udpscope/docs/claude/architecture.md

# Architecture: threads, the data path and the layers

The two threads and what they share, the four display modes, the display clock, and the layers to keep straight.

Part of [CLAUDE.md](../../CLAUDE.md).

`SPECS.md` is the requirement document. `README.md` documents what was actually built,
including deliberate deviations from the spec and known limitations — read it before
concluding something is a bug.

### Threads

Exactly two: a producer thread in `Acquisition` owns a `DataSource` and calls
`source->poll(50)`; the GTK main thread owns every widget and GL context. They share
`ScopeModel` under `ScopeModel::mutex()`, taken once per DATA frame by the producer and
once per frame by the UI (`tick()` in `src/ui.cpp`) around `RowSet::bind()`/`acquire()`.
Samples themselves cross via `plot::data::RingBuffer` (single producer, single
consumer) and never through the lock. Cross-thread scalars are atomics on `ScopeModel`
(`trigger_enabled_`, `trigger_armed_`, `trigger_latches_`, `trigger_time_`,
`event_`, `generation_`). Two of them are a pair: the producer writes `capture_edge_`
before `trigger_pending_` and those two carry release/acquire, so a consumer that sees a
capture open sees which edge it belongs to.

### Sources

What feeds `Acquisition` is a `DataSource`, one of the registered *kinds*.

- The kinds live in `src/source/`. `register_builtin_sources()` fills the registry explicitly --
  there is no static-initialiser registration, which a linker may drop -- and a future `dlopen`
  loader would only call `register_kind()`.
- A `DataSource` delivers into `Acquisition` (a `SourceSink`) from inside `poll()`. `Frame` is
  source-neutral, and its pointers are valid **only during `on_data`**: copy what is kept.
- The GTK thread reaches a source **only** through `DataSource::post(Command)`; the queue is drained
  at the top of `poll()`. A panel never calls a source directly.
- `UdpsSource` wraps `deps/udpsclient` unchanged; `udps_*` names appear only in
  `src/source/udps_*.cpp` (`grep -rn "udps_" src` outside those files is a leak of the old coupling).
- The frame is **single-timebase**. A multi-timebase source (ABScope) will change the interface and
  the data path together; do not bend one frame to carry two clocks.
- A kind's capability panels are pages of `side_stack`, named `source-<id>`. A disconnected source
  leaves them **insensitive, not hidden** (`apply_source_panels` in `src/ui_source.cpp`).
- Session file naming for a non-UDPS kind: the pseudo-host is `<kind>-<identity>` (characters outside
  `[A-Za-z0-9._-]` become `_`) on port **1** (0 is rejected by the session-file parser), the kind is
  stored as `source_kind` and its settings as `source_opt_<key>`. A UDPS file keeps `<host>_<port>` and
  has neither key, so a legacy session is read as UDPS and nothing existing moves.
- **The E-1608 kind has three layers, and only the top one knows the framework.** `e1608_link`
  (discovery, handshake, framed TCP commands, the scan socket) and `e1608_device` (typed commands,
  factory calibration, `decode_scans`) are plain POSIX C++ with no GTK or `DataSource` types; discovery is
  `Scan` (one socket per interface, polled by the caller: the form polls it from a GLib timer on the main
  thread, the source's lookup goes through `discover_broadcast`);
  `e1608_source` is the `DataSource`; `e1608_kind` and `e1608_panels` are the UI. A change to the wire
  never reaches the UI files.
- The E-1608 source runs **everything inside `poll()`** on the producer thread: posted commands, the
  connect state machine (Down, Scanning, Halted), the scan read, the 500 ms status check (its not-running restart is skipped when a hardware trigger is configured), the digital
  input/counter poll and the 20 ms waveform tick. There is no second thread. A connect attempt (discovery,
  handshake, calibration read, scan start) can block `poll()` for as long as the command timeout allows each
  step; that delays only the command queue. Every exception is caught inside `poll()`: an escape would end
  the producer thread, and `drain_commands()` runs inside the same `try` for that reason.
- Its frames are **fixed-size blocks** of about 10 ms (`round(scan_rate / 100)` scans, 1 to 4096), not
  whatever one `recv` returned: `build_stamps` warns once when a frame's sample count differs from the
  first one's. Digital lines and the counter are held across a block (a step trace).
- The E-1608 panels read the board's state from `E1608Readouts`, a set of atomics the producer writes
  and a tick callback throttled to 100 ms reads; a panel never touches the source's internals, and writes
  only through `post()`. `Acquisition::on_event` ignores the detail string and does nothing for `Error`,
  so the source also writes each new error to stderr as `[e1608] ...` (once per distinct message, not per
  retry); that line is where a user sees why a connection failed.
- **The flat UDP kind is three files and no thread.** `flatudp_signals` is the `signals` string codec,
  the validity rules and the decoders, with no GTK and no sockets (the probe round-trips it);
  `flatudp_source` is `FlatUdpSettings` plus the `DataSource`; `flatudp_kind` is the form, a signal-row
  editor (a name, a type, a count and, while the count is above 1, a "Vector as" drop-down per row). The
  datagram has no description, so a list that is wrong is invisible except as dropped datagrams: the size
  check is the only defence, and it is why a wrong-size datagram is counted (`stale`) and reported once per
  size.
- A vector row is shown one of two ways, chosen per row (`Row::wave`, the `w` suffix in the text). As
  **channels** it is `count` scalar channels of one sample per cycle, so the list batches like an
  all-scalar one. As a **waveform** it is one channel with `rows = count` and
  `sampling_rate = rate * count`, one datagram per `Frame` (`num_samples = 1`). The two cannot share a
  list because the model spans a frame by its largest element block and `Frame::value` is 0 past a scalar's
  one value: a scalar beside a vector block would be zero-padded, so `problem()` demands that a list with a
  waveform row be waveform rows of one count, and the form shows that message in the subtitle. The channel
  total is capped at 256.
- Its frames are **fixed-size blocks** like the E-1608's for the same reason (`build_stamps` warns when the
  sample count varies): `round(rate / 100)` datagrams for a list with no waveform row, a partial block
  flushed after 100 ms. `rate` is required because `sampling_rate = 0` makes the model size the ring and
  the ingest decimation for 1 kHz; there is no time-base switch, since `acquisition.cpp` has no stamping
  path that ignores the declared cadence.
- `Disconnected` after 2 s without a *valid* datagram (wrong-size ones do not refresh the clock); it resets
  `Acquisition`'s per-session state, so the next datagram re-anchors rather than reading as a gap, and the
  history survives (`CONFIG` is sent once, from the first `poll()`).
- The form's row editor removes a row from inside its own button's `clicked` emission (GTK keeps the emitter
  alive to the end of the emission, and the handler copies the form and the row out of its context first);
  every handler tests `Form::alive`, which the groups' `destroy` clears, because the chooser may have
  destroyed the sibling widgets the handlers reach by raw pointer.

