Files
plotter/tests/CMakeLists.txt
T
Martino Ferrari 38ef459897
CI / debug (tests + all adapters) (push) Canceled after 0s
CI / coverage (floor + measured table) (push) Canceled after 0s
CI / asan (AddressSanitizer) (push) Canceled after 0s
feat(plot)!: harden pipeline, XY series, dash/markers, touch, packaging
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.
2026-09-30 17:32:06 +02:00

223 lines
11 KiB
CMake

# EGL powers the surfaceless headless context the GL tests assert pixels
# against. It is a *test-only* dependency: plot_core never links or includes it,
# so a machine without EGL can still build the library (-DPLOT_BUILD_TESTS=OFF).
# Prefer CMake's own module, fall back to pkg-config, and fail loudly if neither
# can supply EGL rather than silently dropping the GL tests.
find_package(EGL QUIET)
if(NOT EGL_FOUND)
find_package(PkgConfig QUIET)
if(PkgConfig_FOUND)
pkg_check_modules(EGL_PC QUIET IMPORTED_TARGET egl)
endif()
endif()
if(TARGET EGL::EGL)
set(plot_egl_library EGL::EGL)
elseif(TARGET PkgConfig::EGL_PC)
set(plot_egl_library PkgConfig::EGL_PC)
else()
message(FATAL_ERROR
"plot: PLOT_BUILD_TESTS=ON requires EGL (EGL/egl.h plus libEGL) for the headless "
"OpenGL context used by the GL tests. Install it (e.g. libegl-dev, libegl1-mesa-dev "
"or mesa-libEGL-devel) or configure with -DPLOT_BUILD_TESTS=OFF.")
endif()
find_package(Threads REQUIRED) # std::thread in the push/read race test
add_executable(plot_tests
main.cpp
test_transform.cpp
test_gl_context.cpp
test_primitives.cpp
test_ring_buffer.cpp
test_minmax_pyramid.cpp
test_series.cpp
test_xy_series.cpp
test_ticks.cpp
test_measure.cpp
test_plot.cpp
test_figure.cpp
test_render_smoke.cpp
test_controller.cpp
egl_headless.cpp
)
target_link_libraries(plot_tests PRIVATE plot::core ${plot_egl_library} Threads::Threads)
# The Dear ImGui widget is the one adapter with a test here: its input translation
# (edge synthesis, the hover gate, drag capture, wheel containment) is logic this
# project owns, not a toolkit's. The suite is otherwise ImGui-free, so the source
# and the link are added only when the adapter exists -- which the top-level
# CMakeLists.txt guarantees by adding the adapters before this directory.
if(TARGET plot::imgui)
target_sources(plot_tests PRIVATE test_imgui_widget.cpp)
target_link_libraries(plot_tests PRIVATE plot::imgui imgui)
endif()
# The GLFW host's figure-mode routing -- which row's controller a pointer pixel reaches --
# is this project's logic too, and it needs no window: the cases drive `controller_at()`
# against the rects `Figure::layout()` computes. Guarded on the adapter's target for the
# same reason as the ImGui block above.
if(TARGET plot::glfw)
target_sources(plot_tests PRIVATE test_glfw_host.cpp)
target_link_libraries(plot_tests PRIVATE plot::glfw)
endif()
# The Qt adapter's pointer anchor is deliberately toolkit-free (see
# adapters/qt6/qt_pointer.hpp), so this one piece of that adapter's translation is
# tested without linking Qt -- and without `PLOT_BUILD_QT6`, since no Qt target has
# to exist for it to compile.
target_sources(plot_tests PRIVATE test_qt_pointer.cpp)
# Same for the GTK adapter's input translation (see adapters/gtk4/gtk_input.hpp):
# the device-pixel mapping it derives from GTK's two sizes, the pixel space its drag
# anchor is kept in, the wheel's unit and sign and the modifier mask are this
# project's logic, and a headless X server at an integer scale can exercise only
# part of it.
target_sources(plot_tests PRIVATE test_gtk_input.cpp)
target_include_directories(plot_tests PRIVATE "${CMAKE_SOURCE_DIR}/adapters")
