diff options
| author | Craig Jennings <c@cjennings.net> | 2026-06-14 14:19:45 -0500 |
|---|---|---|
| committer | Craig Jennings <c@cjennings.net> | 2026-06-14 14:19:45 -0500 |
| commit | af2ad1febf87027cbd4408312c1fff2815a9a0ff (patch) | |
| tree | 4a40812ea73abf8954f46ae7781a6d8106d57874 | |
| parent | 269f23a38789190d112b04e8e70c3a6d649193b1 (diff) | |
| download | dotemacs-af2ad1febf87027cbd4408312c1fff2815a9a0ff.tar.gz dotemacs-af2ad1febf87027cbd4408312c1fff2815a9a0ff.zip | |
fix(modeline): drop per-render truename, guard vc fetch against signals
The VC modeline cache rebuilt its key every render, and the key included file-truename, so a stat ran on every redisplay rather than once per refresh as the comment claimed. Now it keys on (file show-remote). A moved symlink target is caught at the next TTL refresh, when vc-backend resolves the link fresh. And cj/modeline-vc-fetch is wrapped in condition-case returning nil, so a git signal on a slow or unmounted filesystem degrades to no-VC-info instead of breaking all redisplay.
| -rw-r--r-- | modules/modeline-config.el | 43 | ||||
| -rw-r--r-- | tests/test-modeline-config-vc-cache-key.el | 60 | ||||
| -rw-r--r-- | tests/test-modeline-config-vc-cache.el | 7 | ||||
| -rw-r--r-- | todo.org | 19 |
4 files changed, 70 insertions, 59 deletions
diff --git a/modules/modeline-config.el b/modules/modeline-config.el index 0e6e5d0fb..21ecd7e47 100644 --- a/modules/modeline-config.el +++ b/modules/modeline-config.el @@ -137,12 +137,12 @@ Uses built-in cached values for performance.") cj/modeline-vc-cache-set-p nil)) (defun cj/modeline-vc-cache-key (file) - "Return the cache key for FILE. -Includes the resolved `file-truename' so that if FILE is a symlink whose -target moves to a different VC tree, the key changes and the cache is not -served a stale backend. The extra `file-truename' is one stat per refresh, -cheap next to the VC calls the cache avoids." - (list file (file-truename file) cj/modeline-vc-show-remote)) + "Return the cache key for FILE: the file path and `cj/modeline-vc-show-remote'. +`file-truename' is deliberately omitted -- the mode-line rebuilds this key on +every render to check cache validity, so a stat here would run per redisplay. +A symlink whose target moves to a different VC tree is picked up at the next +TTL refresh, when `vc-backend' resolves the link fresh." + (list file cj/modeline-vc-show-remote)) (defun cj/modeline-vc-cache-valid-p (key now) "Return non-nil when cached VC data is valid for KEY at NOW." @@ -157,18 +157,25 @@ Return a plist with `:branch' and `:state', or nil when FILE has no VC data. Uses `vc-git--symbolic-ref' for branch names when available (it returns the symbolic ref like \"main\" instead of a SHA when HEAD is on a branch), but falls back to `vc-working-revision' if the internal accessor is missing -- -the symbol is internal and can be renamed or removed between Emacs versions." - (unless (and (file-remote-p file) (not cj/modeline-vc-show-remote)) - (when-let* ((backend (vc-backend file)) - (branch (vc-working-revision file backend))) - (when (eq backend 'Git) - (unless (fboundp 'vc-git--symbolic-ref) - (require 'vc-git nil 'noerror)) - (when (fboundp 'vc-git--symbolic-ref) - (when-let* ((symbolic (ignore-errors (vc-git--symbolic-ref file)))) - (setq branch symbolic)))) - (list :branch branch - :state (vc-state file backend))))) +the symbol is internal and can be renamed or removed between Emacs versions. + +The whole VC probe is wrapped in `condition-case' returning nil. These are +synchronous git calls that, on TTL expiry, run while the mode-line is built; +on a slow or unmounted filesystem a signal here would land in redisplay and +break it. Caching nil degrades to \"no VC info\" instead." + (condition-case nil + (unless (and (file-remote-p file) (not cj/modeline-vc-show-remote)) + (when-let* ((backend (vc-backend file)) + (branch (vc-working-revision file backend))) + (when (eq backend 'Git) + (unless (fboundp 'vc-git--symbolic-ref) + (require 'vc-git nil 'noerror)) + (when (fboundp 'vc-git--symbolic-ref) + (when-let* ((symbolic (ignore-errors (vc-git--symbolic-ref file)))) + (setq branch symbolic)))) + (list :branch branch + :state (vc-state file backend)))) + (error nil))) (defun cj/modeline-vc-info () "Return cached modeline VC data for the current buffer." diff --git a/tests/test-modeline-config-vc-cache-key.el b/tests/test-modeline-config-vc-cache-key.el index ae869f4b8..6ba7985c2 100644 --- a/tests/test-modeline-config-vc-cache-key.el +++ b/tests/test-modeline-config-vc-cache-key.el @@ -1,56 +1,36 @@ ;;; test-modeline-config-vc-cache-key.el --- Tests for VC modeline cache key -*- lexical-binding: t; -*- ;;; Commentary: -;; The VC modeline cache keys on the file. A symlink whose target moves to a -;; different VC