feat: reduce spiral node-leaf subtree churn during new-pane #113

New issue

Closed

opened 2026-04-28 01:58:00 +00:00 by barrettruth · 0 comments

barrettruth commented

2026-04-28 01:58:00 +00:00

Problem

In the harder spiral node-leaf phases, the new pane takes an outer leaf slot and the previous tail gets pushed inward. Those phases are not the same as the easy leaf-node phases and deserve their own treatment.

Blocked by: #101, #102, #109

Proposed solution

Investigate how much of the visible subtree churn in spiral node-leaf phases can be reduced, isolated, or otherwise made less jarring during explicit managed new-pane.

The goal is not to promise perfect elimination. The goal is to separate avoidable churn from the recursive topology changes that may remain necessary.

Context

The current matrix classifies these as local role-shift cases. The strongest surveyed reprojection precedent is tmux-tiling-revamped, which rebuilds the BSP geometry and then performs swap-pane order repair for spiral.

Acceptance notes

Make the first frame as branch-correct as possible.
If some inward subtree motion remains unavoidable, make it explicit enough to test and document.
Do not regress the easier leaf-node phases.

Alternatives considered

Keep full reprojection for all spiral phases.
Treat spiral as all-or-nothing and refuse to separate the easy and hard phases.

## Problem In the harder `spiral` node-leaf phases, the new pane takes an outer leaf slot and the previous tail gets pushed inward. Those phases are not the same as the easy leaf-node phases and deserve their own treatment. Blocked by: #101, #102, #109 ## Proposed solution Investigate how much of the visible subtree churn in `spiral` node-leaf phases can be reduced, isolated, or otherwise made less jarring during explicit managed `new-pane`. The goal is not to promise perfect elimination. The goal is to separate avoidable churn from the recursive topology changes that may remain necessary. ## Context The current matrix classifies these as local role-shift cases. The strongest surveyed reprojection precedent is `tmux-tiling-revamped`, which rebuilds the BSP geometry and then performs `swap-pane` order repair for `spiral`. ## Acceptance notes - Make the first frame as branch-correct as possible. - If some inward subtree motion remains unavoidable, make it explicit enough to test and document. - Do not regress the easier leaf-node phases. ## Alternatives considered - Keep full reprojection for all spiral phases. - Treat spiral as all-or-nothing and refuse to separate the easy and hard phases.