NereusSDR

A cross-platform SDR console for OpenHPSDR radios

NereusSDR is a C++20/Qt6 port of Thetis — the canonical OpenHPSDR / Apache Labs SDR console, itself descended from FlexRadio PowerSDR — carrying its radio logic, DSP integration, and feature set forward to a native cross-platform codebase (macOS, Linux, Windows) with a Qt-based GUI. The Thetis contributor lineage (FlexRadio Systems, Doug Wigley W5WC, Richard Samphire MW0LGE, and the wider OpenHPSDR community) is preserved per-file in source headers and summarized in docs/attribution/THETIS-PROVENANCE.md. Distributed under GPLv3 (root LICENSE), elected under the "or later" grant in upstream Thetis source-file headers (Thetis is GPLv2-or-later). A verbatim copy of GPLv2 ships at docs/attribution/LICENSE-GPLv2 for reference, since several WDSP and ChannelMaster source files explicitly reference v2.

Works with any radio implementing OpenHPSDR Protocol 1 or Protocol 2:

• Apache Labs ANAN line — ANAN-G2 (Saturn), ANAN-7000DLE, ANAN-8000DLE, ANAN-200D, ANAN-100D, ANAN-100, ANAN-10E
• Hermes Lite 2
• All OpenHPSDR Protocol 1 radios — Metis, Hermes, Angelia, Orion, Orion MkII
• All OpenHPSDR Protocol 2 radios

Pre-built binaries for Linux (AppImage, x86_64 + aarch64), macOS (DMG + PKG, Apple Silicon + Intel), and Windows (NSIS installer + portable ZIP, x64) are published as GitHub Releases:

https://github.com/boydsoftprez/NereusSDR/releases

All artifacts are GPG-signed (KG4VCF) via SHA256SUMS.txt.asc. To verify:

gpg --keyserver keyserver.ubuntu.com --recv-keys KG4VCF
gpg --verify SHA256SUMS.txt.asc SHA256SUMS.txt
sha256sum -c SHA256SUMS.txt

Alpha builds: until Apple Developer ID and Authenticode certificates are obtained, macOS users will need to right-click → Open the DMG on first launch, and Windows users will see a SmartScreen warning to click through. Linux is unaffected. See the per-release notes for details.

v0.4.0 status: macOS DMG and PKG remain Apple Developer ID-signed and notarized; Gatekeeper accepts them on first launch. Windows installer remains unsigned (Authenticode certificate pending); SmartScreen "More info → Run anyway" still applies.

Current release: v0.5.2 (2026-05-24). Substantial release on top of v0.5.1 (and v0.5.0 underneath). One major epic plus a new SKU port, a new UI subsystem, and a polish tail: Phase 3P-II External RF accessories (AetherSDR 1:1 PGXL + TGXL baseline, Thetis analog S-Meter port with 4 RX modes, connection robustness with exponential auto-reconnect / keepalive / RTT-correlated ping / full PGXL pairing flow, advanced UI with Fault Log / Tune Memory / TX Interlock Policy / power-cap toast); ANAN-G2E (HermesC10) SKU port (new board enum, hardware profile verified against Thetis v2.10.3.15, codec wrappers, BPF1 algorithm, PA telemetry, P2 RX unblock + crash fix); Applet visibility controller (hamburger menu on AppletPanel banner, View → Containers → Applets show/hide, capability-gated availability, persistence); polish tail (4O3A integration cleanup, TCI live-state fixes + 5 review issues, PS-A persistence, PA profile / quit handling). v0.5.2's 3P-II bench-verification matrix (36 rows) is pending live PGXL + TGXL hardware; the ANAN-G2E matrix (12 rows) is pending live G2E hardware. v0.5.1 (2026-05-15) was the patch before: three ship-blocking release-artifact fixes (Windows rade.dll, macOS x86_64 Qt6::WebSockets, HL2 + Win11 waterfall slider persistence), three correctness fixes (orphan .bak recovery, connect-state-stuck-green, VOX prime-on-connect), and two CodeQL pipeline maintenance fixes. v0.5.0 detail below.