tree must invalidate the cache, so the key includes the -;; resolved `file-truename', not just the symlink path. +;; The VC modeline cache keys on the file path and the `cj/modeline-vc-show-remote' +;; flag only. `file-truename' is deliberately NOT in the key: it would run on +;; every redisplay (the mode-line rebuilds the key each render to check validity), +;; and a moved symlink target is picked up at the next TTL refresh anyway, since +;; `vc-backend' resolves the link fresh. The per-render stat isn't worth it. ;;; Code: (require 'ert) -(require 'cl-lib) (add-to-list 'load-path (expand-file-name "modules" user-emacs-directory)) (require 'modeline-config) -;;; Normal Cases - -(ert-deftest test-modeline-vc-cache-key-includes-truename () - "Normal: the cache key includes the resolved truename of the file." - (let ((f (make-temp-file "cj-mlkey-"))) - (unwind-protect - (should (member (file-truename f) (cj/modeline-vc-cache-key f))) - (delete-file f)))) - -;;; Boundary Cases - -(ert-deftest test-modeline-vc-cache-key-changes-when-symlink-target-moves () - "Boundary: re-pointing a symlink to a new target changes the cache key. -The symlink path is identical both times; only its truename differs, so a -key that ignored the truename would serve a stale VC backend." - (let* ((dir (make-temp-file "cj-mlkey-dir-" t)) - (target-a (expand-file-name "a" dir)) - (target-b (expand-file-name "b" dir)) - (link (expand-file-name "link" dir))) - (unwind-protect - (progn - (write-region "" nil target-a) - (write-region "" nil target-b) - (make-symbolic-link target-a link) - (let ((key-a (cj/modeline-vc-cache-key link))) - (delete-file link) - (make-symbolic-link target-b link) - (let ((key-b (cj/modeline-vc-cache-key link))) - (should-not (equal key-a key-b))))) - (delete-directory dir t)))) +(ert-deftest test-modeline-vc-cache-key-is-file-and-show-remote () + "Normal: the key is (FILE SHOW-REMOTE), with no per-render file-truename stat." + (let ((cj/modeline-vc-show-remote nil)) + (should (equal (cj/modeline-vc-cache-key "/x/y.el") '("/x/y.el" nil))))) + +(ert-deftest test-modeline-vc-cache-key-tracks-show-remote () + "Boundary: toggling show-remote yields a different key (separate cache entry)." + (should-not (equal (let ((cj/modeline-vc-show-remote nil)) + (cj/modeline-vc-cache-key "/x/y.el")) + (let ((cj/modeline-vc-show-remote t)) + (cj/modeline-vc-cache-key "/x/y.el"))))) (ert-deftest test-modeline-vc-cache-key-stable-for-same-file () - "Boundary: the key is stable across calls for an unchanged file." - (let ((f (make-temp-file "cj-mlkey-stable-"))) - (unwind-protect - (should (equal (cj/modeline-vc-cache-key f) - (cj/modeline-vc-cache-key f))) - (delete-file f)))) + "Boundary: the key is stable across calls for an unchanged file + show-remote." + (let ((cj/modeline-vc-show-remote nil)) + (should (equal (cj/modeline-vc-cache-key "/x/y.el") + (cj/modeline-vc-cache-key "/x/y.el"))))) (provide 'test-modeline-config-vc-cache-key) ;;; test-modeline-config-vc-cache-key.el ends here diff --git a/tests/test-modeline-config-vc-cache.el b/tests/test-modeline-config-vc-cache.el index b6aafbfbe..dab755442 100644 --- a/tests/test-modeline-config-vc-cache.el +++ b/tests/test-modeline-config-vc-cache.el @@ -98,5 +98,12 @@ (should (text-property-any 0 (length rendered) 'mouse-face 'mode-line-highlight rendered))))) +(ert-deftest test-modeline-config-vc-fetch-swallows-vc-errors () + "Error: a signal from the VC backend is swallowed (returns nil) rather than +propagating into the mode-line redisplay path, where it would break all redisplay." + (cl-letf (((symbol-function 'file-remote-p) (lambda (&rest _) nil)) + ((symbol-function 'vc-backend) (lambda (&rest _) (error "git boom")))) + (should (null (cj/modeline-vc-fetch "/tmp/project/file.el"))))) + (provide 'test-modeline-config-vc-cache) ;;; test-modeline-config-vc-cache.el ends here @@ -195,6 +195,16 @@ Spec draft: [[file:docs/theme-studio-palette-generator-spec.org][theme-studio-pa Build a constraint-first palette generator for Theme Studio: start from bg/fg, generate editable palette columns in OKLCH, preview candidate columns before applying them, and apply proposals by append/replace/regenerate modes while preserving stable column ids and existing assignments where possible. V1 is palette-only, not automatic face assignment; generator output becomes normal editable columns after apply. +*** TODO [#C] theme-studio palette generator source modes for base-only vs ground-aware palettes :feature: +:PROPERTIES: +:LAST_REVIEWED: 2026-06-14 +:END: +Tentative follow-up from walking through the generator algorithms. Consider splitting the current =source: palette= behavior into two explicit source modes, names TBD: +- =base color palette= — current behavior; use non-ground base color columns only and ignore bg, fg, ground spans, and color spans. +- =ground and base palette= — use bg/fg plus non-ground base color columns as generator anchors, useful when colored ground endpoints should shape fill-gap or harmony choices. + +This may be cancelled if the extra distinction makes the generator harder to understand. Before implementing, decide final names and whether ground-aware source should include only bg/fg or also ground span steps. + ** TODO [#B] theme-studio save button does not overwrite the current theme file :bug: :PROPERTIES: :LAST_REVIEWED: 2026-06-13 @@ -1031,8 +1041,10 @@ From the 2026-06 config audit, =modules/help-config.el=: - =:111= — auto-mode-alist maps .info to an interactive command that KILLS the buffer mid find-file — programmatic =find-file-noselect= of any .info destroys buffers and pops Info windows. Drop the entry; keep the explicit command. Zero test coverage on this module (the two broken paths are exactly the untested ones). Fixed 2026-06-13: (1) extracted the save/cancel/open decision into a pure =cj/--info-open-plan= and routed =cj/open-with-info-mode= through it — no more =cl-return-from=, declining cancels cleanly; (2) deleted the dead =:hook= and the empty =:preface=; (3) dropped the destructive =auto-mode-alist= .info entry (kept =cj/open-with-info-mode= as an M-x command and =cj/browse-info-files= on C-h i). New test-help-config.el covers the planner (open / save-then-open / cancel) — 3 green; module loads clean. Stale daemon state (the .info auto-mode entry + the bogus info-mode-hook entry) cleared by hand so the running session is correct without a restart. Interactive Info open + find-file-no-longer-destructive are a VERIFY. -** TODO [#B] modeline runs synchronous git on the redisplay path, unguarded :bug:solo: +** DONE [#B] modeline runs synchronous git on the redisplay path, unguarded :bug:solo: +CLOSED: [2026-06-14 Sun] =modules/modeline-config.el:173,154,145= — the mode-line :eval calls vc-backend/vc-state/vc-working-revision (synchronous git) on TTL expiry; a slow or unmounted filesystem stalls ALL redisplay. The cache key computes =file-truename= on every render (the "one stat per refresh" comment is wrong), and nothing is condition-case-wrapped, so a signal lands inside the mode-line eval. Defer the truename behind the TTL check; wrap the fetch in condition-case caching nil. From the 2026-06 config audit. +Fixed 2026-06-14 (Approach A): dropped =file-truename= from =cj/modeline-vc-cache-key= (key is now =(file show-remote)=, no per-render stat; a moved symlink is caught at the next TTL refresh when vc-backend resolves the link fresh). Wrapped =cj/modeline-vc-fetch= in =condition-case ... (error nil)= so a git signal on a slow/unmounted FS degrades to no-VC-info instead of breaking redisplay. Rewrote the two truename cache-key tests to assert the cheap key; added a fetch-swallows-vc-errors test. 9 modeline vc tests green; live daemon confirms the 2-element key and nil-on-error fetch. ** TODO [#B] Stale elpa gptel shadows the local fork — likely the gptel-magit root :bug:quick:solo: =elpa/gptel-0.9.8.5= is still installed alongside the =~/code/gptel= fork (=ai-config.el:383=); package activation puts the elpa dir + autoloads on load-path, so which copy wins depends on ordering, and a mixed load (fork .el + elpa .elc) produces "impossible" bugs. =gptel-magit= (elpa) declares gptel as a dependency, so IT may be pulling the stale copy — check this first when working the open "[#B] Investigate gptel-magit not working properly" task. Fix: =package-delete= the elpa gptel + remove from .localrepo so the fork is the only copy on disk. From the 2026-06 config audit. @@ -4429,6 +4441,11 @@ From the 2026-06-11 messenger-unification brainstorm. Google Voice has no offici ** TODO Manual testing and validation Exercised once the phases above land. +*** VERIFY modeline still shows the git branch and state +What we're verifying: the VC-cache simplification didn't change what the modeline shows on a normal repo. Fixed in modules/modeline-config.el (live in the daemon after reload). +- Open a file inside a git repo +- Glance at the mode-line VC segment +Expected: the branch name and state still render as before (e.g. "main" with the usual state face). The change only drops a per-render stat and guards against git errors; normal display is unchanged. *** VERIFY info-mode open is non-destructive and cancels cleanly What we're verifying: opening a .info file no longer auto-kills the buffer, and the explicit cj/open-with-info-mode prompt cancels cleanly on decline. Fixed in modules/help-config.el; stale daemon state already cleared, so this also survives a fresh restart. - find-file a .info file (e.g. one under elpa) — it should open as an ordinary buffer, not vanish into Info |
