`Style::marker_below_samples` (50 by default, `0` for never) draws a `marker_px` dot on every
*raw* sample when the visible window holds at most that many, in the trace's own colour. Zoomed
in far enough to see individual samples, the polyline between them stops being the shape a reader
is looking at -- a handful of points across a whole panel ARE the picture -- and a dot makes each
one a place rather than a bend in a line. Level 0 only: above it the geometry is pyramid buckets,
which are not points anyone can find.
Test: a_sparse_window_marks_its_samples -- the same flat series at ten samples and at a hundred.
Ten gives exactly ten dots, evenly spaced (their centres are checked against the panel's own
step, so "they are the samples" is asserted rather than "something was painted"); a hundred is
past the threshold and gives none. The dots *group* into runs of columns (a marker is `marker_px`
wide), which is how the test reads one marker per sample -- counting columns read three.
Cursors took `Style::line_width_px` and were always solid, so a host had no way to place a
*readout mark* on the data (a measured level, a peak, a pulse's edge) without it reading as
another instrument framing the plot. `set_cursor_stroke(id, width, dash)` gives it its own, the
same shape as `set_cursor_color()`: per cursor, because one plot holds both kinds.
`width <= 0` means the style's default for that cursor (`handle_width_px` while hovered or
grabbed, `line_width_px` otherwise), and a hover can only make the stroke *thicker* -- the cue
that the pointer has hold of something must not be the thing that hides it. `dash.off <= 0` is
solid, the convention a trace's, the crosshair's and the grid's dash already share.
Test: a_cursor_stroke_can_be_thinner_and_dashed -- a default cursor and a thin dashed one in the
same frame. The dashed column alternates ink and gap (153/151 at the 4/4 the test asks for)
while the solid one has no gaps; the ink laid down over each stroke's whole run of columns is
less for the thin one. That last assertion is deliberately *not* a single-column peak: a cursor
is a data position and is not snapped, so the solid stroke's ink split across two columns and
read *lighter* than the thin one -- which is how this test first failed.
grid_major 0x2A3038 -> 0x39414C, still well under axis_line's 0x8A94A6, so the three-way
hierarchy the eye reads stays minor < major < axis. Already fully opaque, so "more
prominent" meant a brighter colour and nothing else.
The tests that classify grid pixels take their colours from the style, so they followed
the change; the one hardcoded mention (grid_pixel()'s doc comment) is updated, and the
minor's relationship to the major is now stated as an order of magnitude rather than the
fifth the old colours happened to give. 361 cases, 0 failed.
Subdivisions used to be drawn exactly like the labelled lines -- same width, same
opacity -- so a plot dense enough to need them read as two competing sets of ticks.
The minor class is now three things the style states explicitly:
- grid_minor_width 1.0 -> 0.75, a sub-pixel logical width. At 1x that is expressed as
reduced coverage through the anti-aliased ramp; on a HiDPI display the line is
literally narrower.
- grid_minor_dash_on/_off (new, 3/3): dashed by default, off == 0 draws it solid, the
same convention as the crosshair's and a trace's dash. The dash is a fragment-shader
discard, so no geometry is added -- the vertices are the line either way.
- grid_minor carries alpha (0x1C2126 at 50%), and the line shader applies u_color.a,
so it blends into the background instead of sitting on it.
Together that is about a fifth of the ink a major line leaves along its length, which
is what gives the eye one grid to read instead of two. The major class is untouched:
it stays the structure.
Tests: a new pixel test (the_minor_grid_is_thinner_dashed_and_dimmer_than_the_major)
asserts a major column is inked top-to-bottom while a minor one alternates, and that
the minor's ink sits closer to the background than the major's. Three existing tests
counted the minor class as "most of eight rows" of its exact colour; two now take
their floor from the style (minor_line_ink: width x alpha x duty), and the width test
states the dash and alpha away because widths are its subject. 361 cases, 0 failed.
Renderer::draw_traces picked the sample window with (x - x0()) / dx(), and dx() is
documented as a *nominal* spacing -- for a series whose x is a streamed timestamp it
is only the mean of the retained window. The window it computed therefore landed
inside the visible range, and the two end segments, the ones that enter and leave the
frame, were never emitted: the left and right margins of a zoomed-in view came out
empty. A wide enough drift dropped the trace entirely (last <= first).
It asks Series::neighbours() now -- a binary search over the real x, the same call the
crosshair reads through, costing the same. A uniform series answers the same question
through the same call, so it is one path. dx is still read for the samples-per-pixel
density estimate, where a nominal spacing is exactly right.
Test: a_sparse_window_between_samples_still_inks_both_margins -- a series whose dx()
lies, with the window strictly between two samples, so nothing inside the window is
sampled and only the crossing segments can ink it. With the grid walk reverted it
stages nothing at all (visible=0, gathered=0, covered=0/24); with the fix it stages the
three bracketing samples and covers 24/24 columns.
A host zooming into a sub-second window cannot label it absolutely: Time
formatting uses three significant digits, so 12.345671s and 12.345672s are the
same string and a 1 ms step around a multi-second time is unlabelled. The axis
gains an `origin`: every value it writes is `v - origin`, and the ticks are
generated on the shifted grid, so the window reads 0, 500us, 1ms.
