Bill W:
Chad G:
By step twelve, a video file was no longer enough. The app needed to know what that file represented.
We created a canonical metadata model around each recording: recording ID, session ID, start and end times, duration, platform, local location, cloud identifier, upload state, testing-project association, bookmarks, file size and basic video properties. File names would remain human-readable, but the database – not the filename – would be the source of truth.
That distinction matters. A cloud upload might later allow the local file to be deleted, but the recording’s history should survive. The original timestamp still had to drive Today and monthly filters. Bookmarks and project associations still had to point to the same recording.
This was one of those unglamorous development steps that quietly makes everything else possible. Once recordings had durable identities, the rest of the app could stop treating them as anonymous MP4 files.
@@@