Phase 3J-1: TCI v2.0 WebSocket server. External programs (WSJT-X, JTDX, FreeDV, Quisk, ESDR3, N1MM, Log4OM, contest software) can now drive NereusSDR over Thetis-compatible TCI. Setup > CAT/Network > TCI Server configures bind interface, port, and sensor intervals; Tools > TCI Server opens the log viewer; the bottom-bar TCI indicator shows live state. The audio pipeline negotiates 8 / 12 / 16 / 44.1 / 48 kHz with per-stream resampling, so FreeDV 8 kHz, Quisk, and JTDX 12 kHz all work end-to-end. 15 closeout items shipped after the initial port stabilized the on-bench behavior against real clients.

Phase 3J-2: Spot system, FreeDV Reporter, PSK Reporter. Seven spot-source clients in one place: DX cluster, RBN, WSJT-X UDP, SpotCollector / DXLab UDP, POTA HTTPS, FreeDV Reporter Socket.IO, PSK Reporter IPFIX. Tools > Spot Hub (Ctrl+Shift+S) opens a 10-tab modeless dialog; Tools > FreeDV Reporter (Ctrl+Shift+R) opens the 14-column live station view with TX/RX highlights, QSY support, and 2-hour idle auto-removal. Spots render on the panadapter with collision-avoidance stacking and click-to-tune, colored by a DXCC 4-tier resolver against cty.dat plus the operator's ADIF log.

Phase 3R: RADE as a true peer mode (RX + TX end-to-end). RADE is wired as a first-class DSP mode (DSPMode::RADE_U / DSPMode::RADE_L) and not as a DIGU pretense, a virtual audio bus, or a slice-mute hack. RX decodes through a dedicated RadeChannel; the VFO flag gains a mode-aware SNR row and shows the EOO-decoded speaker callsign when known. TX is end-to-end: TxWorkerThread feeds the RADE encoder and sendTxIq carries the 24 kHz stereo modem output. RadeApplet docks in the right column when RADE is the active mode. Vendored radae_nopy (BSD-2-Clause) plus Opus with LPCNet/FARGAN add about 9 MB to the binary on every platform; neural-net weights are compiled into librade, so no external model file ships.

Bench-fix tail. Wire / parser / UX gaps that surfaced when 3J-1 + 3J-2 + 3R drafts hit real radios, real DX clusters, real WSJT-X feeds, and real WSJT-X TCI cycles. Highlights: first-MOX audio-volume seed, ten missing spot-client lifecycle wires, the DXSpider parser, FreeDV Reporter row-highlight / Socket.IO ACK push / VFO freq-publish throttle, the RADE callsign / idle-clear timer, and cross-source spot dedup via SpotModel::dedupIndexFor.

Earlier releases. v0.4.0 (2026-05-08) shipped a substantial minor release on top of v0.3.2 with five pieces of work landing together: 3M-4 PureSignal (the big new feature: feedback DDC plumbing on Protocol 1 and Protocol 2, calcc.c + iqc.c vendored verbatim from Thetis, PsForm dialog at Tools → PureSignal, AmpView modeless dialog, two-tone IMD overlay on the spectrum, bottom-banner FB+PS indicator pair, enabled on every supported P1 and P2 SKU including HL2 and plain Hermes), a Display + DSP-Options refactor (WDSP avenger() and detector() ports, new Setup → DSP page with 18 controls and RX/TX combo split, in-place filter resize, Filter Impulse Cache, Thetis-faithful FFT slider with 7 windows, NF-aware grid, SpectrumPeaksPage, Multimeter page, SettingsSchemaVersion v5 migration), 3M-3a-iv anti-VOX cancellation feed (closes the v0.3.2 gap so the Anti-VOX gain control actually works; full grpAntiVOX trio on Setup → Transmit → DEXP/VOX), live-apply sample rate (no disconnect needed; 12-step Thetis-faithful coordinator routes through WDSP's SetXcmInrate path; HL2 P1 384 kHz parity), and the AF Gain rewire (KM4BLG) + VAX bus calibration (AF slider through SetRXAPanelGain1 closes a long-standing distortion bug; VAX tap inverse-scales by 1 / afGain so digital-mode apps stay calibrated). v0.3.2 (2026-05-05) shipped DEXP/VOX speech processing end-to-end for the first time, the HL2 mi0bot RF/Tune slider maturity push, full Setup → PA parity + the PA over-drive safety hotfix for high-gain finals (notably the ANAN-8000DLE), and a long tail of persistence and stability fixes. v0.3.1 (2026-05-03) carried forward everything 0.3.0 was supposed to deliver plus per-profile TX bandwidth control, the user-editable filter preset store, TX filter overlay on the panadapter and waterfall, the per-board PA forward-power calibration system, and the Setup IA reshape. HL2 SSB transmit is bench-cleared since v0.3.1's ATT/filter safety audit. Earlier-shipped 3G RX-Epic, 3P-A…I antenna integration, 3O VAX + Linux PipeWire, and the v0.2.x maintenance fixes still apply. 3M-2 CW TX is next up; 3F multi-panadapter, 3H skins, 3K CAT remain not-started.

