9 commits

Author	SHA1	Message	Date
BruebachL	b1c8f156dc	[Game] Add hover zoom and a tabbed card panel to the board (#7700) ... * [Settings] Add hover zoom/panel triggers, delays, and double-confirm * [Game] Smooth board hover-zoom with fade helpers * [Game] Add hover card panel with tabbed actions * [Game] Wire hover zoom/panel triggers through the scene * [Game] Repaint cached zones when the hover zoom moves `TableZone` cuts the hovered card out of its own paint so zone washes, indicators, dimming and the land divider never draw over the zoom. It is a `DeviceCoordinateCache` item, so the card changing geometry never dirtied it and the stale hole stayed transparent, uncovering the black viewport as soon as the zoom crossed a zone edge. - Emit `hoverFootprintChanged(old, new)` from `setCardScale()` and once from `setHovered(false)`. - Repaint every overlapping `TableZone` on that signal. - Clip with the card's painted outline instead of its bounding rectangle, and treat a zoom covering the whole zone as "paint nothing". * [Interface] Keep the hover panel P/T controls out of the way The P/T set/inc/dec rows were always visible and sat under the cursor, so clicking a card to select it could mutate its P/T instead. - Collapse the controls to a pill on the printed P/T box, anchored on the drawn artwork so letterboxed and sideways cards stay aligned. - Open the rows only from the pill and grow them upward, and re-measure the pill on every value so three-digit P/T does not clip. - Close the panel when another card is hovered, and repaint the area it covered so no patch is left behind. * [Interface] Address hover panel review findings * [Game] Fix include order flagged by the format gate The card hover widget header was included before theme_manager.h, which clang-format rejects; the branch's format CI job failed on it. * [Interface] Load the hover panel icons through the themed icon engine The panel still named pre-#7595 assets (arrow_*_green, arrow_right_blue, cogwheel) and built QIcon snapshots from themePixmap(), so the deleted names rendered blank and every other glyph froze one palette. Load every glyph through loadThemedIcon() with the emphasis axis the adoption plan calls for: the category tabs take the primary level as panel wayfinding, the action grid the quiet level because its glyphs label the button text. Semantic tints keep their family and follow the same quiet level; Clone keeps its sticker treatment. * [Icons] Add an Informational semantic color family Face-down play and the exile zone are states that are neither a gain nor a loss, and the pre-#7595 art painted them blue (arrow_right_blue). #7595 dropped that hue, and the icon engine has had only Positive and Negative since, so both states fall back to neutral ink. Add the family with the same three-level ladder the other semantic roles use: authored roles per theme, scheme defaults otherwise, an emphasis mapping, the palette-editor rows and the generator seed. The defaults clear the 4.5:1 label floor on both schemes and the ladder's saturation ordering; the existing emphasis, hue and preview tests now cover the new family. Preview sheets that name roles resolve an unauthored one through the same default the client paints. * [Interface] Split the hover panel icons by emphasis and semantics Category tabs only navigate between pages, so they take the quiet level; the action grid is the surface the player works in and takes the primary level. Both stay line glyphs: a filled treatment reads as a second state on every cell. Hue marks an outcome where the glyph cannot. Play, clone and to hand are Positive; reduce life and graveyard Negative; play face down and exile Informational. Direction-only moves (top and bottom of library) stay neutral because the arrow already carries the meaning. * [Icons] Add Hidden and Marked semantic color families Two game concepts had no hue of their own. Hidden information (a face-down play, peeking at a card face) now has violet, and the marks a player leaves on a card (annotation, draw arrow) now share amber instead of splitting blue and red for the same job. Both families follow the existing three-level ladder: authored roles per theme, scheme defaults otherwise, an emphasis mapping, palette-editor rows and generator seeding. The defaults clear the 4.5:1 floor on the representative platform surfaces and keep the saturation ordering the emphasis tests assert; the hue tests pin violet's blue-red-green dominance and amber's red-green-blue one, and every preview sheet now renders the two families. The hover panel grid preview is added alongside: it renders the real pages (section headers, two-column cells, glyph over label) so the whole map can be judged at once instead of glyph by glyph. * [Icons] Add a Movement semantic color family A card changing position inside the library is neither a gain, a loss nor information, so it had no hue. Movement names it with teal and covers the three library placements (top, bottom, X from top). It follows the existing ladder: authored roles per theme, scheme defaults otherwise, an emphasis mapping, palette-editor rows and generator seeding. The defaults clear the 4.5:1 floor on the representative platform surfaces, keep strong > base > soft saturation, and keep green > blue > red so teal cannot be mistaken for the green gain family. * [Interface] Give every hover panel action a semantic color The quick actions are the panel's whole point, so a neutral glyph made the most-used buttons the hardest to scan. Every action now names a game fact with its hue: gain, loss, information, hidden information, a mark on the card or a move. The line glyph stays identical, so the colour is the only difference between neighbours. The grid is also ordered by family so a row reads as one idea: the two state toggles, the two gains, the two marks, the two hidden-information actions and the two losses each share a row. Conditional actions (Peek, Unattach) sit at the end of their pair, so a missing one leaves a hole at the section edge instead of splitting a row. In a zone, Play and Play Face Down stay adjacent on purpose: the same action, one of them hiding information. The move tab leads with the three library placements (Movement), then the destinations. actionGridIcon is removed; semanticActionIcon is now the only grid ink. The