tan-cli 0.7.0__py3-none-any.whl

tan/templates/vendored/sensor/E1M-V2N101/CMakeLists.txt

@@ -0,0 +1,8 @@
+# SPDX-License-Identifier: Apache-2.0
+
+cmake_minimum_required(VERSION 3.20)
+
+find_package(Zephyr REQUIRED HINTS $ENV{ZEPHYR_BASE})
+project(i2c_master LANGUAGES C)
+
+target_sources(app PRIVATE src/main.c)

tan/templates/vendored/sensor/E1M-V2N101/README.md

@@ -0,0 +1,112 @@
+# i2c-master
+
+Discrete I2C master that reads a known device at a known
+address.  Pattern: open the bus, init the chip driver, loop
+reading the sensor every second, close cleanly.
+
+Contrasts with [`examples/peripheral-io/i2c-scanner`](https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-scanner) -- the
+scanner *probes* every 7-bit address for ACKs without knowing
+what's behind them; this example *reads* a known sensor.
+
+## What this shows
+
+* `alp_i2c_open()` -- open `BOARD_I2C_SENSORS` at 400 kHz Fast-mode.
+  The `<alp/board.h>` alias resolves to `ALP_E1M_I2C0` on the E1M
+  EVK and `ALP_E1M_X_I2C0` on the E1M-X EVK.
+* `bmp581_init()` -- chip-driver probe (CHIP_ID check).
+* `bmp581_set_sampling()` -- configure oversampling, ODR, and power mode.
+* `bmp581_read_raw()` + `bmp581_compensate()` -- one pressure+temperature read per second.
+* Error handling: NACK on probe -> graceful exit with diagnostic.
+* `bmp581_deinit()` + `alp_i2c_close()` -- clean shutdown.
+
+## Hardware
+
+The BMP581 barometer is the on-board sensor on `BOARD_I2C_SENSORS`
+on both the E1M EVK (U14) and the E1M-X EVK, per
+[`metadata/boards/e1m-evk.yaml`](https://github.com/alplabai/alp-sdk/blob/v0.17.0/metadata/boards/e1m-evk.yaml) and
+[`metadata/boards/e1m-x-evk.yaml`](https://github.com/alplabai/alp-sdk/blob/v0.17.0/metadata/boards/e1m-x-evk.yaml)
+`i2c_devices:`.
+
+7-bit address depends on the SDO strap:
+
+| SDO strap | Address | On this bus                          |
+|-----------|---------|---------------------------------------|
+| VDDIO     | 0x47    | E1M EVK U14 + E1M-X EVK (this default) |
+| GND       | 0x46    | (not populated on these EVKs)          |
+
+Both supported EVKs strap SDO to VDDIO, so `BMP581_ADDR_7BIT`
+works unchanged across them (see
+[`metadata/boards/e1m-evk.yaml`](https://github.com/alplabai/alp-sdk/blob/v0.17.0/metadata/boards/e1m-evk.yaml),
+BMP581 datasheet BST-BMP581-DS004 s5.6, or
+[`include/alp/chips/bmp581.h`](https://github.com/alplabai/alp-sdk/blob/v0.17.0/include/alp/chips/bmp581.h)).
+
+> **#1269:** this example used to target the TMP112 temperature sensor,
+> but TMP112 lives on **BRD_I2C**, not on `BOARD_I2C_SENSORS` -- opening
+> `BOARD_I2C_SENSORS` and probing for TMP112 NACKs on real hardware.
+> BRD_I2C is a separate controller instance regardless (on the E1M-AEN
+> family it is SoC I2C0, function C -- #1848), so simply repointing the bus_id
+> would still not reach a TMP112 there. See