• Radio connection — every OpenHPSDR P1 and P2 board. ANAN-G2 (Saturn / Protocol 2), ANAN-10/10E/100/100B/100D/200D, Hermes, Hermes Lite 2, Angelia, Orion, Metis, Red Pitaya — all discover, connect, stream I/Q, demodulate through WDSP, and persist per-radio settings keyed by MAC. BoardCapabilities registry drives per-board DDC count, ADC count, BPF, Alex filter, and sample-rate support; HardwareProfile engine overlays Thetis clsHardwareSpecific.cs behavior for ambiguous discovery bytes (user-selectable radio-model override in ConnectionPanel).
• Sample-rate wiring — P1 48 / 96 / 192 kHz (+ 384 on Red Pitaya); P2 48 / 96 / 192 / 384 / 768 / 1536 kHz. Per-MAC persistence under hardware/<mac>/radioInfo/. Inline reconnect banner on RadioInfoTab when the selected rate differs from the active wire rate.
• Full Display parity — Setup → Display (Spectrum / Waterfall / Grid & Scales) is wired end-to-end to the renderer on both the QPainter fallback and the QRhi/Metal/Vulkan/D3D12 GPU path. 47-control verification matrix at docs/architecture/phase3g8-verification/README.md. Per-band grid state persists across all 14 bands (160m–6m + GEN + WWV + XVTR) via the first-class Band enum on PanadapterModel.
• Clarity adaptive display — Clarity Blue waterfall palette, ClarityController with cadence / EWMA / deadband, NoiseFloorEstimator percentile, Reset-to-Smooth-Defaults button, Re-tune button + Clarity status badge in the spectrum overlay panel, per-band Clarity memory. Zoom persistence across restarts.
• Full RX DSP parity — 10 WDSP feature slices wired end-to-end through SliceModel → RadioModel → RxChannel → WDSP: AGC advanced (threshold / hang / slope / attack / decay), EMNR (NR2), SNB, APF (SPCW), 3-variant squelch (SSB / AM / FM), mute / audio pan / binaural, NB2 advanced, RIT / XIT client offset, frequency lock, mode containers (FM OPT / DIG / RTTY). Per-slice-per-band persistence under Slice<N>/Band<key>/*.
• VfoWidget rewrite — 4-tab layout (Audio / DSP / Mode / X-RIT), 4×2 DSP toggle grid, AGC 5-button row, S-meter level bar with dBm readout and cyan→green gradient, mode containers, tooltip coverage test. AGC-T ↔ RF Gain bidirectional sync prevents audio-breaking gain runaway.
• Auto AGC-T — NoiseFloorTracker feeds the Auto-threshold timer with MOX guard and agcCalOffset; AUTO button toggles auto-mode; right-click the AGC-T slider opens Setup directly on the AGC/ALC page.
• Step attenuator + ADC overload — StepAttenuatorController with Classic + Adaptive auto-attenuation modes, hysteresis, per-MAC persistence. P1/P2 adcOverflow signal from frame parsers, OVL status badge in RxApplet, per-model preamp items from Thetis SetComboPreampForHPSDR.
• Container / meter system — GPU-rendered meter engine (QRhi 3-pipeline), 31 MeterItem types, 38+ ItemGroup presets (S-Meter, Power/SWR, ALC, ANANMM 7-needle, CrossNeedle, Magic Eye, History, SignalText, TX bar meters), full Thetis-parity Container Settings Dialog (3-column layout, per-item property editors), MMIO external-data subsystem (UDP / TCP-listen / TCP-client / Serial transports; JSON / XML / RAW formats).
