Changelog
Back to All
improved

Weekly Changelog: Monday, August 25

4 days ago by Rohan Ramchand

Today is Monday, August 25. Last Monday (August 18) was Never Give Up Day, and Tuesday (August 19) was National Potato Day. If you've ever seen a potato trying to sprout in a kitchen cabinet, with no possible hope of reaching soil or water, you know that celebrating one is very much a celebration of the other.

Last week was an engineering onsite week, where we had the whole team out to SF to work together for five days. Highlights included all-you-can-eat sushi; stumbling upon the excellent Argonaut bookstore in Nob Hill, where Rohan got to hold a book of Gregorian chant published in 1661, and Elangeni purchased a terrifying clown painting; and team bonding over tea tree face masks.

We also spent last week planning out some major pieces of work, including a redesign of our checkout, that'll be dropping in the next 2-4 weeks. As a result, the changelog is a little shorter this week.

We made a few changes to our external APIs and reporting to give our users more access, more easily:

  • Julia added purchase metadata (passed in via externalMetadata to POST /checkout) to the payout report
  • Elangeni added a contact[email] field to the POST /checkout endpoint to allow prefilling the user's email
  • Julia added information about the payment method and processor to our purchase and refund webhooks and APIs
  • Jinting added a new webhook, payment.failed, that fires when a payment fails, so you can be alerted in real time

We cleaned up our direct-checkout flow to make checking out with Neon from in game even faster:

  • Shreyas removed the email field from the Apple and Google Pay forms, instead pulling it from the wallets directly

And, as always, a grab bag of other miscellania:

  • Shreyas updated our handling of a few Razorpay errors to show the appropriate error to the user where relevant
  • Elangeni continued her work on improving our processing of longer-running asynchronous payments from last week
  • Jinting updated our fraud model to take into account more user-level information for better analysis