85aa85a35767dc7f171a02636747de9a4bcd77ae
19
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 admin
|
85aa85a357 |
The GLES backend keeps one glyph atlas per GL context, split into pages, so a surface that is created again does not rasterise and upload its text again
Every GLES canvas and sub-canvas had a glyph atlas of its own, a 2048×2048 texture created empty with the canvas, so every surface that was torn down and built again, and every scroll viewport inside it, rasterised and uploaded all of its glyphs on its first frame. crustace's app launcher pays this on every opening: the sliding subsurface and its scroll sub-canvas are destroyed when the launcher closes, and the interactive overlay that replaces the subsurface at the top has scroll sub-canvases of its own. With `LTK_PERF_RASTER=1` on a PinePhone Pro (RK3399, Mali-T860, panfrost) the first raster of the subsurface took 175 to 330 ms of CPU and the first full raster of the overlay 175 to 260 ms, on every opening, with the GPU done 3 ms later; the user saw a delay of two or three tenths of a second before the launcher followed the finger and a stall of about one more when it reached the top. On a Librem 5 (i.MX8MQ, GC7000L, etnaviv) the same rasters took 18 to 40 ms. Panfrost makes each glyph upload expensive. The glyphs of a line are written into the atlas with `glTexSubImage2D` after earlier lines were drawn from it, and when a pending batch still samples a texture that is written to, Mesa 25.0 allocates a new buffer for the texture and copies the whole old one into it with the CPU (`panfrost_ptr_map` in `pan_resource.c`), 4 MiB for the old atlas, read from write-combined memory. Etnaviv flushes and waits for the GPU instead (`etna_transfer_map`), whose cost does not depend on the size of the texture. The new `gles_render/glyph_atlas.rs` holds one `GlyphAtlas` per GL context, found through `shared_glyph_atlas` as the image cache is through `shared_image_cache`, and every canvas and sub-canvas keeps an `Rc` to it, so glyphs rasterised by one surface are resident for the next. The context is held as a `Weak`, so the atlas of a context that was dropped is forgotten instead of being found again by a new context allocated at the same address. The atlas is made of 512×512 pages, created as glyphs need them, up to 8 MiB in all; a glyph larger than a page gets a page of its own, up to the 2048 px of the old atlas, and those pages are freed when the atlas is reset, the regular ones kept and rewound. A write into a page that a pending draw samples then costs panfrost a copy of 256 KiB, not of 4 MiB. The glyph cache moves into the atlas with `GlyphEntry`, which records the page of each glyph, and its key; the cap of 8192 glyphs before a reset is now for the context as a whole, not for each canvas. `draw_text` uploads through the atlas and groups the quads of a line by page, one draw call per page instead of one per line, which only differs when the glyphs of a line are split across pages. `GlesCanvas::new` and `sub_canvas` no longer create an atlas texture each, and dropping a canvas no longer deletes one. `docs/architecture.md` says that glyphs are cached per GL context too and how the atlas is paged. Measured with `LTK_PERF_RASTER=1`, opening and closing the launcher a dozen times. On the PinePhone Pro the subsurface raster of the first opening went from 330 to 95 ms and of the later ones from 175–215 ms to 2–10 ms, with peaks of 17 and 35 ms, and the first raster of the overlay from 175–260 ms to under 1.5 ms; the other frames stay at 2 to 13 ms. On the Librem 5 the subsurface raster went from 22.7 ms on average, 85.6 ms on the first opening and over 15 ms in 9 of 14 openings, to 13.2 ms, 69.7 ms and 3 of 14, and the other frames from 7.7 ms on average and 43 ms at most to 6.8 ms and 28.5 ms. Not done. The first raster of a session still uploads its glyphs interleaved with the draws that sample them; uploading all of them before the first draw would need a second pass over the view on every frame. There are no tests for the atlas. The image cache still identifies its context by address only. |
||
|
admin
|
3edd822918 |
Toplevel events name the window, apps hear the pointer hover a surface and where their blurred cards were laid out, and rows can align to the bottom
A shell could not tell the windows of an app apart, follow the mouse over one of its overlays, or tell its compositor where the translucent cards of a surface ended up. crustace's floating dock needs all three, to activate and close single windows, to magnify the icons under the pointer and to have forge blur behind its pill and its Quick Settings cards.
`ToplevelEvent::Opened` carries the handle's `identifier`, the compositor's token for that one window, and its `title`, both from `ForeignToplevelInfo`; `toplevel_opened` builds the event for the new and the updated handle. Matches on `Opened { id, app_id }` have to add the fields or `..`.
`App::on_pointer_hover( SurfaceTarget, Option<( x, y )> )` is called on every mouse motion over a surface, in its physical pixels, and with `None` from the pointer's `leave`, which ltk did not handle before. `SurfaceFocus::target` gives the `SurfaceTarget` of a focus.
`Container::blur_behind()` marks a container whose rect and largest corner radius, capped at half its shorter side, the layout pass records in surface coordinates through the `DrawCtx::origin` offset of the sub-canvases, clipped by the viewports it sits in. `commit_draw_ctx` keeps them in `FrameState`, and after a frame where they changed `draw_frame` hands them to `App::on_blur_regions( SurfaceTarget, &[ ( Rect, f32 ) ] )`, for the main surface and each overlay.
`Row::align_bottom()` pins the children to the bottom edge, so taller siblings grow upward. `docs/architecture.md` lists `on_pointer_hover` among the input hooks.
Not done. `on_blur_regions`, `blur_behind` and `align_bottom` are not in the documentation, and there are no tests for them. The regions are reported after the frame is committed, so a compositor that receives them over IPC is one frame behind a surface whose cards move.
|
||
|
admin
|
a63e905a7a |
A text field keeps the cursor where an edit left it unless it asks to be pinned to the end, which only the time picker does, and a hook for a window that loses the keyboard
A field built with `select_on_focus( true )` kept its cursor at the end of the value after every insertion and deletion. That was meant for the time picker, whose `on_change` normalises what was typed, "2" becoming "02", so that a cursor placed at the end of the typed text would land in the middle of the normalised one. But the flag was also the only way to have a field select its whole value on focus, so every field that wanted that got the pinning too, and in an address bar it is a defect: tapping inside the address to correct a word, the cursor was placed where the finger was, and the first character typed sent it to the end, so the rest of the edit went to the wrong place. The two behaviours are now separate. `select_on_focus` only selects the value when the field takes the focus. The new `TextEdit::pin_cursor_to_end( bool )`, `false` by default, keeps the cursor at the end of the displayed value through the `usize::MAX` sentinel, as before, for fields whose `on_change` rewrites the value. It reaches the runtime in a new `pin_cursor_to_end` field of `WidgetHandlers::TextEdit`, set where the snapshot is built in `Element` and copied in its `Clone`, and the four places in `insert_delete.rs` that anchored the cursor, for an insertion, a delete forward, the deletion of a selection and the move after a replacement, read it instead of `select_on_focus`. The time picker's value fields ask for it, so they behave as they did. A field that uses `select_on_focus` only to select on focus now edits with an ordinary cursor. The one other user known is a dialog in `crate`, which gets the ordinary behaviour; if its `on_change` normalises the text it has to ask for `pin_cursor_to_end( true )` as the time picker does. Documentation of the previous commit and of what building crate showed. `docs/widgets.md` says that a `pressable` loses the tap and the right click to an interactive widget inside it even when that widget has no handler, as a `list_item` without `on_press`, so its content should be built from pieces that take no input; and that the hint of `Button::tooltip` is an xdg-popup beside the button in a window, placed by the compositor and taking no grab, and an overlay spanning the output on a layer surface. `docs/theming.md` gains "Application icons": a desktop entry can point at an installed theme icon, the Figma glass effects resvg does not reproduce, a `backdrop-filter` in a `<foreignObject>` ignored and `plus-lighter` rewritten to `screen` by `decode_svg_bytes`, and the advice to flatten them; the table row of `decode_svg_bytes` mentions the rewrite and that its pixels come back premultiplied. `docs/cookbook.md` says that requesting the focus for the text field that already has it puts the cursor at the end of its value without a second `on_focus`, which is how an application moves the cursor after replacing the value, and that Tab and the vertical arrows reach `App::on_key_with_modifiers` before the field, while Escape reaches it after ltk has closed any transient UI and taken the focus off the field. Losing the keyboard. An application that repeats keys on its own timer, as prompt does for its terminal, never heard that the window lost the keyboard, so a key held while the focus went elsewhere kept repeating until the next key, since its release went to the other window. The new `App::on_keyboard_leave()`, a no-op by default, is called from the keyboard's `leave`, right after ltk stops its own repeat, for the application to stop its repeat and forget the modifiers it kept. Documentation of the previous commit's hooks and widgets. `docs/architecture.md` lists `on_mouse_button` with `MouseButtonEvent` and `MouseButton`, `on_modifiers` and `on_keyboard_leave` among the input hooks. `docs/widgets.md` describes `ListItem::ink( label, secondary )`, which draws a row's label, subtitle, trailing text and hover and press tints in colours of its own for a surface the theme does not know about, and says that `icon_button` keeps a bitmap's proportions, fitting one that is not square into its size and centring it rather than stretching it. Not done. There is no test for the new flag, as `select_on_focus` has one in `text_edit/tests.rs` for its builder and its propagation to the handler, and the documentation of the widget does not mention it. |
||
|
admin
|
a0ceef51ef |
WallpaperBundle::blurred; the on-screen keyboard at the right size on a scaled output, keys as tall as they are wide, a terminal row that fits and a Spanish symbols layer that does not squeeze the rest
The keyboard of the previous commit was drawn and tried on a desktop at scale 1. On the Librem 5, at scale 2 and hosted by forge, it showed what that could not: letters twice their size, a terminal row whose modifier labels did not fit, keys shorter than they were wide, and a Spanish numbers layer with a gap at each side. This fixes those, and adds the wallpaper helper the shell and the compositor now use for their frosted panels. A frosted wallpaper. Crustace decoded the theme's lockscreen image at full size a second time for the launcher and the quick settings, and forge decoded it again for the app switcher, blurred it with a floating-point Gaussian at 1920 px and composed a copy per panel size; the switcher's decode alone left some 60 MB in a thread arena that glibc never gives back. A backdrop that is blurred needs no resolution. `WallpaperBundle::blurred( divisor, sigma )` returns a bundle `divisor` times smaller on each side, reduced with a triangle filter and blurred with `imageops::fast_blur`, the box approximation that works in place on the 8-bit pixels; the renderer scales it back up with linear filtering, and the blur hides what the reduction lost. It is an ordinary bundle, so `for_size` crops it for portrait as it does the original. Two tests: the size is divided and never reaches zero, and a flat colour stays that colour. `docs/architecture.md` mentions it next to `for_size`. Text twice its size. `Panel::draw` passed the canvas a font size already multiplied by the output scale, and the canvas multiplies every font size by its `dpi_scale` once more, as it must for widget text, which is sized in font space. At scale 1 the two agree; at 2 the labels came out at four times the area. The keys and the variants popup now pass `TEXT_SIZE` and `ACTION_SIZE` unscaled, and `draw_centred` computes the baseline offset from the size in physical pixels, `size * dpi_scale()`, so the label stays centred on the key. Radii, key geometry and icons were already right: they are physical pixels and never go through `dpi_scale`. The runtime's built-in panel and forge's both draw through `Panel::draw`, so both had it. Keys as tall as they are wide. Rows were a fixed 46 px high whatever the width, so on a phone the keys came out wide and flat. The new `key_unit( width, scale )` is the width one key unit takes when the widest row of the layer fills the panel, the width of a `q`; `set_rect` uses it as both the width unit and the height of every row, the terminal row included, and the variants popup, which was already as tall as its key, follows. The panel's height therefore depends on its width: `height( scale )` becomes `height( width, scale )`, `place_osk` passes the surface width, and `KEY_HEIGHT` goes. The architecture document says so. In landscape the keys grow with the screen's width; a cap on the height has not been needed yet. The terminal row. Esc, Tab, Ctrl and Alt were one unit wide and their labels did not fit. They are 1.5 units now, and to keep the row at 13 units `\` and `~` leave it: the row is Esc, Tab, Ctrl, Alt, the four arrows, `-`, `/` and `|`. Neither is lost: `\` is a long press on `/`, next to `?`, and `~` one on `-`, next to `_` and `–`; `|` takes over the variants `\` had, `` ` ``, `^`, `$`, `#` and `*`, beside its own `&` and `!`. The arrows are drawn at 14 instead of the 22 of the other action icons, which filled the narrow key. The built-in bar's test checks `|` in the row and `\` and `~` among the variants. The Spanish symbols layer. Its third row carried `¡`, `!`, `¿` and `?`, 12 units against the 10 of every other row; since all rows share the key unit of the widest, the numbers and `@ # € %` rows came out narrower and centred, with a gap at each side, and switching to English and back looked right only because that layout's row has 10. `¡` and `¿` leave the row and become long presses on `!` and `?`, as on Android, and stay among the variants of `.` where they already were. The new test `shipped_layers_are_no_wider_than_their_base` reads `en.json` and `es.json` and fails when any row of any layer is wider than the widest row of the base layer, so a layout file cannot bring this back unnoticed. |
||
|
admin
|
a1a4d51677 |
An on-screen keyboard of ltk's own, driven by layout files; TextEdit::purpose for the row a terminal needs; the wheel scrolls a multiline field; a notebook page fills its height
Text entry on a phone comes from `zwp_text_input_v3`: a focused field enables the protocol and the compositor's input method, squeekboard today, draws a keyboard and commits text through it. Eydos wants to drop that process, so the keyboard has to come from somewhere else, and the natural place is the toolkit that already draws everything else and knows which field has the focus and what it is for. This commit puts one in ltk, as data files plus a panel any host can drive: ltk's runtime shows it under a focused text field when no input method is there, and forge will draw the same panel in its chrome and deliver the keys to the focused client, which is the second half and is not in this commit. Along the way the demo that exercises it exposed two gaps in the widgets, fixed here too.
Layouts are data. `keyboards/<language>.json`, one file per layout, searched in `LTK_KEYBOARDS_DIR` when set, then `$XDG_DATA_HOME/ltk/keyboards` or `~/.local/share/ltk/keyboards`, then `/usr/share/ltk/keyboards`; `osk::layout::search_paths()`, `available()` and `load( name )` are the API, and the English layout is compiled in as the last resort. A file has `name`, a `label` for the layout-switch key, `layers` and `variants`. A layer is a list of rows; a row a list of keys; a key is a bare string for the text it types, or an object with an `action` (`shift`, `backspace`, `enter`, `space`, `layer`, `layout`, `hide`), the target `layer`, a `label`, and a `width` in key units, `1` by default, which share the row's width so a row of nine keys comes out centred under one of ten. `base` is mandatory; a missing `shift` layer is derived from it by uppercasing. `variants` maps a key's text to what a long press offers, and the shifted form of a key uses its base's, uppercased. A key may also be a named key, `{ "key": "Escape" }`, `Tab`, the arrows, `Home`, `End`, `Page_Up`, `Page_Down`, `Delete`, `Insert`, `F1` to `F12`, or a sticky modifier, `{ "modifier": "ctrl" }` or `alt`. `en.json` and `es.json` ship, with symbol layers and the accented variants each language needs. The files go into the registry copy through `make install` and into `ltk-theme-default` under `/usr/share/ltk/keyboards`.
The terminal row. A terminal needs Escape, Tab, Ctrl, Alt, the arrows and the shell's symbols, and no other application does, so those are not a layout but a row laid above any language's layout: `keyboards/bar-terminal.json`, made of the named keys and modifiers above plus `-`, `/`, `|`, `\` and `~` with their variants. Files named `bar-*.json` are rows, not layouts, and `available()` leaves them out, so the layout-switch key does not cycle onto one; the first version listed it and a tap on that key made the keyboard vanish behind a duplicated bar. Which applications get the row is decided by what the field says it is for: `TextEdit::purpose( InputPurpose::Terminal )`, snapshotted into the handlers like `secure`, which the built-in panel reads through `set_purpose` and which also goes out as the `terminal` content purpose of `zwp_text_input_v3`, the same one VTE-based terminals declare, so a compositor hosting the panel can pick the row from the protocol; for a client that declares nothing it can decide by `app_id`.