preview renders the real pages so the whole map can be judged at once. * [Icons] Use quiet emphasis for glyphs that label rather than act The search and info glyphs inside the search fields, and the combo-box chevron, label or afford an action rather than performing one. At the default secondary weight they compete with the actual controls next to them. Paint them at quiet emphasis so they stay legible without claiming attention. Sites: deck editor, visual database display, visual deck storage, set manager, settings search, in-game zone view, and the dropdown chevron. * [Icons] Restore sticker treatment to the add, remove and delete actions The pre-#7595 add and delete glyphs were filled shapes with an outline. After the line migration they read as ordinary toolbar glyphs, losing the weight their verbs had. Give the add and delete family the sticker treatment again. - Deck editor Add card: primary positive sticker. It is the one action the database panel exists for. - Visual database Clear filters: negative sticker, matching the delete family. - Filter dock Clear all: negative sticker, matching Clear one. * [Icons] Give sharing and pinning their semantic families Sharing is an informational act: the app already tints reveal and make-public with the informational family, so the deck storage share action, the replay share-code action and the visual deck storage share button follow it. The printing-selector pin badge marks a card as pinned. Paint it with the marked family, the same one used for a mark left on a card, instead of leaving it neutral. * [Game] Dismiss the hover panel before opening a modal dialog The hover panel is a Qt::ToolTip window, i.e. a popup: it is always on top and holds the input grab. Actions on the panel that open a modal dialog (Set power/toughness, Set annotation, Set counter value, token creation) therefore opened the dialog behind the panel, where it could not be seen or activated. Hide the panel synchronously and defer the action to the next event-loop turn, so the popup grab is released before the dialog's event loop starts. The panel asks the scene to dismiss it through a new dismissRequested signal, keeping hover timers and the covered-card fade on the scene's side. Reported from the P/T overlay SET button; the same path covers the counter tab and the token tab. * [Icons] Refine VDE and VDD emphasis and semantics The visual database filter section and the syntax help are information aids, so paint them with the informational family at the default emphasis: the filter section glyph and the search field's help glyph. The help glyph leaves quiet emphasis because it is a clickable action, not a field label. The visual deck editor's display-type toggle is the one control that changes the whole view, so it takes primary emphasis. The search glyphs stay quiet; only the help action changes. * [Icons] Raise the visual database filter controls to primary The six quick-filter buttons are the controls the filter section exists for, so paint them at primary emphasis instead of the default weight. The sort glyph carries the same informational family as the filter glyph, in both the visual database toolbar label and the visual deck editor's sort-criteria button. * [Icons] Keep the display options sort glyph neutral The sort button beside the display-mode toggle is not part of the informational family. Only the visual database toolbar's sort glyph carries informational; the display options button returns to the default neutral weight. --------- Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>	2026-11-08 21:51:31 +01:00
BruebachL	c7e8422808	[Icons] Author icon emphasis colors as palette roles (#7789) ... * [Icons] Author icon emphasis colors as palette roles The emphasis levels were computed from the accent at paint time, so a theme could not vary them: every theme got the same derived strong and soft tones, and a semantic hue was pushed toward the accent whether the author wanted that or not. Give each family three authored roles instead. Icon/IconStrong/IconSoft carry the neutral glyph ink and Positive/Negative gain the same Strong and Soft companions. Emphasis becomes a lookup inside the family, and a caller that names a variant pins that level at every emphasis. The paint path returns the authored color verbatim; only the filled treatments keep their saturation floor, because carrying a disc or a fill is a treatment decision. Shipped palettes author the full ladder for both schemes; the palette editor exposes the new rows and the generator writes them. System keeps deriving its neutral ink from the platform palette, since it ships no palette of its own. * [Icons] Keep the authored emphasis in filled treatments The saturation floor was added while the emphasis levels were still computed: it rescued the pale dark-scheme tones once a whole shape carried them. With the levels authored, the floor overrides the author instead: a quiet level jumped to the secondary saturation as soon as a sticker or a badge painted it, so the ladder only existed for line glyphs. The authored tone is now the top of the fill gradient and the treatment only darkens the bottom for depth, so every treatment renders exactly the level the palette stores. The badge glyph ink no longer guesses the scheme from the tone's value: an authored quiet level is a dark color on the dark scheme too, so the scheme is passed in from the surface palette. The preview sheet renders every shipped palette theme, three glyphs per family, and each row prints the authored tone and the darker bottom. * [Icons] Order the emphasis ladder by chroma, not contrast Contrast cannot order the ladder. On a dark surface the highest-contrast tone is the palest one, so a 7:1 primary came out near-white while the secondary kept the family's color; the ladder read inverted, and for the shipped semantic greens and reds the primary level was already satisfied by the secondary, so the two levels resolved to the same hex. Emphasis is chromatic intensity now. Quiet recedes toward the surface at the 3:1 non-text floor, secondary keeps the family color at the 4.5:1 text floor, and primary is the most chromatic tone the same floor allows. Every level stays legible on the window and the button surface, and the primary is a strictly more saturated member of its family on both