# The optional runtime-font backend brings its own cases, and they exist only when
# its target does: the header, the class and the archive are all absent otherwise,
# so there would be nothing to compile them against (see PLOT_BUILD_TTF_FONT in
# cmake/plot_options.cmake). Guarded on the target rather than on the option, like
# the adapter cases above, so a target that exists for any reason gets its cases.
#
# Two paths are handed over as definitions because a case needs them and should
# not guess: the core archive the "no stb_truetype in plot_core" case inspects
# (a generator expression, so it is this build's own archive) and the cmake the
# umbrella check is re-run with.
if(TARGET plot::font_ttf)
target_sources(plot_tests PRIVATE test_true_type_font.cpp)
target_link_libraries(plot_tests PRIVATE plot::font_ttf)
target_compile_definitions(plot_tests PRIVATE
PLOT_CORE_ARCHIVE="$<TARGET_FILE:plot_core>"
PLOT_CMAKE_COMMAND="${CMAKE_COMMAND}")
# The cases compare a bake against the rasteriser's own glyph bitmaps, which is
# the only reference that can tell a complete cell from a clipped one, and for
# that the test needs the vendored header's *declarations* (its implementation is
# already in the binary through plot::font_ttf). This is the optional target's
# own test, compiled only when that target exists, and it is the one target
# outside src/text that sees third_party/: the isolation this project promises --
# plot_core's archive and every public header under include/plot/ -- is what
# plot_core_stays_free_of_stb_truetype keeps asserting, and it is unaffected by
# which include paths a test binary has.
target_include_directories(plot_tests PRIVATE "${CMAKE_SOURCE_DIR}/third_party")
endif()
# Same switches the library is built with: the cost tests read the counters
# behind them (see include/plot/data/detail/{cover,merge}_count.hpp).
target_compile_definitions(plot_tests PRIVATE PLOT_PYRAMID_COUNT_COVERS
PLOT_PYRAMID_COUNT_MERGES
PLOT_RING_STATS
PLOT_COUNT_MEASURE_RECOMPUTES
# The committed source root, so the
# atlas-generator case can run
# tools/fonts/gen_atlas.py --check
# against the committed atlas.
PLOT_SOURCE_DIR="${CMAKE_SOURCE_DIR}")
plot_apply_warnings(plot_tests)
add_test(NAME plot_tests COMMAND plot_tests)
# The GL tests create a real EGL context, and llvmpipe's teardown leaks driver
# allocations that are not this library's. The suppressions file names those
# libraries and nothing else (see cmake/plot_asan_suppressions.txt); a report from
# anything under include/plot/ or src/ is still an error. The environment is set
# on the test rather than in a preset so `ctest --test-dir build/asan` gets it
# however it is launched, and print_suppressions makes the counts visible -- a run
# must show what it suppressed, not hide it.
if(PLOT_ASAN_ENABLED)
set_tests_properties(plot_tests PROPERTIES ENVIRONMENT
"ASAN_OPTIONS=detect_leaks=1:print_suppressions=1:suppressions=${CMAKE_SOURCE_DIR}/cmake/plot_asan_suppressions.txt")
endif()
# --- public headers: standalone compilation and umbrella coverage -------------
#
# The plan's Global Constraints require every header under include/plot/ to
# compile standalone, and include/plot/plot.hpp to include every public one
# (never an adapter). Both were verified by hand and nothing kept them true, so
# a header added later could break either silently. These targets make the build
# enforce both and the ctest entries below make the suite fail when the checks
# do. The header list is globbed at configure time (CONFIGURE_DEPENDS, so adding
# or removing a header re-globs) -- a hand-kept list would defeat the point.
file(GLOB_RECURSE plot_public_headers CONFIGURE_DEPENDS
RELATIVE "${CMAKE_SOURCE_DIR}/include"
"${CMAKE_SOURCE_DIR}/include/plot/*.hpp")
list(SORT plot_public_headers)
