TechPrufer
Get started
Atlassian
Atlassian logoAtlassian · Feature trail

Sync State Indicators

3 posts · Mar 2017

What this trail covers

Atlassian products display sync state indicators to communicate data synchronization status to users. This includes visual cues for queued, syncing, and synced states across boards, cards, comments, and attachments. The system aims to provide clear feedback without overwhelming the user, ensuring they are aware of data consistency. This post details the implementation of background downloading for Trello's mobile applications, focusing on optimizing bandwidth and battery usage through scheduled syncs, and introduces robust failure handling mechanisms for network requests, including temporary/permanent error categorization, exponential backoff, idempotence via client-generated UUIDs, and a 'last-writer wins' conflict resolution strategy for most data types.

All Atlassian trails
  1. Displaying sync stateThis post details the design and implementation of sync state indicators for Atlassian products. It addresses the challenge of balancing user feedback frequency with UI clutter, settling on displaying indicators in four key areas: boards, cards, comments, and attachments. The implementation involves a database table to track model states and a pragmatic approach to handling parent-child relationships by storing parent IDs for boards and cards to efficiently query sync status.
  2. Sync is a two-way streetThis post details the implementation of background downloading for Trello's mobile applications. It introduces strategies for optimizing bandwidth and battery usage by scheduling syncs at most twice a day, skipping unchanged data by querying last activity timestamps, and intelligently prioritizing data downloads (e.g., starred boards, recently viewed boards). A priority-based download queue is described, differentiating between urgent (app open) and less urgent (notifications) syncs, with the ability to remove items from the queue if no longer needed. The post acknowledges that this system is not fully realized and aims for a single, standardized download system across the architecture.
  3. Sync failure handlingThis post details Trello's sync failure handling mechanisms for mobile applications. It covers categorizing network errors into temporary and permanent, implementing exponential backoff for temporary errors with retry caps, and ensuring idempotence through client-generated UUIDs to prevent duplicate actions. It also discusses a 'last-writer wins' conflict resolution strategy for most data and a method for reverting unsynced data by allowing future server GET requests to overwrite local changes.