Commit Graph

1195 Commits

Author SHA1 Message Date
ac2ab1443c Added fe_apply_theme_to_toplevel() for:
Standard GTK file chooser fallback dialog in src/fe-gtk/gtkutil.c

        Font chooser dialog in src/fe-gtk/setup.c

        Color chooser dialog in src/fe-gtk/setup.c

        About dialog in src/fe-gtk/menu.c
2026-02-25 23:18:19 -07:00
ce5128e4fb Added a theme_name null guard in the Adwaita/Yaru workaround condition inside create_input_style() so g_str_has_prefix() is only called when theme_name is non-null, while preserving existing behavior for non-null theme names.
Left allocation/free flow unchanged; theme_name is still freed exactly once at the existing cleanup point after the reload block.
2026-02-25 22:52:40 -07:00
c37faa1492 Refactored the Preferences color-page edit-source model to use an explicit snapshot pointer (setup_color_edit_source_colors) and a helper that selects light vs dark snapshot palettes, so the page preview is decoupled from the runtime colors[] palette. This affects selector refresh, page initialization, and cleanup lifecycle.
Updated edit-target and dark-mode combo callbacks on the color page to stop applying global palette mode directly; they now only switch/refresh the Preferences page source palette.
Changed color-chooser behavior so setup_color_response_cb() writes edits to the selected target snapshot via palette_user_set_color / palette_dark_set_color, then refreshes only page widgets (no runtime palette apply). Also updated dialog initialization to read from the current edit-source snapshot.
Added palette snapshot accessors (palette_user_colors, palette_dark_colors) in the palette API and implementation so Preferences can render from light/dark snapshots safely without mutating runtime palette state.
Runtime palette changes remain on the real apply paths (theme/dark-mode application), not edit-target toggles.
2026-02-25 22:48:55 -07:00
685989fa25 Updated fe_system_prefers_dark() to remove the old !has_theme_name gate and always read gtk-application-prefer-dark-theme whenever that property exists. Theme-name matching is still kept as an additional signal.
Added deterministic dark-mode precedence by combining explicit signals (theme_name_prefers_dark || property_prefers_dark), while preserving Windows system preference as another dark-enabling signal when available. This ensures “dark if any explicit dark signal is true.”
Re-checked the relevant callers (fe_auto_dark_mode_changed() and fe_apply_theme_for_mode()): they consume fe_system_prefers_dark() output and do not depend on the removed !has_theme_name conditional behavior.
2026-02-25 22:42:43 -07:00
cbc474477b Updated gtkutil_menu_icon_theme_variant() to select the menu icon theme variant from the app’s effective dark-mode state first (fe_dark_mode_state_is_initialized() + fe_dark_mode_is_enabled()), and only use GTK/theme-name heuristics as fallback before app state is initialized. This preserves the existing light-asset fallback behavior in menu icon loading logic.
Added a new frontend helper declaration fe_dark_mode_state_is_initialized() in the GTK frontend header so callers can check whether dark-mode state is ready.
Implemented dark-mode initialization tracking in fe-gtk.c via a new static flag, set when auto mode state is externally set and during fe_init(), and exposed it through fe_dark_mode_state_is_initialized().
2026-02-25 22:31:52 -07:00
51f8795d1a Updated create_input_style() so input selection CSS now prefers palette-derived COL_MARK_FG / COL_MARK_BG colors directly (as #RRGGBB from sel_* values), instead of defaulting to @theme_selected_*.
Added a guarded fallback path that only uses theme lookup/string conversion when palette selection components are invalid (isfinite checks fail), with palette values as secondary fallback there too. This keeps theme colors out of the primary path while preserving robustness for malformed data.
Kept needs_reload dependent on last_sel_* tracking, so changing selection colors in Preferences still triggers immediate CSS regeneration (including across light/dark mode).
2026-02-25 22:25:35 -07:00
faacd95dfc Added setup-local state for palette editing target (setup_color_edit_dark_palette) plus Light/Dark option strings, so palette editing is no longer coupled to runtime AUTO dark-mode detection. This keeps the choice in Preferences state only.
Added a new Editing: combo control near the dark-mode selector in setup_create_color_page, with a callback that switches between Light/Dark palette targets and refreshes swatches immediately using the same refresh path as dark-mode UI updates.
    Updated the dark-mode combo callback to refresh color swatches based on the explicit editing target (not runtime dark-mode detection), preserving existing runtime theme behavior while editing persisted overrides.
    Updated setup_color_response_cb to write color changes to palette_dark_set_color vs palette_user_set_color based on the new edit-target state.
    Initialized the editing target in setup_create_color_page from explicit Dark mode selection only (hex_gui_dark_mode == ZOITECHAT_DARK_MODE_DARK), avoiding AUTO runtime detection for editing context.
2026-02-25 21:28:08 -07:00
ed02b21228 Extended setup_dark_mode_menu_cb so after setup_menu_cb it now computes the selected dark-mode state via fe_dark_mode_is_enabled_for(setup_prefs.hex_gui_dark_mode), swaps the in-memory editor palette with palette_apply_dark_mode(...), and immediately refreshes color swatches. This keeps the editor model synchronized with the dark-mode combo selection.
Added setup_refresh_color_selector_widgets() to iterate color_selector_widgets, resolve each button’s stored color index, and re-run setup_color_button_apply against colors[] so all selectors repaint to the active palette.
    Updated setup_create_color_button to register each widget’s palette index (zoitechat-color-index), enabling correct per-button repaint during mode switches.
    Preserved existing persistence behavior in setup_color_response_cb (palette_dark_set_color vs palette_user_set_color) so changes remain non-destructive until existing OK/Save flow persists them.
2026-02-25 21:22:56 -07:00
d321717da8 Hardened create_input_style() so selection-color CSS no longer hard-depends on @theme_selected_*: it now first tries symbolic theme colors via gtk_style_context_lookup_color(), and only falls back when unavailable.
Added a fallback path that derives selection colors from palette values (COL_MARK_FG/COL_MARK_BG) and then samples selected-state colors from a GTK style context when possible, emitting explicit color strings for CSS.
Ensured fallback regeneration is tied to the same reload flow by extending needs_reload tracking with selection palette channels and persisting those last-seen values.
Kept Preferences > Colors foreground/background behavior intact for normal input rendering; only selection-specific CSS is conditionally overridden. Base text/background/caret rules remain unchanged while only text selection uses the new selection logic.
2026-02-25 20:59:07 -07:00
6310ab245c Updated fe_apply_theme_for_mode() to immediately reapply fe_apply_theme_to_toplevel() across all currently-open GTK toplevel windows (gtk_window_list_toplevels()), so mode switches update existing windows without reopening them.
Added/kept lifecycle call sites for main and detached/channel windows, with comments explaining that .zoitechat-dark / .zoitechat-light classes are required for GTK CSS selectors. This includes top/tab windows and server list/edit windows.
Wired dialog/toplevel creation paths in src/fe-gtk/ to call fe_apply_theme_to_toplevel() (without duplicating platform logic), including common prompt/message dialogs and feature-specific dialogs (join, notify, setup, ignore/ban, etc.), each with the requested brief comment near the new call site.
2026-02-25 20:39:48 -07:00
5952006662 Refactored fe_system_prefers_dark() to always allow gtk-application-prefer-dark-theme to be read as a fallback (when no explicit theme name is set), removing the previous suppression tied to app-written state. This preserves priority for explicit dark theme names and platform-native preference checks before the GTK property fallback.
Added a short in-code comment explaining why app-written GTK property state must not permanently block future reads in AUTO mode.
Removed the now-unneeded app_set_prefer_dark tracking behavior from fe_set_gtk_prefer_dark_theme(), so writes no longer disable later fallback reads.
Confirmed fe_auto_dark_mode_changed() still prevents reapply loops by early-returning when enabled == auto_dark_mode_enabled before calling theme/apply logic.
2026-02-25 20:32:44 -07:00
e4cb453915 Added a centralized theme-class application in gtkutil_window_new() so any toplevel created through this helper gets fe_apply_theme_to_toplevel() during construction. This covers most non-session windows consistently at creation time.
Updated Preferences window creation to explicitly apply the toplevel theme after gtk_widget_show_all() in setup_window_open(), matching your requested timing there.
Added explicit theme application for non-session windows that bypass gtkutil_window_new() and use direct gtk_window_new() in server list UI (editserv and servlist).
Extended mg_apply_setup() to iterate all current GTK toplevels (gtk_window_list_toplevels()) and reapply fe_apply_theme_to_toplevel() during setup/theme reapply flows, ensuring existing windows stay consistent after theme changes. Palette override widgets were not touched.
2026-02-25 20:27:15 -07:00
361e35de7f Updated chanview_apply_theme() in src/fe-gtk/chanview.c to apply the palette colors unconditionally (for tree view chanview), instead of branching on raw dark-mode enum checks. This keys behavior off the already-resolved palette output, which matches how setup_apply_to_sess() / palette_apply_dark_mode() flows are intended to work.
Preserved font handling exactly as before (input_style->font_desc when available).
Added inline rationale comment clarifying that AUTO/light should continue using palette-managed colors and should not revert to GTK theme defaults in a way that clears custom colors.
2026-02-25 20:23:01 -07:00
bbde2e5578 Added a new internal state flag, app_set_prefer_dark, and now set it whenever fe_set_gtk_prefer_dark_theme() writes gtk-application-prefer-dark-theme, so AUTO detection can distinguish app-authored writes from system/user signals.
Reworked fe_system_prefers_dark() to prioritize non-app-written sources in AUTO flow:

        first: gtk-theme-name dark-variant heuristic,

        then (Windows): native platform detection via fe_win32_try_get_system_dark,

        only then fallback to gtk-application-prefer-dark-theme when no theme-name signal was available and the app has not written that property.

    Kept manual LIGHT/DARK behavior unchanged by preserving the existing mode application path (fe_apply_theme_for_mode() still drives fe_set_gtk_prefer_dark_theme() directly).

    Verified fe_auto_dark_mode_changed() and fe_init() signal wiring still trigger AUTO palette/theme refreshes on system theme changes (notify::gtk-theme-name and existing notify callback flow remain in place).
2026-02-25 20:03:20 -07:00
97c6f36b20 Adjusted create_input_style so it no longer globally forces color on #zoitechat-inputbox, reducing over-constraining while still applying the user-selected background and caret color where intended. This keeps non-text-node state rendering more theme-driven.
Kept user override support for baseline text color and caret color on the input’s text node (#zoitechat-inputbox text).
Added targeted state-aware text rules for :focus and :backdrop (to preserve readability when theme state styling changes), plus a softened :disabled text rule instead of broad global color forcing.
Added explicit selection-node handling (#zoitechat-inputbox text selection) using theme selection tokens so selected text remains readable and aligned with Adwaita/Yaru-style theme behavior.
Preserved the existing Adwaita/Yaru background-image workaround logic, so fallback behavior remains compatible with those themes.
2026-02-25 19:43:19 -07:00
deepend-tildeclub
2bf32fc770 Merge pull request #90 from ZoiteChat/code-cleanup
Code cleanup
2026-02-25 18:16:54 -07:00
3e095c3c33 Removed two redundant inline comments in the file dialog response/done handlers, leaving the behavior unchanged and the control flow cleaner.
Removed an old commented-out helper function block (gtkutil_treeview_get_selected_iter) that was dead/commented code.
2026-02-25 18:07:26 -07:00
3445d946bc Removed a redundant inline comment in icon loading/scaling logic, leaving behavior unchanged.
Removed a redundant inline comment in the Windows sysinfo volume-name guard path, with no functional impact.
Removed a redundant inline comment in WinRT toast setup while keeping the same notification/audio attribute behavior.
2026-02-25 18:05:21 -07:00
e3f932aa6b Updated the Win32 window-state handling to include shellapi.h and query taskbar auto-hide/taskbar edge state via SHAppBarMessage, so maximized windows can leave a 1px reveal edge for the auto-hidden taskbar.
Added monitor-aware work-area adjustment when maximized (MonitorFromWindow + GetMonitorInfo) and shrink by 1px on the detected taskbar edge (including bottom edge), then apply that rectangle with SetWindowPos. This addresses the “mouse at bottom edge should unhide taskbar” behavior.
Kept the existing HWND_NOTOPMOST safeguard in place after the adjustment, preserving prior not-topmost behavior.
2026-02-25 14:48:18 -07:00
7cfa3b3513 Added a Windows-only helper (mg_win32_allow_autohide_taskbar) that gets the native HWND and calls SetWindowPos(..., HWND_NOTOPMOST, ...) for non-fullscreen window-state transitions, so a maximized ZoiteChat window no longer blocks the taskbar auto-hide reveal edge.
Hooked that helper into mg_windowstate_cb so it runs whenever the main window state changes (while preserving existing maximize/fullscreen preference handling).
2026-02-25 14:18:23 -07:00
81c691491c Removed three redundant inline comments in joind_ok_cb that duplicated obvious control flow, leaving behavior unchanged and improving readability. 2026-02-25 13:55:44 -07:00
6a02f578b6 Removed obsolete commented-out lines from fe_cleanup() so the function no longer contains dead commented code.
Removed an unnecessary explanatory comment above the unconditional early return in log_handler(), leaving only the active behavior.

