max / goingson

Hand the drift sweep to witchbroom, and say what the tables now owe "Nothing automates this" was true when it was written and stopped being true today. `witchbroom`'s `described-drift` check reads both tables in this header: the pairing table for what to watch, the drift table for what has been answered. That makes two rows load-bearing that were previously prose. A new described module needs a pairing row in the commit that declares it. A commit touching a counterpart without the description needs a row here saying why, and an empty Closed cell reads as a finding rather than as a note to self. Acknowledges `5e16cc1`, which added the `raised` class across six JS files and changed nothing a description can say. GoingsOn infra 2bf42978.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: Max Johnson <me@maxj.phd> · 2026-08-17 18:21 UTC

Signed with PGP, not checked

Commit: 675e88001563617f73c2dda4e6ba15d77850962c

Parent: a4362aa

1 file changed, +17 insertions, -5 deletions

M src-tauri/src/quasi/mod.rs +17 -5

			@@ -126,6 +126,7 @@
126	126		//! \| [`board`] \| `0df3488`, both markers on the card \| [`Availability::marker`] \|
127	127		//! \| [`projects`] \| `0df3488`, both markers on the dashboard card \| [`Availability::marker`] \|
128	128		//! \| [`day_planning`] \| `0df3488`, the unblocks marker in the pool \| [`Availability::frees_marker`] \|
	129	+	//! \| [`projects`], [`tasks`] \| `5e16cc1`, a CSS class rename across six files \| presentational only: nothing a description can say changed \|
129	130		//!
130	131		//! Clean: [`contacts`], [`settings`], [`weekly_review`], [`emails`],
131	132		//! [`monthly_review`], [`problems`]. Four of those now have a commit touching
			@@ -135,11 +136,22 @@
135	136		//! marker went into `weekly-review-render.js` and into [`weekly_review::focus`]
136	137		//! together, which is the only reason it is still on this line.
137	138		//!
138		-	//! Nothing automates this. The sweep is a person running `git log` against
139		-	//! two paths, so it is only as current as the last time someone thought to.
140		-	//! What would close it is a check that fails when a JS file changes without its
141		-	//! described module changing, which is `witchbroom`-shaped work rather than
142		-	//! anything this module can do to itself.
	139	+	//! The sweep runs this now, and both tables above are its input.
	140	+	//! `witchbroom`'s `described-drift` check reads the pairing table for what to
	141	+	//! watch and the drift table for what has been answered, then reports every
	142	+	//! commit since a module was added that touched one of its JS counterparts and
	143	+	//! left the module alone. It runs on goingson only, nightly, and never on the
	144	+	//! four by-decision files or the ten non-screens, because those appear in no row
	145	+	//! of the pairing table.
	146	+	//!
	147	+	//! Two obligations follow, and neither is optional if the check is to stay
	148	+	//! readable. A new described module needs a row in the pairing table in the same
	149	+	//! commit that declares it, or the check reports `mapping-incomplete` naming it.
	150	+	//! And a commit that touches a counterpart without changing the description
	151	+	//! needs a row here saying why: either the description is behind and this is the
	152	+	//! fix, or the change was presentational and nothing a description can say moved.
	153	+	//! An empty Closed cell is a drift written down and not answered, and reads as a
	154	+	//! finding.
143	155
144	156		use std::sync::Arc;
145	157