A test and code-analysis round of 0.2.72 to 0.2.76 (suites at every tag, sanitizers, ESP32 and GitHub builds, host
harnesses on both boards, the editor in a browser, best practice checked against ESPHome, LVGL, Home Assistant and
Vue) found these, fixed here with Max's choices for the product questions:
- Firmware 0.2.65: the second hand of a single analog clock moves again; on a page that is one full on/off tile the
waking push also switches it; on the Guition a flick is never a tap and a finger that slides off a tile doesn't tap
it; Previous/Next react to a firm press; the same Go to page tile may sit on several pages; a layout the screen has
no memory for is refused instead of crashing; 64-bit redraw mask for 48 tiles; Montserrat 14 no longer built.
- App: Save refuses a layout too long for one message or too big for eight pages; an offline screen keeps its
firmware version from the device registry; HA refusals reach the editor; 128 KB requests; changelog out of the
live stream; cached hashed assets; tini as init and a clean stop; changelog parsing without asterisks or crashes.
- Editor: clicking the open screen keeps unsaved edits, edits during a save stay unsaved, a Page row and phone
dragging to other pages, repeated page tiles, limits from the server, HA's error messages, a type-checked build.
- tools/check.sh + GitHub Actions + a CYD flash budget; docs match 48 tiles and the new editor.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The Previous and Next bars ran their tap through the tile guard, whose 600 ms
same-tile window (so one contact never fires a tile twice) dropped the second
and third tap of a quick Next, Next, Next. They now use accept_repeat, the
-/+ keys' rule since 0.2.25: every clean tap counts, only a contact within
150 ms of the last accepted tap on the same button is a bounce. The page fill
already dropped a fill under way when the next page was asked for. Firmware
0.2.61; test_cyd_ui.cpp covers the page buttons; host harness and check builds
in docs/TEST_RESULTS_0272.md.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A light's colour slider on Studio 1 sometimes sent red on release. Measured
live: an automation with a state trigger on a six-lamp group (no `to:`) set
"Standby after" to its current value after every colour change. The screen
reported the setting, ESP Screen Manager saved the layout and resent layout,
header and every tile state, and the screen drew its whole page again,
reading no touch input meanwhile. The GT911's stray (0, 0) contact then
became the finger's last position, and LVGL 9.5 set the slider from it on
release. Both red commits left 75-85 ms after such a resync.
The app no longer saves or resends a setting event whose value the layout
already has. The four setting numbers return before the preference write
and the event when nothing changed. A layout message that changes nothing
on the screen no longer runs layout_changed() and refresh_all(); the tile
states after it redraw their own tiles. The Guition wraps the LVGL touch
read callback with cyd::GhostTouch, which repeats the previous read for a
pressed (0, 0), and the edge swipe skips that sample too. Both slider
handlers log a jump to an end on release.
Checked with 214 Python and 14 C++ tests, generate_packages --check, and
ESP32 builds of both Easy profiles. Firmware 0.2.47.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The CYD restarted under load: a slider fill rounded less than its track made
LVGL draw the fill into a buffer of its own size on every redraw (15-20 KB), and
when that allocation failed the refresh retried until the task watchdog fired.
Track and fill now share one radius, and tests/test_layer_free.py keeps every
layer-making style out of the profiles, the packages and the firmware headers.
A tile keeps only the fields every card uses; climate modes, select options,
weather, sun and timer times, the media title and the vacuum rows moved to an
Extra block that exists only while a state needs one, and the state copy became
a fingerprint. runtime_tiles::model went from 24,296 to 8,216 bytes.
The manual profile that bound tiles to fixed entities in the YAML is gone:
331 TILE* substitutions, DYNAMIC_TILES/TILE_COUNT/DIRECT_ACTIONS, the 80 Home
Assistant subscriptions, its tap handlers and scripts, its own vacuum card, its
guides and helper scripts, and the five stale header copies in the repository
root that two C++ tests were compiling instead of the components. esphome config
of both Easy profiles is unchanged by that cleanup.
tile_scroll and the wake overlay no longer light up on a tap, and the debug
sensor min_free reports the lowest free heap since boot. On the bench CYD static
RAM went from 32.0% to 22.8% and free heap from 64-77 KB to about 101 KB, with
78 KB as the lowest point over thirteen hours without a restart.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Everything a user sees or reads is now English: both boards' labels,
status replies and logs, the ESP Screens editor and Alerts cheatsheet,
and the README and guides (docs/TILES.md, CALIBRATING.md, ACCEPTANCE.md
and TROUBLESHOOTING.md replace the Dutch file names). Labels follow
Home Assistant's own English names (Heat/Cool modes, Heating/Cooling
actions, Away, Returning to dock, weather conditions), dates use English
abbreviations and the top bar writes numbers the English way. Docs use
HA 2026.9's menu names (Settings -> Apps, App store, Settings -> Tools).
Unused legacy files (home-like.yaml, buttons.yaml, a.out) are removed.