Removed an unneeded section marker comment before fe_idle() to keep surrounding code cleaner.
2026-02-25 02:34:50 -07:00
fd48c97209 Removed four redundant inline comments in src/fe-gtk/urlgrab.c to clean up readability without changing behavior (treeview enum declaration area, right-click selection block, headers visibility call, and URL list overflow trimming block). 2026-02-25 02:09:35 -07:00
b3a0288fd9 Cleaned up stale inline comments in the tab scrolling handlers by removing outdated explanatory text while keeping behavior unchanged (tab_left_is_moving / tab_right_is_moving reset paths).
Removed an old commented-out gtk_container_set_border_width(...) line that was dead/commented code in cv_tabs_init.
Simplified the tab-switch condition by removing a legacy commented expression from if (is_switching).
2026-02-25 02:01:11 -07:00
bb11c4676b Cleaned up src/fe-gtk/meson.build by removing an outdated commented-out WinRT build block in the Windows section, leaving only active build logic.
Removed an inline TODO comment from the gnome.compile_resources call to keep the Meson configuration concise and focused on current settings.
2026-02-25 01:47:32 -07:00
ef2befa698 Cleaned up redundant comments in src/fe-gtk/fkeys.c by removing section-divider noise and a stale XXX line near key-action helpers, while keeping useful function-level documentation.
Removed repetitive inline comments from return 2; statements in tab-move/history key actions, since behavior is already clear from the surrounding code.
Removed an unnecessary separator comment before replace_handle() for a cleaner transition between functions.
2026-02-25 01:35:30 -07:00
6c9aeac36f Cleaned up unneeded inline comments in swtype by removing index/reserved annotations that didn’t add value, while keeping the data unchanged.
Removed a stale commented-out ST_HEADER line from tabs_settings so the settings table only contains active entries.
Removed a dead commented-out call from setup_apply_trans to reduce noise in the function body.
Removed redundant trailing declaration comments from proxy_user and proxy_pass since the types are already explicit.
2026-02-25 00:55:37 -07:00
48f7693eac Removed redundant comments in the notification plugin implementation to clean up the code without changing behavior (including comments around color stripping, alert gating logic, and special-event hook registration). 2026-02-25 00:53:10 -07:00
c4951a06f3 Cleaned up unneeded commented-out debug lines in maingui.c (old printf/puts and an obsolete commented call), without changing behavior or logic flow.
Committed the cleanup on the current branch with commit message: Remove stale debug comments from maingui.
2026-02-25 00:29:52 -07:00
618aecd349 Fixed a crash-prone memory management path in Linux URL launching by only freeing the previous environment vector when g_environ_unsetenv() returns a different pointer. This avoids freeing memory that may still be owned/returned as-is by GLib, which matches your segfault symptoms when opening links from the AppImage build.
The fix is in fe_open_url_inner() right where LD_LIBRARY_PATH and LD_PRELOAD are sanitized before spawning xdg-open.
2026-02-24 20:29:48 -07:00
3ed66f9945 Updated Linux URL-opening logic in fe_open_url_inner to sanitize runtime linker env vars (LD_LIBRARY_PATH, LD_PRELOAD) before launching external URL handlers, which is aimed at preventing /bin/sh symbol lookup failures in AppImage-like environments.
Changed the flow to prefer xdg-open first (with sanitized env) and only then fall back to gtk_show_uri, instead of trying GTK first under a potentially broken runtime environment.