`origin_follows_range` re-takes the origin from the axis' own range on every
refresh -- the mode a *moving* window needs, since the edge is derived from the
newest sample at layout time and a value the host wrote a frame earlier is a
frame of scroll stale.
`format_time` now scales its sub-second values by prefix, the way `format_si`
does on a y axis: the unit that keeps the mantissa in [1, 1000). A fixed
millisecond unit printed 0.005ms where 5us is the reading, and 1e-05ms where
10ns is. An exact zero prints `0`, the same convention `format_si` states. The
origin subtraction's dust is snapped by a *relative* epsilon, so a tick on the
origin reads 0 rather than 1.78e-12ms.
Tested by `ticks_are_written_relative_to_an_axis_origin`,
`a_range_origin_puts_the_zero_tick_on_the_left_edge` and the extended
`format_time_shapes`; the suite is 359 cases, 0 failed.
A host that draws the result of a computation as a cursor -- a ruler at a level
it measured, a marker at a time it found -- must not let the user drag it: the
value belongs to the measurement, not to the pointer. Parking it outside the
axis and re-placing it every frame does not achieve that, because a drag is held
for the gesture's life and the marker follows the pointer between frames, so the
readout *looks* draggable however often it is put back.
`Cursor::selectable` defaults true; the controller's hit test skips an
unselectable cursor, so it is neither hovered nor grabbed and its hit radius
goes to whatever is behind it. Set through `Plot::set_cursor_selectable()`.
Tested by `an_unselectable_cursor_is_neither_hovered_nor_grabbed` in
tests/test_controller.cpp.
A *readout* cursor -- a marker a host keeps for something it computed, rather
than an instrument the user placed -- has to be told apart from its neighbours,
and the cursor API could set everything about one except its colour. The setter
keeps the id and ignores a stale one, like the setters beside it.
Tested by `a_cursor_can_be_recoloured_by_id`, and the suite is 356 cases, 0
failed.
The figure-mode cursor test covered the *vertical* marker, which is one time
shared by every row. A horizontal ruler is the other half: its y is on that
row's own axis, so two rows' rulers are two different values, and every row's
has to be drawn in its own band. A host that puts two rulers on every plot
(`udpscope`'s cursors pane) needs exactly this, and had no test for it.
It asserts the strokes only -- the ruler's value label goes through the text
path, which this case is not about, so it uses the harness's no-atlas backend.
A host whose annotation is a readout -- the point where a cursor crosses a
trace -- has to move it every frame, and `add_annotation()` only appends:
the id would change every frame and the annotation count grow without bound.
The setter keeps the id, ignores a stale one exactly as the cursor setters
do, and pairs with parking (`visible = false`) since there is no
`remove_annotation`.
Tested by `an_annotation_can_be_moved_by_id` in tests/test_plot.cpp.
A range straddling zero never lands *on* zero: the walk starts at
`ceil(min / stride) * stride`, which is not representable whenever the stride is
not a power of two -- 0.1 and 0.2 are the everyday ones -- so the tick that means
zero arrives a few ULP off and `format_si` labels it from six prefixes down.
`[-1, 1]` read `-1 -500m -139a 500m 1`. Deriving the tick by multiplication
rather than accumulation does not help, because the error is already in the
first term.
Every tick is an integer multiple of the stride, so the smallest magnitude a
non-zero tick can have is the stride itself, while the residue is a few ULP of
the range -- ten orders of magnitude apart. Anything inside a millionth of the
stride is therefore the origin and nothing else. It is stored as exactly `0.0`
rather than merely labelled so: `Axis::refresh()` re-formats a time axis from
`tick.value`, where the same residue reads `3.19e-13ms`. The walk variable is
left alone, since `minor_index_is_major` counts indices from `first_major` and
both operands have to stay on the walk's own grid.
The two narrow-y-title cases were calibrated on the defect -- the bogus `-55.5a`
was a six-character label, and its width is what spent the chrome allowance and
made the titled and untitled frames comparable. They are re-seated on a
milli-scale range, whose `-200m` is wide for a real reason, and the zero-width
strip the frame comparison depends on is now asserted rather than assumed.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Text was point-sampled (`GL_NEAREST` plus `floor(v_uv * u_tex_size)`), which
at a non-integer scale *deletes* texels instead of blending them. The library's
own default -- a 12 px style over the 13 px embedded bake -- dropped one of
every 8 columns and one of every 17 rows of every glyph, so whole strokes went
missing and which ones depended on where the unsnapped pen fell. `kTextFS` now
does a 2x2 bilinear fetch clamped to the instance's own cell: the clamp keeps
the no-bleed guarantee the sampler mode protects (the atlas is a tight strip of
cells with no padding), and at scale 1 with a whole pen the weights are exactly
0, so it stays as crisp as the point sample.
Grid and axis lines were staged in data space and mapped in the vertex shader,
so a 1 px line's centre landed on an arbitrary fraction -- x = 100.0 lights two
columns at half coverage, x = 100.5 lights one crisply, and a frame mixed both.
`u_snap` rounds an axis-aligned segment's ink *leading edge* onto a pixel
boundary, which reduces to integer+0.5 for odd widths and to an integer for
even ones with no case analysis. It is on for the chrome that frames a plot and
off for anything drawn at a data position, since the rounding moves a line by
up to half a pixel. Diagonals are untouched.
