forgejo_admin/tmux-custom
1
0
Fork 0
Code Issues 1 Pull requests 1 Projects Releases Packages Wiki Activity Actions

Revert .tmux.conf to minimal baseline (closes #1) #2

Open
forgejo_admin wants to merge 1 commit from 1-revert-to-minimal-config into main
pull from: 1-revert-to-minimal-config
merge into: forgejo_admin:main
Conversation 1 Commits 1 Files changed 1 +3 -89
forgejo_admin commented 2026-04-10 16:24:01 +00:00
Owner
Copy link

Closes #1

Summary

Restores .tmux.conf to the exact 22-line state from abeb159 (initial commit). Forward revert — no history rewrite.

Changes

All additions stacked across 5fd40bb, df57813, 11a3cda, 8a8fcf5 are removed from .tmux.conf:

  • TPM + plugins (tmux-sensible, resurrect, continuum, pain-control, prefix-highlight, catppuccin)
  • OS-detection if-shell wrappers (status-right battery, clipboard)
  • vi-mode clipboard binds (pbcopy / xclip / clip.exe)
  • split-window + new-window cwd inheritance
  • after-split-window and pane-exited auto-rebalance hooks
  • renumber-windows on
  • < / > swap-window bubble-sort binds + jump-to-back/front scripts
  • catppuccin window-status format overrides
  • automatic-rename off

Final file diff: .tmux.conf -89 / +3 lines. Result is byte-identical to abeb159:.tmux.conf.

Why: the layered additions produced bugs that outweighed their value.

  1. Ctrl-b c landed new windows at the end of the stack instead of next to the current window.
  2. pane-exited rebalance hook fought manual pane layouts — closing one pane shuffled the rest.
  3. Plugin stack pulled in opaque state (resurrect + continuum auto-restore) that made the config hard to reason about.
  4. < / > swap binds needed a follow-up kludge (jump-to-back/front shell scripts) because the first version broke at the boundaries.

Going back to the 22-line baseline so additions can be re-introduced deliberately, one per PR, only if they earn their keep.

Test Plan

  • After merge, pull on host: cd ~/tmux-custom && git pull forgejo main
  • Reload: tmux source-file ~/.tmux.conf
  • Verify no TPM/catppuccin errors in tmux message bar
  • Verify Ctrl-b c still works (stock tmux behavior — will still append to end until a deliberate follow-up ticket adds -a)
  • Optional cleanup: rm -rf ~/.tmux/plugins to remove TPM leftovers

Review Checklist

  • .tmux.conf is byte-identical to abeb159:.tmux.conf (verify with git diff abeb159 HEAD -- .tmux.conf → no output)
  • History is preserved — no reset --hard or force-push
  • Commit message records the why (bug list)
  • No files outside .tmux.conf touched
  • No re-addition of removed features smuggled in

Related Notes

  • Issue: #1
  • Reverts effects of: 5fd40bb (TPM + plugins), df57813 (OS detection), 11a3cda (</> fix), 8a8fcf5 (boundary insert)
  • Baseline restored: abeb159
  • Stale branch 1-fix-window-move-insert is untouched and can be cleaned up separately if desired
  • Out of scope: re-adding clipboard, plugins, or window helpers — future tickets per addition

