docs(scroll): describe the fixed-step model and document frame_hold

Since #545 a crisp speed from scroll_config.configure() makes the helper
advance a fixed whole-pixel step per presented frame with no clock, and the
display manager's frame hold is part of the speed. The docs still described
the removed wall-clock model:

- scroll_config's module and configure() docstrings said speed is applied
  in time-based mode and that omitting the hold "falls back to fractional
  pixels"; omitting it actually runs the scroll frame_hold times too fast.
- SCROLL_PERFORMANCE.md said ScrollHelper accumulates elapsed time in both
  modes, and read a 20 ms stats median as missed refreshes although that
  is a healthy 50 px/s (hold 2) scroll. It now explains the fixed step,
  the hold-dependent healthy median, that target_fps plays no part, and
  that a hand-added scroll_pixels_per_second loses to a schema-default pair.
- PLUGIN_API_REFERENCE.md documented set_scrolling_state(is_scrolling)
  without frame_hold; it now documents the parameter (core 3.4.0) with a
  configure() + set_scrolling_state example.
- update_scroll_position/set_scroll_speed and set_scrolling_state
  docstrings say the same.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
Chuck
2026-09-17 08:56:37 -04:00
co-authored by Claude Opus 5
parent e62ec164f8
commit ec57b35448
5 changed files with 143 additions and 38 deletions
+46 -7
View File
@@ -514,21 +514,59 @@ self.display_manager.draw_text_with_icons(
For plugins that implement scrolling content, use these methods to coordinate with the display system.
#### `set_scrolling_state(is_scrolling: bool) -> None`
#### `set_scrolling_state(is_scrolling: bool, frame_hold: int = 1) -> None`
Mark the display as scrolling or not scrolling. Call when scrolling starts/stops.
Mark the display as scrolling or not scrolling, and set this scroll's frame
pacing. Call it when a scroll starts (calling it on every scroll frame is fine)
and with `False` when it stops.
**Parameters**:
- `is_scrolling` (bool): True if currently scrolling, False otherwise
- `frame_hold` (int, default 1): how many panel refreshes each pushed frame is
held for (clamped to 1-255; ignored when `is_scrolling` is False, which
resets it to 1). Pass the `frame_hold` of the settings
`src.common.scroll_config.configure()` returned. Added in core 3.4.0.
**Why `frame_hold` matters**: `scroll_config.configure()` snaps the speed to
one the panel can show in whole pixels and sets the `ScrollHelper` to advance a
fixed number of pixels on every presented frame -- no clock is consulted. The
panel presents frames at its refresh rate divided by the hold, so the hold is
part of the speed. Omit it and a 50 px/s scroll (1px every 2nd refresh on a
100 Hz panel) runs at 100 px/s. The hold is not applied by `configure()`
because it must not outlive the scroll: plugins share one display manager.
**Example**:
```python
from src.common import scroll_config
from src.common.scroll_helper import ScrollHelper
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.scroll_helper = ScrollHelper(
self.display_manager.width, self.display_manager.height, self.logger)
# ...later, hand it content with self.scroll_helper.set_scrolling_image(img)
self.scroll_settings = scroll_config.configure(
self.scroll_helper,
plugin_config=self.config,
global_config=self.global_config,
display_manager=self.display_manager,
plugin_logger=self.logger,
)
def display(self, force_clear=False):
self.display_manager.set_scrolling_state(True)
# Scroll content...
self.display_manager.set_scrolling_state(False)
self.display_manager.set_scrolling_state(
True, frame_hold=self.scroll_settings.frame_hold)
self.scroll_helper.update_scroll_position()
self.display_manager.image = self.scroll_helper.get_visible_portion()
self.display_manager.update_display()
if self.scroll_helper.is_scroll_complete():
self.display_manager.set_scrolling_state(False)
```
Don't pace the loop with `time.sleep()`: `update_display()` blocks on the
panel's vsync, which is what paces a scroll. See `docs/SCROLL_PERFORMANCE.md`
for choosing a speed.
#### `is_currently_scrolling() -> bool`
Check if the display is currently in a scrolling state.