`text_pixel`'s threshold drops to 160 because filtered glyphs legitimately peak
at 0.844 coverage at 12/13; the 180 threshold found text only while the sampler
was also deleting a column and a row of it.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Breaking: Series::gather emits x relative to x0() and takes a GatherMode;
consumers reconstruct absolute x in f64 before mapping (series.hpp documents
the contract).
Bugs, each pinned by a fail-before regression test:
- render: f32 staging quantised epoch-time x into 128 px columns; every
data-space draw now stages to_space(x) - to_space(origin) with the origin
folded into the offset uniform in f64 (epoch/zero-origin twins rasterise
identically)
- ticks: major test was ULP-fragile and blanked epoch axes; majors are an
integer-index test and are never truncated by the minor cap
- gather: LOD envelope emitted buckets straddling [begin,end); ceil/floor
bucket edges + raw lead/trail spans
- live follow: log x axes held a linear span and zoomed every frame; span and
offset now live in axis space (plot + both figure link-group paths)
- input: a zero-notch wheel disarmed auto_scale on every axis; dy==0 pans x,
factor == 1 / inf refused in wheel and drag zoom
- render: the single text flush painted legend names over the crosshair
readout panel; two batches, under flushed before the panel
- data: NaN dropped from min/max unless it seeded the fold; non-finite
samples are now excluded from every aggregate and from cached bounds
Robustness: viewport normalisation (no clamp UB), non-finite range refusal,
pyramid level cap at 57 (no shift/div UB through public gather), rounded ring
capacity documented + pre-sized, GL blend func and UNPACK_ALIGNMENT
save/restore, Renderer::release_gl() teardown, u64 buffer epochs, annotation
captions clamped, button-aware GLFW row capture, cached figure row rects (no
per-event layout), ruler-pair measurement refused, span-nonfinite guards in
axis_ends.
Features: XY series (sorted x, binary-search neighbours/snap, LOD over sample
index); StreamSeries::freeze(); ArraySeries::push(); per-trace dash and
markers (scatter at width 0) with auto or literal stride; GatherMode
Envelope/Nearest/Mean via bucket_sum/sumsq; Style::line_aa_px; per-glyph text
colour (readout values in trace colour); wheel-dx x pan; arrow-key pan;
Escape cancels a drag and restores captured ranges; modifier-masked box zoom
(Shift both, Ctrl x, Alt y); double-click reset via PointerEvent::time_s;
ImGui key repeat unified; pinch gestures on Qt and GTK; pixelDelta wheel path
and DevicePixelRatioChange handling; GNUInstallDirs, SOVERSION,
SameMinorVersion, generated plot/version.hpp, relocatable plot.pc, licence
install, FILES_MATCHING, GitHub Actions CI (debug/coverage/asan).
Suite 283 -> 350 cases, 0 failed; coverage 93.9% (3730/3971) over 24 files,
floor 80/90 held; ASAN preset suite green; all four demos run headless (the
GTK scripted-resize check needs a window manager and fails identically on the
parent commit); readme_main.cpp committed and compiles with the README
command; pkg-config and find_package consumption verified from a relocated
install prefix.
`remove_region()` now clears or decrements the published `hot_.region` with the
cache it re-indexes, so a frame after a removal highlights the region the handle
actually names. The new case covers both branches (a handle above the hole
follows it down, the removed region's handle is cleared with its zone), and the
coverage document is regenerated by the command that measured it: 95.2% of src/,
every file at or above 80%.
Three smaller findings at the plot/figure seam:
* `Plot::remove_region()` re-indexed `region_stats_` but left `hot_.region`
naming the old index, so a frame after a removal highlighted the wrong
region's bound and an open drag edited a different region. The published
handle (and its zone) is now cleared or decremented with the cache, and the
contract documents that a host must not remove a region mid-drag.
* `Figure`'s `kMinRowHeightPx` was applied as a *device*-pixel floor while
chrome is logical, so at 2x the guarantee was half as strong exactly where
the chrome is twice as large. It is now a logical constant converted through
the row's own `ui_scale()`, 1:1 for a non-positive scale as `Style::device()`
does.
* `Figure::link_x()` refused a member whose axis 0 was `XFollow::Live`, but
nothing refused `set_x_follow(0u, XFollow::Live)` *after* linking. The group
then wrote its range into an axis that reinterpreted it against its own live
edge, read it back as an external change, and tracked one member instead of
the union. `Plot` now carries a linkage flag (`set_x_link_member()`), set by
`link_x()` and cleared by `unlink_row()`, and `set_x_follow()` refuses `Live`
on a linked axis.
Tests: arming `Live` on a linked member leaves it `Off` and unlinking restores
the ability to arm it; the row-height floor is 24 and 48 device pixels at 1x
and 2x respectively.
Three findings reachable by ordinary use:
* `Controller::axis_zoom()` divided by `Plot::ui_scale()` directly, though
`Style::device()` documents a non-positive scale as host state that must be
tolerated. With `ui_scale == 0` and a negative drag delta, `logical_dy` was
`-inf`, the factor `+inf`, and the guard `!(factor > 0.0)` did not reject it:
`zoom_x_axis`/`zoom_y_axis` computed `anchor + (bound - anchor) * inf`, and
`set_y_range()` stored it verbatim and cleared `auto_scale` -- an axis left
non-finite until an auto-fit. A negative scale silently inverted the zoom.