### Data path

`ScopeModel` holds **one `BufferStore`** (`live()`): one shared time-base ring plus one
value ring per channel, sized by `plan_partition()` to cover `--max-window` seconds
within the byte budget. The producer always writes it, whatever the trigger is doing —
that is what guarantees a full pre-trigger history on the *first* latch after arming.
The spec's second buffer (item 1) is the latched `EventSnapshot`, which is already an
immutable copy of the captured window; see README for why two live rings were dropped.

What a trace binds to is `ScopeModel::display_mode()`, one of four states, and
`RowSet::bind()` takes it as an argument rather than recomputing it:

- `Live` — no trigger: the live ring, on the clock of the **newest sample**
  (`display_epoch() == -newest`), so the axis is a fixed `[-window, 0]` and x reads as an age.
- `Capturing` — an edge has been accepted (so the pre window was full) and the post
  window is still filling: the live ring, on the coming event's own clock
  (`display_epoch() == -edge`), inside an axis already pinned to `[-pre, +post]`. The
  post side fills in view and the marker sits at x = 0 throughout.
- `Held` — the frozen `EventSnapshot` of the last event, whose stamps are already edge-relative,
  so the display epoch is 0.
- `Waiting` — armed, nothing captured yet: blank, with the live epoch (there is nothing bound,
  but the axis is still a live one).

**Disarm during a capture publishes what the capture got to**, then disarms: the producer
cuts the window at the newest sample and that partial snapshot becomes the held event, so
the picture on screen at the moment of the Disarm stays on screen. `engine.disarm()` clears
`pending_`, which drops the display out of `Capturing`, and with nothing published the view
fell back to the *previous* event — or to blank on a first capture. No event fired, so
`trigger_latches()` is deliberately not bumped; `ScopeModel::event_seq()` counts published
snapshots instead, and the frame tick keys its event framing on *that*. Keep the two
distinct: the latch count is what the probes assert "how many events fired" with.

