🖥️ Creating Routines and Jobflows via GUI

This guide explains how to define and run routines and jobflows using the platform’s visual interface. This GUI-based approach complements the code-driven method by offering an intuitive, form-based experience for designing, configuring, and launching complex workflows.

---

📌 Routine Creation in the GUI

Instead of writing wkube.py, the GUI provides a form to define a routine as a reusable routine entity.

🧾 Metadata Form

When creating a new routine:

• You fill out a structured form that mirrors all the required WKubeTask fields:
  • name, base_stack, command, etc.
• These fields are one-to-one mapped with the programmatic API.

🛠️ conf Configuration via JSON Schema

• A separate section of the form is generated dynamically using a JSON Schema.
• This allows you to define any number of additional configuration parameters.
• When the form is filled, values are converted into a conf dictionary — just like in code — and passed as * environment variables* to the container.

🔄 Explicit Data Mappings

The GUI provides explicit input fields for:

• input_mappings
• output_mappings

These support semicolon-separated mapping syntax and help users wire up storage paths easily.

---

📸 Routine Creation Form

---

🚀 Launching a Routine in the GUI

Once a routine entity is saved, you can launch it in multiple ways:

Option 1: From the Routine List

• Navigate to the Routines section.
• Click the “Run” button on any saved routine.
• A dispatch form appears:
  • Fields from the metadata and conf are editable.
  • You can override mappings and configuration at launch time.

Option 2: From the File Explorer

• Select one or more files or folders.
• Right-click and choose “Run with Routine”.
• This injects special mapping keys:
  • selected_files
  • selected_folders
• These are resolved into full mappings automatically when the job runs.

📸 Routine Launch Form

---

🧩 Visual Jobflow Builder

The platform provides a graphical interface for designing and launching Jobflows.

🎛️ Jobflow Construction

• Drag and drop previously saved routine entities into a visual canvas.
• Connect them with:
  • Child relationships
  • Callback relationships

This builds an acyclic execution graph where tasks can be executed in sequence, in parallel, or conditionally.

🧱 Aggregator (Spanner) Nodes

• You can add a dummy aggregator node to group child routines.
• These spanners don’t run code — they are used for structuring or parallelizing flows.

---

🔁 Parallel Execution with Multipliers

• Parallelism is controlled by a multiplier parameter (typically a CSV file).
• This file can be:
  • Mapped manually in the flow configuration
  • Or more conveniently: selected in the file explorer at launch time

When launching from the file explorer, the selected file is automatically used as the multiplier source for spanner/aggregator nodes.

---

📸 Jobflow Creation View

---

✅ Summary

Feature	Code-Based	GUI-Based
Define Routine	wkube.py and WKubeTask	Form-based entity creation
Configure conf	Python dict	JSON schema–driven form
Data Mappings	input_mappings, output_mappings	Explicit fields in form
Launch Routine	accli dispatch	Run from Routine list or File Explorer
Build Jobflow	Define via task graph in code	Drag-and-drop interface
Control Parallelism	Manual mapping	File-based multiplier at launch
Spanner/Aggregator Support	Holder WKubeTask()	Dummy node in Jobflow builder

GUI workflows provide a user-friendly, modular, and reusable way to define and orchestrate jobs, especially helpful for less code-heavy users or collaborative teams.