oss-estimator

Set of tools to get info about GitHub project that is needed to estimate the cost of running it.

Tools

Data used

• gharchive.org, GitHub events archive from 2014 until now
• Direct clone of GitHub repos (~140GB of data)
• cncf/gitdm, CNCF affiliation of developers

Example

$ go run ./cmd/estimator-sloc open-telemetry/opentelemetry-ebpf
+-------------------+
| open-telemetry/op |
| entelemetry-ebpf  |
+---------+---------+
| SLOC    | 4295    |
| Commits | 267     |
| PR      | 110     |
| HEAD    | df4e6e6 |
+---------+---------+

$ du -hs _work/
137G	_work

Stats

Languages: BASH, Bazel, C, C Header, C#, C++, Cython, Dart, Erlang, Go, Go Template, Haskell, Java, JavaScript, Kotlin, Lua, Objective C, Objective C++, PHP, Perl, Protocol Buffers, Python, Ruby, Rust, Scala, Swift, TypeScript, Zig.

Effort

Effort is rough estimate how much developer-years it would take to write some project from scratch in proprietary, enterprise environment (i.e. in a big company).

Current model is very simple: we estimate that average developer writes 12k SLOC per year.

TODO:

• Take into account COCOMO (not suitable for FOSS?)
• Estimate on commits and PRs (can vary a lot)
  • Average developer makes 3 commits per year
  • Exclude robot commits and PRs, like dependabot
• Research papers on FOSS effort estimation if proposed model is applicable

Aggregates

Some organizations are highlighted in bold, they represent additional aggregated statistics.

CNCF

Organizations that are CNCF projects (graduated, incubating, sandbox).

K8s

Repositories in kubernetes and kuberentes-sigs organizations.

Organizations

Repositories

CNCF Projects