`osk::Panel` is a state machine a host drives. It holds the loaded layouts, the current layer, shift as off, one-shot or locked, the sticky modifiers, the purpose and its extra row, the geometry of all of it inside the rectangle the host gave it with `set_rect( rect, scale )`, and the press in flight. The host feeds it `press_down`, `press_move`, `press_up` and, from its own timer, `tick`, each returning an `Event`: `Text`, `Backspace`, `Enter`, `Hide`, `Key( keysym, modifiers )` for what text cannot carry, an arrow, Escape, Tab, or a character while Ctrl or Alt is held, or `Changed` when only the panel's look changed. A key types on release; a press held 450 ms opens its variants, which the finger slides onto and the release picks; backspace fires on press and repeats every 60 ms after the same delay; shift is one-shot and locks on a double tap within 350 ms; a modifier stays down for the next key and a character typed under it goes out as a `Key`, since text cannot carry Ctrl. `next_deadline()` says when `tick` next has work, `height( scale )` what the panel needs, `press_down` answers `None` for a press outside it so the host treats that as its own. `draw( canvas )` paints it with the active palette; the action keys show theme icons rather than glyphs, `keyboard/shift`, `keyboard/backspace`, `keyboard/enter`, `general/down-simple` for hide, `accessibility/languages` for the layout key, the `general` arrows for the terminal row, rasterised once per size and tint, because the first version drew `⇧` and `🌐` and the fallback chain has no glyph for the latter, which showed as a question mark; a layout key with no label of its own shows the active layout's. Nothing in `Panel` knows about text fields or Wayland, which is what lets a compositor host it. Fifteen tests cover the layout parser, the derived shift layer, variants following case, the bar filtering, and the panel's taps, one-shot and locked shift, variants, repeat, layer and layout switches and the terminal row.
The runtime shows it. `event_loop/osk.rs` puts a `Panel` on a surface when a text field takes the focus and `builtin_osk` holds: the compositor advertised no `zwp_text_input_manager_v3`, or `LTK_BUILTIN_OSK=1` is set, which is how to try it on a desktop that has an input method, or the app returns `true` from the new `App::builtin_keyboard()`. The panel is placed at the bottom of the surface by both renderers, and the widget tree is laid out in what is left above it, so the focused field is never covered. Presses, motion and releases from the pointer and from touch are offered to the panel before the gesture machine, a press on the already focused field brings a hidden panel back, and the run loop's timeout now takes the minimum of the long-press, tooltip and panel wakeups so variants and repeat fire on time. `Text` goes through the same insertion as a physical key, `Key` through the same dispatch, with Ctrl applied for its duration, `Enter` submits or inserts a newline in a multiline field, `Hide` keeps the focus and takes the panel away; losing the focus removes it. The layout is picked from `$LANG`. `ltk::InputPurpose` is exported, and `docs/architecture.md` gains the section that says all of this, with the compositor's part sketched.
The demo. `examples/inputs` becomes a notebook with the login page it had and a Shell page: one multiline field that holds a transcript, prompt included, declared a terminal, so the row appears; Enter runs the line under the last prompt (`echo`, `ls`, `pwd`, `clear`), the arrows move through the transcript, and a Ctrl combination interrupts the line the way a terminal shows it. Escape quits only from the login page, because a key the app answers never reaches the field, which is what first made the arrows dead.
Two widget gaps the demo found. `Notebook` gave its page the page's preferred height, so a page that fills its space, a scrolled list or a flex, got none and showed nothing: the active page is now wrapped in `flex` and gets the height under the strip. And the multiline field had no way to scroll: it chose its first visible line from the cursor alone on every draw, so the transcript ran under the keyboard and the wheel did nothing. `text_edit/scroll.rs` gives it a `TextScroll` the runtime keeps per field beside the cursor and the selection, in the frame state, the draw context and the `core` runtime alike: the first line asked for by hand and the cursor at that moment. The view holds there while the cursor stays put and, once it moves, follows it by the least it has to, as an editor does; without a request the result is exactly the old rule. `Window::resolve` computes the visible lines once for drawing, hit testing and the wheel, so they cannot disagree, and replaces the copy each of the first two had. In `on_pointer_axis` the vertical motion is offered to the multiline field under the pointer first, three lines per notch from `value120` or `discrete`, touchpad pixels accumulated until they add up to a line, and only when the field cannot move that way does it go to a `scroll` around it. `TextScroll` is public but opaque, since `Element::draw` is public and needs it; `Default` is the cursor-following behaviour. Six tests cover the hold, the follow, the clamping, the notch and the accumulation. A read-only field is not interactive and does not take the wheel; touch drags inside a field do not scroll it yet.
Not done. Forge does not host the panel yet, so on the phone squeekboard is still what appears. The panel's look is a first pass, sized in logical units with the theme's palette, to be restyled from mockups. Only `en` and `es` ship.
|
||
|
admin
|
2af281520b |
External text input for fields ltk does not own; TextEdit::color and TextEdit::on_focus
Two additions driven by apps that host another UI inside an ltk surface: trust, whose pages have their own form fields, and rustdroid, which projects Android views into ltk widgets. External text input. ltk only ever enabled `zwp_text_input_v3` for its own `TextEdit`s, so nothing told the compositor that an editable inside an embedded web view had the focus: the on-screen keyboard never came up inside a page, which makes a browser unusable on a phone without a hardware keyboard. Three hooks on `App` close that. `external_text_input() -> Option<u64>` asks for the input method on behalf of such a field: while it returns `Some( serial )` and no ltk text widget holds the focus, the runtime keeps a text-input enabled, and the compositor raises its keyboard. `on_external_text_commit( text )` and `on_external_text_delete( before, after )` receive what the input method sends, the latter in UTF-8 bytes as the protocol counts them. Keys such as Backspace and Enter keep arriving as key events, which hosts already forward. The request is a serial rather than a `bool` so that it can be renewed. A compositor that folds its keyboard while the field keeps the focus, which forge does on back, has nothing to observe when the user taps that same field again, because the text-input never stopped being active; a change of serial makes the runtime send `enable` again through `reenable_text_input`, the same signal ltk now sends for its own fields. The host chooses what bumps it; trust counts taps on the page, so dragging to scroll does not bring a folded keyboard back. `AppData::sync_external_text_input`, in `text_editing/ime.rs`, is polled after every batch of `update`s next to `sync_fullscreen`. An ltk text widget always wins: while one has the focus the function only clears its record, and `set_focus` manages the text-input as before. When that widget loses the focus `set_focus` destroys the text-input, and the next poll finds the request still standing with no text-input and enables a new one, so the two owners hand it back and forth without either knowing about the other. The `zwp_text_input_v3` dispatch routes `commit_string` and `delete_surrounding_text` to the app only when the external request is held and no ltk text field is focused; that second check covers the one iteration between a widget taking the focus and the next poll. The empty-string filter on `commit_string` is folded into the `Option` it already lived in. `TextEdit::color( c )`. The colour of a field's value always came from the theme. A field placed over a surface the theme knows nothing about, an Android `EditText` with its own text colour on its own background, could end up unreadable. `color` sets the value's colour in both draw paths, single-line and multiline; the placeholder keeps the theme's, and a field without it draws exactly as before. `TextEdit::on_focus( msg )`. `App::on_text_input_focused( bool )` says that some text field is focused, not which. An app that mirrors focus into another model needs to know which: rustdroid keeps Android's single focused view in step, and could only learn the field from its first `on_change`, so `View.isFocused()` stayed false until the user typed. A `WidgetId` cannot carry that, being a `&'static str` while such handles are created at run time. A per-widget message can, since the app builds it and puts in it whatever it needs. It rides the `WidgetHandlers::TextEdit` snapshot, and `set_focus` pushes it when the focus actually moves to the field, from a press, Tab or `App::take_focus_request`, and not again for a press on the field that already holds it. Every other match on that handler variant uses `..`, so the new field touches its definition, its clone and the snapshot in `element.rs`. Documented in the `App` rustdoc, the trait surface list of `docs/architecture.md`, the `text_edit` entry of `docs/widgets.md` and the changelog. Neither path has a test: both need a live Wayland connection and a compositor to exercise. |
