2026 beta — known limitations¶
This file is the single authoritative list of operator-visible limitations the 2026 beta ships with. Every line traces to an explicit owner decision on the MOR-1410 hardware-conformance gate (scope freeze of 2026-08-16: bucket B = release-note scope) or to a standalone owner ruling cited inline.
Process rule (owner decision, 2026-08-17): a ticket may move to bucket B (release-note scope) only together with its line in this file. A bucket-B reclassification without a corresponding line here is incomplete.
None of the items below produces false transmit state or routes RF incorrectly; that class of defect is release-blocking by definition and is not on this list.
Browser microphone capture¶
- Browser voice TX requires a secure browser context.
getUserMedia()is available to the Web UI only over browser-trusted HTTPS or a local loopback origin such aslocalhost; a private-LAN HTTP URL is not a loopback origin. Use Core TLS setup with--tls-certand--tls-keytogether. Use a certificate valid and trusted for the server hostname. The--tlsoption alone generates self-signed TLS and does not establish browser certificate trust. Open the resultinghttps://URL. Alternatively, access a loopback endpoint through an SSH tunnel. This limitation concerns microphone capture and voice TX; it does not make all control or RX operations unavailable over HTTP.
Receive-control feedback and precision¶
- PBT values across acquisition gaps: the last confirmed value is kept in software, but the owner-present IC-7300 observation is still pending. (MOR-1692)
- Combined RF/SQL control does not track the gesture locally. Values update only after canonical radio readback (~1 s), which makes precise placement awkward; the SDR-screen skin's rendering of the same control tracks correctly. (MOR-1693)
- Filter Shape and grouped Notch choices show pending feedback in software, but the owner-present IC-7300 rerun is still pending. (MOR-1689)
- IC-7300 has no APF. Advanced Manual (11a) omits
16 32, and a read-only remote-testbed GET NAK'd twice with documented16 22DATA as the adjacent control. Any build or profile advertising CW APF for IC-7300 exposes a nonfunctional control. (MOR-2144)
FTX-1 specific¶
- Manual-notch position on the FTX-1 is shown as a position, not in Hz. The radio's own display does the same; the CAT code range is bounded to 001..320 since #3480/#3526. (MOR-1680, owner ruling 2026-09-17)
- IF Shift, CW Pitch and NR level on the FTX-1 follow the CAT lattices in software (20 Hz, 10 Hz to 1050 Hz, 0–10) but have not been re-run on the radio since the change. (MOR-1681, MOR-1682, MOR-1678 — hardware rerun pending)
- Attenuator and preamp cannot be set for the SUB receiver. The FTX-1 CAT
RA/PAcommands have no SUB form; the controls are shown disabled with the hint "Not available on this receiver" while SUB is active. (MOR-2511, owner ruling 2026-09-17)
rigctld write handling during transmit¶
- While the radio is known to be transmitting, rigctld silently drops
state-changing writes in the mode, VFO, and split families and answers
RPRT 0(success) instead of an error. This deliberately mirrors hamlib's own core, which returns success without writing while PTT is on: a non-zero return makes WSJT-X treat the situation as a hard rig-control failure — it de-keys, stops the auto-sequencer, and on a repeated error shows a modal dialog. Under unknown or stale RF state the seat still refuses truthfully with an error; the drop applies only to known transmit. A third-party client that writes while the radio is transmitting is told the write succeeded even though the radio did not change; it discovers the real value on its next read. (MOR-1881) - For up to about a second after an unkey, the seat may still read the radio as transmitting and drop a mode/VFO/split write the same way. (MOR-1881; owner ruling 2026-09-17 on MOR-1892: not to be narrowed further for the beta)
- Frequency and RIT/XIT are exempt from both limitations above. Bench measurement showed both radios accept and apply these writes while keyed, so the two families are classified TX-SAFE and are always applied, in any RF state, including the up-to-a-second post-unkey window. This also resolves the case previously recorded here of a WSJT-X "Fake It" split-mode dial restore issued right after unkey being swallowed: that restore is a frequency write and is no longer subject to any drop or refusal. (MOR-1940)
rigctld raw CI-V (w) timeouts¶
- A raw
wframe the radio never answers is now reported asRPRT -5(timeout) instead of an empty success. Nothing is known about whether such a frame was applied, so reporting success was a lie. Clients that treated the old empty success as "sent OK" will now see an error on frames that time out.wis a raw CI-V escape hatch for diagnostics and advanced tooling, not part of the frequency/mode/PTT command flow a logging or digital-mode client uses for normal operation. (MOR-1882)
Dual-receiver topology¶
Dual-receiver hardware certification is not part of this beta. The bench holds an IC-7610 (returned 2026-09-14) and an FTX-1; the IC-7610's dual-watch, dual-scope and simultaneous MAIN/SUB audio-routing paths were not re-run for this beta and remain covered by automated profile fixtures and fail-closed tests only. What was accepted on hardware: single-receive SUB operation on the FTX-1 — frequency, mode, width, S-meter, AF/RF/squelch, repeater shift, NB/NR, notch, IF shift, NARROW and AGC read from the SUB receiver, and AGC and NB writes measured landing on SUB with MAIN unchanged (MOR-2511, 2026-09-18). Do not treat any other dual-receiver path as hardware-certified.