schemes. The reformatted palettes move the two upper levels onto the shared floor and leave the quiet level alone. * [Icons] Make the emphasis ladder structural on every theme The chroma ladder landed with four P1s from review: reachContrast picked its pole by background lightness instead of by what each pole can reach, the generated neutral ink was solved on the window surface only, Fusion's neutral family was still the green accent, and quiet sat below the text floor while disabled reused the palette gray. Pick the pole by achievable contrast, solve fallback and preview inks on the harder of the window and button surfaces, re-author Fusion's neutral trio in slate blue, and keep quiet on the shared text floor by muting its saturation to 45% instead of blending it into the surface. Disabled is now derived from the quiet ink, so it is structurally dimmer than every enabled level on every theme. Re-author all soft roles at a 4.60 target with a rounding margin, rebuild the per-theme preview generator the rebase dropped, and add a Disabled row to every treatment sheet. * [Icons] Render preview glyphs at the sheet's device pixel ratio The preview sheets paint on a 2x painter but rasterized each glyph at its logical size, so every icon was upscaled on the way into the file and the sheets read as blurry. Render each glyph at device size and tag the pixmap with the ratio, so one source pixel lands on one file pixel. * [Icons] Keep the System semantic ladder on the text floor System authors no palette of its own, so its semantic family is fixed data measured against whatever surfaces the platform provides. The ladder that landed with the authored roles reused the old contrast-ordered values: PositiveSoft and NegativeSoft sat at 2.76–3.90:1 on common light and dark platform surfaces, and the strong level was less saturated than the secondary, which is the inversion the chroma rule removes. Re-author the six default semantic colors per scheme at the same 4.60 target the shipped themes use, against representative platform surfaces, and mirror them in System's palette files. The new DefaultSemanticFamilyMeetsTheLadderOnPlatformSurfaces test holds the floor and the saturation order there, and shippedPalette's early return now says why System is outside the authored-palette assertions. * [Icons] Apply the CI clang-format output to the new tests * [Icons] Adapt the treatment preview to the authored roles The preview file is inherited from the treatment PR. Here the palette already authors the fill tone and the badge takes an explicit scheme, so the generator stops re-saturating the role and passes the scheme the sheet renders. * [Icons] Author neutral System icon ink The System theme derived icon ink from the platform palette, which rendered blue under the default palette. Author Icon/IconStrong/IconSoft explicitly: a black ladder in light mode, a white ladder in dark mode. Pin authorship, achromaticity and ordering in the render test. * Address review findings	2026-11-08 19:09:02 +01:00
BruebachL	3e2884e6e1	[Icons] Paint neutral glyphs with icon ink and add an emphasis axis (#7788) ... * [Icons] Paint neutral glyphs with icon ink instead of body-text ink Neutral icons resolved `currentColor` straight to `QPalette::WindowText`. A glyph is not a word: it is a 16-24px mark drawn with a 2-unit stroke, so antialiasing thins it until it reads lighter than the text beside it, in the one color the palette reserves for reading. The result is that icons recede into the chrome exactly where they should anchor it. Add `theme_icon_ink`, one rule with one job: mix a share of the theme accent into the neutral text color, then hold the result to a contrast floor. The accent share gives the glyph chroma the text does not have, so it reads as an icon rather than as leftover text, and it is derived from `QPalette::Highlight` rather than a new token. The floor keeps a theme with a weak text color from producing an illegible glyph. Only the neutral label roles (`WindowText`, `Text`) are inked. Semantic tints are hues the call site chose on purpose, disabled uses the disabled group, and the selected-row swap to `HighlightedText` is unchanged. The WCAG contrast math moves from `theme_icon_badge` to the new unit, so badge ink decisions and the floor share one implementation. Tests: the ink rule against every shipped palette, the weak-text floor, and the palette-following render tests now ask the rule instead of restating its output. A preview sheet sweeps the accent share at 0-35% over window and button surfaces for both schemes. * [Icons] Add an emphasis axis for icon ink Tint says which hue, style says how much shape, and nothing said how loud. Call sites that have a hierarchy (a card hover panel's nav strip, the visual database filter toolbar) could only pick between "the theme color" and "another theme color", so "this is the important one" had no way to be expressed and every glyph in a surface arrived with the same weight. Add `IconEmphasis` as the third axis: Quiet, Secondary, Primary. It travels inside `IconTint` instead of becoming a fourth `loadThemedIcon` parameter, so every existing call site keeps working and its look, and emphasis reads as part of the same "how should this glyph be inked" decision as the tint. - Primary lifts a neutral glyph to `QPalette::Highlight`, so the surface's own action color names the primary action instead of a second gray token, and pushes both neutral and tinted glyphs to a 7:1 floor. - Quiet blends toward the surface and stops at the 3:1 WCAG non-text floor, for glyphs that label a control rather than act (the magnifier in a search field, a section marker). - Secondary is the previous color verbatim. The default ink borrows enough accent to sit off the palette's pole (5.4:1 on light, 6.4:1 on dark), so the 7:1 primary floor is measurably above it in both schemes instead of being lost against a near-black or near-white default. The tests pin that order where it exists, and the hue shift still separates the levels at a glance. Disabled glyphs and the Selected -> HighlightedText swap are unchanged at every emphasis level. * [Icons] Render the emphasis sheet for every shipped theme The preview only rendered one theme, so a reviewer could not see that the emphasis levels behave the same way across Fabric, Fusion, Leather, Plasma and VelvetMarble. It now writes one sheet per theme and scheme: the neutral, positive and negative families at all three levels, three glyphs per family, as line, sticker and badge at 16 and 24 px, on the window and button surfaces. * Address review findings	2026-11-08 19:09:01 +01:00