• VAX audio routing — NereusSDR-native multi-channel audio bus. IAudioBus abstraction with 5 platform backends (CoreAudio HAL plugin on macOS, PulseAudio pipes / pactl on Linux, PortAudio on Windows). First-run VAX dialog auto-detects Windows virtual-cable families (VB-Audio / VAC / Voicemeeter / Dante / FlexRadio DAX); MasterOutputWidget in the menu bar; Setup → Audio sub-tabs (Devices / VAX / TCI / Advanced); per-slice VAX channel assignment on the VFO Flag, persisted under Slice<N>/.
• SSB voice transmit on every supported board, including Hermes Lite 2 — TxChannel, mic input pipeline (Pc / Radio / Composite sources), MOX state machine, I/Q output on Protocol 1 and Protocol 2. TX speech processing chain: 10-band parametric TX EQ, TX Leveler, TX ALC (3M-3a-i); CFC multi-band compressor, CPDR companding/drive ratio, CESSB controlled-envelope SSB, Phase Rotator (3M-3a-ii). Per-profile TX bandwidth (FilterLow/FilterHigh on every mic profile, TxApplet TX BW spinboxes, debounced WDSP path). 21 factory mic profiles ported verbatim from Thetis; profile manager with Save / Save-As / Delete; two-tone IMD test mode; VOX / DEXP / Anti-VOX. The HL2 ATT/filter safety audit closed in v0.3.1.
• Connection workflow (Phase 3Q) — single state-machine-driven Disconnected → Probing → Connecting → Connected → (LinkLost | Disconnected). Unicast probe reaches radios across Layer-3 VPN tunnels (WireGuard / ZeroTier / Tailscale). 16-SKU model picker organized by silicon family in the Add Radio dialog. Auto-connect-on-launch with per-radio toggle. Spectrum disconnect overlay (fade + click-to-recover) replaces the v0.2.x "frozen spectrum" mystery state.
• Status-bar chrome — title-bar ConnectionSegment shows [state dot] [▲ tx Mbps] [RTT ms] [▼ rx Mbps] [♪ audio] with hover tooltip and right-click menu. Receive-info BadgePair ladder drops in priority order on narrow windows (mode + filter never drop). StationBlock clickable radio-name anchor. AdcOverloadBadge (yellow > 0, red > 3, 2 s auto-hide). CPU System / App right-click toggle. SVG icon system on StatusBadge. Min-filtered RTT for accurate sub-millisecond LAN ping readout.
• Hermes Lite 2 configuration surface — new Hermes Lite Options tab (I2C control, I/O pin state), N2ADR HERCULES toggle writing all 13 SWL pin-7 entries, signed −28..+32 dB step-attenuator range, 13 SWL bands × 7 pins matrix, full per-MAC persistence. Bigger gaps elsewhere in the app remain; this expands a previously-thin HL2 surface.
• App polish — Help → About NereusSDR (version / Qt / WDSP / GPG fingerprint / heritage credits), 💡 AI-assisted issue reporter in the menu bar corner (structured prompts, submits to the bug_report.yml / feature_request.yml GitHub templates), radio-model override persistence, P1 full 17-bank C&C round-robin, NetworkDiagnosticsDialog 4-section health grid.
• Packaging — release.yml prepare → build×3 → sign-and-publish pipeline. macOS DMG and PKG are now Apple Developer ID-signed and notarized in v0.3.0+; Windows installer remains unsigned (Authenticode certificate pending). All artifacts GPG-signed via SHA256SUMS.txt.asc. Per-platform artifacts: Linux AppImage (x86_64 + aarch64), macOS DMG + PKG (Apple Silicon + Intel), Windows portable ZIP + NSIS installer.