+> [`examples/v2n/v2n-temp-sensor`](https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/v2n/v2n-temp-sensor) for the
+> V2N-only BRD_I2C/TMP112 pattern instead.
+
+## Build
+
+```bash
+# Standalone, native_sim (emul I2C; bmp581_init NACKs cleanly):
+# writes ./generated/alp.conf, which west reads below (#866)
+tan generate --target zephyr-conf --core m33_sm --sdk-root "$ALP_SDK_ROOT" --output generated/alp.conf
+west build -b native_sim/native/64 . \
+    -- -DEXTRA_CONF_FILE=generated/alp.conf -DEXTRA_ZEPHYR_MODULES=$ALP_SDK_ROOT
+west build -t run
+
+# On real silicon, point -b at the SoM's Zephyr board target.
+# Example for E1M-V2N101:
+west build -b alp_e1m_v2n101_m33_sm/r9a09g056n48gbg/cm33 . -- -DEXTRA_CONF_FILE=generated/alp.conf
+west flash --host <board-ip>
+```
+
+## Expected output
+
+Real hardware (BMP581 populated, room temperature, sea-level-ish pressure):
+
+```
+[i2c-master] open BOARD_I2C_SENSORS @ 400 kHz
+[i2c-master] bmp581_init @ 0x47 -> 0 (OK)
+[i2c-master] sample 0: 96234 Pa, 23.625 degC
+[i2c-master] sample 1: 96231 Pa, 23.687 degC
+[i2c-master] sample 2: 96233 Pa, 23.625 degC
+[i2c-master] sample 3: 96230 Pa, 23.687 degC
+[i2c-master] sample 4: 96232 Pa, 23.750 degC
+[i2c-master] done
+```
+
+native_sim (emul I2C, no BMP581 registered):
+
+```
+[i2c-master] open BOARD_I2C_SENSORS @ 400 kHz
+[i2c-master] bmp581_init @ 0x47 -> -5 (populated? right address?)
+[i2c-master] done
+```
+
+## Troubleshooting
+
+* **`bmp581_init -> -5`** (ALP_ERR_IO / NACK).  Either the chip
+  isn't populated on your board, the address is wrong (see
+  table above), or the bus is held low (missing pull-ups, stuck
+  slave).  Run [`examples/peripheral-io/i2c-scanner`](https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-scanner) to confirm what
+  ACKs.
+* **`alp_i2c_open failed`** (NULL return).  The `alp-i2c0` DT
+  alias isn't set -- check your board overlay or, for
+  native_sim, that `CONFIG_EMUL=y CONFIG_I2C_EMUL=y` and the
+  overlay we ship are picked up.
+* **Garbled readings.**  Wrong baud (bitrate) for the bus
+  capacitance.  Drop from 400 kHz to 100 kHz to confirm.
+
+## Reference
+
+- [`<alp/peripheral.h>`](https://github.com/alplabai/alp-sdk/blob/v0.17.0/include/alp/peripheral.h) I2C surface.
+- [`<alp/chips/bmp581.h>`](https://github.com/alplabai/alp-sdk/blob/v0.17.0/include/alp/chips/bmp581.h) -- driver API.
+- [`examples/peripheral-io/i2c-scanner/`](https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-scanner) -- discovery companion.
+- [`examples/peripheral-io/i2c-slave/`](https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-slave) -- slave-mode companion, built on the `alp_i2c_target_*` surface.
+- [`examples/v2n/v2n-temp-sensor/`](https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/v2n/v2n-temp-sensor) -- TMP112-on-BRD_I2C companion, V2N-only.
+- BMP581 datasheet (BST-BMP581-DS004, rev 1.13).