BruebachL	e81ee02c24	[Icons] Tint semantic glyphs and add badge and sticker treatments (#7775) ... * [Icons] Keep semantic red and green on themed glyphs The one-glyph cutover recolored every icon from a single palette role, so a disabled deck, a not-ready player and a remove button all lost the red or green that carried their meaning. Add Positive and Negative application colors beside the accents, ship scheme-appropriate defaults for every theme, and resolve them through ThemeManager. Icons now take an IconTint: a palette role for structural glyphs, or a semantic role for states and add/remove actions. A disabled semantic icon falls back to the disabled foreground; on a selected row a semantic tint keeps its color only while it stays legible on the highlight, and a palette tint keeps the existing WindowText -> HighlightedText swap. * [Icons] Render a semantic-tint sheet for every shipped theme The PR showed two sheets for one theme, which cannot show that the shipped reds and greens stay legible on every palette. Each shipped theme and scheme now gets its own sheet: both families, three glyphs, at 16 and 24 px, on the window and button surfaces the contrast is measured against, with the disabled tone beside each family. The generator lives in its own test file so the preview work does not have to move the branches above it. * [Icons] Add a badge icon treatment Restore the filled-disc look the ready/not-ready icons had before the one-glyph cutover, without bringing back palette-locked art. IconStyle::Badge composes the existing sentinel SVG at paint time: a gradient disc, a dark ring, and the glyph re-weighted and inked for contrast against the tone. Line stays the default and is unchanged. Only the two state icons opt in. The add/remove icons keep the line treatment because their old art was a filled glyph with an outline and no disc, so badging them would invent a look the UI never had. Disabled badges fall back to the line glyph so an unavailable control still reads as unavailable. * [Icons] Add a sticker icon treatment The pre-#7595 add, remove and delete icons were filled shapes with a black outline and no disc. A tint cannot bring that back, and the badge disc is a different icon language, so add a third value to the existing style axis. IconStyle::Sticker draws the sentinel glyph twice: a near-black rim at the band width plus one rim on each side, then the tone pass clipped to the glyph and filled with a vertical gradient. Both treatments keep the disabled fallback to the line glyph. The fill rule moves into theme_icon_fill: the semantic tones are tuned for text, so a filled shape gives its tone a saturation floor, and a dark-scheme tone is pulled down rather than up. IconTint::resolve() applies the same floor to line glyphs, and the badge disc uses the same fill, so a line, a badge and a sticker next to each other carry the same green or red instead of a pale line beside a saturated glyph. The badge's glyph ink follows the scheme: white on light-scheme tones, as in the pre-#7595 art, and the saturated tone darkened to an outline-dark ink on dark-scheme tones, where the shared saturated disc is light enough that white would sit near 2.7:1. The new test pins both cases so the glyph cannot silently flip. resolveSvgStrokeWidth() moves the stroke-width rewrite into libcockatrice_utility so the badge and the sticker share it; the badge's local copy is gone. The sticker is opted into at 17 call sites: increment, decrement and remove_row in the deck editor, settings pages, token dialog, server dialogs and replay/deck storage tabs. The visual database's clear-filter button keeps its trash can (icons/delete) in the line treatment: that glyph is new in #7595, but it is the clearer metaphor, and a sticker rim closes its lid gap at toolbar sizes. The preview generators also gain a semantic-tint sheet for #7775 and a line/badge/sticker family strip, so the three PR descriptions can show the same rendered colors. * [Icons] Render authored theme SVG art without the SVG image plugin `QPixmap("theme:file.svg")` needs Qt's SVG *image-format* plugin. Debian builds that as the separate `qt6-svg-plugins` package, which is only a Recommends of `qt6-svg-dev`, so a `--no-install-recommends` install (CI, minimal user installs) does not have it. Authored theme art then loaded as a null pixmap: the two authored-art render tests failed on Debian 13 and on the Ubuntu sanitize leg while passing on a developer machine whose Qt bundles the format. Render authored SVG assets through QSvgRenderer instead, which is already linked for the sentinel path and needs no plugin. Raster assets keep the QPixmap loader. The test suite now hides the plugin directories, so the dependency cannot come back unnoticed on platforms that ship the plugin. * [Icons] Render a badge/sticker sheet for every shipped theme The PR showed one theme's badge and sticker sheets, which cannot show that the filled treatments read on every palette. Each shipped theme and scheme now gets its own sheet: both semantic families, three glyphs, as line glyphs, stickers and badges, at 16 and 24 px, on the window and button surfaces the contrast is measured against, with the disabled fallback under each family. The generator lives in its own test file so the preview work does not have to move the branches above it. * [Icons] Show the trash can with the sticker treatment again The sticker sheet kept the delete glyph out of the sticker rows and explained why: at toolbar sizes the rim closed the lid gap. The icon is used with the sticker treatment again for destructive actions, so the sheet now shows it and the exception note is gone. * [Icons] Apply the CI clang-format output to the treatment preview clang-format 18 wraps the family label and the glyph x offset differently from the local clang-format 23. The preview test takes the CI output so the bottom of the stack passes the style job on its own. * [Icons] Rewrite the icon role docs in plain language Address review: drop the AI-cadence prose from the IconStyle/IconTint and treatment headers, use ///< for the enum member docs, and move theme_icon.cpp's anonymous namespace above the IconTint definitions. No behavior change. * Address review findings --------- Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>	2026-11-08 19:09:01 +01:00
