Why Chengwu Low-Code Platform Chose LogicFlow for Business Process Orchestration

Why a Process Orchestration Engine Is Needed

Traditional form-driven platforms only support linear flows such as submit → approve → end. Modern enterprise systems require conditional branching (e.g., amount > 100k requires GM approval), parallel processing (multiple departments approve simultaneously), automated nodes (API calls, notifications), sub-process calls and flow jumps, instant effect after changes without code deployment, and full version control with history traceability. These demands require a complete graphical logic orchestration platform, not just an approval engine. The Chengwu platform therefore introduced a full stack combining flowcharts, node business logic binding, and backend orchestration scheduling, with LogicFlow as the core graphical layer.

LogicFlow Overview: Developer-Friendly Flowchart Engine

LogicFlow is an open-source flowchart orchestration tool released by DiDi in 2020, positioned as a "developer-friendly flowchart framework." It uses SVG rendering, encapsulates common nodes, edges, anchors, and an event system, and allows deep customization of node behavior and styling. Key features include built-in basic primitives (start, decision, user, system, end nodes), flexible edge connection with anchor control and conditional lines, comprehensive event mechanism (click, connect, move, delete), high customizability (custom nodes, context menus, floating buttons), JSON-based data structure for serialization with backend, and explicit business semantic model emphasizing that every graphic element carries business meaning. Unlike G6 or ReactFlow which focus on graphical expressiveness, LogicFlow is built for "business process modeling" with a semantics-first approach.

Comparative Evaluation of Alternatives

The team compared LogicFlow against BPMN.js, ReactFlow, and AntV G6 across several dimensions:

License: All four are permissive (MIT or Apache 2.0).

Rendering: LogicFlow and BPMN.js use SVG; ReactFlow uses HTML + Canvas; G6 uses Canvas.

Process Semantics: LogicFlow has built-in common nodes with business orientation; BPMN.js follows the strict BPMN standard; ReactFlow and G6 lack process semantics.

Customization: LogicFlow offers high customization via node/menu/command registration; BPMN.js is constrained by the BPMN standard; ReactFlow has high customization but no process semantics; G6 provides rich graphical customization.

Learning Curve: LogicFlow is easy to adopt; BPMN.js has a steep curve; ReactFlow is moderate; G6 requires extensive custom work.

Low-Code Fit: LogicFlow works well with schema-driven configuration; BPMN.js expresses redundancy; ReactFlow lacks data-driven capability; G6 requires building an entire model layer from scratch.

Conclusions: BPMN.js suits standardized organizations like banking or government with strong compliance needs. ReactFlow fits free-form graph interactions such as graph database visualization. AntV G6 is a graph engine, not a process engine, requiring massive semantic supplementation. LogicFlow best fits a low-code platform that must express real business logic.

LogicFlow Application Practices in Chengwu Platform

1. Node Type Extensions

Beyond official nodes, the team extended:

API Node: Calls external APIs with configurable request parameters and response handling.

Condition Node: Supports boolean expressions or variable value matching.

Human Task Node: Binds user groups to initiate task processing.

Amis Form Node: Links Amis components to nodes for data collection.

Script Execution Node: Configures Pebble template syntax expressions for dynamic computation.

Notification Node: Pushes to WeChat, SMS, or internal notification modules.

2. Visual Process Editor

The team wrapped LogicFlow with:

Double-click node to open configuration panel (Drawer + JSON editor).

Context menu with custom commands: copy node, set condition, set as start node.

Integration of @logicflow/extension plugins for alignment lines, snap lines, keyboard shortcuts.

Auto-layout plugin to improve diagram aesthetics.

Multi-language support: node labels and fields internationalized.

3. Integration with Backend Execution Engine

After saving the process JSON, the backend performs "interpretive execution":

Each node type registers a corresponding handler class.

State machine drives process flow.

Node execution status supports write-back (success/failure/retry).

Combined with organizational permission system for role-based authorization during execution.

Supports version rollback and process diagram history comparison.

4. Multi-Tenant and SaaS Adaptation

Each application/organization maintains independent process definitions, supporting:

Application variable placeholders (e.g., ${app.id}).

Condition branches using Pebble templates for complex judgments.

Multi-language and tenant style isolation.

Export/import process templates as "tenant initialization packages."

Summary: Why LogicFlow Is the Ideal Choice

LogicFlow won because:

Open developer-oriented architecture with extreme flexibility.

Supports process semantic modeling for complex business abstraction.

Native JSON streaming definition eases backend integration.

Mature plugin system enables rapid secondary development.

Active ecosystem, frequent maintenance, comprehensive documentation.

Most critically, LogicFlow is not a general-purpose graph tool but purpose-built for "business process diagrams." Compared to BPMN's complexity, ReactFlow's freedom, and G6's low-level nature, LogicFlow hits the sweet spot of clear semantics, flexible customization, and developer ergonomics, perfectly matching the low-code platform's dual demand for graphical expression and executable deployment.