tan/templates/vendored/sensor/E1M-V2N101/board.yaml

@@ -0,0 +1,57 @@
+# board.yaml -- per-app project configuration for the
+# i2c-master example.
+#
+# Reads the BMP581 barometer on BOARD_I2C_SENSORS via the chip-
+# driver header.  Discrete master that talks to a known device at
+# a known address -- contrasts with https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-scanner which
+# probes every 7-bit address for ACKs.
+#
+# Customer workflow: `tan init --from-example peripheral-io/i2c-master`, change `som.sku` to
+# your MPN, edit the `preset:` (or use an inline board) if you're on a non-EVK board,
+# and `tan build`, which plans alp.conf from this file; twister and a bare
+# west build read the pre-generated generated/alp.conf instead
+# (`tan generate --target zephyr-conf`, see README.md).
+som:
+  sku: E1M-V2N101
+
+preset: e1m-x-evk
+supported_boards:
+  - e1m-evk
+  - e1m-x-evk
+
+# E1M pads this app uses (subset of the EVK preset's `e1m_routes:`).
+pins:
+  - { e1m: E1M_X_I2C0, macro: XEVK_I2C_BUS_SENSORS, doc: "On-board sensor + IO-expander + INA236 bus (ICM-42670, BMI323, BMP581, TCAL9538, INA236)." }
+cores:
+  m33_sm:
+    app: ./src
+    # Zephyr peripheral subsystem exercised by this example.  The
+    # loader translates the entry to CONFIG_I2C=y in alp.conf.
+    peripherals:
+      - i2c
+
+# BMP581 is populated on every member of the AEN, V2N, V2N-M1
+# families per alp-sdk's metadata/chips/bmp581.yaml (not part of
+# this scaffolded project) -- so this example is Ring 2
+# (chip-bound, multi-family) per
+# alp-sdk's scripts/check_example_portability.py classification
+# (also not part of this scaffolded project).
+#
+# NOTE (#1269): TMP112 was here previously, but TMP112 lives on
+# BRD_I2C, not on BOARD_I2C_SENSORS -- see alp-sdk's
+# metadata/boards/e1m-evk.yaml i2c_devices: and
+# metadata/boards/e1m-x-evk.yaml i2c_devices: (not part of this
+# scaffolded project), neither of which lists a TMP112 on the sensor
+# bus this example opens. BRD_I2C
+# is a SEPARATE controller instance from BOARD_I2C_SENSORS regardless
+# (on the E1M-AEN family, BRD_I2C is SoC I2C0, function C -- #1848), so
+# repointing the bus_id would still not reach a TMP112 there. BMP581 is
+# the part that is actually on BOARD_I2C_SENSORS on both supported
+# boards, at the same address (0x47) on each -- see
+# https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/v2n/v2n-temp-sensor for the V2N-only BRD_I2C/TMP112 pattern
+# instead.
+chips:
+  - bmp581
+
+diagnostics:
+  log_level: info

tan/templates/vendored/sensor/E1M-V2N101/boards/native_sim_native_64.conf

@@ -0,0 +1,10 @@
+# SPDX-License-Identifier: Apache-2.0
+#
+# native_sim has no real I2C controller; pull in the emul I2C
+# driver so the bus enumerates at boot and bmp581_init can probe
+# (which returns NACK because no BMP581 is registered -- exactly
+# the path the example's error-handling branch exercises).
+# Auto-loaded by Zephyr when building for native_sim/native/64.
+
+CONFIG_EMUL=y
+CONFIG_I2C_EMUL=y

tan/templates/vendored/sensor/E1M-V2N101/boards/native_sim_native_64.overlay

@@ -0,0 +1,28 @@
+/*
+ * Copyright 2026 Alp Lab AB
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * One emulated I2C bus exposed via the alp-i2c0 alias.  No emul
+ * targets are attached so bmp581_init returns NACK -- but the
+ * open / probe / close path runs end-to-end and the harness
+ * latches on the `[i2c-master] done` marker.
+ */
+
+#include <zephyr/dt-bindings/i2c/i2c.h>
+
+/ {
+    aliases {
+        alp-i2c0 = &i2c0_emul;
+    };
+
+    soc {
+        i2c0_emul: i2c@100 {
+            compatible = "zephyr,i2c-emul-controller";
+            status = "okay";
+            reg = <0 0x100 4>;
+            #address-cells = <1>;
+            #size-cells = <0>;
+            clock-frequency = <I2C_BITRATE_FAST>;
+        };
+    };
+};

tan/templates/vendored/sensor/E1M-V2N101/prj.conf

@@ -0,0 +1,5 @@
+# SPDX-License-Identifier: Apache-2.0
+#
+# Intentionally empty.  Every CONFIG_* this app needs is derived
+# from board.yaml by `tan build` and layered in via
+# EXTRA_CONF_FILE (see README.md: `tan generate --target zephyr-conf`).

