The diagrams conflated the SIMtrace2 hardware with the sniffer that reads it
and omitted the original Osmocom simtrace2-sniff: they showed
simtrace2-pysniff feeding the GSMTAP listener with the hardware only as its
subtitle, and a second, unexplained "SIMtrace2 hardware / direct USB" box.
The diagrams now show the device -> sniffer -> GSMTAP relationship explicitly:
SIMtrace2 hardware is read either by simtrace2-pysniff or by the original
simtrace2-sniff, and FX2 hardware by the sigrok_iso7816_stream decoder; the
sniffers/decoder stream GSMTAP-SIM to the UDP :4729 listener. The direct USB
path is drawn separately from the device to the server and is labelled
"pysniff code in-process": --capture direct uses the bundled
simtrace2_pysniff package (DirectSniffer -> SniffSession) inside the server
process, not the simtrace2-pysniff CLI, and bypasses GSMTAP. PCAP/COMPRION
import, SQLite and the browser UI keep their places.
The READMEs' alt text, the first-run capture-mode notes and the Usage command
comments now name the original simtrace2-sniff as a GSMTAP source and say the
direct mode runs the bundled sniffer code inside the server.
Both languages; 558 Python + 80 Node tests; runtime zip still carries both
diagrams and the README image guard passes.
Both READMEs gain a capture-and-analysis diagram right under the screenshot:
simtrace2-pysniff and sigrok_iso7816_stream feed the GSMTAP UDP :4729
listener, SIMtrace2 hardware can be read directly over USB, PCAP/COMPRION
traces come from files, and simtrace-analyser-server decodes/stores into
SQLite and serves the web UI. The SVGs carry title/desc and role=img, use
only inline presentation attributes (no scripts, no external refs) and render
without clipped text; the RU text matches the EN layout.
The runtime zip includes both diagrams so the embedded READMEs keep working,
and a new drift guard checks that every image a README references is present
in the archive. RU wording: "устанавливаемое PWA" (neuter agreement).
558 Python + 80 Node tests.
catarrh