||
|
admin
|
96a9e88051 |
Back as a navigation request: App::on_back, BackAtRoot, NavStack and docs/navigation.md; renew the text-input enable on a press; fit an oversized window_size_hint to the screen
Three changes that came out of making the shell's back button work across Eydos the way it does on Android. The first is the feature; the other two are gaps it exposed. Back is no longer Escape. forge turns its back gesture and button into the `XF86Back` key for the focused client, and ltk treated that key as an alias of `Escape`: close transient UI, else drop the focus and reach `on_key_with_modifiers`. That is "cancel", and it is all an app could get: nothing navigated, and nothing ever left the app. `XF86Back` is now a back request with its own order, one layer per request, top-most first: transient UI the runtime owns (open xdg-popup overlays, the built-in context menu, a text selection, the top-most pressable carrying an `on_escape`, i.e. a dialog's cancel), then text-input focus, then the new `App::on_back`, then `on_key_with_modifiers( XF86_Back )` so an app that bound the key itself keeps receiving it (crustace does, for its launcher), then `App::back_at_root`. The first step is what Escape already did; it moves out of `handle_key_escape` into `AppData::dismiss_transient` in the new `input/keyboard/back.rs`, which both paths call, so Escape behaves exactly as before. Text focus comes before the app on purpose: with a keyboard covering half the screen, back means "put the keyboard away", and navigating instead would discard what the user was typing. A compositor that folds its keyboard itself, as forge now does, only lets the next request through, which makes this step the fallback for the ones that do not. `App::on_back` returns a `BackResponse`: `Message( msg )` with the message that undoes the top-most piece of navigation state, `Unhandled` at the root, or `Consumed` for a host whose back stack lives elsewhere and cannot be inspected ahead of time. `Consumed` exists for rustdroid: an Android app keeps its activities, fragments and back callbacks in its own code, so the host can only hand the request over. `App::back_at_root` then picks `Minimize` (the default), `Close` (through `on_close_requested`, which can still veto) or `Ignore`. Minimising is what a root activity does on Android since version 12 and what suits an app the user comes back to: a browser keeps its tabs, a call goes on. It is also the single point where the two stacks meet: the navigation stack belongs to the app's model, the task stack to the compositor, and `xdg_toplevel.set_minimized` is the one standard request between them, so no custom protocol carries navigation state out of the process and clients that are not ltk apps keep working. Only `ShellMode::Window` surfaces react at the root; a panel or a lock screen that minimised itself would be a bug. Three consequences of back being a navigation request rather than input: the key leaves the default `App::key_repeats` set, since a held button would empty the whole stack; it is no longer reported through `App::on_raw_key`, so a host forwarding raw keys into an embedded web view or guest toolkit stops leaking it as a keystroke; and the mouse's back button (`BTN_SIDE` / `BTN_BACK`) enters through the same `handle_back`. `focused_text_input` becomes `pub( crate )` for the new module. The answer is derived, not recorded, and that is the design decision worth writing down. The runtime keeps no navigation stack of its own. Its entries would have to be messages, which are not serialisable (the trait puts no serde bound on `Message`, by design), so it could not survive a relaunch; and a list of "things to undo" drifts the first time the user leaves a screen some other way, through its own close button or a link, leaving a back press that does nothing visible. An answer computed from the model cannot disagree with it. This is also what ties back to the session persistence added earlier: navigation state is both what `on_back` reads and what `save_state` should persist, and keeping the two the same is what makes back retrace the same places after a session restore or a crash, with nothing extra to serialise. Persist less and a restored app opens on the right screen with nothing behind it; persist more and it restores a menu the user had closed. `NavStack<R>` packages that rule for the common case, an app made of screens. It is a plain value the app keeps in its model, never seen by the runtime: `push`, `pop`, `replace` (the sideways move back should not retrace), `pop_to_root`, `reset`, `current`, `current_mut`, `iter`, `depth`, `can_pop`, and `back( msg )`, which answers `on_back` directly. It is stored as a root plus the routes above it rather than as a `Vec`, so "never empty" holds by construction, including after deserialising an old or damaged state file, and it derives serde whenever `R` does. Six unit tests cover it, a serde round trip among them, and its rustdoc example is a complete `App` where one object answers `on_back`, `save_state` and `restore_state`. Documentation, since the point is for app authors to take navigation state into account. The new `docs/navigation.md` is the reference: transient UI versus navigation state and the test that separates them, the back order and the reason for each step, why the answer is derived, how it lines up with session state, `NavStack`, apps that embed a web view (whose page history is navigation state too, to be persisted with ltk-webkit's `session_state`), hosts with a foreign back stack, the root, who owns which stack, and an author's checklist. `architecture.md` gains a group in the trait surface and a paragraph in the state patterns, `onboarding.md` and the README index link to it, and in the cookbook the multi-screen recipe now uses `NavStack` with `on_back` and persists the stack without serde, which keeps it compiled by the Markdown doctest harness (that harness links `ltk` alone, so the serde example in `navigation.md` is marked `ignore` and compiled as `NavStack`'s own doctest instead). `BackResponse`, `BackAtRoot` and `NavStack` are exported at the crate root and in the `ltk::window` facade. Behaviour change for existing apps, also recorded under Changed: a window app that implements neither `on_back` nor the key is now minimised on back at its root, where nothing happened before; `BackAtRoot::Ignore` restores that. With a text input focused, back only drops the focus and the app sees the next request, where it used to get the key in the same press. Renew the text-input enable on a press. Moving the focus between two text fields, or pressing the one already focused, only committed the refreshed content type on the existing `zwp_text_input_v3`. `activate_text_input` now sends `enable` first, as GTK does when a focused entry is clicked. A compositor that folded its on-screen keyboard while the field kept the focus, which is what back now does, has no other way to learn that the user wants it again: the text-input never stopped being active, so there was nothing for it to observe. ltk sends neither surrounding text nor a cursor rectangle, so the only state `enable` resets is the content type set right after it. Fit an oversized `window_size_hint` to the screen. The hint is pinned as `min_size == max_size` before the first commit, when no output is known yet, so an app asking for a phone-shaped 720x1440 window on a 1080p desktop got a toplevel taller than the screen; rustdroid carried a dead branch that tried to avoid this from a screen size nothing could give it. The first configure is the earliest point the hint can be checked, and the protocol has the right input for it: `xdg_toplevel.configure_bounds`, exposed by sctk as `WindowConfigure::suggested_bounds` and ignored until now. The new `event_loop/window_fit.rs` holds `fit_within`, which scales a size down keeping its aspect ratio until it fits and leaves one that fits untouched (four unit tests), and `AppData::fit_size_hint`, run once behind the new `pending_size_hint_fit` latch: it uses the compositor's bounds when sent and otherwise 90 % of the first output's logical size, since panels and decorations are unknown from the client side. When the hint has to shrink, `min_size` (and `max_size` for a non-resizable window) is re-pinned to the fitted size, which is kept in `fitted_size_hint` and replaces the app's hint as the fallback for later configures that carry no size; without that the window would grow back on the next one. On that one configure the fitted size wins over the compositor's `new_size`, which was derived from the oversized pin. `App::window_size_hint` documents the rule with the worked example (720x1440 opens as 486x972 on a 1080p output). The back order needs a live Wayland connection to exercise and has no test; the unit tests cover `NavStack` and `fit_within`. |
||
|
admin
|
53e7ccc46e |
Keep the GLES path when the scale arrives first, stop re-rendering what does not change, and let the compositor skip what the surface covers