tooomm	aa0c8dc7cb	[CI] Pin used GH Actions (#7716) ... * pin used actions plus some reordering and harmonization across files * [CI] Link the version string and RNG stub into the test targets The distribution libgtest.a now used by the Linux images makes the linker pull more translation units out of libcockatrice_network than the fetched googletest did. Those units reference VERSION_STRING and the global rng, which several test targets never defined, so the instrumented Ubuntu 26.04 legs failed at link: undefined reference to `VERSION_STRING' undefined reference to `rng' Add the generated version string source wherever a test links the network umbrella, and define the rng stub in the tests that pull the server game classes without declaring it. This matches what the neighbouring server tests already do. A full Debug build and ctest suite in the ci-ubuntu2604 image (system libgtest.a) is green: 841 tests pass. Fixes the nightly failures introduced by #6894. --------- Co-authored-by: Lukas Brübach <44814898+BruebachL@users.noreply.github.com>	2026-11-08 15:45:54 +01:00
BruebachL	627c7a83aa	[Icons] Migrate UI glyphs to palette-token SVGs (#7595) ... * [Icons] Migrate UI glyphs to palette-token SVGs Runtime inversion is not a theming system: it rotates hue, breaks duotone art and mishandles alpha, and a QPixmap snapshot cannot follow a palette change. Redraw the top-level UI glyph set on a 24x24 grid with `currentColor` as the paint token, drop Inkscape/Sodipodi cruft, and delete the icons whose metaphor was duplicated (cogwheel/gear -> settings, scale_balanced -> scales, update.png -> reload). Load the glyphs through `loadThemedIcon` so the engine resolves the token from the palette on every request, and let `themePixmap` use the same resolution for one-shot pixmaps. `loadColorAdjustedPixmap` is gone; every caller now keeps a QIcon, which `SettingsButtonWidget::setButtonIcon` accepts instead of a QPixmap. Content art (mana, counters, countries) keeps its authored colors. Add a render test covering aspect ratio, palette flips, icon modes and high-DPI pixmaps. * [Tests] Keep the icon render test QApplication alive until exit Creating it as a function-local static tears it down before Qt's library statics at exit, and libQt6Network then runs QObject destructors against freed memory — a SEGV the sanitizer leg reports after the tests have already passed. Leak the application instead, the usual pattern for a test binary. * [Icons] Render lock and dropdown glyphs through the icon engine Both generators rendered the sentinel SVG to a fixed height x height pixmap at DPR 1. The pre-migration path rendered larger and let QIcon scale it, so on a 2x display the games-list lock (13px) and the banner dropdown arrow (24px) came back crisp; the new path hands the widget a DPR-1 bitmap that gets upscaled and blurred. - The games list asks the themed engine for `icons/lock` directly, so the item view paints it at the device pixel ratio and the lock also follows HighlightedText on a selected row. - The dropdown goes through `loadThemedIcon(...).pixmap(size, dpr)`, which keeps the glyph's aspect ratio and renders at the requested ratio. - LockPixmapGenerator has no consumer left, so it and its cache are gone. --------- Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>	2026-11-04 18:39:41 +01:00
BruebachL	8822ade49d	[Tests] Register each test case with CTest instead of each test binary (#7399) ... * [Tests] Register each test case with CTest instead of each test binary Registering a test cost four edits across a dozen CMakeLists files: add_test, add_executable, a conditional add_dependencies on gtest, and target_link_libraries. The conditional branch was the trap -- it is only needed when GTest was fetched and add_subdirectory'd, and omitting it yields undefined gtest symbols at link time -- so it was easy to leave out and impossible to notice until the link failed. Fold all four steps into cockatrice_add_test() and migrate every test directory onto it, with a cockatrice_add_manual_test() wrapper for the three benchmarks that are built but deliberately kept out of ctest. The calls now read as a declaration of what a test is rather than a transcript of how to register one. While rewriting the registrations, replace add_test(NAME x COMMAND x) with gtest_discover_tests(). Naming the binary as a single CTest entry meant a CI failure reported only which binary failed, and ctest -R Suite.Case could not select an individual case. Discovery raises the suite from 35 opaque entries to 469 real ones: ctest -R ThemeMigrationTest.IsIdempotent now runs exactly that case, and ctest -L server / -LE bench work off labels applied at registration. Give every test a 60s TIMEOUT. CTest's implicit default is 1500s, so one hung socket wait would stall CI for 25 minutes, and single_instance_manager_test already does real QLocalSocket waits. Dropping per-case granularity costs wall-clock time, since the suite becomes process-spawn bound rather than CPU bound: serial ctest went from 3.8s to 30.8s. Run it with -j in CI to recover it -- 7.6s at -j4, and 3.5s at -j8, both at or below the old serial time. The binaries are parallel-safe: the only test claiming a fixed resource scopes its local socket name to its own pid, so concurrent copies cannot collide. Also delete dummy_test, which asserted 1 == 1 and only proved the harness was wired up -- something the other 468 cases already establish. No test is added, removed, or changed by this commit; the 469 cases are the same ones the 35 binaries contained. * [Tests] Make QtTest available to test targets QtTest was not among the modules test targets could link, so the Qt model and signal contracts had no way to be verified from a test: QAbstractItemModelTester and QSignalSpy both live there. Add Test to _TEST_NEEDED in cmake/FindQtRuntime.cmake. That is the single place the union of test-only Qt modules is declared, so this is the only file that needs to change to make the module reachable from every test target. Nothing uses it yet. This lands the capability