🤖 Generated with Claude Code

Closes #1 ## Summary Restores `.tmux.conf` to the exact 22-line state from `abeb159` (initial commit). Forward revert — no history rewrite. ## Changes All additions stacked across `5fd40bb`, `df57813`, `11a3cda`, `8a8fcf5` are removed from `.tmux.conf`: - TPM + plugins (tmux-sensible, resurrect, continuum, pain-control, prefix-highlight, catppuccin) - OS-detection `if-shell` wrappers (status-right battery, clipboard) - vi-mode clipboard binds (pbcopy / xclip / clip.exe) - split-window + new-window cwd inheritance - `after-split-window` and `pane-exited` auto-rebalance hooks - `renumber-windows on` - `<` / `>` swap-window bubble-sort binds + jump-to-back/front scripts - catppuccin window-status format overrides - `automatic-rename off` Final file diff: `.tmux.conf` -89 / +3 lines. Result is byte-identical to `abeb159:.tmux.conf`. **Why:** the layered additions produced bugs that outweighed their value. 1. `Ctrl-b c` landed new windows at the end of the stack instead of next to the current window. 2. `pane-exited` rebalance hook fought manual pane layouts — closing one pane shuffled the rest. 3. Plugin stack pulled in opaque state (resurrect + continuum auto-restore) that made the config hard to reason about. 4. `<` / `>` swap binds needed a follow-up kludge (jump-to-back/front shell scripts) because the first version broke at the boundaries. Going back to the 22-line baseline so additions can be re-introduced deliberately, one per PR, only if they earn their keep. ## Test Plan - [ ] After merge, pull on host: `cd ~/tmux-custom && git pull forgejo main` - [ ] Reload: `tmux source-file ~/.tmux.conf` - [ ] Verify no TPM/catppuccin errors in tmux message bar - [ ] Verify `Ctrl-b c` still works (stock tmux behavior — will still append to end until a deliberate follow-up ticket adds `-a`) - [ ] Optional cleanup: `rm -rf ~/.tmux/plugins` to remove TPM leftovers ## Review Checklist - [ ] `.tmux.conf` is byte-identical to `abeb159:.tmux.conf` (verify with `git diff abeb159 HEAD -- .tmux.conf` → no output) - [ ] History is preserved — no `reset --hard` or force-push - [ ] Commit message records the *why* (bug list) - [ ] No files outside `.tmux.conf` touched - [ ] No re-addition of removed features smuggled in ## Related Notes - Issue: #1 - Reverts effects of: `5fd40bb` (TPM + plugins), `df57813` (OS detection), `11a3cda` (`<`/`>` fix), `8a8fcf5` (boundary insert) - Baseline restored: `abeb159` - Stale branch `1-fix-window-move-insert` is untouched and can be cleaned up separately if desired - Out of scope: re-adding clipboard, plugins, or window helpers — future tickets per addition 🤖 Generated with [Claude Code](https://claude.com/claude-code) 
Closes #1

Restores .tmux.conf to the exact 22-line state from the initial commit.
Forward revert — no history rewrite.

Removed (stacked across 5fd40bb, df57813, 11a3cda, 8a8fcf5):
- TPM + plugins: tmux-sensible, tmux-resurrect, tmux-continuum,
  tmux-pain-control, tmux-prefix-highlight, catppuccin/tmux
- OS-detection if-shell wrappers for status-right and clipboard
- vi-mode clipboard binds (pbcopy / xclip / clip.exe variants)
- split-window + new-window cwd inheritance binds
- after-split-window and pane-exited auto-rebalance hooks
- renumber-windows on
- < / > swap-window bubble-sort binds and the jump-to-back/front
  boundary-insert scripts layered on top of them
- catppuccin window-status format overrides
- automatic-rename off

Why: the layered additions produced bugs that outweighed the value.
Ctrl-b c landed new windows at the end of the stack instead of next
to the current window, the pane-exited rebalance hook fought manual
layouts, and the plugin stack pulled in opaque state (resurrect +
continuum) that made the config hard to reason about. Each iteration
piled more kludges on: the < / > swap binds needed a follow-up
"insert at boundaries" patch because the first version broke at the
edges. Going back to the 22-line baseline so additions can be
re-introduced deliberately, one at a time, only if they earn their
keep.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
forgejo_admin commented 2026-04-10 16:25:25 +00:00
Author
Owner
Copy link

PR #2 Review

DOMAIN REVIEW

Tech stack: tmux dotfile (shell-adjacent config). Single-file forward revert.

