Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.lagerdata.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Features

  • Concurrent J-Link probes on a single Lager Box. Two J-Link probes plugged into one Box can now run independent debug sessions side-by-side. The box-side service in box/lager/debug/service.py resolves each debug net’s J-Link USB serial from its VISA address and allocates a deterministic per-probe slot (read from saved_nets.json via NetsCache). Slot N owns a three-port window — GDB 2331+3N, SWO 2332+3N, telnet 2333+3N — plus RTT base 9090+2N. The auxiliary -swoport and -telnetport are passed explicitly to JLinkGDBServer so its defaults of 2332/2333 can’t collide with another slot’s GDB port. The service passes -select USB=<serial> to JLinkGDBServer and -SelectEmuBySN <serial> to JLinkExe, writes per-serial PID and log files, and narrows pkill so disconnecting one probe no longer tears down the other. The CLI’s --gdb-port default is now None rather than 2331, so the box’s allocator is honored unless an explicit port is requested; the effective gdb_port returned by the box is printed on connect. start_box.sh publishes the widened 2331-2342 Docker port range and secure_box_firewall.sh admits the same range, so existing hardened boxes need a firewall refresh to use the new slots. Nets without a parseable serial (legacy single-probe setups) fall back to slot 0 / GDB 2331 / RTT 9090 / /tmp/jlink_gdbserver.pid and continue to work unchanged.
  • RIGOL DP811 power supply detection. lager instruments now lists the DP811 alongside the DP821 and DP832. The DP811 shares VID:PID 1ab1:0e11 with its siblings, so it is identified by USB serial prefix (DP8H or DP81 → Rigol_DP811) in both box/lager/http_handlers/usb_scanner.py and cli/impl/query_instruments.py, and is added to the serial-disambiguated bucket so a generic VID:PID match cannot misclassify it as a DP821.
  • Multiple concurrent viewers per webcam stream. Each /stream connection used to open its own cv2.VideoCapture against /dev/videoN, which V4L2 serves exclusively — the second viewer either failed or got blank frames. The streamer subprocess in box/lager/automation/webcam/service.py now spins up a single daemon capture thread on the first viewer that owns the device and broadcasts encoded JPEG frames to a shared buffer guarded by a threading.Condition. Any number of viewers can now subscribe to the same stream concurrently. Stop and re-start each webcam to pick up the regenerated streamer script.

​
Bug Fixes

  • LabJackADC.input() no longer inherits sticky AIN register state from a previous tool. ADC reads previously called ljm.eReadName with zero AIN register configuration, inheriting whatever the previous tool left in AIN_RANGE, AIN_NEGATIVE_CH, AIN_RESOLUTION_INDEX, and AIN_SETTLING_US. T7 register state persists in device RAM until USB power-cycle, so if a previous tool left an AIN in differential mode with a floating negative channel, every read saturated at ~10.10 V regardless of the actual signal — indistinguishable from a real wiring fault. Safe defaults (RANGE=10.0, NEGATIVE_CH=199, RESOLUTION_INDEX=0, SETTLING_US=0) are now written once per (handle, channel) tuple before the first eReadName, cached in a class-level set. Config-write failures are logged but do not raise.

​
Improvements

  • Webcam capture forces MJPEG so two cameras can share a USB 2.0 bus. Default OpenCV negotiation picked YUYV (uncompressed, ~150 Mbps at 640×480 30fps), which doesn’t leave room for a second camera on the same bus — the kernel rejected VIDIOC_STREAMON with “Not enough bandwidth for altsetting”. MJPEG is roughly 5× smaller and fits two cameras comfortably; the FourCC is now set before width/height/fps so negotiation honors it.

​
Installation

To install this version: 
pip install lager-cli==0.17.0
To upgrade from a previous version: 
pip install --upgrade lager-cli

​
Resources

View Release on PyPI