fix: place the guided tour's card from its allocation, and count GTK warnings

The tour measured its card with a height-for-width request whose for_size
came from the natural width minus the margins the tour sets for positioning.
A measure reports margins *included*, so that for_size landed a margin short
of the minimum the box answers with and GTK warned once per frame:

  Trying to measure GtkBox 0x... for width of N, but it needs at least M

(about 9/s for as long as the tour was up). The card now takes its size from
gtk_widget_compute_bounds() once a frame has run -- its allocation rectangle,
not gtk_widget_get_width(), which is only the content box -- and falls back
to a measure with no for_size at all only on a frame with nothing allocated.

GTK is built with structured logging, so its messages reach the log writer
and never the handler API; no existing probe can read one. The exit dump now
counts them (`gtk warnings:`, must read 0) through a writer function chained
to g_log_writer_default(), which leaves stderr unchanged.

docs/claude/invariants-gtk-and-chrome.md

@@ -511,6 +511,23 @@ The `Help` submenu, the contextual help's "?" and the guided tour, and the traps
   request did not help -- the margin alone was the loop). Take the margins off a measure that feeds a
   position back. `hdl`/`bd`-style CSS padding is *not* part of it, and neither is the other axis's
   margin, so it is `margin-start + margin-end` on the horizontal and top/bottom on the vertical.
+- **A widget's allocation is not `gtk_widget_get_width()`, and a measure is never a position.** GTK's
+  `gtk_widget_get_width()`/`get_height()` return the widget's **content** box: `size_allocate`
+  subtracts the CSS border and padding before storing them (measured on the tour card, which carries
+  `border 1px` + `padding 12px`: `get_width()` reads 254 where the rectangle on screen is 280), so a
+  card *placed* from it overhangs its own size and covers the target it is meant to sit beside.
+  `gtk_widget_compute_bounds()` is that rectangle -- border and padding in, margins out -- and is what
+  the probe's geometry and the help's figures already read. And `gtk_widget_measure()` cannot stand in
+  for it: its size includes the widget's margins (the bullet above), and asking for a **height with a
+  width** (`for_size`) on a widget whose margins were then subtracted from that width makes GTK warn
+  `Trying to measure GtkBox ... for width of N, but it needs at least M` **once per frame**, because
+  the for_size lands one margin short of the minimum the widget answers with. The guided tour's card
+  therefore takes its size from `compute_bounds()` once a frame has run and falls back to a measure
+  **with no for_size at all** (`-1`, the form GTK measures without that check) only on a frame with
+  nothing allocated yet. The warning reaches no probe -- GTK is built with **structured logging**, so
+  it goes through the log *writer* and never `g_log_set_handler()`/`g_log_set_default_handler()` --
+  which is why the exit dump counts GTK's own warnings through a writer function (`gtk warnings:`,
+  which must read 0) as the one reading that can assert the class away.
 - **The overflow menu's `Help` submenu is a two-page `GtkStack`, and neither of the obvious
   alternatives works.** Rebuilding the rows -- the plot menu's own shape -- destroys
   `app.layout_toggle`, which points into the main view and which `sync_layout_editor()` and the `L`

docs/claude/probe-readings.md

@@ -410,6 +410,17 @@ Two lines of the self-check output are assertions in disguise:
   against the same config directory reads `auto=0` — and `requested=1` is the **chooser's** "Guided tour"
   (`Config::tour_pending`) having been carried into this run instead, which is `auto=0` because it is not
   the auto-start.
+- `gtk warnings: N` (in the exit dump, beside `help layer:`) — **GTK's own log at warning or above,
+  counted for the whole run, and it must read `0`.** It is counted through a log *writer* function
+  because GTK is built with structured logging: its messages go through the writer and never through
+  `g_log_set_handler()`/`g_log_set_default_handler()`, which is exactly why no probe can read one —
+  the message is written by the toolkit from inside a call of ours and is stored in no widget. The
+  first class to need it was `Trying to measure GtkBox … for width of N, but it needs at least M`
+  (once per frame, from a card measured for a height with a width a margin short of its own minimum;
+  see the allocation bullet in
+  [invariants-gtk-and-chrome.md](invariants-gtk-and-chrome.md)). The writer chains to
+  `g_log_writer_default()`, so every message still reaches stderr exactly as before and the documented
+  baselines are read the same way.
 - `help figures: trigger_head=X,Y,W,H trigger_summary=… trigger_panel=… signal_list=… side_switcher=… side_stack=… window_button=… status_label=… header_bar=… plot=…` (in the exit
   dump, beside `help layer:`) — the **window-coordinate** rectangle of each surface the help's annotated
   figures label, `x,y,width,height`, or `-1,-1,0,0` for a surface that is **not on screen**. It is

