Skip to content

waze-on-cluster: the cluster map's video buffer given to another video on the cluster while it shows - #222

Closed
TheoneLT83 wants to merge 1 commit into
robertnisipeanu:mainfrom
TheoneLT83:pr/map-video-yield
Closed

TheoneLT83 wants to merge 1 commit into
robertnisipeanu:mainfrom
TheoneLT83:pr/map-video-yield

Conversation

@TheoneLT83

Copy link
Copy Markdown

The display has two video buffers; with the cluster map and the HUD map both on, a third video on the map target (a Turn View camera) was refused by cluster-render ("too many video buffers").

While /run/turn-view/map-yield is fresh (another tool touches it every second while its video is on the cluster; at most 3 s old), the cluster map drops its layer, keeping its newest picture, and makes it again once the file goes or goes stale (MapFeed::set_yield_video, MapFeeds::set_cluster_yield). Nothing changes while no tool writes the file.

Test: map_feed_test.cpp (the layer dropped and made again, the map showing again).

Screenshots: none: nothing of this app's screens changes.

AI: Claude Code, Claude Opus 5.5 (ported from TheoneLT83/Cluster-apps, waze-on-cluster 0.10.14 there, where its Turn View copy writes the file). Tested: make test-cpp on this branch: the new checks pass; shown again: the newest picture's part first failed once and passed on three reruns (a timing test that also flakes without this change). Not tested on a car from this branch. Human review: TheoneLT83 asked for this pull request; the code was not reviewed line by line by a person.

🤖 Generated with Claude Code

Merge Idrive7 assistant versions

…o on the cluster while it shows

The display has two video buffers; with the cluster map and the HUD map both on, a third video (a Turn View camera)
was refused. While /run/turn-view/map-yield is fresh (touched every second, at most 3 s old) the cluster map drops
its layer, its newest picture kept, and makes it again once the file goes or goes stale.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RaQUWvi4djHaPEqip7eCr8
@robertnisipeanu

Copy link
Copy Markdown
Owner

Thanks for this. Closing it: sharing the cluster between apps should be done properly in cluster-render, not through a signal between two specific apps that the map watches.

cluster-render already handles several videos at once. Clients can have up to 8 video layers together; the display holds the buffers of two, and past two, unturned videos at their frame's size are composed into one (README "Layers", PROTOCOL.md "Client to service"). Layers are ordered by class and z, and an exclusive layer hides the ones below it while it is drawn and brings them back when it goes (README "Arbitration"). Another video on the cluster can use those as they are, and the map needs to know nothing about it. If something there doesn't cover a real case, the place for it is cluster-render's arbitration, for every client.

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.

2 participants