Repository navigation
ForceDirectedLayout/README.md publishes the pre-tuning defaults and then asserts they are the measured ones #391
Description
Activity
Triage
Priority: Medium Effort: Small
What needs to be done:
ForceDirectedLayout/README.md's tuning table still shows the six pre-tuning default values that commitd99d704moved (and that CLAUDE.md's "Tuned defaults" table correctly documents), with prose directly underneath asserting the numbers were measured. The staleMinRepulsionDistance = 50.0is actively harmful to copy — CLAUDE.md records that value as what "used to let bodies crowd" — and its accompanying comment now contradicts the actualMinimumRepulsionClampbehaviour.Suggested next steps / acceptance criteria:
- Sync README:106-124 to
LayoutSettings.Defaults, and add a note thatMinRepulsionDistanceis floored internally (per theLayoutSettings.csinvariant this repo documents elsewhere). - Also correct
ImGui.NodeEditor/README.md:126-130's stale illustrative values in the same pass, per the issue's secondary note. - Consider whether the defaults table can be generated or test-asserted against
LayoutSettings.Defaultsso it can't drift out of sync a second time (worth a follow-up issue if out of scope here).
Blockers / dependencies: None identified. Docs-only change, no code risk.
Generated by Claude Code
- Sync README:106-124 to
- linked a pull request that will close this issueSync ForceDirectedLayout and NodeEditor README tuning values with current defaults #399
on Sep 13, 2026
What happens
Commit
d99d704("Tune the layout defaults against the benchmark corpus") moved six settings and updatedLayoutSettings.cs,PhysicsSettings.csandCLAUDE.md— but notForceDirectedLayout/README.md, whose last commit is the earlier1da8962.The README's "Tuning" block still carries every pre-tuning value:
ForceDirectedLayout/LayoutSettings.cs:131-150)RepulsionStrength = 600_000.0(README:107)900_000.0LinkSpringStrength = 0.5(README:108)0.1RestLinkLength = 225.0(README:109)50.0DirectionalBias = 0.5(README:110)4.0LinkFlatteningStrength = 0.5(README:111)3.0MinRepulsionDistance = 50.0(README:117)5.0Those six are exactly the six rows of CLAUDE.md's "Tuned defaults" table — the README is publishing its was column as current.
The paragraph immediately below, at README:127, then says:
Why it matters
This is the NuGet package README for
ktsu.ForceDirectedLayout, so it is the primary reference for anyone outside the repo.MinRepulsionDistance = 50.0is actively harmful to copy: CLAUDE.md records that a value that high is what "used to let bodies crowd". The README comment beside it — "so touching bodies push hard, not infinitely hard" — also now contradictsLayoutCore.cs:287'sMinimumRepulsionClamp, which is what actually keeps the law finite sinceaebc684.So a reader following the README gets the pre-tuning behaviour the corpus was built to fix, with a paragraph underneath assuring them the numbers were measured.
Suggested fix
Sync README:106-124 to
LayoutSettings.Defaults, and add a note thatMinRepulsionDistanceis floored internally.Secondary:
ImGui.NodeEditor/README.md:126-130showsRepulsionStrength = 1_200_000.0/LinkSpringStrength = 0.5/RestLinkLength = 225.0. That block reads as illustrative rather than as a defaults listing, but it is stale in the same direction and worth correcting in the same pass.Worth considering whether the defaults table can be generated or test-asserted, since this is the second document that has to move whenever a default does.