docs/claude/verification.md

@@ -24,7 +24,10 @@ touching each `*.gresource.xml` under `src/` and rebuilding; a third `-Wmissing-
 or any other warning is a finding. (The fifth warning that used to appear here was an unused
 `output_built` in `math_probe()`; it was **deleted** as dead code — redundant with the
 `output_after_add` already printed on the same line — which dropped the baseline from five to
-four.) And an exit-dump run emits **no** `Gtk-CRITICAL` at all.
+four.) And an exit-dump run emits **no** `Gtk-CRITICAL` at all, nor any `Gtk-WARNING`: the dump's
+`gtk warnings:` line counts GTK's own log (through a writer function — GTK is built with structured
+logging, so the handler API never sees its messages) and must read `0`; see
+[probe-readings.md](probe-readings.md).
 
 An *incremental* build reads **one** of those three: the generated GResource `.c` is not recompiled
 when only `src/` changed, so its `-Woverlength-strings` does not reappear.

src/ui.cpp

@@ -5646,9 +5646,35 @@ void collect_spins(GtkWidget* w, std::vector<double>* out) {
 /// dialogs built and destroyed in one run.
 
 
+/// GTK writes its own log messages -- `Trying to measure GtkBox \u2026 for width of N, but it needs at
+/// least M` is one -- from inside a call of ours, so they reach stderr and never a widget: no probe
+/// can read one, and the class is otherwise invisible to the self-check. The exit dump therefore
+/// carries a count of them, which is the one reading that can assert the class away (`gtk warnings:
+/// 0`). GTK is built with structured logging, so its messages go through the log *writer* and not
+/// through the handler API (`g_log_set_handler()` never sees them); the writer is chained to rather
+/// than replaced, so every message still reaches stderr exactly as before and the documented
+/// baselines are read the same way they always were.
+int g_gtk_warning_count = 0;
+GLogWriterOutput count_gtk_log(GLogLevelFlags level, const GLogField* fields, gsize n_fields,
+                               gpointer user_data) {
+  for (gsize i = 0; i < n_fields; ++i) {
+    if (g_strcmp0(fields[i].key, "GLIB_DOMAIN") != 0) continue;
+    if (g_strcmp0(static_cast<const char*>(fields[i].value), "Gtk") == 0 &&
+        (level & (G_LOG_LEVEL_WARNING | G_LOG_LEVEL_CRITICAL | G_LOG_LEVEL_ERROR)) != 0) {
+      ++g_gtk_warning_count;
+    }
+    break;
+  }
+  return g_log_writer_default(level, fields, n_fields, user_data);
+}
+
 int run_ui(int argc, char** argv, const Config& cfg, ScopeModel& model, Acquisition& acq,
            RowSet& rows) {
   (void)argc;
+  // Count GTK's own log messages for the run (see `count_gtk_log`). Installed before anything can
+  // log -- and it must be: a log writer function may only be installed once, and GTK does not install
+  // one, so this is the whole of the run's logging rather than a replacement.
+  g_log_set_writer_func(count_gtk_log, nullptr, nullptr);
   // Installs libadwaita's stylesheet, which the shortcut overview's `AdwShortcutsDialog`
   // needs to be drawn at all. Idempotent, and `choose_source()` may already have run it.
   adw_init();
@@ -6040,6 +6066,10 @@ int run_ui(int argc, char** argv, const Config& cfg, ScopeModel& model, Acquisit
     // guided tour stood when the last capture ran. Both are captured on the tick (`capture_help_state`),
     // so neither asks a destroyed widget anything.
     std::printf("[udpscope]   help layer: %s\n", app.dbg_context_help.c_str());
+    // GTK's own log at warning or above, counted for the whole run (see `count_gtk_log`). A probe
+    // cannot read one of these -- they are written by the toolkit, not stored in a widget -- so this
+    // count is the reading that catches the class, and the run must print `0`.
+    std::printf("[udpscope]   gtk warnings: %d\n", g_gtk_warning_count);
     // The bounds the help's annotated figures draw their crop boxes from: window coordinates, or
     // -1,-1,0,0 for a widget that is not on screen (`is_visible()`, not `get_visible()` -- a widget
     // under a hidden parent keeps its own flag true). Measured on the tick, after a frame.