Diff verification against PR claims:

  • Byte-identity to abeb159: Diff math checks out. +3/-89, .tmux.conf only. The three additions are (a) the simplified set -g status-right "#(pmset -g batt | grep -Eo '[0-9]+%%')" line replacing the OS-detection if-shell pair, (b) the bare setw -g mode-keys vi at EOF (the old file had the same line but it was swept out as part of the larger copy-mode block removal, then re-added clean). Context lines at top (status-interval 5, status-left "#S ", MouseDown1StatusLeft choose-session) are preserved unchanged. No smuggled re-additions.
  • Single file: changed_files: 1, only .tmux.conf. Confirmed.
  • Removed stack matches PR summary: TPM + 6 plugins, OS-detection if-shell wrappers, vi-mode clipboard binds (v/V/y/Y/Enter/MouseDragEnd1Pane), cwd inheritance for "/%/c, after-split-window + pane-exited rebalance hooks, renumber-windows on, </> swap-window binds, @continuum-restore, @resurrect-capture-pane-contents, automatic-rename off, catppuccin window-status format overrides, TPM init line. All accounted for.
  • No force-push evidence: mergeable: true, forward revert commit on branch 1-revert-to-minimal-config. History preserved.
  • Commit message / PR body records why: Bug list is in the PR body (4 numbered bugs: Ctrl-b c append, pane-exited layout fight, opaque plugin state, swap bind boundary kludge). Good.

BLOCKERS

None.

NITS

  • The restored baseline hardcodes pmset -g batt in status-right, which is macOS-only. On Linux hosts this silently produces an empty status-right. This is inherent to abeb159 (the revert target), not introduced by this PR, so it is correctly out of scope. Flagging as discovered scope: consider a follow-up ticket to either (a) drop status-right entirely from the baseline, or (b) re-introduce a minimal OS-aware battery line as its own deliberate PR per the "earn their keep" philosophy in the PR body. Not required for merge.
  • Both old and new files end with \ No newline at end of file. Pre-existing condition, matches abeb159, no action needed.
  • Test Plan item "Verify Ctrl-b c still works (stock tmux behavior — will still append to end until a deliberate follow-up ticket adds -a)" correctly acknowledges the original bug #1 symptom is only partially addressed by the revert (it removes the plugin-layer complexity but not the stock tmux append behavior). Consider filing that follow-up ticket now so it doesn't fall off.

SOP COMPLIANCE

  • Branch named after issue (1-revert-to-minimal-config)
  • PR body follows template (Summary, Changes, Test Plan, Review Checklist, Related Notes)
  • Related section references issue #1 and the reverted commits
  • No secrets committed
  • No files outside .tmux.conf touched
  • Forward revert only, no history rewrite

PROCESS OBSERVATIONS

Clean, surgical revert. Exemplary PR hygiene for a dotfile change: explicit byte-identity claim, verifiable diff math, bug-driven rationale, and a stated philosophy (one deliberate addition per follow-up PR) that prevents the same layering problem from recurring. Change failure risk: near zero — this file is developer-local, no CI/deploy surface. Deployment frequency impact: neutral. Documentation gap: minor — the pmset-on-Linux quirk and the not-yet-fixed Ctrl-b c append behavior should each become their own backlog ticket so the scope stays tracked per feedback_discovered_scope_always_tracked.