tan/templates/vendored/sensor/E1M-V2N101/src/main.c

@@ -0,0 +1,186 @@
+/*
+ * Copyright 2026 Alp Lab AB
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * i2c-master -- discrete I2C master that reads a known device at
+ * a known address.
+ *
+ * Pattern: open the bus, init the chip driver, loop reading the
+ * register every second, close cleanly.  Contrasts with
+ * https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-scanner which probes every 7-bit address for ACKs
+ * without knowing what's behind them.
+ *
+ * Hardware: the BMP581 barometer sits on BOARD_I2C_SENSORS on every
+ * E1M and E1M-X EVK, strapped to 7-bit address 0x47 (SDO -> VDDIO)
+ * on both -- see alp-sdk's metadata/boards/e1m-evk.yaml and
+ * metadata/boards/e1m-x-evk.yaml (not part of this scaffolded
+ * project) `i2c_devices:`.  On a brand-new
+ * bring-up you may want to build and run the alp-sdk repo's
+ * https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-scanner example first to confirm which
+ * address ACKs -- it isn't part of this scaffolded project.
+ *
+ * (#1269: this example used to target the TMP112 temperature
+ * sensor, but TMP112 lives on BRD_I2C, not on BOARD_I2C_SENSORS --
+ * opening BOARD_I2C_SENSORS and probing for TMP112 NACKs on real
+ * hardware.  BRD_I2C is a separate controller instance regardless
+ * (on the E1M-AEN family it is SoC I2C0, function C -- #1848), so
+ * simply repointing the bus_id would still not reach a TMP112 there; see
+ * https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/v2n/v2n-temp-sensor for the V2N-only BRD_I2C/TMP112
+ * pattern.)
+ *
+ * What success looks like (real hardware):
+ *
+ *   [i2c-master] open BOARD_I2C_SENSORS @ 400 kHz
+ *   [i2c-master] bmp581_init @ 0x47 -> 0 (OK)
+ *   [i2c-master] sample 0: 96234 Pa, 23.625 degC
+ *   [i2c-master] sample 1: 96231 Pa, 23.687 degC
+ *   ...
+ *   [i2c-master] done
+ *
+ * On native_sim (CI lane) the alp-i2c0 alias maps to the emul I2C
+ * driver -- no BMP581 is registered as a target, so bmp581_init
+ * gets NACKed and the example exits with the diagnostic.  Either
+ * way the [i2c-master] done marker latches the harness.
+ */
+
+#include <stdio.h>
+
+#include "alp/peripheral.h"
+#include "alp/chips/bmp581.h"
+
+/* BOARD_I2C_SENSORS is a portable cross-EVK alias from <alp/board.h>:
+ *   E1M EVK  -> EVK_I2C_BUS_SENSORS  -> ALP_E1M_I2C0
+ *   E1M-X EVK -> XEVK_I2C_BUS_SENSORS -> ALP_E1M_X_I2C0
+ * Rebind it in board.yaml `pins:` to port to another board. */
+#include "alp/board.h"
+
+/* BMP581 7-bit I2C address with SDO -> VDDIO -- the on-board strap
+ * on both the E1M EVK (U14) and E1M-X EVK (see alp-sdk's
+ * metadata/boards/e1m-evk.yaml / e1m-x-evk.yaml, not part of
+ * this scaffolded project, `i2c_devices:`).
+ * SDO -> GND gives 0x46 instead; see the BMP581 datasheet
+ * BST-BMP581-DS004 s5.6 or alp-sdk's include/alp/chips/bmp581.h
+ * (not part of this scaffolded project) if your board straps
+ * differently. */
+#define BMP581_ADDR_7BIT BMP581_I2C_ADDR_HIGH /* 0x47 */
+
+/* Number of samples to take before exiting.  Capped so the
+ * native_sim build doesn't stall the twister harness; real
+ * firmware would loop forever. */
+#define SAMPLE_COUNT 5u
+
+/* Wait between samples.  50 Hz ODR (below) means a fresh reading
+ * every 20 ms; waiting a full second gives a comfortable margin
+ * and prints once per watch-tick which is easy on the eyes. */
+#define SAMPLE_PERIOD_MS 1000u
+
+int main(void)
+{
+	/* Bring up the SDK runtime before anything else -- thin today,
+	 * but future backends rely on it (see <alp/peripheral.h>). */
+	(void)alp_init();
+
+	printf("[i2c-master] open BOARD_I2C_SENSORS @ 400 kHz\n");
+
+	/* Open the bus at 400 kHz (I2C Fast-mode).  BMP581 supports up
+     * to 1 MHz (Fast-mode Plus) per its datasheet, so 400 kHz is
+     * comfortably inside spec; the SDK rounds DOWN to the
+     * controller's closest achievable rate.  100 kHz is the safe
+     * baseline for unknown devices. */
+	alp_i2c_t *bus = alp_i2c_open(&(alp_i2c_config_t){
+	    .bus_id     = BOARD_I2C_SENSORS, /* E1M EVK: ALP_E1M_I2C0; E1M-X EVK: ALP_E1M_X_I2C0 */
+	    .bitrate_hz = 400000,