    Added explicit final warning when both mechanisms fail, and ensured temporary allocations (including xdg_open_path) are released cleanly.
2026-02-24 20:22:54 -07:00
4284cb76c1 Updated Linux URL opening fallback in fe_open_url_inner to launch xdg-open with a sanitized environment instead of inheriting AppImage runtime linker vars. Specifically, LD_LIBRARY_PATH and LD_PRELOAD are removed before spawning, which prevents host /bin/sh and related tools from loading incompatible bundled libraries (the root cause of the rl_trim_arg_from_keyseq symbol error).
Kept the existing behavior of trying gtk_show_uri first and only using xdg-open as fallback; this change only hardens the fallback path and preserves existing flow/logging.
2026-02-24 20:13:48 -07:00
b7693e75ec Updated Linux URL launching in fe_open_url_inner to capture gtk_show_uri errors instead of discarding them, so failures are no longer silent.
Added an xdg-open fallback path when gtk_show_uri fails, which improves behavior in AppImage/sandbox-like environments where GTK/GIO handler resolution can fail.
Added warning logs for both the primary gtk_show_uri failure and fallback failure to make future debugging much easier.
2026-02-24 19:36:36 -07:00
deepend-tildeclub
118ac25059 Change network name check from 'Libera.Chat' to 'Zoite' 2026-02-24 14:22:01 -07:00
b7804fd7eb Fixed the right-click nick info refresh flow in fe_userlist_update() so the submenu can keep updating after the first refresh when WHOIS-derived data is still incomplete. It now tracks a needs_refresh condition and reattaches the submenu show callback when required. This ensures user info can continue to refresh as data arrives instead of stopping after one update. 2026-02-24 08:35:23 -07:00
a79d34bcc0 cleaned up some gtk2 specific comments 2026-02-23 12:42:44 -07:00
eebe11b95b I removed the leftover GTK2 compatibility include path from src/fe-gtk/xtext.h by dropping the GTK_MAJOR_VERSION < 3 / gtkobject.h block, leaving the GTK3 header path (<gtk/gtk.h>) in place. 2026-02-23 12:36:33 -07:00
1b66b136dc Removed the obsolete GTK3 compatibility compile define from the GTK frontend build flags by deleting -DHAVE_GTK3 in src/fe-gtk/meson.build. 2026-02-22 15:31:51 -07:00
4b188192f8 Removed GTK2 compatibility shims from fe-gtk.h by deleting the local HAVE_GTK3 fallback derivation, deleting GTK2 no-op wrappers for expansion APIs, and making InputStyle GTK3-only.
Simplified the Windows theme block in fe-gtk.c from a GTK3-conditional compile guard to a direct Windows guard (#ifdef G_OS_WIN32), matching the GTK3-only codebase direction.

Simplified native file chooser guards in gtkutil.c from WIN32 && HAVE_GTK3 to #ifdef WIN32, removing remaining GTK3 compatibility-condition clutter in that file.
2026-02-22 15:27:28 -07:00
1c5bf9d6b1 Removed GTK2/GTK3 dual-path preprocessor branches across src/fe-gtk by keeping GTK3-native paths only, including the high-density targets you prioritized (xtext.c, setup.c, maingui.c, gtkutil.c, servlistgui.c).
Removed GTK2 compatibility shims from headers (fe-gtk.h, gtkutil.h), leaving direct GTK3-facing declarations/includes only (and platform-specific guards unrelated to GTK2 kept as-is).

Cleaned GTK2-specific comments/wording now that behavior is GTK3-native (menu accelerator note and emoji/userlist comment blocks).
2026-02-22 15:11:35 -07:00
52f547ae13 Deprecated the gtk3 Meson option and effectively forced GTK3 usage (value: true, deprecated: true), plus updated option/help text so users no longer expect GTK2 frontend builds. Also clarified the AppIndicator option text for non-Windows GTK frontend builds.
Updated src/fe-gtk/meson.build to always resolve gtk_dep from dependency('gtk+-3.0', version: '>= 3.22'), removed the GTK2 fallback branch and -DHAVE_GTK2, and kept AppIndicator detection guarded under host_machine.system() != 'windows' as requested. The GTK X11 helper dep is now unconditionally checked within that same non-Windows block (no gtk3 toggle).

Updated Meson summary output to explicitly report the GTK frontend as GTK+ 3.22+ when enabled, reinforcing that GTK2 builds are no longer expected.
2026-02-22 14:42:20 -07:00
deepend-tildeclub
e58d128c73 Implement fallback for tray icon resolution
Add fallback for tray icon if desktop-id fails to resolve.
2026-02-22 14:54:18 +00:00
acb8d2f539 Investigated the XFCE tray issue and confirmed the code path depends on build-time AppIndicator availability: GTK3 uses AppIndicator/StatusNotifier only when Ayatana/AppIndicator headers/libs were present at compile time; otherwise it falls back to GtkStatusIcon behavior. That fallback is less reliable across XFCE setups, which explains why one user could miss the tray icon.
Added a new Meson feature option appindicator (default auto) so this backend is explicitly controllable.

    Updated GTK Meson logic to honor -Dappindicator=... and fail fast if enabled is requested but neither ayatana-appindicator3-0.1 nor appindicator3-0.1 is found (prevents silently shipping a build without AppIndicator tray support).

