It depends: A dialog about dependencies
summary
Argues that not all dependencies are equal and gives concrete metrics and tools to decide which to keep: real incidents like left-pad removal and event-stream malware show risk, installation benchmarks show glue installs in 5–10s versus readr 200–700s, and binary sizes range from <10 KB to >4 GB (Bioconductor). Demonstrates the itdepends::dep_weight / dep_uses / dep_locate workflow to count function use, measure install time and binary weight, visualize heavy deps, and locate calls to remove them.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 2:27Examples of dependencies
- 5:53What should we do
- 7:20Consider your users
- 8:55Illusionary superiority
- 11:16It depends
- 12:29DEP uses
- 14:21DEFWeight
- 17:12Visualising dependencies
- 18:25Removing dependencies
- 19:53Conclusion
related talks
22mYou Don't Know Your Dependencies
01h 01mWhy managing C++ dependencies is hard
38mDependency Management: The Cause of—and Solution to—All Supply...
20mdepstat: Analyzing a Go Project’s Dependencies
25mFinding undeclared & unused dependencies in your notebooks
27mDependency hell - Or, Developers' perception of software dependencies
55mCI Pulled It, Prod Ran It: The Real Risk in Dependency, by GraphsIlkka Turunen
40mCode metrics for your app