+	});
+	if (bus == NULL) {
+		/* No alp-i2c0 alias on this build -> NULL handle.
+         *
+         * Common causes:
+         *   * Board overlay forgot to set the alias.
+         *   * SoM has no I2C0 routed (rare -- it's part of the
+         *     portable E1M baseline).
+         *   * On native_sim without the emul overlay we ship,
+         *     the alias is unset. */
+		printf("[i2c-master] open failed: alp_last_error=%d\n", (int)alp_last_error());
+		printf("[i2c-master] done\n");
+		return 0;
+	}
+
+	/* Initialise the BMP581 driver.  This reads CHIP_ID and verifies
+     * it matches BMP581_CHIP_ID -- catches the "wrong address" case
+     * (NACK on probe) up-front.  If init fails the example exits
+     * cleanly -- maybe the chip isn't populated, maybe the address
+     * is wrong, maybe the bus is held low by another device.
+     * https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-scanner can confirm which devices ACK. */
+	bmp581_t     sensor;
+	alp_status_t s = bmp581_init(&sensor, bus, BMP581_ADDR_7BIT);
+	if (s != ALP_OK) {
+		/* Most-frequent failure modes:
+         *   * ALP_ERR_IO   -- bus error or NACK.  No BMP581 here,
+         *                     wrong address, or pull-ups missing
+         *                     (the bus floats high without them).
+         *   * ALP_ERR_INVAL -- bad argument (NULL ctx or NULL bus).
+         *
+         * Use the alp-sdk repo's https://github.com/alplabai/alp-sdk/tree/v0.17.0/examples/peripheral-io/i2c-scanner
+         * example to enumerate what IS on this bus before chasing a
+         * BMP581 that may not be populated. */
+		printf("[i2c-master] bmp581_init @ 0x%02x -> %d "
+		       "(populated? right address?)\n",
+		       BMP581_ADDR_7BIT,
+		       (int)s);
+		alp_i2c_close(bus);
+		printf("[i2c-master] done\n");
+		return 0;
+	}
+	printf("[i2c-master] bmp581_init @ 0x%02x -> %d (OK)\n", BMP581_ADDR_7BIT, (int)s);
+
+	/* Configure oversampling + ODR + power mode in one call.  x8
+     * oversampling at 50 Hz normal mode is a reasonable general-
+     * purpose default; reach for higher OSR when you want lower
+     * noise at the cost of more power/latency. */
+	s = bmp581_set_sampling(
+	    &sensor, BMP581_OSR_X8, BMP581_OSR_X8, BMP581_ODR_50_HZ, BMP581_MODE_NORMAL);
+	if (s != ALP_OK) {
+		printf("[i2c-master] bmp581_set_sampling -> %d\n", (int)s);
+		/* Non-fatal: the chip stays at whatever mode init left it in. */
+	}
+
+	/* Sample loop: read SAMPLE_COUNT pressure+temperature pairs, one
+     * per second.  Real-life firmware would publish each reading
+     * over MQTT, push to a ring buffer for trend analysis, or
+     * compare against an alert threshold and pull a GPIO. */
+	for (uint32_t i = 0; i < SAMPLE_COUNT; i++) {
+		bmp581_raw_t         raw  = { 0 };
+		bmp581_compensated_t comp = { 0 };
+		s                         = bmp581_read_raw(&sensor, &raw);
+		if (s == ALP_OK) {
+			s = bmp581_compensate(&raw, &comp);
+		}
+		if (s == ALP_OK) {
+			/* Format integer + fractional parts so we avoid float
+             * printf on M-class targets.  temperature_c1000 is
+             * signed -- the fractional part takes the absolute
+             * value so e.g. -1750 milli-C prints as "-1.750 degC"
+             * (not "-1.-750 degC"). */
+			int whole = comp.temperature_c1000 / 1000;
+			int frac =
+			    (comp.temperature_c1000 < 0 ? -comp.temperature_c1000 : comp.temperature_c1000) %
+			    1000;
+			printf(
+			    "[i2c-master] sample %u: %d Pa, %d.%03d degC\n", i, comp.pressure_pa, whole, frac);
+		} else {
+			/* Read/compensate errors during steady-state are rare --
+             * usually a transient bus glitch (EMI, ground bounce).
+             * Log and continue rather than aborting; the next
+             * sample will likely succeed. */
+			printf("[i2c-master] sample %u: read -> %d\n", i, (int)s);
+		}
+		alp_delay_ms(SAMPLE_PERIOD_MS);
+	}
+
+	/* Clean shutdown -- deinit the chip driver, then close the bus
+     * handle (which releases the slot back to the pool). */
+	bmp581_deinit(&sensor);
+	alp_i2c_close(bus);
+	printf("[i2c-master] done\n");
+	return 0;
+}

