pardes.git/test/gui_golden.py, branch main

GUI goldens settle on the window's idle, not two captures that agree

2026-10-01T03:12:16Z

test/gui_golden.py hashed a scene once two captures 400 ms apart agreed; on a loaded machine a slow frame agreed with itself before the scene's change landed and every later hash moved up a scene. A test window now writes a status file into its capture directory each loop step (gui.zig writeTestStatus: step, frames captured, idle — nothing owes a frame, nothing moves, nothing waits on the core's clock), and the harness waits for an idle step two past the one current when it asks: a 9P write has been served by then, so the frame is the scene's. Two things kept a window from ever going idle, both fixed: - Look-hover re-armed forever: resting on a word with nothing to open, the hover wait ran its two frames, found nothing, cancelled, and the next present's pointer refresh started it again: a frame every other frame for as long as the pointer rested there, live too (CPU). The place last refused is remembered (Pardes.look_hover_refused) until the pointer moves off it. - The virtual clock moved only on a wait that timed out, so a steady trickle of events held it still; it now moves to the core's wake too when an event cut the wait short and nothing owes a frame. Captures go to a directory of their own, outside the session's (a file written there every frame is a change in the directory its panes look at). All 17 goldens match the file as recorded at oooopqyv; eight runs, five alongside a ReleaseFast build and three with all 16 cores saturated (load 13-19), were byte-identical, 14 s a run unloaded (was ~30). Tests: the resting pointer lets the core go idle.

Snapshots and GUI goldens run with the default Placement acme

2026-10-01T03:12:15Z

Their clicks follow the panes to where acme's placement puts them; doccol and splitparent stay on Placement pardes, which they cover. Co-Authored-By: Claude Opus 5.5

A column can be empty, as acme's can; new panes go where acme's makenewwindow puts them

2026-10-01T03:12:15Z

Newcol makes an empty column; closing a column's last pane leaves it empty with the keyboard on its tag; Delcol and Joincol alone take a column away; the session's last pane closing quits. The +New stand-in and replaceStillborn / stillborn_joiner are gone. Every new pane goes through exec.placeNew, which follows acme's makenewwindow (active column, empty column whole, blank space, halving the biggest) or, with Placement pardes, the old rules. ColumnTags is gone: column tags are always shown. A grip drag shows acme's box cursor. Co-Authored-By: Claude Opus 5.5

The GUI draws in tiers from the regions; pane chrome moves with its pane

2026-10-01T03:12:15Z

Stage 8 of docs/render-pipeline.md. The frame is drawn in groups: tier 0, one per track in paint order (tiers 2 and 3), tier 4 (notices, then guides and the debug box), tier 5 (bar cursors), each cells, images, decor. Decor (rules, rails, thumbs, grip marks, spines, the workspace and column rules, the bottom band, notice rules, bar cursors) is whole-pixel rects through the new decor.frag over ui.vert, so it carries its track's transition and clip. A closing pane's chrome comes from Surface.previous_regions. Deleted: taglineBaseRgb, topbarPaneBorderHeight, bottomTaglinePresent, frameChromeBg, cellBackgroundIs and the rail inference, paneGripCell's scan, transient_on, PaintPlan. One cover map (coverFrame) is marked from the layers and regions once a frame. New regions column, guide and debug; Surface.chrome is the palette, carried in wire v8 (not shipped yet), so an attached GUI draws the same chrome (test). A per-instance clip replaces the vertical transition's scissor. Goldens: 01-12 byte-identical. 13-17 differ only past the grid: the image pass left its scissor at the grid's size, cutting rule ends, rail feet and the bottom band in the leftover pixels whenever a picture was on screen. 16-debug now shows the debug box, which a context-row layer had hidden. 18-mid-transition is new: virtual clock (PARDES_TEST_CLOCK in the GUI), PanelSlide Newcol with the picture, frame 6 of 12. Also: the GUI sleeps when idle instead of polling every 16 ms, and a minimized or occluded window sleeps through animation. Tracy 'gui frame build' at 200x60: 992/1015 us median before, 989/987 us after. Shared files touched: pardes.zig (one export), detached/client.zig, detached/server.zig, detached/wire.zig (all additive), gui.zig. Not touched: Messages.zig, mouse.zig, tagline.zig, colors.zig, tty.zig, host_io.zig, dump.zig, exec.zig, panes.zig.

Paint a body with context rows once, into its layer, and join it

2026-10-01T03:12:15Z

joinBody replaces the second paint of a file body with compact tree-sitter context rows: the body is painted once into its body layer (more rows than the grid has) and the grid's body rows are copies of the layer's first ones; the rail is the grid's alone, from its region. Which rows are context rows is still decided by a text build before the paint, and a body without them reuses that text and paints straight onto the grid: every other body has no row the grid cannot show, so it gets no layer and no copy. Each part now paints only its own rows. The tag paints the tag rows of every pointer sweep that crosses them (the select sweep in the tag's colours, the look-hover affordance over it, a middle or right sweep over both in its button's tint), which the body used to paint over the tag on the grid alone; the body paints only body rows, and a select sweep begun on the tag marks the body rows it crossed before the text is laid down, as it did. Grid and layers agree, the snapshot goldens are byte-identical, and a test checks the joined grid rows and caret against a plain paint of the same body with the tag on top and at the bottom. The GUI goldens gain a terminal pane scene.

Step core animation by the shell's clock and sleep to the next wake

2026-10-01T03:12:14Z

A shell now answers Host.now, its monotonic clock in ns, and the pump advances core animation to it: one .tick per whole 16 ms frame since core time last stood still. The same animation therefore takes the same time at 60, 120 and 144 Hz, over ssh and after a slow frame (the GUI's re-armed clock ran them ~25% slow at 144 Hz; the tty's post-frame sleep drifted). nextWake() lists every core animation in one place: a frame from now while anything moves, the end of the wait while something only waits (a message lingering, the look-hover delay), null when idle. Shells sleep exactly that long, and a wait is jumped to its end in one step, so an 800 ms linger costs one frame instead of fifty. An overshoot under 1.5 ms after a step is let go: at 60 and 120 Hz every display frame takes exactly one step (a 16 ms frame against a 16.67 ms vsync would otherwise double-step every 24th). The six tick drivers are gone: tty's timer thread only times the wait, interruptibly (a newer, shorter request cuts short a sleep still timing a longer one); the GUI's AnimationClock, web's JS tick bank (pardes_tick now takes rAF's timestamp), the detached server's and the board's ticks; grid mode steps a virtual clock straight to each wake. PARDES_TEST_CLOCK, set by the snapshot harness, gives tty and the detached server the same virtual clock. Every stepped frame is drawn. The old pump never drew the last frame of a fade (a .tick asks for no frame, and the fade was over by the check), so a theme switch stopped at 9/10 of the way until the next input: theme.golden, themesel.golden and the GUI's acme-light scene move to the theme's true colours, and nothing else changes. The frozen previous grid is captured only while a panel transition is chosen, not on every idle frame.

Hash settled GUI frames of sixteen scenes, a local gate for render refactors

2026-10-01T03:12:14Z

test/gui_golden.py drives a hidden pardes-gui through 9P and window-pixel mouse events on its test stdin into sixteen scenes: documents, a notice, sticky context rows with the scroll thumb mid-rail, a body selection, a tag holding the keyboard with a selection and caret, a prompt notice, wide glyphs across a tag's and a body row's right edge, a collapsed pane with focus tint, Tagbottom, column tags, a multi-line tag, an image, a dark and a light theme, and Debug. Each scene waits until two captures agree and is hashed. The goldens are machine- and driver-specific: a local gate, never a CI one.