Every month marks a new goal and new priorities laid out for the weeks to come. During April '22, we've spent some time to solidify our infrastructure, improve the stability of our service, and expand on the robustness that devs would come to expect from Terra. Beyond this, we've also been able to work on improving end users' experience through Terra through additional transparency.
- Our first and most important change towards the beginning of this month fell under the category of infrastructure; some major clients polled us with thousands of payload requests for years' worth of data for a lot of users, which caused a massive backlog of webhooks to be stuck behind them in a queue. This meant that new end users authenticating wouldn't see this reflected in devs' apps, and left a lot of people confused. To improve this, creating more queues based on the type of task - large historical data querying tasks are pushed out to a separate queue, and sending out small payloads like authentication success requests have their own queue. Read more about this here
- Multilingual support. A lot of our customers are building revolutionary apps for developing markets, in various countries where english isn't the primary language and for users who would rather have an interface tailored to the language they are most comfortable in. To this effect, we wanted to improve on UX and offer this opportunity for devs to display authentication in multiple languages (part of this was powered by our heavily multilingual team! 💪 ) We've added support for ES, PT, FI, SV to our widget as well as login screens! Additionally, we can thank our supportive community of developers, who help us out with any languages they'd like to see added.
- Google Fit SDK. UX being one of the things we could afford to spend time on this month, this transitioned over to how we thought of what integrations to build out. A couple months ago some devs brought to our attention some discrepancies between users' activities and what is displayed within their Google Fit app. Additionally, they mentioned that the auth process was much smoother for end users when granting permissions natively in their device as opposed to through an oauth2 screen. We've gone ahead and built out the SDK version of Google Fit for this reason. The SDK allows us to fetch the exact same data users would see within the app, and makes the auth process much more seamless.
More improvements:
- FreestyleLibre integration will now ensure that 2fa is set up appropriately for the account during authentication to ensure that fetching data will work correctly
- Implemented timezone offsets when making requests to the Google Fit REST API to ensure that the requested days correctly correspond to the user's timezone
- Implemented Google daily manual & automatic daily payload tagging for API Mk II, based on whether any manually added samples have been included in the payload
- Improved processing of IDs for specific providers to increase security
- Improved payload deduplication to work better for body and nutrition payloads
What's next?
Our goal has always been to transition health & fitness analytics from a static view of snapshots in time to a dynamic, constantly updating feed of data. So far, we have achieved this to an extent and with some delay between users' data and devs getting access to it. Our next step is to now provides live streams of data from users' wearables directly to our customers, so that their apps can act on it within milliseconds, capture and process that data. Our first step is the last point mentioned above. Through our Google Fit SDK we taking a first step in this direction, as this is the most straightforward integration of a live data stream for us. As we're writing this, we are working on updating our backend servers to support these live streams of data, as well as preparing the SDK updates that will interact with this new infrastructure and deliver instantaneous feeds of data to you all.
Stay tuned for more
