Repository navigation
Conversation
|
Hi, Having some abstracted UI concepts would be ideal, and to have different display driver helpers. I will hopefully tackle this soon. |
|
Totally agree with you on this. And I'm happy to see you want to keep things in control. Also, maybe displays are out of scope of the meshcore lib and should be kept apart ... I find it good to have a lib that only cares about the mesh itself I've added this sample on my github account : https://github.com/fdlamotte/MCChatScreen so it can be built by others. I'll try to keep it in sync it with the base example ... To use my sample, I just put my projects at the same level as MeshCore, platformio.ini references your repo and there is a symlink to the ed25519 lib (since it is only 240k I should have just made a copy) I let you close this PR if you agree Regards, Florent |
* variants/linux: use linux_base for platformio * variants/linux: use linux_base for platformio
* variants/linux: use linux_base for platformio * variants/linux: use linux_base for platformio
… gate Partial implementation of plans/MONITORING_UNIFICATION.md, scoped to what's verifiable without hardware: - MonRing: new MON_DEBUG kind + DebugRecord struct (folded-in RLOG shape, not yet wired to any real call site), RLOG_KIND_MASK/RLOG_CONT_BIT continuation-bit constants, appendDebug()/debugCount(). Fixes a real bug found while adding this: the eviction switch never masked off a continuation-flagged kind byte, silently leaking the per-kind resident counter. - GET_MONRING header: adds route_count (a pre-existing gap -- MON_ROUTE's own counter was never wired into this header) and debug_count. - ProfileLog: persisted per-role enable/disable gate (default disabled, changing today's always-on behavior), PrefsTlvKey 55, wired at boot alongside monring.setConfig(). Native test coverage added for all of the above; full native suite (8 suites) and the default/companion/repeater firmware builds all pass. Deliberately not done in this pass (see plan's Progress section for detail): the ms-offset time model (meshcore-dev#1) and PSRAM-allocation reordering (meshcore-dev#2), both firmware-timing changes needing hardware to verify safely; retiring xportlog.py and its dependents (meshcore-dev#6, meshcore-dev#12-14); the MON_DEBUG root-cause view (meshcore-dev#16); any hardware verification (meshcore-dev#10, no device available). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
…the cal-emission provenance map later conversion tasks cite: (a) BOARD: meshcore-board (NIP-100, 500 cards, read via kanbanstr-cli card list this session) conversion work items located — n4s-t_91f4019e 'Convert fitted k/T0 into parser literal strings' (backlog), n4s-t_2656f57c 'Convert fitted k/T0 into parser representation' (backlog), n4s-t_0017d836 literal inventory, n4s-t_291d4094 final deliverable, n4s-t_2e041ab8 char-by-char verify; upstream producers n4s-t_3f930320 fit (done, committed ab17ec5..a9d7265 = the bench/atcut-fit/ lineage), n4s-t_d6484f25 parser contract (done, 2075f92), n4s-t_c11f2cbe syntax pin (done, pin felixfelix-bot/MeshCore feat/passive-mapper @ 3fdfbc8), n4s-t_8f776e31 re-derive (backlog); NO card carries linked PR/branch/sha via CLI (list exposes Title/ID/Status/Rank only — searches listed: board-list grep meshcore|atcut|fit|cal|freq|clock|tcxo over 1119 boards, card-list greps cal|atcut|T0|fit|emit|provenance + convert + parabola|fitted over 500 cards); dispatch state checked not assumed: t_2656f57c WAS dispatched (session 20260914_194259_6b43c6, interrupted before artifact, zero conversion commits in git log --all grep), t_91f4019e never dispatched (0 session hits). (b) EMITTERS (18 sites, ordered, every one path:line + snippet + language): REPO main@abe0d36f sweep over all refs (origin main/dev/default-scope/feature-nicerf-lr2021/fix-2739/flood.max.advert/gh-pages/t-impulse-ui + rev-list --all max-count=400) matched exactly 7 files all in bench/atcut-fit + bench/dryrun-ambient — canonical builder fit_atcut.py:154-157 'return "cal %.6g %.2f" % (k, t0)' citing the fork kit as format source; dispatch :389-393 (print after gates); anti-emitter no-fit branch :400-402 'VERDICT: DO NOT ISSUE cal'; echo render :396; self-test :430; verify_cal_domain.py:25,62-63 renders 481,601 candidate lines via same builder (:86,90 token round-trips, :32 shape regexes); cal_cli_harness.cpp:64-79 verbatim firmware cal branch replica (consumer contract), :120-146 parser test vectors, :200-205 CR/LF wire vectors; sample_calshow.py:190,193 fixtures only; FORK bench kit reachable ONLY as dangling commit d7ae93f (git fsck: sole dangling; ancestor 3fdfbc8 also in store) — calib.py:48-50 'return "cal %.6g %.2f"' (format origin), thermal_bench.py:141-143 set_cal the ONLY site that writes the line to a serial wire, :236 dry-run mock