Measured on a Librem 5 (Cortex-A53, GC7000Lite, 720×1440 at scale 2) with a frame trace on both sides of the socket: the shell's main surface repainted in 110 to 136 ms, quick settings drew at 20 fps with the GPU saturated, and every panel open uploaded its wallpaper and icons again. Most of it was the toolkit doing work the frame did not need. Backend choice. When the compositor announces the output scale before the first `configure`, `scale_factor_changed` allocated an SHM pool for the surface and `on_configure` took that pool as the sign that the software path had been chosen, locking the surface to tiny-skia for its whole life. Every ltk main surface on a scale-2 phone ran in software while its overlays, which inherit the scale from the main and never receive that event, ran on the GPU. The scale handler now only reallocates a pool that already exists; before the first configure neither path is chosen. Wallpapers. `WallpaperBundle::for_size` returned the crop at the source resolution, a 1103×2160 slice of a 4K image for a 720×1440 panel, 10 MB uploaded per texture and a hundred milliseconds each time. The crop is now downscaled to the surface (bilinear, once per size, cached), also for landscape surfaces smaller than the source. Viewport. `Element::Viewport` created a fresh sub-canvas every frame, on GLES a framebuffer, a 2048² glyph atlas and empty image and glyph caches, so a panel clipped to a height that animates re-uploaded its wallpaper and re-rasterised every glyph on every frame. Without a bottom fade the clip is now a scissor on the parent canvas (a mask on the software backend): no offscreen pass, no blit, and the child's textures stay in the parent's caches. `local_viewport` is honoured through a temporary override of the layout viewport. Viewports with a fade keep their sub-canvas, which, like the scroll sub-canvases, is now kept across frames and resized in place instead of reallocated when its size changes. Caches. The image texture cache moves from the canvas to the GL context: every canvas and sub-canvas shares one content-keyed LRU (32 MiB), so an overlay or subsurface that is destroyed and recreated finds its images resident. A GLES canvas dropped with an overlay is pooled in `AppData::canvas_pool` and handed to the next overlay, with its compiled programs and glyph atlas, which took 130 to 290 ms to build on that GPU. Subsurface slots keep their scroll and viewport sub-canvases between rasters. Surfaces. A `container.surface()` with shadows costs a Gaussian drop-shadow pass, a fill and one pass per inset shadow, all full-quad fragment shaders; a dozen glass cards were 40 ms of GPU per frame for chrome that never changes. `Canvas::fill_surface` on GLES now renders the chrome once per size, corners and style into a texture held by the context cache (`gles_render/surface_cache.rs`, 16 MiB LRU) and blits it, drawing directly only for surfaces with `PlusLighter` or `Overlay` insets, which depend on the backdrop. Opaque regions. ltk never called `wl_surface.set_opaque_region`, so the compositor blended every ltk surface and repainted everything underneath a full-screen homescreen. A surface whose `background_color` is opaque now declares itself opaque, and the new `App::overlay_opaque_region( id )` lets an overlay declare the physical rects it paints opaque, such as a panel column over an opaque wallpaper. `sync_opaque_region` sends the rects only when they change. Frame pacing. `eglSwapBuffers` blocked 13 to 37 ms per frame waiting for Mesa's own throttle callback; ltk paces itself on `wl_surface.frame`, so `make_current` sets the swap interval to 0. A subsurface move is now applied only when its parent has no frame callback outstanding, and the parent commit that applies it requests the next callback: a drag commits once per vblank with the latest finger position instead of once per input sample, without the app throttling samples itself. While the app reports `subsurface_motion_only`, the main's callback, and the callback of an overlay that hosts a subsurface, no longer force a full repaint; they mark the surface and the reconcile keeps the cadence with a bare commit when nothing moved. Swipe releases arm a settle the same way instead of dirtying every cache. Two hooks, `on_overlay_frame_committed` and `on_subsurface_committed`, let an app hand a panel over between its host and a sliding subsurface on real commits rather than loop ticks. Lines. A diagonal `draw_line` on GLES stamped one `fill_rect` per pixel of length, 190 draw calls for a spinner and over 2000 for a pattern-lock stroke; it is now a single rotated quad through the rect shader, with round caps and antialiased edges. A pixel test covers the scissor viewport, `for_size` gets a downscale test, and `docs/architecture.md`, `docs/widgets.md` and `docs/backends.md` describe the new hook, the viewport paths, the shared caches and the line rasteriser. |
||
|
admin
|
5b8ae35830 |
[ltk — src/input/gesture/mod.rs, src/input/gesture/tests.rs, src/widget/handlers.rs, src/input/keyboard/dispatch.rs, src/render/image.rs, docs/architecture.md, CHANGELOG.md] Keep a tap on a button while the finger rolls inside it, treat XF86Back like Escape, and convert ARGB8888 by whole pixels
Any travel past the slop turned a press into a vertical or horizontal drag and its release skipped the button, so a thumb that rolled on a large key lost the tap. `GestureState::on_move` now stays idle while the pressed widget fires on press (the new `WidgetHandlers::has_press`) and the finger is still inside its rect; the swipe machinery takes over from the original press point once the finger leaves. Presses inside a scroll viewport are unaffected, since a list of full-width rows is scrolled by dragging on them. Two tests cover the tap that survives motion inside the button and the swipe that arms once the finger leaves it; the architecture notes and the changelog describe the rule. The compositor now sends XF86Back for the Android back gesture, an inward swipe from either side edge. With a text input focused that keysym fell into `handle_key_default` and was swallowed, so an app never saw it. `XF86Back` now dispatches through `handle_key_escape`, so every ltk app gets the back gesture with Escape's layering: open popups, then the context menu, then a selection, then a dialog's cancel, and finally the focus drop and `on_key_with_modifiers` with the keysym, where an app can act on it. `SoftwareCanvas::write_to_wayland_buf` swapped red and blue with a byte-by-byte loop, unrolled by hand over sixteen bytes. It now reads each pixel as a `u32`, exchanges the two channels with masks and shifts, and leaves the vectorisation to the compiler. Compositors that offer `ABGR8888` never took this path. |
||
|
admin
|
75be800390 |
input, render: keep fast touch typing whole, fire pattern nodes once, and make the software image path cheap
Two ways a tap was lost on a phone. The gesture machine serves one finger: a second finger landing while the first was still down went to the auxiliary slot and never reached it, so with two thumbs every overlapping tap on a keypad vanished. A finger landing on a primary that is still a plain press (GestureState::is_plain_press: no drag, scroll, swipe or slider in flight) now commits that press as a tap at its last position and takes the primary slot; the earlier finger's release reports through on_touch_up, and a second finger during a pattern stroke or a slider drag stays out as before. The touch-up body moves to release_primary so both paths share it. The other loss needed one finger: the 8 px travel that turns a press into a scroll, swipe or drag, and the 6 px stray that cancels a touch long-press, were compared against physical coordinates, so at buffer scale 2 they were 4 and 3 logical px and an ordinary thumb tap became a vertical drag whose release skipped the button. SwipeConfig carries the surface's scale and the machine reads the thresholds through slop() and stray(), which are logical. Pattern grids fired on_node on every motion event within a node's snap radius, and the Slider move outcome requested a repaint each time, so drawing a pattern rebuilt and re-rastered the surface at the touchscreen's motion rate. The machine remembers the node under the finger (last_pattern_node) and emits only on change; the handlers split pattern_node_at from pattern_msg, with pattern_node_msg kept on top of them; a move outcome without a message no longer requests a redraw. SoftwareCanvas::draw_image_data premultiplied the whole source in floating point on every call, so a full-surface backdrop paid a pass over every pixel per frame before the blit. Premultiplied copies now live in a 32 MiB least-recently-drawn cache keyed by size, content fingerprint and alpha, the key the GLES texture cache already used; fingerprint_rgba moves to render::helpers and both backends share it. The entry records whether the source was fully opaque, and an opaque image drawn 1:1 with no clip and full alpha is copied row by row into the canvas instead of going through tiny-skia's bilinear pipeline. docs/architecture.md describes the primary-slot hand-over, docs/widgets.md the once-per-node contract, and the CHANGELOG lists the five fixes. |
||
|
admin
|
492553829e |
widgets, input: PatternGrid for pattern unlock, framed icon buttons, and text keys that bubble past a focused non-text widget; Debian 0.3.0-2