`Capturing` is a **deliberate departure from SPECS:37** ("data visualisation should not
be refreshed till next event"), asked for explicitly; that reading is what the old "never
the live ring mid-capture" rule enforced. What survives of the rule is that `Held` binds
the snapshot *only* — a held event is never quietly replaced by live data — and that the
*axis* does not move during a capture: it is the event's window from the edge onwards,
never a sliding live view.

`display_mode()` must be read **once per frame** and passed around (the tick does this,
and hands it to `bind()`). The producer can open or close a capture between any two
reads, and two reads in one frame put the bind, the epoch, the pin and the status line on
different clocks.

**The display clock is not the stream clock, and `ScopeModel::stream_at_x0()` is the only
conversion.** `display_x = stream_t - stream_at_x0()` in **every** mode, and the offset lives in
one of two places: a snapshot carries its own (`EventSnapshot::x_offset`, frozen with the
picture) and a live view carries it as minus `display_epoch()`. Anything crossing between the
two clocks goes through that one accessor rather than picking the branch itself — the export's
`abs_seconds` column, the pause handler putting a display window back on the ring to cut it, the
measurement marks — because there are four modes and picking the branch by hand gets one of them
wrong. It reads atomics only, so it is safe to call inside an existing `mutex()` guard.
Three things about the live epoch specifically:
- **It is written after `bind()` and `acquire_all()`, not before.** It must be minus the newest
  sample the *traces* can report, and asking before the acquire is one frame of production
  stale — at 1 MS/s over a 50 ms window that is a third of the screen to the right of the axis.
- **There is no feedback loop**, even though the newest stamp is read through the epoch that is
  about to change: `RowSet::live_edge()` reports in the epoch currently in force, so
  `display_epoch() - live_edge()` is minus the *raw* stamp and the fixed point is reached in one
  frame rather than converged on.
- **`Waiting` takes the live epoch too.** Nothing is bound, but the axis is still a live one, and
  an epoch of 0 there would make the axis jump on the frame a capture closes.

**A measurement mark is glued to its sample; a cursor stays on screen.** The two instruments move
in opposite directions under a rolling axis and that is deliberate — a bench scope's cursors sit
where the user put them while the waveform rolls under, and a measurement's mark names the sample
it measured. The cursors need no code: their x *is* a display coordinate. The marks are stored in
**stream** seconds and reconverted every frame by `place_measure_marks()`, since a mark left in
display coordinates would crawl forward at one window per window. `App::measure_window` is
stream-clocked for the same reason, and that one is load-bearing: the recompute trigger is "the
window moved by 2 % of its span", and in display coordinates a live window is a fixed
`[-window, 0]` — it never moves, so the measurements would never be recomputed again.

Layers to keep straight:

- `SignalInfo` — one signal as the CONFIG described it. The unit of enable/disable.
- `ChannelInfo` / `ChannelBuffer` — the unit of acquisition and plotting. A
  `ChannelBuffer` is a `plot::data::Series` over a (value, time) ring pair.
- `SlotSeries` (`src/channel.hpp`) — a trace's stable series object, rebound each frame
  to either a `ChannelBuffer` or an `EventSeries`.
- `RowSet` (`src/plot_host.hpp`) — **one `plot::Figure`** of up to `kMaxRows` rows (added
  and removed one at a time by the layout editor), each a
  borrowed `plot::Plot*` with `kSlotsPerRow` fixed trace slots. The figure owns
  everything the rows have to agree on: the x window (one x link group every row joins),
  the live follow (the *group's* — a member's own axis must be `XFollow::Off`, since
  `Figure::link_x()` refuses a `Live` member), the left gutter and the crosshair.

- `ExportTable` (`src/export.hpp`) — the visible window as columns, collected from the
  snapshot or the live rings and handed to the `.npy`/`.csv` writers. Always the stored
  samples, never the LOD pyramid's buckets.

`src/ui.cpp` is the GTK4 window (toolbar, trigger panel, signal panel, plot box) plus
the `tick()` frame function that binds, updates the trigger state and rewrites the
status line. One `plot::gtk4::PlotArea` in figure mode draws the whole figure. It is by
far the largest file.