refactor(sim): reuse PR33 link-budget helpers - #77
Conversation
a36b7bc to
db3debf
Compare
Komzpa
changed the title
b495d48 to
3f9a5af
Compare
Komzpa
changed the title
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3f9a5afe9d
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
3f9a5af to
6fed8ee
Compare
0ce6eaf to
0a25ca3
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0a25ca3d8a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Signed-off-by: Darafei Praliaskouski <me@komzpa.net>
- treat unparsable map altitude values as absent instead of failing the whole import with an uncaught TypeError/ValueError - import CLIENT_BASE nodes on the router side: current firmware rebroadcasts them like ROUTER_LATE - compare the trig-derived origin longitude with a tolerance in tests
A present origin block that is not a map or lacks a usable lat/lon pair now raises instead of silently dropping the origin and projecting terrain against a different origin than the author intended. Accept the latitude/longitude key spellings alongside lat/lon.
Concurrent runs sharing one cache directory wrote the same fixed temp filename and could atomically publish an interleaved, corrupt tile that then stayed cached. Download and unpack through process-unique temp names and clean up archives after unpacking.
Cache keys carried id(TERRAIN_GRID), so a reused object address after a Config-reusing parse could serve losses computed against an earlier grid. Key the cache on a monotonically increasing per-grid token instead. Also pick the nearest interpolation samples with heapq.nsmallest instead of sorting every sample in this hot path, and use the shared node_antenna_height helper from lib.common instead of a diverging local copy.
Automatic SRTM tile selection evaluated the full link budget for every node pair. Derive one conservative best-case link distance (path loss grows monotonically with distance in every lib.phy model) and skip pairs beyond it before the expensive per-pair work; document that the selection scales with node pairs and recommend --map-limit for broad imports. Also log when a wrapped --map-bbox is discarded for terrain so the substituted extent is not silent, and import node_antenna_height from its canonical module.
TCPInterface connects without a socket timeout, so a stale --nodedb-host blocked the CLI indefinitely. Probe reachability with a bounded connect first so unreachable hosts fail fast with the existing could-not-connect error.
ec0a51e to
e6c3daa
Compare
|
Important Review skippedDraft detected. Please check the settings in the CodeRabbit UI or the ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: You can disable this status message by setting the Use the checkbox below for a quick retry:
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
e6c3daa to
0dca93c
Compare
1 participant
Summary
This is the small reusable optimization slice from the stale #33 after the work was split along the discussion in #39 and #50.
The new review surface on top of #72 is intentionally narrow:
estimate_path_loss()can take an explicit path-lossmodel=without mutatingConfig.MODEL, so scenario or analysis code can compare models without leaking global config stateMeshPacketrecords the receiver IDs it already classified as CAD-detected or sensitivity-sensed while refreshing link budgetsMeshNode.transmit()reuses those receiver-ID lists instead of scanning every node again for flags that were just computedStack context
GitHub shows the full stacked diff against
masteruntil the lower PRs merge. The dependency stack below this PR is #68, #69, #70, #73, #74, #75, #71, and #72. Those carry the map/terrain/clutter/RF/DCR/DTP pieces; this PR only adds the PR33 helper cleanup needed on top of that stack.Why this shape
The useful part from #33 here was not the whole old branch. Maintainer feedback on #33 asked to separate optimizations from Burning Man-specific code, and #39/#50 asked for smaller reviewable PRs.
What is deliberately left out:
lib/signal.py: Sync modem presets with current firmware #33 adds it, but the old branch does not import or call it from the simulator path, so reviving it here would add dead API surface without a callerburningManSim.py: kept out of this PR because it is scenario-specific and depends on the newer DCR/DTP/radio-policy stack; that belongs in the separate feat(sim): add Burning Man DCR/DTP scenario #78 scenario PRValidation
MPLBACKEND=Agg .venv/bin/python -m unittest tests.test_phy tests.test_lora_mesh_cli tests.test_node tests.test_dtp tests.test_packet -v-> 58 tests OKgit diff --check origin/pr/72..origin/pr/77unittestcheck is green