Firmware 0.2.34 names its entities in English. Home Assistant registers
a renamed ESPHome entity under a new entity ID, so the inbox
text.<screen>_tegelinstellingen becomes text.<screen>_tile_settings and
the layout stored under the old ID would be lost. The manager now
recognises the screen by its device, during a run or after a restart,
and moves its layout and update history along; the Update button and
the nightly round follow the screen through the rename. Screens that
still run 0.2.33 or older keep working: both languages are matched.
Main is the only release branch from now on.
Tested on a Home Assistant Yellow (HA 2026.9.1) with both bench boards:
see docs/TEST_RESULTS_0240.md.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
The 0.2.32 edge swipe listened on LVGL's input device, but the LVGL 9.5.0
bundled by ESPHome sends the indev PRESSED and RELEASED without PRESSING,
so the swipe never received a position update: on Studio 1 the log showed
the GT911 press and then nothing. The Guition follows the swipe through
on_touch, on_update and on_release again, the route that worked live in
0.2.24 once the card-close fix was in, with ESPHome's own rotation mapping,
the 45-degree rule, end() on release so a stale armed state cannot fire on
the next touch, and the "genegeerd"/"niet gevuurd" log lines. Firmware
0.2.29 for both boards; the app only moves its target version.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
A firm press on Studio 1 that drifted 73 px was still refused by the touch
guard's movement limit. On a capacitive panel that limit has no job left:
the edge swipe consumes its own tap, sliders capture their own drag, and
LVGL already drops the click when the finger leaves the tile. The guard's
limit can now be 0 (no limit); the Guition profile sets it so, the CYD
keeps 56 px because its resistive panel jumps. Firmware 0.2.28 for both
boards; the app only moves its target version.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Read live on a Guition: every edge swipe was recognised but refused by the
"card open" check, because the light, climate and vacuum cards cleared
their overlay on close (X, tap beside the card, wake from standby) without
forgetting active_entity. One opened card blocked swiping until the next
layout change; the old whole-screen gesture and the CYD had the same gate.
Both boards now close cards through one script, close_cards, which hides
every overlay and clears that state; wake_display uses it too.
The Guition edge swipe now follows LVGL's own pointer events
(lv_indev_add_event_cb: PRESSED begins, PRESSING flips the page and calls
lv_indev_wait_release from inside LVGL) instead of ESPHome's touchscreen
on_update, so the point arrives already rotated. The direction rule is 45
degrees instead of twice as horizontal, and the log now says why a swipe
was refused or how far an unfinished one got. Firmware 0.2.27 for both
boards; the app only moves its target version.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Page swipes were an LVGL gesture anywhere on the screen: 50 px, fast only,
and almost always starting on a tile that lit up as pressed. On the Guition
a touch that starts within 32 px of the left or right edge and travels
40 px inward, clearly sideways, now flips one page (from the right edge
leftwards is next, from the left edge rightwards previous), at any speed
and in every rotation; a swipe that starts in the middle does nothing, so
taps and drags on tiles can no longer change the page by accident. The
whole-screen gesture is removed from the Guition profile; the CYD keeps
it. Firmware 0.2.24 for both boards, the app only reworded the setting.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The touch guard dropped a tap as soon as the finger drifted more than
18 px (under 3 mm) from its very first contact point, on both boards and
regardless of the page-swipe setting; LVGL itself never cancelled these
clicks (no scrollable ancestors, gestures do not suppress clicks). A finger
that flattens or rolls while pressing was enough, and the tap vanished
without a trace in the log.
The limit is now about one centimetre per board (TOUCH_MOVE_LIMIT_PX:
Guition 67 px, CYD 56 px), measured from a reference that settles over the
first four samples; only the contact that started the touch is followed,
so a second finger no longer counts as movement; the minimum contact time
is per board too (TOUCH_MIN_PRESS_MS: 60 ms resistive, 20 ms capacitive).
Every refused tap is logged on INFO with its reason (moved, too short,
already used, bounce). Firmware 0.2.23 for CYD and Guition.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
App 0.2.25 / firmware 0.2.21. The 600 ms touch-guard lockout dropped a second
tap on the same -/+ key; stepper keys now use a 150 ms bounce guard, repeat
while held (3 per second), and still send one call after 700 ms of rest.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
App 0.2.23 / firmware 0.2.19.
- Wide cards of climate, switch, light, fan, vacuum, cover, media_player,
number, select, timer, scene, script and button show Home Assistant style
controls on the right (keys, -/+ pill with debounced send, slider with
handle, toggle, run key); per-tile choice `options.controls` in the editor
with mockup preview; the add-on sends only the set a card really shows.
- Weather card: now + next hours and coming days with rain, from daily and
hourly forecasts; scenes/scripts/buttons show when they last ran.
- Climate card: power key, tap-to-toggle option, fan and swing rows in a
redesigned mode picker (grey backdrop, white chips, accent selection).
- Busy state as a spinner on a translucent sheet; render guards so only
changed styles reach LVGL; dark theme removed from the runtime code.
- supported_features above 1e6 reaches the screen; middle dot in text fonts;
18 control glyphs in the icon fonts; diagnostics services
preview_climate_picker and preview_busy; demo layout --controls.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Max Gramser