EvidenceChapter One
The next generation of the SDLC
Rahul Parundekar
1 entry
Figure evidence
active-repo-dates
FindingChange sets per repository per week, with the number of active repositories above: output grew 7.2 times from the weeks before July to weeks 27 to 39 (early July to late September) because the active repositories went from 1.7 to 10.6 a week, while change sets per active repository barely moved (13.5 to 15.5). Seven repositories were active before July; the seventh became active in the week of 4th of May.
QuestionHow did meaningful repository activity change over the study, and how much of the growth in output came from more repositories being active rather than more work per active repository?
| Measurement | Definition |
|---|---|
| unit | non-Dependabot change sets landing on main, per ISO week |
| active threshold | 1 |
| meaningful threshold | 3 |
| window | 2026-W01 to 2026-W40, through 1st of October; W40 is partial |
| growth split windows | ["W01-W26", "W27-W39"] |
| date basis | The created column records the earliest retained default-branch commit, not independently verified repository creation. infrastructure-environments has older PR history. |
Sourcedetectors.cs_sets; detectors.cs_units; git.commits; record.changeset_facts.