ahead of the model-layer tests that need it, so those can be added as tests alone rather than as test plus build-system plumbing. * [Tests] Correct the note on the oracle test compression defines add_definitions() populates the directory's COMPILE_DEFINITIONS property, which is consumed at generate time, so it reaches every target in the directory no matter where it is called relative to add_executable(). The previous note claimed the flags stayed confined to the benchmark because the find_package block came after it; oracle_importer_test, created above that block, carries both defines too. Verified harmless rather than a live bug: HAS_ZLIB and HAS_LZMA gate code only in oracle/src/pages.cpp and oracle/src/pagetemplates.cpp, and no test target compiles either. The ordering and the leaked defines are both pre-existing, so this only corrects the explanation. * [Tests] Carry one CTest label per test instead of several gtest_discover_tests() can hand CTest exactly one label. GoogleTestAddTests.cmake accumulates its generated script with `foreach(arg ${ARGN})`, so a label list reaches the CTestTestfile as bare `LABELS ci server`, and CTest reads only the first value of a property and drops the rest without a word. The `ci` label that is appended to every registered test was therefore absent from every test that also asked for `server` or `bench`: `ctest -L ci` reported 400 of 469 cases, and `ctest --print-labels` never listed a second label anywhere in the build. Reduce each test to exactly one of `ci`, `server`, or `bench`, which is all the selections need. The three now partition the suite exactly, 400 + 64 + 5 = 469, and `ctest -LE bench` still means everything but benchmarks. A second label is now a configure-time FATAL_ERROR rather than a line that quietly means something else. Anything orthogonal to a test's kind belongs in a helper flag beside the label, so it composes with the group instead of competing for the single slot. Also raise DISCOVERY_TIMEOUT off GoogleTestAddTests' 5s default. A plain debug binary answers --gtest_list_tests in 0.04s, but that budget is thin once several binaries discover concurrently behind a -j build, and being over it fails at the *link* step with a bare "Process terminated due to timeout" that points nowhere near the cause. * [Tests] Link gtest_main on the fetched-gtest path The downloaded-gtest branch set GTEST_BOTH_LIBRARIES to `gtest` alone, while FindGTest sets it to both halves. Every test target links that list rather than picking an entry point, so on this path nothing provides main() and any test source that does not define its own fails at link with `undefined reference to main` pointing at crt1.o, naming neither gtest nor the line responsible. That stayed latent while every test carried its own four-line main(). A test that omits it now fails, and the error points at the C runtime rather than at gtest, which is a long way from the cause. Restoring both halves makes the two paths agree and drops the boilerplate requirement on new test sources. The 18 sources that still define main() keep linking: gtest_main.cc defines nothing but main, so the linker leaves its object in the archive unused rather than reporting a duplicate. * [Build] Run the test suite under ASan and UBSan (#7400) * [Build] Add an ENABLE_SANITIZERS option for ASan and UBSan Adds -DENABLE_SANITIZERS, off by default, building with AddressSanitizer and UndefinedBehaviorSanitizer. The bulk of the work is in the three ways the obvious implementation silently does the wrong thing. Probe the flags on the link step, not just the compile step. check_cxx_compiler_flag() only compiles, so every sanitizer flag comes back unsupported, because linking an ASan object without -fsanitize=address leaves __asan_init undefined. The result is not a build failure but a build that appears instrumented and is not, which is worse. Use check_cxx_source_compiles() with CMAKE_REQUIRED_LINK_OPTIONS instead, and treat an unsupported sanitizer pair as fatal since the cosmetic flags are allowed to degrade. Opt vendored sources out per source, through a cockatrice_disable_sanitizers_for() helper rather than a list of set_source_files_properties() calls at the root. The helper has to exist because CMake resolves a source-file property against the directory scope its target was declared in: setting the property from the root with absolute paths is a no-op, silently, which is why the call sites sit next to the add_library and add_executable that pull the vendored code in. SFMT, minizip, the LZMA SDK and Qxt SMTP are excluded. oracle/src/qt-json is in the history but not in the tree, removed in #7214. Also mark those sources SKIP_PRECOMPILE_HEADERS. The PCH is built with the sanitizer flags, so it defines __SANITIZE_ADDRESS__; a translation unit compiled with -fno-sanitize=all does not and GCC then refuses the PCH, which is a fatal -Werror=invalid-pch in the servatrice and oracle builds. -fno-sanitize=vptr is not optional. Qt constructs QObject subclasses into raw storage and writes the vptr afterwards, so the vptr check reports on essentially every QObject instantiation, and it has no value here because nothing outside moc-generated code inherits QObject. Verified on the resulting build: our own translation units carry ASan and UBSan calls, SFMT.o carries neither, and the full suite of 464 cases passes clean under ASAN_OPTIONS=detect_leaks=0:abort_on_error=1 with UBSAN_OPTIONS=halt_on_error=0. The test suite needs two adjustments to make this option usable at all, both of which land here because they are consequences of COCKATRICE_SANITIZERS existing rather than independent changes. DISCOVERY_TIMEOUT rises from 60s to 300s under sanitizers. An instrumented binary takes 2.8s to answer --gtest_list_tests against 0.04s uninstrumented, and discovery runs several binaries at once behind a -j build. One target is excluded via a NO_SANITIZE flag, server_developer_role_test. Reaching google::protobuf::TextFormat::Printer's destructor aborts on an absl assertion inside system protobuf 33.4 whenever the consumer translation unit is built without NDEBUG, which a Debug sanitizer build is. This reproduces with no Cockatrice code at all, just "TextFormat::Printer p; p.PrintToString(msg, &out);" built with -fsanitize=address, so it is upstream rather than ours. It