A scale that is not positive and finite now means no zoom at all.
* `draw_annotations()` stages its caption into the frame's single text batch,
which `flush_text()` draws *after* `clear_clip()`, so a caption staged while
the axes clip was engaged was rasterised with the viewport clip. An
annotation anchored outside the visible range kept a caption with no cross,
and in a `Figure` -- whose rows tile the figure -- that caption painted into
the neighbouring row. A caption whose anchor is outside `axes` is now skipped
with its cross, rather than splitting the frame's one text draw call (which
would cost a draw call per row and mis-clip the labels staged outside `axes`).
* `dump_frame_ppm()` was the one public entry point not `noexcept`, so a
`std::bad_alloc` crossed the API boundary. Its buffers are now sized before
anything is on disk and a failure is reported as `ErrorCode::OutOfMemory`,
exactly as `TrueTypeFont::from_file()` reports it.
Tests: `ui_scale` 0 and -1 leave the axis range finite and unchanged with
`auto_scale` intact; an annotation anchored outside a bottom figure row's y
range leaves no caption ink in the row above, while an in-range one still
draws; a `static_assert` pins the dump's `noexcept`.
`controller_at()` routed every pointer event by the row the pixel was in,
ignoring the row a drag started in, so a press on one figure row followed by a
motion and release in another delivered the motion and release to a controller
whose `mode_ == None`. The owning controller kept `mode_`, `drag_region_`,
`drag_cursor_` and `drag_fixed_x_` set for ever: `publish_hot()` kept publishing
the grabbed handle, `grabbed_region()` reported a drag that had ended, and a
later gesture re-entering the row resumed the stale edit from the old anchor.
`feed_pointer_event()` now owns the routing: a press captures the row it landed
in and every motion and the release until the button comes up go to that row
(`pointer_target()`), cleared on release and when another figure is set. The
GLFW callbacks build the event and call it, so a windowless host or a test can
drive the routing too.
Two neighbours in the same seam:
* `sync_figure_crosshair()` called the four-argument overload unconditionally,
so a documented vertical-only crosshair (`horizontal == false`) was promoted
to a full cross at the stored y == 0, clamped to the axes edge and made
permanent; the change detection also keyed on x alone, missing a y-only move
to another row at the same x (two rows showing one horizontal line, spec 17).
It now consults `crosshair().horizontal` and compares the hovered row's y.
* Only the row under the pointer republished its hot handles, so a row the
pointer left kept drawing a hovered marker or bound as if it were still
under the pointer, and answered `hovered_region()`/`grabbed_cursor()` for
it. The run loop clears every other row's handles, keeping the captured
row while a drag owns the button, and reads the frame's one layout once
instead of recomputing every row five times.
Test: a press on row A's region, a motion into row B, a release in row B
asserts no controller keeps a drag mode, nothing reports a grabbed handle, and
a later press-and-motion that re-enters row A does not resume the old edit.
The repository had no licence at all, so every source was all-rights-reserved
by default. Add the verbatim MIT text at LICENSE (Copyright (c) 2026 Martino
Ferrari) and a two-line SPDX header to each of the 113 first-party sources --
48 include/plot, 26 src, 10 adapters, 5 examples, 21 tests, 3 tools -- so the
grant travels with each file rather than relying on the root document alone.
Scripts keep their shebang on line 1; the SPDX lines follow it. Genuinely
vendored code is untouched: third_party/stb_truetype.h carries no header of
ours. src/render/embedded_mono_font.cpp is generated, so tools/fonts/gen_atlas.py
emit() now writes the two lines into its header comment and the file was
regenerated -- gen_atlas.py --check passes and the regeneration adds only those
comment lines to the previously committed file.
README's provenance section now states MIT and links LICENSE.
No code, test or public API change.
The figure sync learns which row the pointer is in and that row's local y, so the
shared x goes to every row and the horizontal line only to the hovered one -- a
crosshair that finally marks a row where its y means something. The stroke gains
Style::crosshair_width/_dash_on/_dash_off (logical; 0.75 / 4 / 4), a per-Crosshair
override, and a fragment-shader dash on v_along (off == 0 is solid), anchored to the
axes edge each line is staged from so the phase does not crawl.
Review fix pass 1 on the task 7 work; four findings, all closed.
1. The Qt demo's region sat across the middle of the window, which is exactly
where its own scripted input check pans: after the check's box zoom the press
opened DragMode::MoveRegion, the x range never moved, and the check printed
'input check pan FAIL' -- under an exit 0, because the verdict was printed and
not returned. The region now rides the left sixth (as the GLFW demo's does),
both checks pass, and the verdict is wired to the exit status. The same audit
found two more printed-only verdicts -- the GTK demo's scripted-resize layout
check and its screenshot write (and Qt's) -- now wired too. What stays
printed-only is the screenshot line of the GLFW host and the ImGui widget,
where the dump happens inside the adapter's frame loop.
2. A follow-ing link group was invisible to the input layer: a member's own axis
must be XFollow::Off (link_x() refuses a Live member), so Controller's
x_zoom_anchor() extrapolated a coordinate from the pointer instead of using
the live edge and a wheel over a row's axis strip walked the window back by
19% of its width (behind 0.000000 -> 0.285809 s). The group now publishes its
own edge to every member (Axis::live_owner + Plot::set_x_live_owner, written
by Figure::push_link, cleared when the group stops following or the row
unlinks), and x_zoom_anchor() reads it. Public-header addition, called out.