• TX pipeline 3M-1 (Basic SSB TX) + 3M-3a-i / 3M-3a-ii (TX Processing — EQ / Leveler / ALC / CFC / CPDR / CESSB / Phase Rotator) — shipped in v0.3.0.
• TX pipeline 3M-3a-iii (DEXP/VOX + AMSQ) — shipped in v0.3.2.
• TX pipeline 3M-3a-iv (Anti-VOX cancellation feed) — shipped in v0.4.0.
• TX pipeline 3M-4 (PureSignal) — shipped in v0.4.0. Feedback DDC, calcc.c / iqc.c engine vendored verbatim from Thetis, PsForm, AmpView, two-tone IMD overlay; enabled across the ANAN family + HL2 + plain Hermes.
• TX pipeline 3M-2 (CW TX) — sidetone, firmware keyer, QSK / break-in. Next major epic.
• TX pipeline 3M-3b (FM-mode pre-emphasis) — de-scoped from 3M-3a-ii to FM-mode follow-up.
• Multi-panadapter (Phase 3F) — DDC assignment, FFTRouter, PanadapterStack, RX2 enable. Anti-VOX aamix path also waits on this.
• HL2 IoBoardHl2 (Phase 3L) — completed via Phase 3P-E: I2C TLV queue + 12-step state machine + bandwidth-monitor two-pointer byte-rate compute + NereusSDR throttle-detection layer; P1CodecHl2 now intercepts C&C frames to inject I2C TLV payloads.
• Skin system (Phase 3H), TCI + Spots (Phase 3J), CAT/rigctld (Phase 3K), WAV/IQ recording (Phase 3M-recording).