PatternGrid is a 3×3 drag widget for the unlock pattern. It is stateless: the app owns the visited nodes and passes them back in for drawing; a press or a drag that reaches a node within the snap radius fires on_node with the index, 0 to 8 row-major, and release fires on_release, so the crossing and minimum-length rules stay in the app. Geometry lives in widget/pattern_grid/theme.rs (size, dot and node radii, stroke width, snap radius, ring alpha), and the snap radius is resolved through the widget-scaling mode at layout time like the slider thumb (set_pattern_snap_px). The gesture machine reuses the slider's drag ownership: WidgetHandlers::is_drag_target covers both, so hover painting, pointer capture and release behave the same, and hover_affects_paint no longer special-cases sliders. The a11y tree maps it to GenericContainer, since a drawn gesture has no assistive equivalent and the app must offer another method. Exported as pattern_grid / PatternGrid, documented in widgets.md, theming.md and architecture.md, with tests for node centres, a wide rect, snapping, the builders, the handler snapshot and map_msg, and a Security tab in examples/widgets.rs that shows the drawn path. Icon buttons given a width or height are framed keys: preferred_size honours the explicit size instead of the icon size, and drawing paints the variant's background, border and focus ring, shared with text buttons through paint_frame, then centres the icon. An icon button without a size keeps the circular hover overlay it had. This is what lets the PIN keypad's backspace and OK keys sit in the same circles as the digits. Keyboard routing now matches what App::on_key documents. handle_key_default, Backspace, Delete and Space checked focused_idx.is_some() and sent the key into text insertion whenever any widget had focus, which is a no-op on a button, so a focused button or pressable swallowed typed characters. focused_text_input replaces the five copies of that check; characters, Backspace and Delete stay in a focused text input and reach the app otherwise, Return and Space still go to the focused widget's submit or press message when it has one, and the arrow, Home and End handlers use the same helper. The lockscreen relies on this to take PIN digits with its dots row focused. debian/changelog: 0.3.0-2. |
||
|
admin
|
ccf07de593 |
Session management: xdg-session-management-v1 client, mandatory App::app_id / save_state / restore_state, runtime-managed state persistence and clean exit on signals (0.3.0)
Applications built on ltk had no way to come back where the user left them: the toolkit hardcoded `app_id = "ltk"` on every toplevel, never wrote anything to disk, and died on SIGTERM without a chance to save. This release gives the runtime the whole plumbing and asks each application only for the bytes worth keeping, in the spirit of Android's saved-instance state. The `App` trait gains three mandatory methods, deliberately without default bodies so every application states its position: `app_id()` (reverse-DNS, used for `xdg_toplevel.set_app_id`, the AccessKit application name and the state directory — the `app_id` element of the deprecated `window_config` tuple is now ignored and a one-time warning reports a mismatch), `save_state() -> Option<Vec<u8>>` and `restore_state(Vec<u8>)`. The bytes are opaque; the trait carries no serde bound. Their rustdoc is the contract: when the runtime saves, where the files live, when the bytes come back and when they do not, what must never go in them, and a worked serde_json example. The runtime persists under `$XDG_STATE_HOME/<app_id>/` (falling back to `~/.local/state`): `session.json` holds the compositor session id, a clean-exit marker and the writer's pid; `state.bin` holds the application bytes. Writes are atomic (temp file + rename, mode 0600, directory 0700) and best-effort. State is saved every 30 s when the bytes changed, once after the event loop exits (which covers `on_close_requested`, `requested_exit` and lost connections), and on SIGTERM/SIGINT — a calloop signal source, installed before any thread exists, now turns those into a clean exit of the loop instead of process death. `restore_state` runs synchronously in `try_run` before the window is created and before the first `view()`, and only when the process is relaunched as part of a session restore (`LTK_SESSION_RESTORE=1`, removed from the environment before the app can spawn children) or when the previous run left `clean_exit: false`; a plain launch starts fresh. A second concurrent instance detects the live pid and runs with persistence disabled rather than clobbering the first. The compositor side of geometry restore goes through `xdg-session-management-v1`. Neither wayland-protocols nor sctk ship generated code for it yet, so the XML is vendored under `protocols/` and `wayland-scanner` generates the client module in-tree (`src/protocol/`), resolving the crate names through sctk's reexports so the bindings stay on the crate instances sctk links. Before the first commit of a `ShellMode::Window` toplevel the runtime binds `xdg_session_manager_v1`, calls `get_session(reason, stored_id)` and `restore_toplevel(toplevel, "main")`; the three window-creation paths in `run.rs` are folded into one `make_window` helper so the attach always sits immediately before `commit()`. `created` persists the id, `replaced` destroys the objects and stops persisting. Compositors without the global lose only the geometry half. Layer-shell and session-lock surfaces skip the whole machinery. Every `App` implementor in the tree is updated: the twelve examples (`showcase`, `scroll` and `mini_shell` persist real state; the rest return `None`), both integration tests (`event_loop_flow` gains `save_restore_round_trip`), the in-source and markdown doctests, README, onboarding, cookbook (new recipe "Surviving relaunch: session state") and architecture docs, and the changelog. `src/session_state.rs` carries unit tests over a temporary state directory. `Makefile install` now copies `protocols/` into the cargo registry — without it downstream builds would fail inside the proc-macro — and `debian/copyright` covers the vendored XML. The trait change is breaking, hence 0.3.0. Also fixes the pre-existing `viewport_tests` module in `render/mod.rs`, which used `Length` without importing it and broke `cargo test`. |
||
|
admin
|
d9652dce98 |
accessibility text scale: global font multiplier synced to the desktop's text-scaling-factor
New set_text_scale / text_scale process global (clamped [0.5, 3.0]) multiplied into every resolved font size — Canvas::resolve_font for explicit Lengths and the Physical branch of font_px (the Fluid branch routes through resolve_font) — so the whole tree's text follows the accessibility "large text" factor while geometry stays untouched, mirroring GNOME's text-scaling-factor semantics. Unit test covers fonts-scale-geometry-doesn't. The run loop keeps the factor synced on its own: a watcher thread (event_loop/text_scale.rs) reads org.gnome.desktop.interface text-scaling-factor via gsettings get at startup and streams external changes from gsettings monitor into a calloop channel; on a change the loop stores the factor, invalidates the view caches and repaints main and overlays. Since fonts resolve at paint time nothing else needs rebuilding. Every ltk app tracks the settings slider live with zero app-side wiring, the same way GTK apps follow the key; missing gsettings degrades silently to a fixed 1.0. Embedders driving core::UiSurface (forge) call set_text_scale themselves — the multiplication only runs at widget font resolution, so raw Canvas::draw_text callers keep hand-computed sizes. architecture.md documents the multiplier in the font-space paragraph and CHANGELOG gains the entry. |
||
|
admin
|
e343142347 |
viewport: local_viewport() opt-out of root-viewport inheritance; list_item: height/font_size builders
`Viewport::local_viewport()` resolves the child's viewport-relative (vw / vh / vmin) and fluid Lengths against the viewport's own rect instead of the root layout viewport the sub-canvas inherits since the fluid-resolution inheritance change. That inheritance is right for scroll-like clips (content renders the same size inside and outside), but it is exactly wrong for a fixed-size floating mini-UI — crustace's phone-shaped quick-settings pill pinned to a corner of a desktop-wide surface resolved its vw text and fluid stock geometry against the whole monitor, inflating the content past the pill's fixed clip and cutting off the bottom stripe. The flag pins the sub-canvas's layout viewport to its own size via the new crate-internal Canvas::set_local_layout_viewport, nested sub-canvases keep propagating the pinned value, and a unit test covers the override plus propagation. `ListItem::height( impl Into<Length> )` and `ListItem::font_size( impl Into<Length> )` mirror the Toggle / Radio height() builders: override the theme row height (floored at the label's rendered height so the text never clips) and the primary-label font size (subtitle and trailing keep their theme sizes), letting dense context menus trade the stock touch-target generosity for row density. Docs: widgets.md gains both builder sets, architecture.md documents the layout-viewport inheritance model next to the per-canvas density it parallels, the cookbook slide-in panel recipe notes when the pill needs local_viewport(), and the toggle / radio height() rustdoc demotes its private theme::HEIGHT link to a code span so cargo doc is warning-free again. CHANGELOG entries added. |