Measured after: behind 0.012992 s and the window's right edge bit-identical to
the edge the group published at the gesture.
3. The multiplot summary printed x_link_count() -- a number of groups -- under
'rows in it'. It now reports the members and the group count separately, and
every row line carries its own group id.
4. Host::controller_at() promised to clamp a pixel in a row gap to the nearest
row and only clamped below the last one, so a gap pixel went to row 0. It now
selects by distance, and tests/test_glfw_host.cpp pins it with a 40 px gap --
the first coverage of the host's figure routing, which needs no window.
New cases: tests/test_controller.cpp x2 (the anchor and the hint's lifecycle),
tests/test_glfw_host.cpp x3 (routing, the gap, the single-plot fallback): 229
cases, 0 failed, ASan clean. All five demos exit 0, the three self-checking
modes exit 0, and the amended assertions were confirmed failing against the
pre-fix behaviour.
Adding a grabbable region across the middle of the glfw demo turned the pan
gesture PLOT_DEMO_FOLLOW makes -- and a user's own centre drag -- into a region
drag, which that mode's self-check reported as 'distance behind only by the pan:
NO' and exit 1. The region now rides the left sixth of the live window, so a
press in the middle pans exactly as it always did, and the reason is written
where the bounds are placed.
The multiplot demo also prints its region's last-frame fill rect and bound
columns, so the exit frame's region pixels can be read against the geometry
Plot::layout() measured.
Report: .superpowers/sdd/p2-task-7-report.md
The multi-plot example the phase exists for: three plots stacked in one
Figure, sharing one x link with following armed, each row carrying its own
named signals and its own legend, driven by six StreamSeries at 500 kS/s.
Two cursors and a region with an exact statistics readout reach the direct
manipulation of spec §5 in the same frame, and PLOT_DEMO_SYNC=1 is a
self-checking mode that exits non-zero unless all three rows hold one range
and the shared right edge keeps tracking the newest sample through a zoom, a
pan on another row and more arriving data.
The GLFW host gains the figure mode the demo needs: set_figure() makes the
run loop draw the Figure instead of its own plot and route each pointer,
wheel and key event to the row it lands in, through one Controller per row
(row_controller()). All of it is opt-in; the single-plot path is untouched.
Review of the atlas refactor found one Important and three Minor issues:
- Important: TextAtlas::sync called glGetIntegerv before any available() check,
so a host that synced before plot::gl::load() -- which is legal for a public
class Tasks 2 and 8 build backends on -- segfaulted on a null entry point
instead of receiving the Status the interface promises. It now follows the
handle-constructor convention (available(), then the entry point itself) and
reports Unsupported.
- empty() no longer means "a GL name exists": releasing one is a
glDeleteTextures call, so the no-atlas path forgets the atlas and only resets
the texture when there is a table to release it with.
- end_frame's comment still claimed the frame hands back the unpack alignment
init() set, contradicting the code below it and renderer.hpp.
- The orientation case's texel probe mapped the atlas' u along the cell's +v
axis and vice versa, which is a no-op for its two diagonal probes -- so it
could not fail on a transposed UV mapping, and its own comment misstated the
rule. Probes now follow the shader's rule and include an off-diagonal pair.
- The null-glyph path of add_text (and its twin in walk_text_clipped) was
unreached by any case.
Cases: text_atlas_reports_a_status_when_the_entry_points_are_missing,
add_text_skips_a_codepoint_the_backend_cannot_draw.
The text pipeline stages a glyph's rect *origin* per instance and carries the
rect's *size* once per draw, so a backend whose glyphs have differing rect sizes
-- a TrueType packing trimmed to ink, which is Task 8's shape -- would have drawn
every glyph at whichever size it happened to be told, with nothing downstream
able to tell. Make that constraint structural rather than documented:
- TextBackend::cell_uv() declares the one cell size, and the contract (an atlas
is a fixed-cell grid; a proportional face draws its glyphs *into* a uniform
cell instead of trimming each rect to its ink, and why the pipeline needs it)
is stated on the atlas accessors.
- glyph_is_cell() is the check; add_text()/set_cell() declare the cell from the
backend, and the staging path asserts it for every glyph in a debug build.
A release build draws on with the cell it was told, as before.
- Renderer::flush_text() takes u_glyph_uv from the backend, not from the batch,
so the draw's one rect size can never come from whichever glyph was staged
first.
Final whole-branch review fixes.
1 (Important). Plot::measurement() cached on (series pointer, sample count)
only, so a saturated StreamSeries -- whose size() stays pinned while x0()
advances every acquire() -- froze its readout on the first saturated frame
until a cursor moved. The cache key now carries the series' x0() (new
MeasureSlot::last_x0), so a sliding window recomputes and a static series
still does not: asserted with a new macro-gated counter hook
(plot/detail/measure_count.hpp, PLOT_COUNT_MEASURE_RECOMPUTES, wired like the
pyramid and ring counters, absent from the release library). New case
saturated_stream_measurement_tracks_the_sliding_window: saturated ring,
stationary cursors, ramp in absolute x -- zero recomputes for repeated reads
with nothing moving, one per acquire while it slides.
