This repository represents an incremental rewrite of Tidepool's data visualization code. The first implementation of Tidepool's data visualization can be found in tideline.
We continue to develop the data visualization code in a separate repository from Tidepool's main web application blip for several reasons:
- separation of concerns: the visualization code has distinct requirements, dependencies, and development practices from the rest of blip
- potential use by third parties building their own web applications on top of the Tidepool platform but who want to leverage some of the same data visualization components as those used in blip
Historically, tideline was a separate codebase for one additional reason that is no longer a concern for us: we wanted to keep the visualization code "pure" D3 and in particular wanted to only have a React dependency in the blip codebase to avoid forcing the choice of React as a framework on third parties who might want to use our visualization code. In 2013-2014 when React was not yet the juggernaut framework it is today, this made a lot more sense.
State of the rewrite
As of November, 2016...
Blip depends on @tidepool/viz for:
- Device Settings (new implementation)
- Trends (CGM version as new implementation, BGM version a port of more-or-less equivalent functionality from tideline's code)
Blip still depends on tideline for:
- Basics view
- Daily view
- Weekly view
- data preprocessing (found in "nurseshark" and the TidelineData constructor)
Blip itself handles:
- most of the visualization state management (e.g., checkbox options for BGM version of Trends view)
- navigation from view to view and along the time dimension
The eventual goal is for everything listed above to be handled by code in this repository. That is, eventually all code for all the data views as well as for the data preprocessing and visualization state management, including the navigation from view to view and along the time dimension will live in this repository.
a. The Miso project's d3.Chart was the most promising attempt at a lightweight "framework" for D3 heavily inspired by Backbone's API, but the project's support has unfortunately been sporadic, and it has not yet been updated to support the 4.x version of D3. Nevertheless, we used d3.Chart in the tideline version of the Trends view. ↩