A ticket platform for Linköping University, formerly named Tickle. It has been used in the wild in different shapes at the following events:
- SOF 2015
- LiTHe Blås Luciafest 2015
- LiTHe Blås Luciafest 2016
Bitket is a fairly advanced application, with features including
- flexible discount management, e.g. for union discounts
- login management using Google, Facebook and LiU SSO (ADFS/OAuth2)
- purchasing queue that gracefully handles massive, concurrent load without overshooting the maximum allowed ticket count
- support for variations of ticket types (even with different pricing), e.g. for letting visitors choose beverage at a sittning
- secure payments with credit/debit card through Stripe
- pre-purchase access using unique links, allowing purchase of a defined number of tickets before public release
- entry handling, using visitor's LiU cards or QR codes
Bitket is roughly divided into a backend and a frontend. Both are completely separate applications, giving the potential advantages of independent scaling or sharding.
The backend is a Django application with PostgreSQL as a data store, exposing a REST API. It features integrations to services including (but not limited to)
- Facebook (OAuth2)
- Google (OAuth2)
- LiU ADFS (OAuth2)
- Sesam (LiU integration platform)
- Stripe
- Kobra
Directly visible to the user, the frontend fetches data, partly from the Bitket API, partly from services such as
- Facebook (OAuth2)
- Google (OAuth2)
- LiU ADFS (OAuth2)
- Stripe
This (and user-created) data is then sent to the backend and/or presented in a
meaningful way, using mainly React and Redux. The frontend is rebuilt on every
code change (or deployment) through react-scripts (i.e. Babel+Webpack).
Deployment is handled with Docker, including Compose. Docker builds are performed by Gitlab CI.
[1] Implementing fully managed reselling of tickets, meaning that Bitket takes all the risk of reselling a ticket (also reducing the vector for inordinate prices). This is done by charging the new ticket owner for the ticket while simultaneously refunding the previous owner's part and transferring the actual ticket ownership. This is already partly implemented.
[2] Implementing an asynchronous queue+worker architecture for handling ticket purchases, replacing the existing synchronous flow. This will be realized using Celery as the asynchronous queue/broker and Django Channels as the messaging solution (notifying the client that the purchasing process has been carried out, using Websockets/HTTP2).