2-3, 9. Spec: 4.3 described a GL_LINE_STRIP path that does not exist (there is
one rasteriser -- instanced anti-aliased quads -- and dense_threshold_spp
selects the gather level, not a primitive); 8 named a PLOT_GL_CHECK() macro
that exists nowhere, and stated no exception policy. The table, the field's
comment and 8 now describe the real gather-level threshold, the real
glGetError() drain plus the stores' drop-to-orphan fallbacks, and the real
policy: library failures return Status/bool, allocating entry points propagate
std::bad_alloc.
4. The frame handed back neither GL_UNPACK_ALIGNMENT (init() sets 1 for the
atlas) nor the clear colour (queried once with a new glGetFloatv entry).
end_frame() restores both and renderer.hpp lists them in the residue
contract.
5. draw() kept a host TextBackend for layout while drawing with its own font.
The renderer now installs its font unconditionally -- strips and labels measure
and draw with one backend -- with a debug-build diagnostic if a different one
was set, and Plot::set_text_backend() is documented as the no-renderer path.
New case renderer_measures_the_labels_it_draws asserts layout, measuring and
the drawn label's ink agree.
6-10. draw()'s style overwrite stated at the call site; the load-sensitive
wall-clock bound in test_measure.cpp replaced by a counter assertion (same
interval, same fold count over a 512x length ratio); dump_frame_ppm drops
noexcept and says why it may allocate; error.txt ignored (not this project's,
not generated here) rather than deleted.
Verified: debug and release build clean (-Wall -Wextra -Wpedantic -Wconversion
-Wshadow), 4/4 ctest entries pass, 130 cases pass (was 127), umbrella and
header self-containment checks pass, and no counter symbol in the release
archive.
The linux-release preset inherited PLOT_BUILD_TESTS=ON, so the release build
compiled the pyramid cover/merge and ring upload/wait/slot counters into
libplot_core.a and cmake --install shipped them. Set PLOT_BUILD_TESTS=OFF in the
preset; the suite keeps running in linux-debug and linux-asan, which are
unchanged.
Also pin the two header properties the plan calls Global Constraints and that
nothing guarded: every header under include/plot/ compiles standalone, and
include/plot/plot.hpp includes every public one. tests/CMakeLists.txt globs the
headers at configure time (CONFIGURE_DEPENDS) and generates one TU per header
plus an umbrella TU that constructs Plot, Renderer, data::StreamSeries and
input::Controller; a ctest entry builds the header TUs and runs the umbrella,
and cmake/plot_umbrella_check.cmake fails configure and ctest when a public
header is missing from the umbrella.
The exit-frame self-check compared the y label strip to the pixel and printed a
FAIL at GDK_SCALE=2 where it measured 174 -> 172. Measured (temporary probe,
reverted): the strip is `widest major tick label + padding`, and the tick step
follows the axes length, which the resize changes -- 18 majors over 1440 device px
with "-800m" widest, 8 over 1080 with "-500m", 78 vs 76 px in the bitmap font. The
layout was recomputed; the equality was the wrong assertion. The check now asserts
what actually distinguishes a recompute from a scaled stale layout: the axes and
its strips tile the *current* viewport, and the strip is nearer the previous strip
than that strip scaled by the resize ratio (2 px of drift against a scaled 118).
`on_exit_timer`'s main-context pump can re-enter its own 20 ms timeout, which ran
the exit block twice in one observed run; `on_teardown` is guarded the same way so
the producer join, the window destruction and the quit happen exactly once. A window
closed by hand takes the same ordering through `GtkWindow::close-request`, which
`gtk_window_destroy()` does not emit.
The factor both the viewport and every forwarded input coordinate use came from
`gtk_widget_get_scale_factor()`, which GTK documents as "the next higher integer
value" under fractional scaling. It is now derived by `DeviceMapping::from_sizes()`
from the widget's logical allocation and the device size `GtkGLArea::resize`
reports -- the viewport that signal's default handler sets with
`glViewport(0, 0, w, h)` -- so the viewport, the renderer and input are one value
by construction. The public `int scale_factor()` became `f32 device_scale()`.
`kSurfacePixelsPerNotch` is now 100.0 and documented as a *calibration* with a
stated assumption, not a unit conversion attributed to axis_value120/Qt.
`::unrealize` releases the renderer unconditionally, after making the context
current: the error it used to check says nothing about whether GL names exist
(creation failure is the path that returns before a renderer is built), and
skipping the release left the deletes to the wrapper's destructor, with no
context.
The drag anchor caches its press point and offsets in logical pixels and scales
the difference by the factor in force now, so a scale change mid-drag cannot
become a phantom pan.
The demo stops its producer in `on_teardown`, before destroying the window whose
renderer the stream feeds.
The persistent-mapped instance ring had three slots and recorded a fence per
upload, so a frame recycled a slot whose fence was recorded a few draws earlier
*in that same frame*: from the fourth of the demos' nine uploads on, each one
waited on the frame's own work. It is free on a driver that retires draws
promptly (the GLFW host, 1.07 ms/frame) and ~4 ms of the Qt widget's 7.11 ms
under llvmpipe, and it grows with the trace and cursor counts this library
exists for.
* The ring grows to cover a frame's upload count, the frame's write cursor is
the ring index (so a frame writes 0, 1, 2, ... and never a slot twice), one
fence is recorded after the frame's last draw, and the only wait is at the
start of the next frame. A wait that times out, or a slot written with no
fence ever recorded behind it, drops the store to the orphan path -- the
treatment a refused fence already had -- so nothing is ever written into
storage the GPU may still be reading.
