Plotly Dash

The standard Dash workflow follows a specific lifecycle that any new back end should replicate:

1. Initialization: Create the app object (e.g., app = Dash(...)).
2. Layout Definition: Assign a nested structure of components to the app's layout property (e.g., app.layout = html.Div(...)).
3. Callback Registration: Define logic that connects inputs (like button clicks) to outputs (like graph updates) using decorators or specific methods.
4. Server Execution: Run the web server (e.g., app.run_server(...)).

from dash import Dash, html

# 1. Initialization
app = Dash(__name__)

# 2. Layout Definition
app.layout = html.Div('Hello Dash')

# 3. Callback Registration
@app.callback(
    Output('output-id', 'children'),
    Input('input-id', 'value')
)
def update_output(input_value):
    return f'Input value is: {input_value}'

# 4. Server Execution
if __name__ == '__main__':
    app.run_server(debug=True)

Dash is available in two primary versions depending on your deployment and scaling needs:

Dash Open Source

• Best for running apps on a local laptop or workstation.
• Ideal for individual development and prototyping.
• Does not include built-in tools for easy organizational access or enterprise-grade scaling.

Dash Enterprise

Designed for department or company-wide consumption, offering advanced features for scaling and productivity:

ML Ops & Scaling:

• App Manager: Point-and-click deployment and management without requiring IT/DevOps.
• Kubernetes scaling: Horizontal scaling and high availability via Kubernetes architecture.
• No code auth: Built-in support for LDAP, AD, PKI, Okta, SAML, OpenID Connect, OAuth, SSO, and email authentication.
• Job Queue: Enables asynchronous background processing for heavy computations, moving them out of synchronous Dash callbacks.

Low-Code Productivity:

• Design Kit: UI styling and branding without writing CSS.
• Snapshot Engine: Save/share app views as links or PDFs, or automate email reports.
• Dashboard Toolkit: Drag-and-drop layouts and chart editing.
• Embedding: Native embedding of Dash apps into existing websites without using IFrames.

Enterprise AI & Big Data:

• AI App Marketplace: Pre-built templates for common AI/ML business problems.
• Big Data Integration: Connects to Dask, Databricks, NVIDIA RAPIDS, Snowflake, Postgres, Vaex, and more.
• GPU & Dask Acceleration: Support for GPU and parallel CPU computing.
• Data Science Workspaces: Onboard code editors for Python, R, and Julia.

To prepare a production release for both NPM and PyPI, follow these steps:

1. Build JS: Run npm run build.
2. Create Tarball: Run python setup.py sdist to generate a distribution tarball in the dist/ folder.
3. Verify: Install the generated tarball in a clean environment using pip install <tarball-name>.tar.gz.
4. Publish to PyPI: Use twine upload dist/*.
5. Publish to NPM: Use npm publish.

Important Note on CDN usage: Publishing to NPM makes the JavaScript bundles available on the unpkg CDN. By default, Dash servers components from the remote unpkg CDN. If you do not publish to NPM, you must set the serve_locally flag to True in your Dash application to ensure the components load correctly.

# 1. Build code
$ npm run build

# 2. Create Python tarball
$ python setup.py sdist

# 3. Test tarball (in a new env)
$ pip install dash-html-components-<version>.tar.gz

# 4. Publish to PyPI
$ twine upload dist/*

# 5. Publish to NPM
$ npm publish