||
|
admin
|
806dee5167 |
render, types, ci: resolution-time dp with per-canvas density, bounded GLES image cache, clippy gate, backend capability matrix
Length::dp no longer collapses to absolute pixels at construction: the design value travels in a new LengthBase::Dp variant and the density multiplication happens when the length is resolved. Previously dp( n ) baked in whatever density() returned at view-build time, so correctness across output changes depended on the view being rebuilt after set_density and in that order; now a density change is picked up by the very next paint with no reconstruction. Length::resolve keeps its signature (process density), and the new Length::resolve_with_density takes an explicit factor. dp becomes const in the bargain. Density also becomes overridable per canvas, the first step towards surface-local responsive state. SoftwareCanvas and GlesCanvas carry a density: Option<f32> analogous to the layout_viewport introduced for sub-canvas fluid resolution: None means "use the process global", Canvas::set_density pins a local factor, and sub-canvases inherit it. All canvas-routed resolution honours it — geom_px / font_px for stock-widget design pixels, and the new Canvas::resolve_geom / resolve_font for explicit Length values, which every widget now uses in place of the raw l.resolve( canvas.viewport_layout(), EM ) pattern (row, column, wrap_grid, spacer, container, separator, button, text, rich_text, text_edit, list_item, vslider, image, and the container draw path). Overlay sizing keeps resolving against the main surface with the global density, which is what it describes. New tests cover explicit-density resolution, resolution-time application, the local-over-global override and sub-canvas inheritance. The GLES image texture cache is now bounded. It was content-keyed but unbounded and never evicted, so a stream of distinct buffers — a photo carousel, video thumbnails — grew GPU memory for the lifetime of the canvas. The cache now tracks an estimated byte total (RGBA8, w × h × 4) against a 32 MiB budget and evicts least-recently-drawn textures on insert; the most recent entry is never evicted, so a single texture larger than the whole budget still draws and simply owns the cache until replaced. Drop-time cleanup is unchanged: drain deletes whatever the map holds. CI gains a Clippy step (workspace, all targets, test-support, -D warnings) sharing the build cache of the test job, with make clippy mirroring the invocation locally and CONTRIBUTING listing it. Run make clippy locally before pushing the first time — the gate has not seen the tree yet and pre-existing lints will fail CI until addressed. docs/backends.md formalises the software/GLES capability matrix that was previously scattered across per-method rustdoc: parity set (fills, strokes, text, images, paths, path clips), graceful degradations on software (flat-fill gradients, no shadows, no backdrop blur, hard bottom edge), GPU-only features (external textures), the shared Oklab-fallback limitation, and the cross-backend blit panic. Linked from README, onboarding and architecture's known-gaps list, which now states the parity gaps explicitly. The dp/density prose in architecture.md, lib.rs and the Length rustdoc is updated for resolution-time semantics and the per-canvas override. |
||
|
admin
|
1fd697aa6d |
docs overhaul, orientation API, fluid-sizing fixes, examples made honest
Documentation pass: every claim in docs/ and the meta files was audited against the source and the drift fixed — around ninety corrections. CONTRIBUTING and the CI workflow now run cargo test with --features test-support (the gated test_support module made both the documented commands and the CI build fail to compile), make example becomes make examples, make doctest-md and the debhelper requirement of make clean are documented, and patch shape asks for a CHANGELOG entry. theming.md loses the nonexistent surface.backdrop, gains the real gradient defaults (linear-rgb, oklab), the six slot variants including typography, the ten-field palette, a truthful effects-consumer table, the ThemePreference/from_hour API and a responsive-sizing note; the stale docstrings in src/theme that fed the drift are fixed too. architecture.md's "Known gaps" section is rewritten against reality (multi-touch slots, xdg-activation, a11y live regions and SetValue/Increment/Decrement are implemented), gains a module map, subsurfaces and window-lifecycle coverage, and correct crustace/loginmanager paths. widgets.md fixes the ten factual errors (stateless spinner, toast/combo via overlays(), tooltip hover contract, row has no max_width, scroll axes, multiline text_edit, dialog panic wording) and now states the column() 16 px default padding — the recurring ambush — plus row's differing 0 default and dialog's max_width. onboarding, README and cookbook get the remaining sweep: build/test instructions, complete example lists, img_widget, clipping-parity honesty, ~30 Hz software cap, read_rgba_pixels signature, tab indentation in snippets, and rustdoc-style links that rendered literally are gone everywhere. CHANGELOG is restructured per Keep a Changelog with the missing entries (window_resizable, claims_raw_touch, Row::align_top/fill_height, caret fixes, dependency pins) and the pad_v Added/Changed contradiction resolved. New adaptive-layout API: ltk::orientation() with the Orientation enum, backed by viewport_size()/set_viewport_size — the runtime records the main surface's physical dimensions on every configure, before App::on_resize, so view() can branch a layout on portrait vs landscape without hand-tracking resizes. The portrait rule matches Length::orient (square counts as portrait); embedders driving core::UiSurface call set_viewport_size themselves. Documented in the crate root's responsive-design section and architecture.md. Fluid-vs-fixed sizing fixes in widgets, all the same disease — fluid content inside a fixed-pixel box. TextEdit::fixed_width takes impl Into<Length> (f32 call sites keep compiling as px) and the time picker's digit fields move to Length::fluid( 72.0 ), matching their fluid font so digits can no longer outgrow the box. Dialog::max_width takes impl Into<Length> with a Length::fluid( 480.0 ) default so the card scales with the stock buttons inside it, and the card's interior no longer stacks the column() default 16 px padding on top of CARD_PADDING — that double inset squeezed the action row until its buttons clipped on narrow windows. App::on_pointer_axis now triggers a view rebuild and repaint; previously state mutated in the hook did not paint until the next unrelated event. Examples reworked to be honest demos: responsive's mode/density controls become stock buttons in a grid/column so they follow the modes they demonstrate instead of overflowing; dialog's openers stack vertically, and the example gains the app-level ESC handler so the ESC chain closes an open dialog first and quits second; widgets' tab strip now switches real per-tab pages; carousel gains pointer/touch drag through the horizontal-swipe hooks (crustace's pager pattern), one-tile-per-detent mouse wheel, and snap math driven by the real surface width from on_resize instead of a hardcoded 800; clip_path arranges its cells by ltk::orientation() and sizes them from the counter-axis of the flow. |
||
|
admin
|
ce893ac776 |
responsive fluid/physical scaling, widget-API stabilization, and perf guardrails
Responsive scaling. ltk now offers two first-class ways to size a UI so it adapts across screens, chosen per process via `WidgetScaling { Fluid, Physical }` (`set_widget_scaling` / `widget_scaling`, default `Fluid`). Fluid sizing (`Length::fluid( px )`) makes a design pixel a proportion of the surface's smaller side, calibrated against a reference width (`set_fluid_reference` / `fluid_reference`, 412 px default) and bounded by `FLUID_MIN` / `FLUID_MAX`; physical sizing (`Length::dp( px )`) is a constant-physical-size pixel scaled by display density (`set_density` / `density`). `Length` gains `orient( portrait, landscape )` — resolve one value in portrait, another in landscape — plus `widget( px )`, which picks fluid or dp per the active mode. Canvas exposes `geom_px` (geometry, resolved in physical layout space) and `font_px` (font size, bridging logical / physical per mode) so widgets and apps share one resolution path. Note the rename: `set_design_reference` / `design_reference` became `set_fluid_reference` / `fluid_reference`, and `Length::dp` changed meaning — the old surface-proportional behaviour now lives on `Length::fluid`.
Widgets. Every stock widget resolves its default geometry and font through the widget-scaling mode instead of frozen pixels, so a whole UI scales coherently without per-call units. New size builders where they were missing: `button` gains `font_size` / `height`, `text_edit` gains `height` / `font_size_fluid`, `separator` gains `pad_v`, and assorted widgets accept a `Length` where they previously took only `f32`.
Overlays. `OverlaySpec::size` is now `( Length, Length )` instead of `( u32, u32 )`, resolved against the main surface when the overlay is materialized, so overlays can scale with the display; `Length::px( … )` reproduces the old fixed sizing.