* New test-only counters (PLOT_RING_STATS, the PLOT_PYRAMID_COUNT_* pattern:
free functions, no member, so no consumer's layout changes) pin a frame's
upload and wait counts. The new tests assert the frame's slot offsets are
distinct, that the ring grew past its floor to cover the frame, and that a
frame waits at most once per store -- read from the counters and from
glClientWaitSync itself.
* qt6: print the unresolved entry-point names before the load-failure branch,
which used to return first and leave the one failure this widget can have
reporting nothing but a generic message.
* qt6: the cached pointer anchor is now in logical pixels and a move's delta is
the logical difference scaled by the DPR in effect, so a scale change between
two move events (the window dragged onto another monitor mid-drag) is no
longer a cross-scale jump applied as a pan. The anchor moved into a
toolkit-free unit (adapters/qt6/qt_pointer.hpp) with its own test.
GLFW demo headless: 1.07 -> 0.73 ms/frame. Qt widget under xvfb/llvmpipe:
7.1 -> 3.8 ms at dpr 1.0, 6.7 -> 3.6 ms at 1.25, with both upload paths still
rendering and the demo's scripted pan/box-zoom checks passing at both scales.
Zeroing an unusable wheel amount turned it into a zero-notch zoom: the factor was
exactly one, the range did not change, but the zoom body still ran through
set_x_range/set_y_range, which clear auto_scale unconditionally. One garbage wheel
therefore disarmed auto-fit on every axis. on_wheel now rejects an unusable amount
before the event can mutate anything, keeping the position fallback.
The tests that should have caught it are strengthened: the wheel case now asserts
auto scale is still armed (it was passing either way), and the invalid-position
case checks a pan measured from the last real position with the drag alive, rather
than a range nothing could have moved.
The widget drew a region derived from the child's content rect alone, so a child
the hosting window cut off cleared and painted over the host's pixels; it also
scaled ImGui's (-FLT_MAX, -FLT_MAX) no-pointer sentinel and differenced it against
the last real position, turning a drag out of the window into a ~1e35 pan, and it
forgot the host's button levels across a re-bind, handing the next plot a press
that never happened.
The region is now the content rect intersected with the child's clip rect (a
no-op when nothing clips it), an invalid position is no position update, and the
button levels survive a re-bind. The controller sanitises non-finite coordinates
and deltas for any host, and the widget's input translation now has a committed
test on the headless EGL context.
Findings from the review of the sliding-window change:
- A level >= 1 gather emitted only complete, wholly-retained buckets, so the
geometry stopped at the last bucket boundary inside the range: the newest
samples of a live window -- up to `span - 1` of them, about a pixel at the
level level_for() picks -- were not drawn, on the right edge, which is the
part of a scrolling trace that gets watched, and the oldest retained samples
fell outside the geometry too (the level-1 gather of a sliding window started
51 samples in). The two partial edge buckets are now gathered as raw samples
through copy_raw() and bracketed around the envelope -- the same exactness
split reduce/sum/sumsq already use -- so the emitted geometry spans the
requested range exactly. The pyramid still stores no partial bucket, so append
stays amortised O(1); at 1 MS/s the demo stages 511 extra sub-pixel segments
and still draws at 1.05 ms/frame, and the screenshot's trace now reaches one
column further at each end.
- `drop_before`'s doc claimed a no-op "for an empty pyramid": false, and
backwards -- the restart branch is exactly what an empty pyramid hits, which is
what StreamSeries::acquire() depends on when the ring saturated before the
first acquire. The precondition is `absolute_sample <= window_base()`, and the
restart is documented as what it is.
- The pyramid's memory figure was 3x optimistic: an entry is 24 bytes (MinMax
plus the exact sum/sumsq doubles), so level 0 costs 0.375n bytes and the whole
pyramid ~0.75n -- about 19% of 4-byte raw samples, not the 6% a min/max-only
index would cost. Corrected in the header, spec 4.2 and 4.5, with the trade
stated (the doubles buy O(log n) sum/mean/RMS/integral from whole buckets),
and the plan's Task 3 sketch no longer describes per-level prefix sums, a
level() span or an unaligned-end widening that does not exist.
Tests: `gather_at_a_lod_level_reaches_both_ends_of_the_window` pins that the
first and last emitted points are the oldest and newest retained samples -- with
spike values no bucket envelope can produce -- at level 1, at the level a dense
view picks, and for a static series whose end is unaligned. It fails against the
previous commit (first point at the first whole bucket), and so does the sliding
gather mapping check, which now accounts for the raw edge pairs.
Frame cost was O(capacity x levels) once the ring saturated: acquire() saw the
window base move (it always moves once the ring wraps), cleared the pyramid and
re-appended all 1,048,576 retained samples, and append() walked every live level
with no early stop -- 13,647,936 bucket merges and 73.8 ms per frame at a
1M-sample window, scaling with the window, which is what capped the 1 MS/s demo
at 32-58 ms/frame while it was still filling.
