Changelog

Note: the Flourish API is an enterprise-level bolt-on and not available to all customers.

Add the base_visualisation_id option, allowing a live API visualization to be based on an existing published visualization

Remove code to handle a situation (with custom styles for a company) that can no longer obtain.

You can now omit datasets or column names in more situations and sensible defaults will be used.

If API calls are made while initialisation is still in progress, queue them up and run them after initialisation is complete.

Include the compiled files (in the dist/ directory) in the package published to NPM.

Add asynchronous getState(callback) method, allowing the user to get the current template state. callback should be a function with the signature callback(err, state). err will be null if the getState call succeeds.
Add optional width and height parameters, which allows embedding a Flourish visualization at a fixed size.

A Flourish credit is now shown underneath the visualization unless the owner of the API key in use is a member of a company (i.e. a business customer)

Add two new alternative input formats, neither of which require a column_names value, but instead take a bindings object. These two new formats expect data to be either an array of objects (with column names as keys) or an array of arrays (where the first array is column headers, and the subsequent arrays are flat arrays of values).
Change the default export, so that regardless of whether you're loading this API client library from the CDN or via npm, your code can use Flourish.Live as the constructor constructor. If you're using the code via npm, you should change your import statement to: import Flourish from "@flourish/live-api" and invoke the constructor as new Flourish.Live(opts)
Fetch data from the template endpoints on the backend via a caching CDN.
Apply custom company styles automatically if the API key owner is a member of a company which has custom styles configured. (They can be overridden by using the state property in the opts object that's passed in.)
Improve error messages when the number of columns don't match

Allow single values to be passed as data in situations where the template would accept an array of values – e.g. data.value in the line-bar-pie template.

Fix a bug that caused the column_names option to be ignored on initial render (but not on update).

Change the way the .update() method is called, to make it possible to update the state and/or data independently, or to change the column_names, and to make it possible for us to add support later for changing the template without another breaking change to the calling convention.

Use the default values for the column names on update as well as initialization.

Use sensible default values for the column names.
Apply default options before setting the iframe src so you don't need to explicitly set the api_url option.

Keep messages separate if the same template is used more than once in the same page.