Skip to content

docs(spec): keep the long-press and keyboard-swipe rows with a screen in 3.0 - #12

Open
djbclark wants to merge 5 commits into
brobata:mainfrom
djbclark:spec-keep-long-press-and-swipe-screens
Open

djbclark wants to merge 5 commits into
brobata:mainfrom
djbclark:spec-keep-long-press-and-swipe-screens

Conversation

@djbclark

Copy link
Copy Markdown

Docs only, against the 3.0 spec. No code.

Two preference groups sit in settings-catalog.md §13 under "rows with no screen in 2.x: undecided per row", although their behaviour is kept by the subsystem documents and each has a user asking for a screen:

Wording and placement are yours to change; the point is that the two rows stop being constants nobody chose.

Signed off under CONTRIBUTING.md's agreement.

🤖 Generated with Claude Code

… in 3.0

Both rows sit in the settings catalog's "no screen in 2.x, undecided per
row" bucket, but their behaviour is kept by the subsystem documents and
each has a user asking for it: a held letter typing its capital is what
BlackBerry 10 users expect (brobata#9), and swipe-to-accept has been silently off
on every fresh install since the 1.0.2 screen removal (brobata#11). Give each its
own Keep/Drop row naming the screen it needs, and point the native-provider
decision in the trackpad document at the capture offered in brobata#11.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Daniel JB Clark <djbclark@gmail.com>
…ird-boundary defect

D2 was carried on the maintainer's word and asked for a getevent capture;
here it is from an Elite on V02.00.04 (X 0 to 1079, Y 0 to 599 at the
kernel, X 0 to 1079 and Y 0 to 748.75 as Android reports them to the
IME). With that measured, the native provider's 1440 px third boundaries
are no longer an unverified suspicion: the right third begins at 960 of
1079, so the right-hand suggestion is nearly unreachable (brobata#14). Say so in
section 3.3 and the quirks table, and tell 3.0 to take the boundaries
from the device's declared range.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Daniel JB Clark <djbclark@gmail.com>
@djbclark

Copy link
Copy Markdown
Author

Second commit: the getevent -p capture D2 asked for, from an Elite on V02.00.04, and the third-boundary defect it confirms (#14). Section 3.3 and the quirks table now state the measured range instead of "unverified".

djbclark and others added 2 commits September 21, 2026 09:26
…DPAD_UP key event

Captured on an Elite on V02.00.04 with the native provider enabled: a
flick up on the keys never reaches the IME's decor view as touchpad
motion. system_server synthesizes KEYCODE_DPAD_UP (scanCode 19, deviceId
of the touchPad device, source KEYBOARD) and delivers it to the app and
the IME, which is why a flick moves the caret a line today. Record it as
D7, note in 3.2 that the motion provider cannot fire on the Elite, and
point the Keep/Drop decision at a keycode provider like the existing
swipe-to-delete path (brobata#11).

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Daniel JB Clark <djbclark@gmail.com>
…hpad motion with Scroll Assistant on

The previous D7 attributed a synthetic KEYCODE_DPAD_UP to a flick. It
was a drag in Unihertz's double-tap Cursor Assistant mode. A real flick,
tested end to end today, reaches the IME's focused decor view as
SOURCE_TOUCHPAD motion from the touchPad device, but only while Scroll
Assistant is on; with it off nothing arrives. Eleven flicks were
accepted and all three thirds reached once the boundaries came from the
device's X range. Natural flicks cover 280 to 470 px, so the default
500 px threshold rejects most of them. Record the corrected fact, fix
section 3.2, and propose keeping the native provider (brobata#11, brobata#14).

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Daniel JB Clark <djbclark@gmail.com>
@djbclark

Copy link
Copy Markdown
Author

Fourth commit corrects D7: the earlier D-pad attribution was Cursor Assistant mode, not a flick. The end-to-end test (#11) shows the native provider receives touchpad motion on the Elite while Scroll Assistant is on, with measured flick lengths, and the Keep/Drop row now proposes keeping it.

swipe_to_delete has the same gap as the up-swipe: off by default with
no screen since 1.0.2, and a 500 px distance that measured left flicks
on the Elite (170 to 680 px) mostly fail. Put its two rows on the same
proposed screen and take it out of the undecided bucket.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Signed-off-by: Daniel JB Clark <djbclark@gmail.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant