Outputs

The output of the application is defined by the function passed to dstack.app(). See the function get_chart in the following example:

from datetime import datetime, timedelta
​
import dstack.controls as ctrl
import dstack as ds
import plotly.graph_objects as go
import pandas_datareader.data as web
​
​
def get_chart(symbols: ctrl.ComboBox):
    start = datetime.today() - timedelta(days=30)
    end = datetime.today()
    df = web.DataReader(symbols.value(), 'yahoo', start, end)
    fig = go.Figure(
        data=[go.Candlestick(x=df.index, open=df['Open'], high=df['High'], low=df['Low'], close=df['Close'])])
    return fig
​
​
app = ds.app(get_chart, symbols=ctrl.ComboBox(["FB", "AMZN", "AAPL", "NFLX", "GOOG"]))
​
result = ds.push("faang", app)
print(result.url)

Within such a function, it's allowed to return a single object of the type that is supported. The supported base types include:

• pandas.core.generic.NDFrame (a Pandas dataframe)

• plotly.graph_objs._figure.Figure (a Plotly figure)

• matplotlib.figure.Figure (a Matplotlib figure)

• bokeh.plotting.Figure (a Bokeh figure)

• dstack.markdown.Markdown (a Markdown text, can be created via dstack.md())

Given the supported base types, it's possible to use Pandas, Plotly, Bokeh, Matplotlib, and Seaborn for the output of the application. If you'd like to use something else as an output, please file an issue to the tracker.