src/ui_tour.cpp

@@ -296,19 +296,40 @@ void update_tour(App& app) {
   // whichever column is beside it -- which is, on this window, the side panel for the plot's step and
   // the plot for every step in the side panel.
   //
-  // Its size is measured, never requested -- but `gtk_widget_measure()` reports the size *including
-  // the widget's own margins* (measured: a 280 px box with `margin-start` 453 measures 733). Ours are
-  // the card's position, so reading the measure without taking them off is a feedback loop: the width
-  // it was placed at becomes the width it reports, and the card walks to the far corner and back
-  // between frames.
+  // The card's size is the *rectangle it was allocated*, read the same way the probe reads it and
+  // the help's figures are drawn from: `gtk_widget_compute_bounds()`. Not `gtk_widget_get_width()`,
+  // which is the widget's *content* size -- a border and a padding short of the rectangle on screen
+  // (measured here: 254 against the card's real 280), so a card placed from it overhangs its own size
+  // into the target; that call is read only to tell a frame that has been laid out from one that is
+  // too early. And, once a frame has run, not a measure either:
+  //   * `gtk_widget_measure()` reports the size *including the widget's own margins* (a 280 px box
+  //     with `margin-start` 453 measures 733), and ours are the card's position -- reading one back
+  //     without taking them off is a feedback loop.
+  //   * asking for a **height** with a width (`for_size`) on a widget whose margins are then left out
+  //     of that width is a question GTK refuses: it reports `Trying to measure GtkBox ... for width
+  //     of N, but it needs at least M` once per frame, because the for_size derived from the natural
+  //     width minus the margins lands one margin short of the minimum the box answers with. The
+  //     allocation the widget was actually given cannot be short of it.
+  // A measure survives only as the fallback for a frame with nothing allocated yet (the first frame
+  // the card is shown, before any layout pass), and that fallback is taken with **no for_size at
+  // all**: `for_size == -1` is the form GTK measures without the check that warns.
   int cw = 0;
   int ch = 0;
-  gtk_widget_measure(app.tour_card, GTK_ORIENTATION_HORIZONTAL, -1, nullptr, &cw, nullptr, nullptr);
+  graphene_rect_t card_bounds = GRAPHENE_RECT_INIT(0, 0, 0, 0);
+  if (gtk_widget_is_visible(app.tour_card) && gtk_widget_get_width(app.tour_card) > 0 &&
+      gtk_widget_compute_bounds(app.tour_card, app.tour_overlay, &card_bounds)) {
+    cw = static_cast<int>(std::ceil(card_bounds.size.width));
+    ch = static_cast<int>(std::ceil(card_bounds.size.height));
+  }
+  if (cw <= 0 || ch <= 0) {
+    gtk_widget_measure(app.tour_card, GTK_ORIENTATION_HORIZONTAL, -1, nullptr, &cw, nullptr,
+                       nullptr);
     cw = std::max(kCardWidth, cw - gtk_widget_get_margin_start(app.tour_card) -
                                   gtk_widget_get_margin_end(app.tour_card));
-  gtk_widget_measure(app.tour_card, GTK_ORIENTATION_VERTICAL, cw, nullptr, &ch, nullptr, nullptr);
+    gtk_widget_measure(app.tour_card, GTK_ORIENTATION_VERTICAL, -1, nullptr, &ch, nullptr, nullptr);
     ch = std::max(0, ch - gtk_widget_get_margin_top(app.tour_card) -
                          gtk_widget_get_margin_bottom(app.tour_card));
+  }
   int cx = 0, cy = 0;
   if (h - (hy + hh) >= ch + 2 * kGap) {
     cx = std::clamp(hx, kGap, std::max(kGap, w - cw - kGap));