@@ -998,9 +1036,10 @@ if "weather" in enabled_plugins:
self.display_manager.update_display()
```
3. **Handle scrolling state**: If your plugin scrolls, use scrolling state methods
3. **Handle scrolling state**: If your plugin scrolls, use scrolling state methods,
passing the frame hold `scroll_config.configure()` returned
```python
self.display_manager.set_scrolling_state(True)
self.display_manager.set_scrolling_state(True, frame_hold=settings.frame_hold)
# Scroll content...
self.display_manager.set_scrolling_state(False)
```
+46 -12
View File
@@ -83,8 +83,9 @@ second refresh, instead of half a pixel every refresh (which has no good
rendering, only a choice between blur and judder).
`scroll_config.configure()` snaps the requested speed to the nearest entry on
the ladder and reports the hold that speed needs. It does **not** apply the
hold: the hold belongs to a scroll, not to a plugin's lifetime, and plugins
the ladder, sets the helper to advance that entry's whole-pixel step on every
presented frame (`ScrollHelper.set_pixels_per_frame`), and reports the hold
that speed needs. It does **not** apply the hold: the hold belongs to a scroll, not to a plugin's lifetime, and plugins
share one display manager -- one set at construction is reset the moment any
other plugin finishes scrolling. Apply it yourself when the scroll starts:
@@ -102,10 +103,18 @@ self.display_manager.set_scrolling_state(True, frame_hold=settings.frame_hold)
Passing `display_manager` only lets `configure` read the true refresh rate from
`display.hardware`, which a plugin config cannot see. Skipping the
`set_scrolling_state` call is the mistake that matters: the speed still
resolves, but the panel keeps presenting a new frame every refresh, so a slow
snapped speed falls back to fractional pixels. Pass `snap_to_crisp=False` to
keep an exact requested speed and accept the artefacts.
`set_scrolling_state(True, frame_hold=...)` call is the mistake that matters.
The helper consults no clock in this mode -- it moves the fixed step once per
`update_scroll_position()` call, and `SwapOnVSync` is what paces those calls --
so without the hold the panel presents a new frame every refresh and the scroll
runs `frame_hold` times too fast: 50 px/s (hold 2) plays at 100 px/s.
Pass `snap_to_crisp=False` to keep an exact requested speed and accept the
artefacts. The helper then paces off elapsed time instead of stepping, and the
hold is 1.
The General tab's `target_fps` ("Scroll Frame Rate") plays no part in any of
this: frames are presented at the panel refresh divided by the hold.
Speeds slower than about 20 px/s are stepped no matter what, because a 1-pixel
advance at 20 fps is simply a coarse increment. That is the pixel pitch, not a
@@ -123,9 +132,12 @@ settings = scroll_config.configure(
self.scroll_helper,
plugin_config=self.config,
global_config=self.global_config,
refresh_hz=scroll_config.refresh_hz_from_config(self.global_config),
display_manager=self.display_manager,
plugin_logger=self.logger,
)
# each frame of a scroll (or at least when it starts):
self.display_manager.set_scrolling_state(True, frame_hold=settings.frame_hold)
```
It resolves every config shape in one place, applies the speed, and returns
@@ -154,6 +166,14 @@ it is always present and always wins, so the documented settings become
unreachable. That is a real, shipped bug — see
[ledmatrix-plugins#408](https://github.com/ChuckBuilds/ledmatrix-plugins/issues/408).
The flip side: a `scroll_pixels_per_second` you add by hand is ignored whenever
the plugin's config also carries the pair, which it does whenever the pair has
a schema default. Set the speed through the pair instead.
The sports scoreboards (`src.common.sports_scroll`) are the exception to all of
the above: they read `scroll_settings.scroll_speed` per league as px/s directly,
and their `scroll_delay` is kept for compatibility but ignored for pacing.
If you are writing a plugin: do not give a deprecated key a schema default.
## What was actually wrong
@@ -198,8 +218,16 @@ zero pixels and rendered an identical frame, which dirty-tracking skipped, so
it returned in ~2 ms and the beat repeated. No `scroll_delay` value tunes this
out — a shorter delay just trades stalled frames for periodic double-steps.