Working now:

• OpenHPSDR Protocol 1 and Protocol 2 radio discovery, connection, and per-MAC persistence across the full ANAN / Hermes / Metis / Red Pitaya family
• Per-MAC hardware sample-rate selection (P1 up to 192 / 384 kHz; P2 up to 1536 kHz)
• WDSP 2.10 DSP engine — USB/LSB/AM/CW/DIGI/FM demodulation with full RX DSP parity (AGC advanced, EMNR, SNB, APF, 3-variant squelch, NB1/NB2 advanced, RIT/XIT, mute/pan/binaural, frequency lock, mode containers)
• NNR Standard/Premium noise reduction with per-radio/per-slice persistent tuning, right-click controls, and station-owned model assets
• PureSignal 3 settings, correction assets, and AmpView shared by local and remote GUI sessions; remote actuation awaits R4. See the operator notes and software/hardware verification status
• Per-slice-per-band persistence of DSP state (Slice<N>/Band<key>/*)
• Auto AGC-T with noise-floor tracker + MOX guard
• Step attenuator (Classic + Adaptive auto-attenuation) and ADC-overload OVL badge
• Real-time audio output via QAudioSink (48kHz stereo Int16)
• FFTW wisdom caching with first-run progress dialog; audio device selection and persistence
• GPU-accelerated spectrum + waterfall (QRhi — Metal, Vulkan, D3D12, OpenGL fallback); 4096-point FFTW3 FFT, Blackman-Harris window, 30 FPS, FFT-shift + mirror
• Full Setup → Display wiring — 47 Spectrum / Waterfall / Grid controls live on both render paths; 7 color schemes; per-band grid state across all 14 bands (160m–6m + GEN + WWV + XVTR)
• Clarity Blue waterfall palette + Clarity adaptive auto-tune, Re-tune button, per-band Clarity memory, zoom persistence
• VFO flag widget — 4-tab layout (Audio / DSP / Mode / X-RIT), 4×2 DSP grid, AGC 5-button row, integrated S-meter level bar with dBm readout
• CTUN panadapter — independent pan center and VFO, WDSP shift offsets, bin-subset zoom with hybrid FFT replan, off-screen VFO indicator with double-click recenter
• Filter passband overlay, cursor frequency readout, click-to-tune, scroll-to-tune, filter drag, waterfall pan
• Phase word NCO tuning with Alex HPF/LPF/BPF filters, P1 full 17-bank C&C round-robin
• Dockable / floatable containers with axis-lock, hover-reveal title bar, XML serialization
• GPU-rendered meter engine (QRhi 3-pipeline), 31 MeterItem types, 38+ ItemGroup presets, ANANMM 7-needle with exact Thetis calibration, CrossNeedle dual fwd/rev, Magic Eye, History, Edge mode
• Full Thetis-parity Container Settings Dialog — 3-column layout, per-item property editors (~155 fields), snapshot+revert, container-level Lock/Notes/Highlight/Minimizes/Auto-height, Duplicate, Copy/Paste item settings
• MMIO (Multi-Meter I/O) external-data subsystem — UDP / TCP-listen / TCP-client / Serial transports, JSON / XML / RAW formats, endpoint manager, variable picker, 10 fps polled bindings
• VAX multi-channel audio bus — 5 platform backends, Windows virtual-cable auto-detect (VB-Audio / VAC / Voicemeeter / Dante / DAX), MasterOutputWidget, per-slice VAX channel routing
• Interactive button grids — band (14), mode, filter, antenna, tuning step, macro — with hover/click feedback
• Full UI skeleton — 12 applets, 9-menu bar, 47-page SetupDialog, SpectrumOverlayPanel with 5 flyout sub-panels, status bar
• Help → About dialog + 💡 AI-assisted issue reporter wired to the GitHub issue tracker
• SSB voice transmit + speech processing chain (TX EQ + Leveler + ALC + CFC + CPDR + CESSB + Phase Rotator), 21 factory mic profiles, two-tone IMD test
• Per-profile TX bandwidth control + user-editable filter preset store + mode-aware filter grid + TX/RX filter overlay on panadapter and waterfall
• Per-board PA forward-power calibration (Watt Meter / PA Values setup pages, CalibratedPAPower interpolation, SWR protection at every setTxDrive site)
• Connection workflow with state machine + unicast probe + auto-connect-on-launch + disconnect overlay (Phase 3Q)
• Hermes Lite 2 configuration tabs (Hermes Lite Options + I/O Pin State + N2ADR HERCULES + SWL matrix + signed S-ATT range, all per-MAC)
• Status-bar redesign — title-bar connection segment, drop-priority receive badges, ADC overload indicator, station-name anchor, CPU System/App toggle
• GPG-signed cross-platform builds — Linux AppImage ×2 archs, macOS DMG + PKG ×2 archs (Apple Silicon + Intel; Developer ID-signed + notarized in v0.3.0+), Windows portable ZIP + NSIS installer

Planned (see Roadmap):

• Phase 3M-2 CW TX — sidetone, firmware keyer, QSK/break-in (next major TX epic)
• Phase 3M-3b FM-mode work — pre-emphasis (deferred from 3M-3a-ii)
• Phase 3F Multi-Panadapter — DDC assignment (including PS states), FFTRouter, PanadapterStack, RX2 enable. Anti-VOX aamix path also waits on this.
• Phase 3H Skin System — Thetis-inspired skin format with 4-pan support and legacy-skin import
• Phase 3J TCI + Spots — TCI v2.0 WebSocket server, DX Cluster / RBN clients, spot overlay
• Phase 3K CAT / rigctld — 4-channel rigctld, TCP CAT server
• Phase 3L HL2 IoBoardHl2 — delivered via Phase 3P-E: I2C-over-ep2 TLV queue + 12-step UpdateIOBoard state machine + full bandwidth-monitor port with throttle detection.
• Phase 3M Recording — WAV record/playback, I/Q record, scheduled

See docs/MASTER-PLAN.md for the full implementation plan.