is a flag and not a fourth label because "skip this in the sanitizer leg" has to compose with a test's group rather than compete for the single label slot, and because a flag excludes by not registering the cases at all, which cannot rot the way a label silently can. The target still builds and still runs on the legs without sanitizers. * [CI] Run the suite once under ASan and UBSan Adds --sanitize to .ci/compile.sh, mirroring --test, and turns it on for one entry of the existing Linux Docker matrix: Ubuntu 26.04, which already had a test-enabled leg. That reuses the image, the ccache layer and the existing step shape, so the CI diff is a matrix key plus the argument plumbing. Debian 13 and Fedora 44 keep covering the uninstrumented suite, so nothing is lost by instrumenting one leg. Sanitizers stay off the release and package legs, which build with --release for packaging: instrumenting those would change the shipped binaries' optimization profile and build time for no benefit. macOS and Windows stay uninstrumented too, since ASan is unreliable against the Qt build there and MSVC has no equivalent without /fsanitize=address. --sanitize does not imply --test, so an instrumented build can be produced locally without paying for ctest. When both are given, ctest runs with detect_leaks=0 because Qt and the sanitizer runtimes both leak at exit by design and LeakSanitizer's default buries real findings under hundreds of process-teardown reports, and with UBSan halt_on_error=0 so one run collects many findings instead of stopping at the first. ASan keeps abort_on_error=1, since a bad access leaves the process unusable and continuing only produces a second, more confusing report from the wreckage. * [Build] Fix sanitizer flag leak into release package and make UBSan fail CI - compile.sh always passes -DENABLE_SANITIZERS explicitly so the release-package reconfigure in the same build dir cannot inherit ON from the sanitized debug step (which shipped libasan/libubsan DEB deps). - Sanitizers.cmake resets COCKATRICE_SANITIZERS to OFF when the option is off (CACHE INTERNAL otherwise sticks forever, keeping NO_SANITIZE skips and the 300s discovery timeout). - UBSan now fails the job: -fno-sanitize-recover=undefined at compile time plus UBSAN_OPTIONS halt_on_error=1 (recover mode exits 0, so the old halt_on_error=0 could never fail). --------- Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de> --------- Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>	2026-10-11 14:50:09 +02:00
BruebachL	d603a2bb6a	[Tooling] Apply include-what-you-use across the tree (#7396) ... * [DeckList] Apply include-what-you-use to libcockatrice_deck_list * [IWYU] Apply include-what-you-use across the tree Run one whole-tree IWYU pass over the libcockatrice_* libraries, the Cockatrice client, Oracle, Servatrice and the test suite, so each translation unit includes what it uses directly and forward-declares what it does not need. IWYU only analyses real translation units, never the moc code CMake generates from Q_OBJECT headers, so it strips includes that moc output silently relies on. Where a type is passed by value (or nested in a QList/QMap) in a signal or slot signature, the generated moc code needs the complete type; those includes are kept with an explanatory `// IWYU pragma: keep`. The same holds for Q_DECLARE_METATYPE targets such as QSqlDatabase. A handful of includes are added by hand where callers brace-initialize a const-reference parameter, which also requires the complete type. Applied as a single commit on purpose: stripping includes from the libcockatrice_* headers is precisely what breaks the client, test and server translation units, and the fixes for those live in the same translation units. Splitting by component would therefore leave intermediate commits that do not compile. All 426 modified Q_OBJECT/Q_GADGET headers were verified to be moc-self-contained by compiling each generated moc_*.cpp standalone; several built only because of moc include ordering in mocs_compilation.cpp before the keep-includes were added. * [Tooling] Emit public Qt headers and fix IWYU fallout CI found Three problems the local build could not see, all because a green local build only proves the configuration it was run under. Private Qt headers. IWYU emitted private, lowercase Qt headers, and qtypes.h does not exist at all before Qt 6.9, so the Debian 12 and Ubuntu 24.04 jobs on Qt 6.4.2 failed to compile. The generator only emitted a mapping when the capital header name mirrored the private one (QAbstractAnimation for qabstractanimation.h), so qtypes.h, which is wrapped by QtTypes instead, got no mapping and kept the private spelling. Map it explicitly to <QtTypes>, which is public since Qt 5 and provides the same typedefs either way, and rewrite the 100 affected includes. Do not generalize this to "prefer any Qt* wrapper". The table is generated from whichever Qt is installed, and the wrappers are not equally old: QtLogging only appeared in Qt 6.8 and QtMinMax in 6.5, so that heuristic trades one breakage for another. Every other lowercase header still in the tree has shipped publicly since Qt 5, so those are left alone. Platform-guarded includes. IWYU only sees the preprocessor configuration it is handed, so it dropped an include that only the macOS branch of main.cpp needs, and the macOS jobs failed on an undeclared UrlSchemeEventFilter. Restore the include. Force-compiling every file that carries a platform guard under Q_OS_MAC, Q_OS_MACOS and __APPLE__ reports nothing else, so the macOS branches are covered rather than assumed. shellcheck. Drop a leftover variable and stop assigning the include-what-you-use lookup, which was only ever an existence check. * [Tooling] Map Qt 6.5+ split headers to QtGlobal for Qt 6.4 compatibility IWYU emitted QtTypes, QtPreprocessorSupport, QtVersionChecks, QtAssert, QtMinMax, QtClassHelperMacros, QtEnvironmentVariables and QtTranslation. Qt 6.5 split those declarations out of qglobal.h into separate headers, so neither the Qt* wrapper nor the q*.h private name exists on the oldest Qt we support (6.4.2), and Debian 12 / Ubuntu 24.04 fail with "No such file or directory". qtypes.h is worse still: it only appeared in Qt 6.9. QtGlobal is the one spelling that works across the whole supported range. It has been public since Qt 