API stabilization (toward 1.0). Widget struct fields are now `pub( crate )` — they are configured through builders, not field access — except the value / state types apps genuinely read or construct (`Time`, `Date`, `ComboState`), which stay public. The internal `test_support` helpers move behind a `test-support` Cargo feature (off by default, so third-party builds never see them; ltk's own `make test` enables it). `Separator` drops its `0.0`-means-mode sentinel for `Option<Length>`, so an explicit `pad_v( 0.0 )` is a real flush divider distinct from the mode-following default.
Performance guardrails. Opt-in diagnostics via `LTK_PERF_WARN=1` warn about stuck animations, sustained software-render animation, and low `poll_interval`; software-rendered animation is capped near 30 Hz to spare CPU on machines that fall back off EGL. Apps can override the cap with `App::cap_software_animation`.
Docs and build. The two scaling modes are documented in README, onboarding and architecture, with the earlier gradient / backdrop doc drift cleaned up. The Makefile now ships the `locales/` directory into the packaged crate (fixing i18n keys rendering raw for downstreams), builds the new `responsive` example, and runs tests with `--features test-support`.
|
||
|
admin
|
4a80165428 |
event_loop, a11y, text_shaping: AccessKit AT-SPI2 bridge, cross-app clipboard, xdg-activation, HarfBuzz shaping, multi-touch hooks
Five orthogonal capabilities land together because they share the same `try_run` plumbing: an optional global is bound at startup, a piece of state is added to `AppData`, the run-loop iteration drains an inbox / pushes a frame snapshot, and the public surface gains a small set of opt-in `App` hooks. Nothing here breaks an existing app — every new path degrades to a no-op when the compositor does not advertise the relevant global or when the platform adapter cannot start. AT-SPI2 accessibility via AccessKit. A new `src/a11y/` module owns the platform adapter and the inbound `ActionRequest` channel. `A11yState::try_new` constructs an `accesskit_unix::Adapter`; when the AT-SPI2 daemon is not on the session bus (headless CI, locked-down compositors) the constructor returns `None` and the rest of the pipeline runs unchanged. After every successful `draw_frame`, the run loop builds a fresh `accesskit::TreeUpdate` from `widget_rects` and pushes it through the adapter — main surface plus every visible overlay, each translated to global coordinates via `surface_offset_for` so screen readers report positions in the same frame the user sees. Buttons / toggles / checkboxes / radios / list items / sliders / text edits map to the matching `Role`s; `Click` and `Focus` actions are advertised on every interactive node; inbound action requests are drained at the top of each iteration and translated into a synthetic press / focus on the matching widget. The integration is documented as best-effort in `docs/architecture.md` under "Known gaps and non-goals": hierarchical nesting, per-widget accessible names, live regions and `Action::SetValue` are listed as the natural follow-ups that the foundation now supports but does not yet wire. Cross-application clipboard via `wl_data_device_manager`. A new `src/event_loop/data_device.rs` bridges the existing process-local `clipboard: String` to the Wayland selection. Outbound (Ctrl+C / Cut): after the local clipboard is populated, `publish_clipboard_selection` creates a `CopyPasteSource` offering `text/plain;charset=utf-8` and installs it as the seat's selection; `DataSourceHandler::send` writes the cached string into the fd the peer hands us. Inbound (Ctrl+V from another app): `DataDeviceHandler::selection` asks for the offered text via `WlDataOffer::receive`, spawns a tiny worker thread to drain the read pipe with a 16 MiB cap to prevent paste-bomb DoS, and posts the result back through an `mpsc::Sender` that the run loop drains each iteration into `data.clipboard`. The `clipboard:` field's doc-comment is updated to reflect the new behaviour: process-local when the compositor does not advertise the global, synchronised with the seat selection otherwise. External drag-and-drop reception. The same `data_device` module handles `DragOffer` enter / motion / leave / drop_performed: `on_drop_motion( x, y )` fires while the drag hovers over the surface, `on_drop_leave()` when it withdraws without dropping, and `on_drop_received( x, y, mime, text )` when an external payload (`text/uri-list`, `text/plain`, …) is released on top of an ltk window. The receive path reuses the same worker-thread / channel pattern as the clipboard so the run loop never blocks on the read fd. Three new `App` hooks expose the events with no-op defaults; apps that ignore them get the previous behaviour. `xdg-activation-v1`. The global is bound optionally; when it is present, `try_run` reads `$XDG_ACTIVATION_TOKEN` from the environment, removes it immediately (single-use; preventing leaks into child processes) and stashes it on `AppData::activation_token_pending`. After the first successful configure of the main surface — the earliest point at which `xdg_activation_v1.activate` is meaningful — the token is consumed once and the surface raised to focus. Compositors without the global leave `activation_state` as `None` and the inbound path silently degrades. An `App::request_activation_token` outbound path is reserved on the trait but not yet exercised here. HarfBuzz shaping. A new `src/text_shaping.rs::shape_line` drives both renderers: the logical-order string is run through `unicode-bidi`, split into per-font sub-runs, and shaped through `rustybuzz`. Each `PositionedGlyph` carries the per-font `glyph_id`, the visual advance and the ink offsets — exactly what `fontdue::Font::rasterize_indexed` needs to render Arabic connected forms, Devanagari clusters and CJK shaped glyphs correctly. The GLES atlas is re-keyed on `(glyph_id, size_bits, font_id)` so glyphs from different fonts at the same size no longer collide, and the atlas format is selected per ES profile (`GL_R8` / `GL_RED` on ES3, `GL_LUMINANCE` on ES2) — the fragment shader samples `.r` for both, since `GL_LUMINANCE` replicates the coverage byte into `.r=.g=.b`. Software path follows the same key. New `Cargo.toml` deps: `unicode-bidi = "0.3"`, `rustybuzz = "0.14"`. Multi-touch hooks. `App::on_touch_down / on_touch_move / on_touch_up( id, x, y )` expose the raw `wl_touch.id` of every secondary finger. The first finger to land remains the *primary slot* and is fed through the regular gesture machine (`on_pointer_*`, swipe, scroll, long-press, drag-and-drop). Every additional finger fires the new callbacks instead, leaving the existing single-slot behaviour untouched for apps that do not override them. This is the substrate for app-defined pinch-zoom / two-finger pan; the toolkit itself does not yet ship a built-in pinch gesture (called out in the same "Known gaps" doc section). `event_loop::frame` extracted from `draw/mod.rs`. The `draw_frame` orchestrator and its per-format SHM helper (`pick_shm_format`) move into `src/event_loop/frame.rs`, leaving `draw/` strictly responsible for per-surface paint primitives. The import in `event_loop/run.rs` is rewritten accordingly; `draw/mod.rs` shrinks from 192-line orchestrator to a thin module index. Overlay teardown safety. `AppData::discard_overlay( id )` synchronously removes a destroyed overlay from the map and rewrites every per-device focus that pointed at it (pointer, keyboard, every touch slot), migrating an in-flight long-press drag to the main surface the same way `reconcile_overlays` does. Used by the compositor-driven destruction paths (`PopupHandler::done`, `LayerShellHandler::closed`) where waiting for the next reconcile would leave a window in which `surface()` / `surface_mut()` panic. The non-panicking siblings `try_surface` / `try_surface_mut` are added for callers on async dispatch paths (IME `Done`, tooltip arm) that may race a teardown. Miscellaneous. CI: `master` → `main` to match the actual default branch. `Makefile` adds `cargo run --example dialog` to the examples target. `src/lib.rs` re-exports `widget::scroll::ScrollAxis` so apps can configure a `scroll()` axis without reaching into a `pub(crate)` module. `Cargo.toml` adds `accesskit = "0.17"` and `accesskit_unix = "0.13"`. `docs/architecture.md` gains the "Known gaps and non-goals" section that enumerates the new capabilities, what still ships flat, and what is deferred (per-widget a11y labels, primary selection, intra-process multi-touch gestures, `wp_fractional_scale_v1`). |
||
|
admin
|
bbab5e238d | First commit. Version 0.1.0 |