reply, :515 CLI persist hint, test_calib.py:67 format regression; MyMesh.cpp:170-188 firmware consumer (dispatch/echo/calshow), MyMesh.h:79 help list, SelfCal.h 15-byte blob path; ZERO emitters in src/examples of this repo's reachable tree (git grep HEAD -- src examples test = 0 hits, upstream has no passive_mapper). Classification for sibling cards: can-issue #14 only; build-literal #1/#13 byte-identical; render-only #2/#5/#6/#15/#16; consumer #7-#10/#18; fixture #11. (c) PLAN QUOTES: CAL-CLI-SPEC.md §7 verbatim template 'cal <k> <T0>\r' (k <=6 sig digits, T0 2 decimals, CR terminator), §4 'NO scaling/multiplier/fixed-point applied anywhere', §5 'no decimal-count limit, does no rounding'; REPORT.md §1 'Emission format: cal %.6g %.2f — identical to the fork's bench kit'; CAL-COMMAND.md format-contract section + NO-FIT verdict 'NOT FILLED. Do not issue'; fit_atcut.py docstring Method step 3; t_2656f57c card body recovered verbatim from dispatch session (METHOD steps 1-3 + DELIVERABLE quote). (d) root /home/c03rad0r/repos/meshcore, main @ abe0d36, dangling d7ae93f + pin 3fdfbc8. HUNT ONLY: sources inventoried, nothing validated, nothing issued, nothing fabricated
…ometry as CORROBORATION ONLY — fleet card meshcore:t_a82deb2b Corroboration-only pass; nothing overrides a config or marked value, and NO frequency is derived from geometry anywhere. MATCHING NETWORK — result: NONE on the hub_board_v1 design. Structural finding from connectivity, not from absence of documentation: RF_SUB_868 (net 12) and RF_2G4_2400 (net 13) each carry exactly TWO nodes with ZERO intervening parts (U2 pin 9 -> AE1; U2 pin 10 -> AE2), confirmed independently by hub_board/hub_schematic.net and by the generator hub_schematic.py:408-411. Board census holds ZERO inductors (9 caps present, none on an RF net). Copper is a direct trace: 3 segs 36.745 mm / 4 segs 23.935 mm at 0.8 mm, 2 vias, no switch, no balun, no DC block (SP4T + FEM are PLANNED options only, component-guide 238-250). Vendor NiceRF LoRa2021 datasheet V1.3 (recovered from balloon git cc92c869, pdftotext -layout) specifies pin 9 / pin 10 as "external 50-ohm antenna" with NO matching-network section (grep matching = 0 hits). ANTENNA — NOT a printed PCB element: hub_board_v1 contains ZERO drawn antenna graphics and no antenna-class footprint; the radiator is a hand-soldered external WIRE DIPOLE on a plain pad. Sub-GHz 16.4 cm total, 2.4 GHz 3.1 cm total (wiring-dev-board.csv ANT_SUBGHZ/ANT_24GHZ, FLIGHT-BOARD-PLAN 85-86, component-guide 69-70). Silk (design file only; the hand-wired DUT has no silkscreen): F.SilkS values 868MHz / 2.4GHz, refs AE1 / AE2. Bare wire carries no marking. The 23.9/36.7 mm feed copper is a feed LINE, not a radiator. LABEL DISCIPLINE (card acceptance): every geometry-derived row (#8 #9 #10 #12) carries INFERRED-FROM-ANTENNA and is expressed as a dimension-vs-wavelength RATIO against the band the source itself names (0.95 / 1.00 / 0.50) — no frequency computed or stated from geometry at all. 9 MARKED rows sit alongside. NEW DEFECT FOUND in the corroborating sources: the 3.1 cm element measures 0.50x the nominal lambda/2 its own label claims, i.e. the sources mislabel it lambda/4 as lambda/2 (factor-2) — recorded as a documentation defect, stated at band level only, NOT silently corrected. Divergence 16.4 cm vs 2x8.6 cm disclosed, not reconciled (sibling 708585a A1). Also documents a designator COLLISION: the board's AE1 (868 MHz) is NOT the netlist's AE1 (2.4G) — the generator reused one placeholder ref (hub_schematic.py:214,217); the authoritative anchor is the net name. Scope limit stated explicitly: the "no matching network" claim covers the hub_board_v1 design file ONLY; the DUT's module-internal question is left OPEN (a record claims an onboard match, PCB-AUTO-ROUTE-EXECUTION-PLAN-V2.md:209). HARD RULES respected: no ppm, no calibration, no frequency offset, no error budget, no frequency from geometry, no outbound contact, no probes; physical state UNVERIFIED throughout (no camera, DUT not attached — standing constraints). Consistent with siblings 708585a (rf-port-antenna-config) and 0b7bfe9 (dut-rf-ports); F33 SMA/PA module family excluded as a different board. Duplicate check: no prior deliverable for t_a82deb2b (grep a82deb2b over ~/reports + meshcore = 0 hits; no bench/*match* slug). REVIEW: cold cross-family review by deepseek/deepseek-v4-flash (non-GLM, zero prior context) — pass 1 REJECTED with 4 major findings, all fixed; passes 3 and 4 APPROVED with 0 major. Pass 1's reject and the later approvals are NOT netted out; full pass-1..4 history plus the pass-3 review text are committed alongside (review-cold-cross-family.md). Scripts: tracelen/netcheck/wavelength_a82deb2b.py.
Some code to show received msgs on the display, found this usefull when testing companion app ...