Index the pyramid in absolute bucket space instead. Level l bucket k covers the
samples [k * (kBucket << l), (k + 1) * (kBucket << l)), so a bucket's span never
depends on where the window starts and the window can slide over the aggregate
instead of forcing it to be rebuilt. Each level keeps its complete buckets in a
deque plus the absolute base of its first entry; append() folds a sample into the
single in-progress level-0 bucket and, only when that bucket completes, cascades
upward, stopping at the first level that gains no new entry; a new drop_before()
slides the window by discarding leading buckets and remembers where the window
begins. acquire() is now: drop to the new base, then append exactly what was
produced -- 256 merges and 0.034 ms per frame for 8192 new samples, independent
of the window, and memory bounded by the window (32,767 entries after 50M samples
through a 1M ring, the same as after 10M). A producer that laps the consumer by
more than capacity() between two acquires restarts the window rather than
splicing a gap.
Interface: `level(l) -> std::span<const MinMax>` cannot survive a ring and is
replaced by level_base/level_size/bucket in absolute bucket coordinates, plus
window_base/window_end/drop_before. The *_buckets queries stay in window
coordinates (index 0 = oldest retained sample), so Series keeps its window
semantics for callers; its exactness split now aligns in absolute space
(window_base() + the ceil/floor) because the bucket grid is absolute. The sums
move from per-level prefix sums to per-bucket ones -- the only representation a
sliding front can keep exact -- and StreamSeries::y_bounds() becomes
reduce(0, size()), exact in O(log n + kBucket) per frame instead of a scan of the
window. The pyramid owns the published window now, so there is no second copy of
base/end to keep in step, and gather() emits only buckets that are both complete
and wholly retained.
Tests: test-only merge counters (mirroring PLOT_PYRAMID_COUNT_COVERS, absent from
every non-test build) pin an acquire proportional to production and independent
of capacity, an append cost amortised over 16 live levels, and bounded memory;
brute-force checks pin absolute bucket aggregates across slides, sliding-window
reduce/sum/sumsq over unaligned ranges, and the level-1 gather mapping. The demo
now draws 946 samples/px at 1.03 ms/frame (967 fps) where the saturated window
cost 54.96 ms/frame (18 fps).
Two findings on the pass-1 fix:
- The reserved strips were clamped only to the viewport, so a title wider than a
narrow panel took the rest of it and the axes collapsed: draw_traces() returned
on a zero-width rect (no trace drawn, nothing reported) and a zero-height one
flattened every trace onto a single row. The chrome is now capped at half the
viewport -- the label strip is served first, the title strip takes what is left
of that allowance -- an over-long title is clipped to its strip by a walk that
both measures and stages it, and draw_traces() tests the height as well as the
width.
- Every titled y axis was staged at the same centred position inside one strip, so
a Left/Right pair of titles overprinted. Each titled axis now owns a text line:
the x strip grows one line per titled x axis (x_title_height), and the y runs
stack down the axes-tall strip, which needs no extra width for it.
Tests: `narrow_panel_with_a_long_y_title_still_draws` and
`short_panel_with_an_x_title_still_draws` pin the floor (the axes keeps half the
panel and the trace still covers the columns, with the title clipped into its
strip), `titled_axis_pairs_get_their_own_rows` pins one text row per titled axis
in both directions, and `layout_stacks_one_title_line_per_titled_axis` pins the
strip sizing. All fail against the pass-1 sources.
Four review findings on the Task 7 renderer:
- InstanceStore::epoch() is derived from gl::Buffer::generation() -- advanced by
every allocation, deletion and move -- plus the slot the instances were written
to, so the orphan fallback of grow_persistent() can no longer leave the vertex
array's attribute pointers bound to a deleted buffer: the epoch reports the new
identity and upload() re-points them before the draw.
- PlotLayout reserves y_title_area/x_title_area outside axes (sized from the text
metrics) and draw_titles() draws in them, so a title can no longer land on the
data, whatever the envelope does.
- draw() computes plot.layout() once and hands it to the helpers, instead of six
times per frame, each of which rebuilt the ticks and the grid.
- A refused fence is no longer dropped: mark_drawn() falls back to the orphan path
and reports the failure rather than rotating through an unsynchronised slot.
New smoke cases force both refusals through the GL entry points (glMapBufferRange,
glFenceSync) and assert the pixels plus the live transfer mode; the title case
asserts the title colour is absent from the bands inside axes and present in both
strips. All three fail against the pre-fix sources.
Every pixel-to-data conversion in the controller now maps through the
same rectangle Plot::x_transform/y_transform are built from:
- pan() derives the scale from layout().axes.w/.h, so a drag moves the
grabbed data point by exactly the drag distance (the viewport model
stranded it behind the pointer by the label-strip fraction).
- apply_box_zoom() clamps the selection to layout().axes and sets each
range from the transforms (x_transform().to_data_x(l/r), y_transform()
.to_data_y(bottom/top)); the selection's bottom edge is the y minimum.
- end_drag() takes the release event and refreshes the selection from its
coordinates before applying, so a host that reports only press and
release still frames the box instead of degrading it to a click.
- axis_zoom() uses a per-pixel sensitivity (200 px == one decade) rather
than reusing the wheel's per-notch constant, so a pixel drag is no
longer treated as a wheel notch.
Tests now assert the grab-and-drag invariant through the transform (the
old pixel-fraction assertions certified the viewport defect), add a
press+release-only box zoom, and pin the axis-zoom magnitude.
Martino Ferrari