5, defines these declarations itself up to Qt 6.4, and includes the split headers from Qt 6.5 on, so redirect all eight private headers to it and collapse the affected includes in the tree. Curated overrides now take precedence over the name-mirroring rule, which cannot know about Qt version floors because the mapping is generated from whatever Qt is installed. That rule was mapping qtversionchecks.h to <QtVersionChecks> even with an override present, because the override was only consulted as a fallback. A curated key that matches nothing is now a hard error rather than a silent fallback, and the redirected headers are no longer advertised as the home of a symbol. Verified every Qt include in the tree resolves against the real qt6-base-dev 6.4.2 package: 15 distinct includes, 0 missing. * [Tooling] Map QOverload to QtGlobal and drop the QMetaMethodArgument include Two more headers IWYU emitted do not exist on Qt 6.4.2: QOverload was split out of qglobal.h in Qt 6.5, so on 6.4.2 it is declared in qglobal.h directly and the standalone header is a hard error. QtGlobal covers both, so map qoverload.h to it like the other 6.5+ split headers. QMetaMethodArgument is never referenced anywhere in the tree. It is a bare forwarder for qobjectdefs.h that our generator advertised as a class symbol, which is how IWYU came to add the include at all. The symbol mappings are generated from header names, so they cannot tell a forwarder that happens to be capitalised from a real class; removing the unused include is the fix, and the generic case stays a known limitation of the generator. Extending the audit to every Q* header (an earlier pass only matched headers starting with "Qt" or lowercase q*.h, so it missed all of them) against the real qt6-base-dev, qt6-multimedia-dev, qt6-declarative-dev, qt6-svg-dev and qt6-websockets-dev 6.4.2 packages: 341 distinct Qt includes, 0 missing. * [Game] Give moc the complete CardItem type in player_actions.h player_actions.h forward declares CardItem but passes QList<CardItem *> by value in actPlay/actPlayFacedown/actHide. moc instantiates the metatype for those parameters, and on Qt 6.4 that expands to sizeof(CardItem), so Ubuntu 24.04 fails with "invalid application of 'sizeof' to incomplete type 'CardItem'". Newer Qt tolerates the incomplete type, which is why only the 6.4.2 jobs failed. Include the defining header with an IWYU pragma keep, matching how the other MOC-required includes in this area are handled. * [Game] Give moc the complete ServerInfo_Card type player_logic.h, game_scene.h, view_zone.h and view_zone_widget.h forward declare ServerInfo_Card but pass QList<const ServerInfo_Card *> through signal and slot signatures. moc instantiates the metatype for those parameters, which Qt 6.4 expands to sizeof(ServerInfo_Card), so the connect in game_scene.cpp fails on Debian 12 and Ubuntu 24.04 with "invalid application of 'sizeof' to incomplete type 'const ServerInfo_Card'". Newer Qt tolerates it, so only the 6.4.2 jobs failed. Include the generated protobuf header with an IWYU pragma keep, following the pattern already used for ServerInfo_Player and ServerInfo_User in game_event_handler.h. * [Client] Complete the types Qt 6.4 needs for connect and moc Qt 6.4 expands QMetaType::fromType on the normalised argument list of a PMF connect and on moc-generated metatypes, which requires sizeof on every signal and slot parameter. Newer Qt tolerates forward declarations there, so only the Debian 12 and Ubuntu 24.04 jobs failed, with "invalid application of 'sizeof' to incomplete type" for QMouseEvent, QPrinter and GameReplay. Replace the forward declarations with the real includes: - QMouseEvent in the nine widget headers that pass it through signals - QPrinter in abstract_tab_deck_editor.cpp, for the actPrintDeck lambda - game_replay.pb.h in tab_replays.h. GameReplay is a protobuf message, so the project's game/replay.h only forward declares it and cannot supply the complete type. Found by recompiling all 1918 project TUs against the Debian Qt 6.4.2 headers, which reproduces those two CI jobs locally in one pass. * [Client] Complete Release and QMenu for Qt 6.4 moc and connect Continuing the Qt 6.4 complete-type sweep: - window_main.cpp connects ClientUpdateChecker::finishedCheck to MainWindow::checkClientUpdatesFinished, whose Release * parameter made moc 6.4 instantiate the metatype. GameReplay turned out to be a protobuf message, so Release likewise needed release_channel.h in the three headers declaring those signals: window_main.h, dlg_update.h and client_update_checker.h. - moc 6.4 also needs sizeof(QMenu) for the QList<QMenu *> carried by TabSupervisor::setMenu and MainWindow::updateTabMenu. Verified by regenerating moc output with the Debian moc 6.4.2 for all 434 Q_OBJECT/Q_GADGET classes and compiling it, plus recompiling all 1612 source TUs, against the Qt 6.4.2 headers: no remaining failures. --------- Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>	2026-10-08 11:13:50 +02:00
BruebachL	014983b5bc	[Theme] Carry Default theme directory customizations over to System (#7385) ... * [Theme] Carry Default theme directory customizations over to System The Default -> System rename rewrote the stored theme name but left the user profile's themes/Default directory behind, so anyone who had customized zone graphics, stylesheets or palettes there silently lost them. Migrate the directory alongside the name. Rename it when System does not exist yet, otherwise merge file by file without overwriting what System already has, then drop the legacy directory. Users who already ran the rename get their customizations back on the next start, and a hand-made Default folder of a user on another theme is left untouched. * [Tests] Give theme_migration_test an explicit main and gate it on WITH_CLIENT The build machines that cannot find GTest fall back to the downloaded googletest, which links the bare gtest library without gtest_main. Every other test defines its own main() for that reason; this one did not, so Debian 13, Fedora 44 and Ubuntu 26.04 failed to link it. The test compiles client sources, so -DTEST=1 -DWITH_CLIENT=0 has no way to build it. Gate the subdirectory on WITH_CLIENT like the other client-source test. --------- Co-authored-by: Lukas Brübach <Bruebach.Lukas@bdosecurity.de>	2026-10-03 13:43:25 +02:00