RBN Filtering (1.0.9)

[n9jr]

My filters:
[spots]
reject/spots 1 info POTA
accept/spots 1 rbn call n9jr
accept/spots 2 rbn call N9JR
accept/spots 3 ny n
accept/spots 4 by a
accept/spots 5 by v
accept/spots 6 by w

The expectation is I should only see RBN spots for N9JR. With RBN enabled I see this:

DX de WZ7I: 10115.0 K0ACP CW 8 dB 22 WPM CQ 1742Z
DX de W3RGA: 18084.0 EG1PLC CW 8 dB 23 WPM CQ 1742Z
DX de AC0C: 14062.8 K7WA CW 4 dB 23 WPM CQ 1742Z
DX de WA7LNW: 18080.4 SM7FBJ CW 6 dB 21 WPM CQ 1742Z
DX de WZ7I: 14026.0 R3ZBI CW 11 dB 25 WPM CQ 1742Z
DX de AA4PA: 14024.6 W0KO CW 15 dB 23 WPM CQ 1743Z
DX de AC0C: 14028.9 W1AW/0 CW 20 dB 23 WPM CQ 1743Z
DX de AC0C: 14024.6 W0KO CW 27 dB 22 WPM CQ 1743Z
DX de W1NT: 14024.6 W0KO CW 5 dB 22 WPM CQ 1743Z
DX de WC2L: 14024.6 W0KO CW 14 dB 23 WPM CQ 1743Z
DX de WB6BEE: 14024.6 W0KO CW 22 dB 23 WPM CQ 1743Z
DX de WA7LNW: 14024.5 W0KO CW 42 dB 23 WPM CQ 1743Z
DX de VE7CC: 14024.6 W0KO CW 8 dB 23 WPM CQ 1743Z
DX de W6YX: 14024.6 W0KO CW 20 dB 23 WPM CQ 1743Z
DX de W6YX: 14053.0 K2JVB CW 6 dB 24 WPM CQ 1743Z
DX de WS2C: 14024.7 W0KO CW 4 dB 22 WPM CQ 1743Z
DX de VE6WZ: 14024.5 W0KO CW 38 dB 23 WPM CQ 1743Z
DX de W3RGA: 14024.6 W0KO CW 24 dB 23 WPM CQ 1743Z

[AI3I]

Implemented locally for 1.0.9: added a shared spot filter evaluator for human and RBN spots. Legacy RBN-scoped rules such as accept/spots ... rbn ... are now honored for RBN ingress without general spotter accepts opening the full RBN stream; global rejects still apply. Added telnet and direct-ingest regression coverage and validated with full suite: pytest -q (512 passed). Pending staging RBN/filter validation before close.

[n9jr]

Almost 100% working. RBN filtering now works. The issue now is how the RBN information is displayed.

From DXSpider I receive a summary of RBN spots.
DX de N2CR-#: 14024.1 N9JR CW 8dB Q:9* Z:3,4,5 1523Z

PyCluster delivers the raw data.
DX de WC2L: 14024.1 N9JR CW 11 dB 21 WPM CQ 1523Z
DX de W3RGA: 14024.1 N9JR CW 14 dB 21 WPM CQ 1523Z
DX de W8WWV: 14024.1 N9JR CW 15 dB 21 WPM CQ 1523Z
DX de W6YX: 14024.1 N9JR CW 10 dB 21 WPM CQ 1523Z
DX de N2YCH: 14024.2 N9JR CW 21 dB 21 WPM CQ 1523Z
DX de AA4PA: 14024.1 N9JR CW 22 dB 21 WPM CQ 1523Z
DX de K3PA: 14024.1 N9JR CW 33 dB 21 WPM CQ 1523Z
DX de W1NT: 14024.1 N9JR CW 20 dB 21 WPM CQ 1523Z
DX de N2CR: 14024.2 N9JR CW 8 dB 21 WPM CQ 1523Z
DX de WB6BEE: 14024.1 N9JR CW 28 dB 21 WPM CQ 1523Z
DX de VE6JY: 14024.2 N9JR CW 31 dB 21 WPM CQ 1523Z
DX de TI7W: 14024.1 N9JR CW 13 dB 21 WPM CQ 1523Z
DX de K4PP: 14024.1 N9JR CW 19 dB 21 WPM CQ 1523Z
DX de W4KAZ: 14024.1 N9JR CW 18 dB 21 WPM CQ 1523Z
DX de K5TR: 14024.1 N9JR CW 34 dB 21 WPM CQ 1523Z
DX de WZ7I: 14024.1 N9JR CW 29 dB 21 WPM CQ 1523Z
DX de KM3T: 14024.1 N9JR CW 17 dB 21 WPM CQ 1523Z
DX de ZF9CW: 14024.2 N9JR CW 26 dB 21 WPM CQ 1523Z

[AI3I]

Implemented in the 1.0.9 worktree: RBN telnet output now uses summarized skimmer-style text instead of raw WPM/CQ strings. Live RBN delivery is normalized to mode/dB/Q/Z form, and show/rbn now groups matching reports by minute/frequency/call/mode into a single summary line with skimmer count and CQ zones. Added focused show/rbn coverage.