    Updated AppImage CI configure flags to force -Dappindicator=enabled, which should prevent AppImage artifacts from lacking StatusNotifier/AppIndicator tray support in XFCE-like environments.
2026-02-22 06:58:24 -07:00
a459d0a086 fixed docs url, removed extra help menu. 2026-02-21 19:17:31 -07:00
0796b96908 Added a Win32 existing-instance fallback so --existing with no URL/command now sends a dedicated __WIN32_TASKBAR_TOGGLE__ command to the running ZoiteChat instance, instead of doing nothing. This gives taskbar/shortcut invocations an explicit toggle action.
Updated the Win32 WM_COPYDATA handler to recognize that toggle command and perform minimize/restore behavior:

visible window → iconify

hidden window → show/present
while preserving existing behavior for normal remote commands.
2026-02-18 13:28:17 -07:00
e3f624d2bb Fixed the GTK3 window-state handler so minimizing via taskbar no longer immediately deiconifies the window again. Specifically, I removed the conflicting gtk_window_deiconify(wid) call from the iconify path in mg_windowstate_cb, while preserving the existing tray-hide behavior and window state bookkeeping. This restores expected toggle behavior when clicking the taskbar icon on a maximized window. 2026-02-18 13:11:04 -07:00
7568c73062 Updated the Network List Connect button icon on GTK3 to use the bundled zc-menu-connect asset, which avoids theme-dependent missing network-connect behavior.
Added a Help menu _Update entry with a dedicated zc-menu-update icon constant, so Windows builds can display an appropriate update icon in the Help menu.

    Extended icon mapping/fallback logic so emoji/update icons resolve correctly across stock/icon-name paths (zc-menu-emoji, zc-menu-update) including GTK2 stock fallback compatibility.

    Added zc-menu-emoji to the input entry fallback icon list so the emoji affordance has a deterministic built-in fallback when theme emoji icons are missing.

    Added new update and emoji icon assets in both SVG and PNG for light/dark variants, and registered the PNG resources in the gresource manifest for runtime loading.
2026-02-18 01:30:45 -07:00
6ce7de23a3 Switched menu resource loading to use bundled PNG assets (instead of SVG paths) for zc-menu-* icons, so menu icon resolution is consistent and no longer depends on SVG/icon-theme availability on different platforms (including Win32 builds).
Updated the GResource manifest so all menu icons now come from data/icons/menu/{light,dark} PNG files and are embedded with to-pixdata preprocessing, ensuring the app uses one unified internal icon set across platforms.

    Generated and added PNG assets for both light and dark menu icon variants under data/icons/menu/ to back the new resource paths used by the GTK menu code.

    Fixed menu icon resource detection to fall back from PNG to SVG when probing zc-menu-* entries, so menu item icon lookup no longer fails just because one format is unavailable at runtime.

    Updated GTK menu icon loading to use gdk_pixbuf_new_from_resource (instead of ..._at_scale) and added explicit fallback order across variant/light and PNG/SVG resources, which addresses the “red no-entry square” missing-icon behavior you reported.

    Kept compatibility with your previous menu resource setup while making loading more resilient on Win32 and other environments where resource/pixbuf behavior differs.
2026-02-18 01:16:03 -07:00
dbc517de99 Updated channel list icon button creation to always use gtkutil_image_new_from_stock(..., GTK_ICON_SIZE_MENU) so it follows the same zc-menu-* icon resolution path on all GTK targets instead of mixing direct stock-image calls.
Updated the GTK2 channel list context-menu icon path to use the same helper, removing direct gtk_image_new_from_stock usage for menu icons.

    Updated the GTK2 spell-entry popup icon path to use gtkutil_image_new_from_stock, aligning it with the same menu icon set/mapping pipeline used elsewhere (data/icons/menu).
2026-02-18 00:57:58 -07:00
e214c76bdf Unified GTK3 menu icon resolution so menu-sized icons now consistently map to the bundled zc-menu-* icon set (from data/icons/menu) even when callers pass freedesktop icon names (like edit-copy, go-next) instead of legacy stock IDs. This removes the mixed behavior where some items came from system themes while others came from bundled resources.
Added a new icon-name→zc-menu-* mapping helper and wired it into gtkutil_image_new_from_stock() as a fallback path for GTK_ICON_SIZE_MENU, after the existing stock-name mapping.
2026-02-18 00:52:49 -07:00