tan/templates/vendored/sensor/E1M-V2N101/testcase.yaml

@@ -0,0 +1,52 @@
+# SPDX-License-Identifier: Apache-2.0
+
+sample:
+  name: i2c-master
+  description: |
+    Per-peripheral example for I2C master.  Reads the on-board
+    BMP581 barometer at a known 7-bit address via
+    BOARD_I2C_SENSORS.  Builds on both EVKs: E1M EVK uses E1M_I2C0;
+    E1M-X EVK uses E1M_X_I2C0.  Contrasts with
+    examples/peripheral-io/i2c-scanner (discovery vs known-address read).
+
+tests:
+  alp_sdk.example.i2c_master.e1m_x_evk:
+    extra_args: EXTRA_CONF_FILE=generated/alp.conf
+    platform_allow:
+      - native_sim
+      - native_sim/native/64
+    integration_platforms:
+      - native_sim/native/64
+    extra_configs:
+      - 'CONFIG_COMPILER_OPT="-DALP_BOARD_E1M_X_EVK"'
+    tags:
+      - alp-sdk
+      - example
+      - i2c
+      - sensors
+      - v2n
+    harness: console
+    harness_config:
+        type: one_line
+        regex:
+          - "\\[i2c-master\\] done"
+  alp_sdk.example.i2c_master.e1m_evk:
+    extra_args: EXTRA_CONF_FILE=generated/alp.conf
+    platform_allow:
+      - native_sim
+      - native_sim/native/64
+    integration_platforms:
+      - native_sim/native/64
+    extra_configs:
+      - 'CONFIG_COMPILER_OPT="-DALP_BOARD_E1M_EVK"'
+    tags:
+      - alp-sdk
+      - example
+      - i2c
+      - sensors
+      - aen
+    harness: console
+    harness_config:
+        type: one_line
+        regex:
+          - "\\[i2c-master\\] done"

