Built a trigger, and then removed it

The gap looked closable from the OOT side, so I built it: a level-crossing detector in work() that, on a rising edge, simply stops writing into the tensor for a while. Superluminal keeps re-reading the same bytes, so the trace holds still. Demo worked.

It’s still wrong. A real trigger captures it grabs a fixed window framed around the crossing, some samples before, some after, and shows you that record. Mine never captured anything, it just covered the lens after the interesting thing had already gone past. Same look, different semantics, and shipping a look-alike is worse than shipping nothing. It’s gone.

A bug I did not expect

While building a qtgui flowgraph to compare against, it refused to generate and so did GNU Radio’s own bundled qtgui_vector_sink_example.grc, which has never seen a cyberether block. Simply having our custom workflow installed breaks generate_options resolution for unrelated flowgraphs. Traced it as far as the value already being wrong immediately after import_data(), before validation runs. Not root-caused yet, and it’s the loose end I most want closed.

Where this leaves the repo

  • A friction log from real signals instead of guesses.
  • One feature built, demoed, and deleted on purpose.
  • One real compatibility bug found, and honestly, still open.

Next: giving the controls somewhere proper to live.

Repo: gr-cyberether — main Mentors: Luigi Cruz (CyberEther), Håkon Vågsether (GNU Radio).