# One generated TU per header, each holding only that header's #include. An
# OBJECT library: this is a compile check, with no link step, so any header that
# needs something it does not itself include fails the build here.
set(plot_header_check_sources "")
foreach(plot_header IN LISTS plot_public_headers)
string(MAKE_C_IDENTIFIER "${plot_header}" plot_header_slug)
set(plot_header_check_source "${CMAKE_CURRENT_BINARY_DIR}/header_checks/${plot_header_slug}.cpp")
set(plot_header_check_content
"// Generated by tests/CMakeLists.txt: <${plot_header}> must compile on its own.\n#include <${plot_header}>\n")
file(GENERATE OUTPUT "${plot_header_check_source}" CONTENT "${plot_header_check_content}")
list(APPEND plot_header_check_sources "${plot_header_check_source}")
endforeach()
# The generated version header is public (the umbrella names it) but is not
# under the source include root -- the top-level CMakeLists.txt configures it
# into <build>/include/plot/version.hpp -- so the glob above cannot see it. It
# gets the same standalone-compile TU by hand, and reaches the header through
# plot::core's build-tree include interface.
file(GENERATE
OUTPUT "${CMAKE_CURRENT_BINARY_DIR}/header_checks/plot_version.cpp"
CONTENT
"// Generated by tests/CMakeLists.txt: <plot/version.hpp> must compile on its own.\n#include <plot/version.hpp>\n")
list(APPEND plot_header_check_sources
"${CMAKE_CURRENT_BINARY_DIR}/header_checks/plot_version.cpp")
add_library(plot_header_self_containment OBJECT ${plot_header_check_sources})
target_link_libraries(plot_header_self_containment PRIVATE plot::core)
plot_apply_warnings(plot_header_self_containment)
# The umbrella on its own: the generated TU includes only <plot/plot.hpp> and
# constructs one of each layer's entry type, so an umbrella that misses a header
# -- or that is unusable without further includes -- fails to build. The test
# runs the linked binary, so a build that produced nothing usable still fails.
set(plot_umbrella_check_source "${CMAKE_CURRENT_BINARY_DIR}/umbrella_check.cpp")
file(GENERATE OUTPUT "${plot_umbrella_check_source}" CONTENT
"// Generated by tests/CMakeLists.txt: the umbrella must be usable by itself.
#include <plot/plot.hpp>
int main() {
plot::Plot plot;
plot::Figure figure(2);
plot::Renderer renderer;
plot::data::StreamSeries series(16, 0.0, 1.0);
plot::input::Controller controller(plot);
(void)figure;
(void)renderer;
(void)series;
(void)controller;
return 0;
}
")
add_executable(plot_umbrella_check "${plot_umbrella_check_source}")
target_link_libraries(plot_umbrella_check PRIVATE plot::core)
plot_apply_warnings(plot_umbrella_check)
# Coverage check on the umbrella. Included here so a public header missing from
# plot.hpp stops the configure; the ctest entry below runs the same script again
# so the check is also a visible test. The generated version header is passed in
# both times: the script must require it of the umbrella, and must look for the
# file in the build tree where configure_file() put it.
set(PLOT_INCLUDE_DIR "${CMAKE_SOURCE_DIR}/include")
set(PLOT_GENERATED_INCLUDE_DIR "${CMAKE_BINARY_DIR}/include")
set(PLOT_EXTRA_PUBLIC_HEADERS "plot/version.hpp")
set(PLOT_UMBRELLA "${CMAKE_SOURCE_DIR}/include/plot/plot.hpp")
include("${CMAKE_SOURCE_DIR}/cmake/plot_umbrella_check.cmake")
add_test(NAME plot_headers_self_contained
COMMAND ${CMAKE_COMMAND} --build "${CMAKE_BINARY_DIR}" --target plot_header_self_containment)
add_test(NAME plot_umbrella_covers_public_headers
COMMAND ${CMAKE_COMMAND}
-DPLOT_INCLUDE_DIR=${CMAKE_SOURCE_DIR}/include
-DPLOT_GENERATED_INCLUDE_DIR=${CMAKE_BINARY_DIR}/include
-DPLOT_EXTRA_PUBLIC_HEADERS=plot/version.hpp
-P "${CMAKE_SOURCE_DIR}/cmake/plot_umbrella_check.cmake")
add_test(NAME plot_umbrella_check COMMAND plot_umbrella_check)