VERDICT: APPROVED

## PR #2 Review ### DOMAIN REVIEW Tech stack: tmux dotfile (shell-adjacent config). Single-file forward revert. Diff verification against PR claims: - **Byte-identity to `abeb159`**: Diff math checks out. +3/-89, `.tmux.conf` only. The three additions are (a) the simplified `set -g status-right "#(pmset -g batt | grep -Eo '[0-9]+%%')"` line replacing the OS-detection `if-shell` pair, (b) the bare `setw -g mode-keys vi` at EOF (the old file had the same line but it was swept out as part of the larger copy-mode block removal, then re-added clean). Context lines at top (`status-interval 5`, `status-left "#S "`, `MouseDown1StatusLeft choose-session`) are preserved unchanged. No smuggled re-additions. - **Single file**: `changed_files: 1`, only `.tmux.conf`. Confirmed. - **Removed stack matches PR summary**: TPM + 6 plugins, OS-detection if-shell wrappers, vi-mode clipboard binds (v/V/y/Y/Enter/MouseDragEnd1Pane), cwd inheritance for `"`/`%`/`c`, `after-split-window` + `pane-exited` rebalance hooks, `renumber-windows on`, `<`/`>` swap-window binds, `@continuum-restore`, `@resurrect-capture-pane-contents`, `automatic-rename off`, catppuccin window-status format overrides, TPM init line. All accounted for. - **No force-push evidence**: `mergeable: true`, forward revert commit on branch `1-revert-to-minimal-config`. History preserved. - **Commit message / PR body records why**: Bug list is in the PR body (4 numbered bugs: `Ctrl-b c` append, pane-exited layout fight, opaque plugin state, swap bind boundary kludge). Good. ### BLOCKERS None. ### NITS - The restored baseline hardcodes `pmset -g batt` in `status-right`, which is macOS-only. On Linux hosts this silently produces an empty status-right. This is inherent to `abeb159` (the revert target), not introduced by this PR, so it is correctly out of scope. Flagging as **discovered scope**: consider a follow-up ticket to either (a) drop `status-right` entirely from the baseline, or (b) re-introduce a minimal OS-aware battery line as its own deliberate PR per the "earn their keep" philosophy in the PR body. Not required for merge. - Both old and new files end with `\ No newline at end of file`. Pre-existing condition, matches `abeb159`, no action needed. - Test Plan item "Verify `Ctrl-b c` still works (stock tmux behavior — will still append to end until a deliberate follow-up ticket adds `-a`)" correctly acknowledges the original bug #1 symptom is only partially addressed by the revert (it removes the plugin-layer complexity but not the stock tmux append behavior). Consider filing that follow-up ticket now so it doesn't fall off. ### SOP COMPLIANCE - [x] Branch named after issue (`1-revert-to-minimal-config`) - [x] PR body follows template (Summary, Changes, Test Plan, Review Checklist, Related Notes) - [x] Related section references issue #1 and the reverted commits - [x] No secrets committed - [x] No files outside `.tmux.conf` touched - [x] Forward revert only, no history rewrite ### PROCESS OBSERVATIONS Clean, surgical revert. Exemplary PR hygiene for a dotfile change: explicit byte-identity claim, verifiable diff math, bug-driven rationale, and a stated philosophy (one deliberate addition per follow-up PR) that prevents the same layering problem from recurring. Change failure risk: near zero — this file is developer-local, no CI/deploy surface. Deployment frequency impact: neutral. Documentation gap: minor — the `pmset`-on-Linux quirk and the not-yet-fixed `Ctrl-b c` append behavior should each become their own backlog ticket so the scope stays tracked per `feedback_discovered_scope_always_tracked`. ### VERDICT: APPROVED
This pull request can be merged automatically.
You are not authorized to merge this pull request.
View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u origin 1-revert-to-minimal-config:1-revert-to-minimal-config
git switch 1-revert-to-minimal-config

Merge

Merge the changes and update on Forgejo.

Warning: The "Autodetect manual merge" setting is not enabled for this repository, you will have to mark this pull request as manually merged afterwards.

git switch main
git merge --no-ff 1-revert-to-minimal-config
git switch 1-revert-to-minimal-config
git rebase main
git switch main
git merge --ff-only 1-revert-to-minimal-config
git switch 1-revert-to-minimal-config
git rebase main
git switch main
git merge --no-ff 1-revert-to-minimal-config
git switch main
git merge --squash 1-revert-to-minimal-config
git switch main
git merge --ff-only 1-revert-to-minimal-config
git switch main
git merge 1-revert-to-minimal-config
git push origin main
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
forgejo_admin/tmux-custom!2
No description provided.