[AI3I]

Current status: implemented and pushed in 013b696. Live validation needed: RBN filtering should still limit the stream as before, and live/show/rbn output should now use summarized mode/dB/Q/Z text rather than raw per-skimmer WPM/CQ lines. Please compare against DXSpider-style output for readability.

[n9jr]

DX de WS3W-#: 14011.2 N9JR CW 6dB Q:1 Z:5 1816Z
DX de VE3KG: 28028.3 TI5/VA3RA 1816Z
DX de W1NT-#: 14011.2 N9JR CW 17dB Q:1 Z:5 1816Z
DX de KD7EFG-#: 14011.2 N9JR CW 36dB Q:1 Z:5 1816Z

It is in a format most users are accustomed to. But they are not grouped to a summary.

[AI3I]

Implemented in the 1.0.9 worktree: live telnet RBN delivery now uses a short aggregation window per user/session after filters are applied. Multiple skimmer hits for the same minute/frequency bucket/callsign/mode are collapsed into one DXSpider-style summary line with a -# spotter, mode/dB, Q:, and combined Z: zones. This addresses the remaining behavior where live RBN output was still one raw WPM/CQ line per skimmer. Added focused live aggregation coverage plus existing show/rbn coverage. Validation: full suite passes (pytest: 531 passed).

[n9jr]

Still not consolidating the data

DX de N2EPE-#: 14005.6 N9JR CW 8dB Q:1 Z:5 1802Z
DX de VE6AO-#: 14005.6 N9JR CW 2dB Q:1 Z:5 1803Z

And N9JR-2 reports
DX de W3LPL-#: 14005.6 N9JR CW 13dB Q:9 Z:3,4,5 1802Z

[AI3I]

Follow-up fix in the current worktree: live telnet RBN aggregation now groups reports across a five-minute bucket instead of only the exact one-minute boundary, while still separating by session, rounded frequency, callsign, and mode. The displayed time remains the earliest actual report in the group. Added regression coverage for adjacent-minute skimmer reports collapsing to one summary. Full local test suite passes: 537 passed.

[n9jr]

It is still not summarizing the RBN spots.

N9JR-4

DX de NU4F-#: 14005.6 N9JR CW 8dB Q:6* Z:5 1836Z
DX de K4PP-#: 14005.6 N9JR CW 15dB Q:5* Z:5 1836Z
DX de K5TR-#: 14005.6 N9JR CW 23dB Q:3 Z:5 1836Z
DX de W1NT-#: 14005.6 N9JR CW 6dB Q:1 Z:5 1836Z

N9JR-2
DX de KM3T-#: 14005.6 N9JR CW 15dB Q:9 Z:3,4,5,8 1836Z

[AI3I]

Implemented in the current 1.0.9 worktree: live RBN batches are no longer prematurely flushed by user commands while the batch dwell window is still open, preventing same-minute reports from splitting into multiple partial summaries. Live aggregation also now collapses already-summarized upstream -# reports for the same call/frequency/time bucket into a single summary line and carries forward Q counts from upstream summaries.

Verification: added regressions for command-time RBN dwell, summarized upstream -# collapse, and existing raw RBN grouping; full pytest passed with 548 passed. Pending commit/deploy and live comparison against N9JR-2/DXSpider output.

[n9jr]

Closer.
N9JR-2 reports:
DX de WA7LNW-#: 18074.1 N9JR CW 4dB Q:7* Z:3,4,5,8 1511Z

N9JR-4 reports:
DX de WB6BEE-#: 18074.1 N9JR CW 10dB Q:3 Z:5,8 1511Z
DX de AA4PA-#: 18074.1 N9JR CW 18dB Q:2 Z:5 1511Z
DX de WA7LNW-#: 18074.1 N9JR CW 4dB Q:2 Z:5 1511Z

The WB6BEE reports zone 5 and 8 so some RBN consolidation is happening,

[AI3I]

Related follow-up implemented locally for 1.0.11, but this does not fully close the remaining aggregation-comparison thread.

Today's RBN work tightens classification/filtering by treating PC11/PC61 raw/protocol RBN markers as RBN even when the visible comment is plain mode text such as FT8. That prevents RBN-marked peer spots from bypassing RBN ingest/display policy as normal cluster spots, and public websocket spots now honor the same stored RBN filters as /api/spots.

Remaining validation for this issue is still the live aggregation comparison against N9JR-2/DXSpider-style summaries: confirm whether same frequency/call/time RBN hits collapse into the expected single -# summary with combined Q/Z values.