docs: note that the README's Alamofire row predates the resolver pool change
The README no longer presents the 2026-10-14 Alamofire row (1.90 s, 668 MiB) as the fork's current result.
Why
That row predates the resolver pool change (#506), which starts the pool part-way through on mid-size projects. Afterwards the fork's Alamofire index peaks about 925 MiB, not 668, and the mid-size comparison found no clear wall-time difference from upstream:
| Source | Build | Alamofire full index |
|---|---|---|
| README row, 2026-10-14 run | fork ba972275 | 1.90 s, 668 MiB |
| 2026-10-15 mid-size comparison, three runs per engine | fork f405d86b vs upstream c71a3d03 | fork 1.44 s, 1,042 MiB; upstream 1.38 s (no clear change), 791 MiB (+31.7%) |
| Fresh check, three runs, Node 24, no upstream run, load 1.55-1.82 | deployed 559cc097 | 1.31-1.61 s, 919-930 MiB |
Change
- A note under the full-index table says the Alamofire row predates the change and gives the two later results.
- The "Slower to index some projects" bullet names the 2026-10-14 run for the Alamofire +12% and says the later comparison found no clear wall-time difference.
- The 2026-10-14 chart, its 74-comparison tally and the benchmark doc are left as they are: they describe that run.
README rows checked
Alamofire row and bullet: changed as above. gin, pretix, CPython, discourse, supabase and n8n rows: not re-measured since #506/#518 (the 2A check found no clear change for pretix and supabase), so left unedited and unmarked.
Checks
On head: readme-sync 6 passed; check:agent-docs exit 0.
Summary by cubic
Updates the README's Alamofire benchmark so the 2026-10-14 figure is no longer presented as the fork's current result. That run predates the resolver pool change (#506); the README now adds a note with the two later results and clarifies the later wall-time comparison found no clear difference.