2]
        I1 --> I2
        I2 --> E1
        I1 --> I3
        I3 --> F1
    end

    subgraph Main Workflow 4
        direction TB
        J1[Trigger: Tag creation]
        J2[Call Reusable Workflow 1]
        J3[Call Reusable Workflow 2]
        J1 --> J2
        J2 --> E1
        J1 --> J3
        J3 --> F1
    end

Next Steps/Accepted Response

No response