Conversation
|
@Pyohwan please use less of the GenAI commentary and keep comments to a minimum in code and the details of the problem too. we don't have the time to read pages of GenAI justification blurb and testing scenarios to state a simple fact or problem. |
4576fa4 to
b50fd94
Compare
|
Trimmed the code comment and PR description. Thanks for the feedback. |
|
In the snippet, grep is running inside the for loop for every card, and then piped to head -1. It is cleaner and more efficient to let the loop emit all output, then pipe once into grep and head. |
|
Good catch, fixed - grep now runs once after the loop instead of once per card. |
b50fd94 to
26b304c
Compare
…st one Ssystem-splash's fallback (used when batocera-boot.conf has no display.rotate entry yet) grabbed whichever display.rotate.<connector> line came first in the sysconfig file, regardless of which connector was active. Boards defining more than one such line (e.g. a secondary panel that needs rotating while the primary doesn't) get whatever the file ordering happens to be - on the ODROID-M1, an HDMI-only boot picked up the DSI panel's rotation and rendered the boot logo sideways. Fix: detect the active connector (same batocera-drminfo approach S65values4boot already uses) and match its own display.rotate.<connector> key, falling back to a bare display.rotate= default, then to the old first-match behavior only if the connector can't be detected yet.
26b304c to
4e8a0c7
Compare
|
There is just one last small detail you should bring back: whitespace stripping at the end. |
|
This is already handled - there's an unconditional |
|
actually we probably need to handle this better... try this... let me know |
BOOT_CONF's own display.rotate.* lookup still grabbed whichever key happened to appear first in the file, the same bug MODEL_CONF had. Extend the active-connector match to both files consistently instead of only fixing it in one. Co-Authored-By: dmanlfc <noreply@github.com>
|
Tested this live. Fake key ordering in One thing I noticed: I don't think the |
Ssystem-splash's fallback (used when
batocera-boot.confhas nodisplay.rotateentry yet) grabbed whicheverdisplay.rotate.<connector>line came first in the sysconfig file, regardless of which connector was active. Boards defining more than one such line (e.g. a secondary panel that needs rotating while the primary doesn't) get whatever the file ordering happens to be - on the ODROID-M1, an HDMI-only boot picked up the DSI panel's rotation and rendered the boot logo sideways.Fix: detect the active connector (same
batocera-drminfoapproachS65values4bootalready uses) and match its owndisplay.rotate.<connector>key, falling back to a baredisplay.rotate=default, then to the old first-match behavior only if the connector can't be detected yet.Verified on hardware (HDMI-only, secondary panel disabled): with the fix, rotation resolves to none and the logo renders upright; reverting just the connector-detection block back to the old first-match logic reproduces the sideways logo.