[EPIC] Schedule programme data model without ORM/DB

The recent programme data is continuously fetched from the Steering API. To avoid any disruption in play-out (e.g. in cases where the API is not available due to network outages) the programme is also cached locally.

Engine uses a SQLAlchemy model rebuilding an internal representation of the programme and relevant schedules for play-out.

Having an ORM and relatively heavy PostgreSQL server just for this single purpose is some unnecessary overhead. Therefore lean ways of caching should be evaluated. This POC aims to:

1. Cache the raw JSON endpoint data
2. Rebuild the ORM using simple POPOs
3. Use hashing strategies to easily detect updates, where applicable

Sub Tasks

• Extend API data fetchers to transparently cache... (#124 - closed)
• Deserialize API response using Steering/Tank Op... (#129 - closed)
• Scheduling domain objects based on POPOs (#133 - closed)
• Refactor API Fetcher
• POC for TimetableService as a lightweight replacement for ProgrammeService
• Diffing between current and planned schedule using JSON file cache
• Refactor timetable renderer for work with new domain model
• Test and integrate timetable and scheduler with new domain model
• Refactor scheduling module to improve testability
• Check if uses of ProgrammeService.engine.engine_time() can be relocated, to reduce dependencies
• Remove all references of SQLAlchemy and PostgreSQL
• Add some more test cases
• Documentation: Update README.md
• Update docs.aura.radio + Docker Compose settings

Merge Requests

• ORM-less scheduling (!35 - merged)
• ORM-less engine (aura!40 - merged)

changed milestone to %1.1

added P2 label

assigned to @david

marked this issue as related to #98 (closed)

mentioned in issue #98 (closed)

changed milestone to %1.0-alpha2 — Precise Pheasant

mentioned in commit 17e99d87

added To Do label

added 1 deleted label

added Doing label and removed To Do label

mentioned in commit 776d38a9

mentioned in commit b3a6b193

marked this issue as related to #89 (closed)

Since this requires lots of refactoring and testing the finalization will be done as soon there are less high-priority tasks or blocking defects in the ORM and sync layer. Possibly think about splitting it into sub-tasks

changed title from POC: Schedule programme data model without ORM/DB to [EPIC] Schedule programme data model without ORM/DB

added EPIC label

removed Doing label

mentioned in issue #124 (closed)

marked this issue as related to #124 (closed)

changed milestone to %1.0-alpha3 — Playful Platypus