Achieving Economies of Scale in Cloud IDEs
summary
Cloud IDEs today scale roughly linearly because language tools expect local files and each workspace pulls its own dependencies; empirical example: a Theia project shows 850MB node_modules vs 85MB packages, a tenfold wasteable footprint. The proposed fix is to pre-compute indices (navigate, search, proposals) and publish LSIF/monikers into a shared program-database service so navigation and search answer network queries instead of reindexing per workspace. Idea-stage only: some components exist but no complete, deployed prototype yet.
AI-written summary · about
chapters (13)
- 0:00Intro
- 1:34Economies of Scale
- 2:36The Problem
- 5:39The Question
- 5:57What does an IDE do
- 7:16Navigate
- 9:29Search
- 10:34Proposal
- 11:51Compile and Link
- 13:02Unit of Precomputation
- 14:19What we need
- 16:30Caveats
- 19:44Summary
related talks
25mWeb-based Tools and IDEs: Accomplishments and Opportunities
26mScaling Enterprise Development with Cloud IDEs
45mContaining Your Development – Why Cloud IDEs Are the Future
46mBuilding the IDE Golden Path
27mHow browser IDEs are enabling the transition of development to the cloud
10mHow my team built a Theia based Cloud IDE
37mTextual and Graphical Languages for the Cloud Era
36mIf, when and how? - Strategies towards web-based tooling