Skip to content

A cell for every mouse cursor, and the display read back by Print Screen - #24

Merged
jwinarske merged 2 commits into
mainfrom
m4-cursors-app
Nov 17, 2026
Merged

jwinarske merged 2 commits into
mainfrom
m4-cursors-app

Conversation

@jwinarske

@jwinarske jwinarske commented Nov 17, 2026 •

Copy link
Copy Markdown
Owner

The cursor's shape was the item M4's exit report left for a person to look at.

  • apps/cursors: a cell for each of the framework's 35 system cursors, labeled with its kind and the cursor theme's name. Looked at on the Pi 5 with the M705: the cursor changes in each cell
  • App lane test every_system_cursor_is_asked_for_by_its_theme_name: a script's mouse goes into each cell; the framework asks for all 35 and each name is the Linux embedder's for that kind. run-app --input now prints each ask as mouse cursor: NAME
  • run-app --kms --capture out.png: Print Screen writes what the display shows to a PNG, cursor included, read back from the card through drmkit-capture. Later captures are numbered (out-2.png). The key is not told to the framework while --capture is given
  • docs/m4-status.md, docs/m4-exit.md, README, CHANGELOG

Two things the first capture found, both fixed outside this repository and pinned here:

Both are merged and pinned here in the second commit: drmkit 0.1.5 and emblema 7d4253a. Built from the pins, the capture on the Pi 5 has the frame and the cursor, and the cell reads back 2D3748.

- apps/cursors: a cell for each of the framework's 35 system cursors,
  saying its kind and the cursor theme's name. Looked at on the Pi 5
  with a real mouse: the cursor changes in each cell
- every_system_cursor_is_asked_for_by_its_theme_name (app lane): a
  script's mouse into each cell; the framework asks for all 35 and the
  engine's name for each is the Linux embedder's. run-app --input
  prints each ask as `mouse cursor: NAME`
- run-app --kms --capture out.png: Print Screen writes what the display
  shows, the cursor's plane with the frame's, to a PNG, through
  drmkit-capture, and is then no key of the framework's
- docs: the look and what the first capture found. Red and blue were
  exchanged in every frame scanned out on the Pi 5 (emblema named a
  ten-bit buffer XR30 that is XB30), fixed in emblema's repository, and
  drmkit-capture could not read a ten-bit plane, fixed in drmkit's.
  Neither fix is in the versions pinned here yet
- review: the capture's path is the one the user named and no other;
  each capture takes the card's descriptor from the seat and drops it;
  reading the planes is on a thread of its own, off the devices' and
  the frame's. Print Screen is taken from the framework only when
  --capture asks. One dependency more, drmkit-capture, pure Rust

Signed-off-by: Joel Winarske <joel.winarske@linux.com>
- drmkit crates at 0.1.5: drmkit-capture reads a ten-bit plane, so a
  capture on the Pi 5 has the frame under the cursor
- emblema at 7d4253a: a ten-bit buffer is named XB30 to KMS, which is
  what its bits are. Red and blue were exchanged on the Pi 5's screen
- docs: both as pinned. Built from the pins, a cell drawn 2D3748 reads
  back from the Pi 5's card as 2D3748
- review: the lockfile moves the drmkit and emblema packages only.
  emblema's one change since the last pin is the fourcc; offscreen
  frames do not pass through it, and the goldens run here are unchanged

Signed-off-by: Joel Winarske <joel.winarske@linux.com>
@jwinarske
jwinarske merged commit 7bf45e6 into main Nov 17, 2026
11 checks passed
@jwinarske
jwinarske deleted the m4-cursors-app branch November 17, 2026 17:21
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