`ScrollHelper` now accumulates elapsed time in both modes at the same
configured speed, so position stays proportional to real time.
A crisp speed configured through `scroll_config` no longer consults a clock at
all. Once `SwapOnVSync` blocks until the panel has taken the frame, the frame
count is a truer clock than `time.time()`, so the helper advances a fixed whole
number of pixels per presented frame (`set_pixels_per_frame`) and the display
manager holds each frame for `frame_hold` refreshes. Every frame moves the eye
by the same amount.
The time-based path remains only for callers that set a speed directly or pass
`snap_to_crisp=False`. There, frame-based mode no longer steps either: it
advances by elapsed time at `scroll_speed / scroll_delay` px/s.
## Diagnosing a juddery scroller
@@ -222,11 +250,17 @@ p95 10.11ms max 12.03ms min 7.98ms | stalls 0 (0.0%) skips 0 (0.0%)
Reading it, on a 100 Hz panel:
A healthy median is the refresh period times the scroll's frame hold: 10 ms
for a hold of 1 (100 px/s), **20 ms for 50 px/s** (hold 2), 30 ms for 33.3 px/s.
A 20 ms median on a 50 px/s scroll is the hold doing its job, not missed
refreshes. The `Scroll configured:` log line gives the hold (`1px every 2
refreshes`).
| you see | it means |
|---|---|
| median 10 ms, p95 within ~0.5 ms of it | healthy — locked to the panel |
| p95 or max at 20/30/50 ms | frames missing refreshes — per-frame work is overrunning, or a background thread is holding the GIL |
| non-zero **skips**, or a median *below* 10 ms | **duplicate frames** — the swap was skipped because the image did not change, so the frame never waited on vsync. The scroller is advancing less than one pixel per frame. |
| median = refresh period × hold, p95 within ~0.5 ms of it | healthy — locked to the panel |
| p95 or max a whole refresh period or more above that median | frames missing refreshes — per-frame work is overrunning, or a background thread is holding the GIL |
| non-zero **skips**, or a median *below* the expected one | **duplicate frames** — the swap was skipped because the image did not change, so the frame never waited on vsync. The scroller is advancing less than one pixel per frame, which a crisp fixed-step scroll never does; look for a plugin pacing off time or not passing the hold. |
| non-zero **stalls** | frames past 1.5× the median, which is the measure of judder that survives averaging |
`stalls` and `skips` are both counted against that window's own median, so they
+30 -13
View File
@@ -22,12 +22,27 @@ shimmer) or repeat frames (which reads as judder). :func:`resolve` warns when
the requested speed will not divide evenly, because that is a real display
artefact and not a rounding detail.
Speed is always expressed to the helper as pixels per second and applied in
time-based mode. Frame-based stepping gates motion on a wall clock at
``1/scroll_delay`` steps per second; plugins set ``scroll_delay`` to the frame
period, which puts that comparison exactly on its own threshold and makes the
step count flip on sub-millisecond jitter. Accumulating elapsed time keeps
position proportional to real time instead.
How the speed is applied
------------------------
:func:`configure` snaps the requested speed to a :class:`CrispSpeed` -- a whole
number of pixels per presented frame, with each frame held for ``frame_hold``
panel refreshes -- and puts the helper in fixed-step mode
(``ScrollHelper.set_pixels_per_frame``). In that mode the helper consults no
clock: every ``update_scroll_position()`` call moves exactly that many pixels.
``SwapOnVSync`` blocks until the panel has taken each frame, so the frame count
is the clock, and the speed on the panel is::
px/s = refresh_hz / frame_hold * pixels_per_frame
That makes the hold part of the speed. The caller must pass
``settings.frame_hold`` to ``display_manager.set_scrolling_state(True, ...)``;
a caller that omits it is presented every refresh and scrolls ``frame_hold``
times too fast. The refresh is the panel's (``display_manager.refresh_hz``,
i.e. ``display.hardware.limit_refresh_rate_hz``) -- never the global
``target_fps``, which no longer paces anything.
With ``snap_to_crisp=False`` there is no fixed step: the helper is set to
px/s and advances by elapsed time, and the hold is 1.
"""
from __future__ import annotations
@@ -330,17 +345,18 @@ def configure(
) -> ScrollSettings:
"""Resolve the config and apply it to ``scroll_helper``.
Applied in time-based mode: see the module docstring for why frame-based
stepping is not used. ``hasattr`` guards keep this usable against older
ScrollHelper builds that a plugin may be running on.
Frame-based mode is switched off, the (snapped) speed is set, and with
``snap_to_crisp`` the helper steps a fixed whole-pixel amount per presented
frame -- see the module docstring. ``hasattr`` guards keep this usable
against older ScrollHelper builds that a plugin may be running on.
:param display_manager: consulted for the panel's refresh rate only (it can
see display.hardware; a plugin cannot). The frame hold is NOT applied
here -- see the note in the body. The caller must pass
``settings.frame_hold`` to ``display_manager.set_scrolling_state(True,
...)`` when it starts scrolling, or a sub-refresh speed still presents
a new frame every refresh and the motion falls back to fractional
pixels.
...)`` when it scrolls. The helper moves its fixed step on every call,
so without the hold each step is presented every refresh and the
scroll runs ``frame_hold`` times faster than the resolved speed.
:param snap_to_crisp: move the requested speed to the nearest speed the
panel can show in whole pixels. On by default because a speed that does
not divide evenly has no good rendering, only a choice of artefacts.
@@ -357,7 +373,8 @@ def configure(
# refresh to fill in target_fps, pixels_per_frame and the judder warning,
# so deriving it afterwards described a 100Hz panel to everyone running at
# 60 -- and with snap_to_crisp=False nothing downstream corrected it, so
# set_target_fps() paced the helper to 100 FPS on a 60Hz panel.
# the settings and the helper's (informational) target_fps said 100 FPS on
# a 60Hz panel.
hz = _coerce(refresh_hz)
if hz is None and display_manager is not None:
hz = _coerce(getattr(display_manager, "refresh_hz", None))
+13 -4
View File
@@ -282,7 +282,14 @@ class ScrollHelper:
def update_scroll_position(self) -> None:
"""
Update scroll position with high FPS control and handle wrap-around.
Advance the scroll by one presented frame and handle wrap-around.
With a fixed per-frame step (set_pixels_per_frame, which
scroll_config.configure sets for a crisp speed) every call moves
exactly that many pixels and no clock is read; the caller's
vsync-blocking swap, held for ``frame_hold`` refreshes, sets the rate.
Otherwise the position advances by elapsed time at the configured
speed.
"""
if not self.cached_image:
return
@@ -823,10 +830,12 @@ class ScrollHelper:
def set_scroll_speed(self, speed: float) -> None:
"""
Set the scroll speed.
Set the scroll speed, and leave fixed-step mode.
In time-based mode: pixels per second (typically 10-200)
In frame-based mode: pixels per frame (typically 0.5-5 for smooth scrolling)
In time-based mode: pixels per second (clamped to 1-500).
In frame-based mode: pixels per ``scroll_delay`` seconds (clamped to
0.1-5), still applied by elapsed time as scroll_speed / scroll_delay
px/s.
Args:
speed: Scroll speed (interpretation depends on frame_based_scrolling mode)
+6
View File
@@ -1415,6 +1415,12 @@ class DisplayManager:
pixel every second refresh, which is how a scroll runs at half the
refresh rate without fractional pixel positions.
The hold is part of the scroll's speed. A ScrollHelper configured by
``scroll_config.configure()`` advances a fixed whole-pixel step per
presented frame and reads no clock, so pass the returned
``settings.frame_hold`` here: a scroll that leaves it at 1 is
presented every refresh and runs ``frame_hold`` times too fast.
The hold is set here rather than once at plugin construction because
it must not outlive the scroll that asked for it: plugins share one
display manager, so a hold left set by whoever scrolled last would