# Ubuntu 24.04+ / Debian
sudo apt install qt6-base-dev qt6-base-private-dev \
  qt6-multimedia-dev qt6-shadertools-dev qt6-svg-dev qt6-websockets-dev \
  cmake ninja-build pkg-config \
  libfftw3-dev libgl1-mesa-dev \
  libasound2-dev libjack-jackd2-dev \
  libpipewire-0.3-dev libssl-dev

# Arch / CachyOS / Manjaro
sudo pacman -S qt6-base qt6-multimedia qt6-svg qt6-websockets \
  cmake ninja pkgconf fftw \
  alsa-lib jack2 pipewire openssl

# macOS (Homebrew)
brew install qt@6 ninja cmake pkgconf fftw

The bundled PortAudio is built with PA_USE_ALSA=ON and PA_USE_JACK=ON on Linux, so the ALSA and JACK development headers are required at compile time even if you don't use those audio backends at runtime. libpipewire-0.3-dev (≥ 0.3.50) is strongly recommended on PipeWire-default distributions (Ubuntu 24.04+, Fedora 39+, Arch) — without it the Linux audio path falls back from the native libpipewire-0.3 bridge to the older pactl route.

openssl / libssl-dev (≥ 3.0) is hard-required (find_package(OpenSSL 3.0 REQUIRED COMPONENTS Crypto)) since Remote Daemon R2 Task 17: src/core/security/CertificateStore links libcrypto directly to mint nereusd's self-signed TLS certificate, because Qt6 has no certificate-generation API. macOS resolves this through Homebrew's openssl@3 formula with no extra hints; it is not keg-only (verified via brew info --json=v2 openssl@3 fix round 3), its files are ordinary live symlinks into ${HOMEBREW_PREFIX}/lib and .../include, the same path already searched for every other Homebrew library. The macOS Intel release row (no arm64 keg to link against) and both Windows rows source OpenSSL differently; see the find_package(OpenSSL) block in CMakeLists.txt and vcpkg.json for the full per-platform acquisition story.

No manual setup required. CMake auto-downloads fftw-3.3.5-dll64.zip on first configure and drops fftw3.h / libfftw3-3.dll / libfftw3-3.def into third_party/fftw3/. Requires network access on the first cmake -B build run; offline builds need to pre-populate those three files by hand.

git clone https://github.com/boydsoftprez/NereusSDR.git
cd NereusSDR
cmake -B build -G Ninja -DCMAKE_BUILD_TYPE=RelWithDebInfo
cmake --build build -j$(nproc)
./build/NereusSDR

The build produces two binaries. NereusSDR is the GUI. nereusd is the headless daemon added in remote-daemon R1: it links NereusCore only, no GUI object code, and is guarded by tst_core_has_no_gui_includes. It is installed separately so the GUI's release artifacts are unaffected:

cmake --install build --component nereusd

A plain cmake --install deliberately installs neither nereusd nor its systemd unit. Note that release.yml is the only workflow that runs cmake --install and it triggers only on v* tags, so the daemon's install path has no PR-CI coverage. Always pass --profile <name> when running nereusd by hand, or it writes the same settings the GUI reads.

On first run, NereusSDR generates FFTW wisdom (optimized FFT plans). This takes ~15 minutes and shows a progress dialog. The wisdom file is cached for subsequent launches.

See docs/MASTER-PLAN.md for the full implementation plan and docs/project-brief.md for the project brief.

PRs, bug reports, and feature requests welcome! See CONTRIBUTING.md for guidelines.

Development environment: NereusSDR is developed using Claude Code as the primary development tool. We encourage contributors to use Claude Code for consistency. PRs must follow project conventions, pass CI, and include GPG-signed commits.

NereusSDR stands on the shoulders of these projects:

• Thetis — The canonical Apache Labs / OpenHPSDR SDR console (C# / WinForms). NereusSDR's feature source.
• AetherSDR — Native FlexRadio client (C++20 / Qt6). NereusSDR's architectural template.
• WDSP — Warren Pratt NR0V's DSP library. The signal processing engine.
• OpenHPSDR — The open-source high-performance SDR project and protocol specifications.

NereusSDR is free and open-source software licensed under the GNU General Public License v3.

NereusSDR is a derivative work of Thetis licensed under the GNU General Public License. It is not affiliated with or endorsed by Apache Labs, FlexRadio Systems, ramdor/Thetis, or the OpenHPSDR project.