Supply Chain - Data Flow & State Management #

Supply Chain (SC) is a data flow and state management architecture.

Table of Content #

• Idea
• Goals
• Getting Started
• Features
  • Effective Asynchronous Workflow
  • Priorization
  • Animations
  • No Unneccessary Updates
  • Build-In Test Methods
  • Timeout Protection
  • Visualize the Node Graph
• Features and bugs

Idea #

• A node receives components from one or more suppliers.
• It produces a product and delivers it to customers.
• Doing so complex data flow graphs can be created.
• A supply chain manager SCM coordinates the process.

[Basic Principle]

Goals #

• ✅ Efficiently manage application state
• ✅ Visualize application data flow & dependencies
• ✅ Smoothly animate state transitions
• ✅ Data caching
• ✅ Prioritze updates

Getting Started #

• Checkout the example
• Checkout node_test.dart
• Checkout scm_test.dart

Features #

Effective Asynchronous Workflow #

• Our workflow consists of three stages: Nomination, Preparation and Production.
• In the first stage nodes are nominated for production
• The SCM collects nominated nodes for a short time, e.g. a frame or a event loop cycle.
• In the second stage, the SCM prepares a dependency graph. Nodes are staged for production.
• In the final step nodes are instructed to produce depending on dependencies and priorities.

Priorization #

• Node can have a priority.
• Nodes with higher priorities are processed first.
• If a node has a high priority, its suppliers will a high priority too.
• Currently a realtime and a frame priority is offered.
• Nodes with realtime priority will produce in the same run loop cycle.
• Nodes with frame priority will produce at the next frame / tick.

Animations #

• Scm offers a tick() method.
• Tick will nominate all nodes with frame priority
• These node are instructed to provide a new product on every frame

No Unneccessary Updates #

• All products produced by suppliers have a hash
• SCM uses suppliers to estimate if a node needs to produce or not
• No hash changes, no production.
• Additionally SCM builds a dependency graph

Build-In Test Methods #

• We break the paradigm not to mix production and test code
• Scm can be set to test mode using the isTest constructor param
• Use methods like testRunNormalTasks, testRunFastTasks etc. to control the test flow

Timeout Protection #

• I a node does not produce in a given time window, it will be skipped
• Thus nodes cannot block the production process

Visualize the Node Graph #

• Use SupplyChain.graph to get a Graphviz graph that can be turned into an svg file

Features and bugs #

Please file feature requests and bugs at GitHub.