Design Suggestions for XML++
summary
Argues for a pragmatic XML++ focused on tooling and semantics: calls out the XML bomb: 15 lines → 3 GB vulnerability, the proliferation of competing grammars, and slow DOM parsing for large spreadsheets. Demonstrates loading Relax NG into a graph engine (Relax NG → Apache TinkerPop) and exporting GraphML for visualization, proposes runtime models that behave like JSON maps/sequences instead of raw DOM, and reports a working Java toolchain plus plans for a Rust binary model. Emphasizes missing grammar conversion and annotation tooling.
AI-written summary · about
chapters (13)
- 0:00Introduction and background
- 1:28XML design flaws and solutions
- 3:33Simplifying XML grammars
- 6:43Reviving XML tools and projects
- 7:26Early XML work and lessons learned
- 9:58Runtime models and DOM limitations
- 11:00Moving beyond DOM to binary models
- 12:47Semantics and graph-based parsing
- 15:34Improving XML tooling and annotations
- 17:21Use cases and generated libraries
- 17:42Electronic invoices and semantics
- 19:28Key takeaways and future directions
- 20:36Q&A session
