Cloud Programming Languages and Infrastructure From Code: An Empirical Study
summary
Controlled experiment with 40 participants compares Wing (cloud programming language) and Nitric (SDK) on a URL-shortener: participants implement handlers, trace generated Terraform, and answer comprehension questions. Results show similar effectiveness, Wing is moderately slower but correctness rises with extra time, and tracing generated infrastructure back to code is hard for both approaches. Thematic analysis (21 themes, 178 codes) highlights poor onboarding and docs, security and cost worries, and missing tooling for code→infrastructure traceability and local cloud simulation.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 2:10Infrastructure as code evolution
- 5:40Nitric and Wing evaluation
- 8:14Research questions and design
- 10:27Participant demographics
- 14:00Task design and example
- 18:01Concept comprehension findings
- 21:43Participant feedback analysis
- 25:38Barriers to adoption
- 26:42Conclusion and future work
- 27:26Q&A
related talks
01h 00mPanel: Programming Languages in the Era of the Cloud
55mThinking Beyond IaC:an OSS Approach to Cloud Agnostic Infr...
39m.NET: Infrastructure as Software
32mSoftware development reimagined
37mTextual and Graphical Languages for the Cloud Era
24mBeyond Code Completion: The Shift of AI Programming from Pair to Peer
48mSponsored Session: Your Next Workstation Is In The Cloud
25mWeb-based Tools and IDEs: Accomplishments and Opportunities