Paper detail

BarQL: Collaborating Through Change

Applications such as Google Docs, Office 365, and Dropbox show a growing trend towards incorporating multi-user live collaboration functionality into web applications. These collaborative applications share a need to efficiently express shared state, and a common strategy for doing so is a shared log abstraction. Extensive research efforts on log abstractions by the database, programming languages, and distributed systems communities have identified a variety of optimization techniques based on the algebraic properties of updates (i.e., pairwise commutativity, subsumption, and idempotence). Although these techniques have been applied to specific applications and use-cases, to the best of our knowledge, no attempt has been made to create a general framework for such optimizations in the context of a non-trivial update language. In this paper, we introduce mutation languages, a low-level framework for reasoning about the algebraic properties of state updates, or mutations. We define BarQL, a general purpose state-update language, and show how mutation languages allow us to reason about the algebraic properties of updates expressed in BarQ L .