tan/version.py

@@ -0,0 +1,102 @@
+# SPDX-License-Identifier: Apache-2.0
+# The Python port ships as tan v0.5.0 (maintainer decision, 2026-07-29).
+#
+# It must NOT reuse 0.4.0: that is the version of the SHIPPED Rust release, and
+# alp-sdk-vscode pins SUPPORTED_CLI_VERSION = "0.4.0" and gates its refetch on
+# isCliBehind(), which returns false on equality. A Python binary claiming
+# "0.4.0" therefore never triggers an upgrade, while isNativeTanVersionOutput()
+# happily ACCEPTS it -- an accepted CLI whose commands do not exist yet, which
+# is strictly worse for a user's project than a rejected one.
+#
+# Pre-1.0 SemVer puts a break in the MINOR, and a language rewrite is a break.
+#
+# THIS IS A DEVELOPMENT VERSION, and that is the point (tan-cli#377). It sat at
+# a flat "0.5.0-rc4" for the whole post-rc4 wave, which is the version of an
+# ALREADY-PUBLISHED tag (v0.5.0-rc4 -> fbbea03d): the released binary and every
+# build of this tree answered `tan 0.5.0-rc4` while behaving differently, so no
+# bug report could name which one it was on. `version_check.py --not-released`
+# now refuses that state, so this cannot silently recur next cycle.
+#
+# Why `-rc5.dev0` and not the obvious `0.5.0-dev`: `dev` sorts BELOW `rc` in
+# SemVer AND in PEP 440, so `0.5.0-dev` would sort below the v0.5.0-rc4 this
+# tree is built on top of -- a newer build claiming to predate the release it
+# came after. `0.5.0-rc5.dev0` sorts rc4 < rc5.dev0 < rc5 < 0.5.0 in both
+# ecosystems. It does NOT promise an rc5 will be cut: PEP 440 reads `X.devN` as
+# "before X", and the next release being 0.5.0 (above rc5) keeps that true
+# either way. Build metadata (`0.5.0-rc4+dev`) was rejected for the opposite
+# reason -- SemVer 2.0.0 sec. 10 requires precedence to IGNORE it, so it would
+# have sorted EQUAL to the published tag.
+#
+# At release: drop the pre-release entirely (`0.5.0`), date CHANGELOG.md's
+# `## [0.5.0]` heading, and bring python/pyproject.toml (PEP 440 spelling)
+# with it -- `python/scripts/version_check.py --selftest --self
+# --not-released` checks all three together.
+#
+# 0.5.1: patch fix for the 27 camelCase issue codes v0.5.0 shipped in
+# `contract/issue-codes.json` (`doctor.boardYaml`, `support-
+# bundle.gdbserverBackend`, ...) -- all `"reserved"`/`"consumer": "none"`, so
+# the rename is free. See CHANGELOG.md's `## [0.5.1]` entry.
+#
+# v0.5.1 published 2026-08-05, so this tree moves on the same day: leaving
+# TAN_VERSION at "0.5.1" is the exact state `--not-released` exists to refuse,
+# and it turned `version-identity` red on every open PR the moment that tag
+# appeared. `0.5.2-rc1.dev0` is the spelling that check prescribes -- a `.devN`
+# tail on the NEXT pre-release -- and it promises neither a 0.5.2 nor an rc1:
+# PEP 440 reads `X.devN` as "before X", so a 0.6.0 next release still sorts
+# above it. Its CHANGELOG home is `## [0.5.2] — Unreleased`, because a
+# development version documents the release it heads for, never itself.
+#
+# v0.6.0-rc1 published 2026-08-14 (tan-cli#764): this is the SAME recurrence,
+# one cycle later -- `dev`'s tip carried the published tag's exact version, so
+# `--not-released` turned `version-identity` red on every PR opened against
+# `dev` (#762, #763), for a reason unrelated to either diff. `0.6.0-rc2.dev0`
+# is the same spelling this file has used twice before -- a `.devN` tail on
+# the NEXT pre-release -- and carries the same non-promise: PEP 440 reads
+# `X.devN` as "before X", so neither an actual rc2 nor the eventual 0.6.0 is
+# guaranteed by this string, only that both sort above it. CHANGELOG home is
+# `## [0.6.0] — Unreleased`, per `release_target()`.
+#
+# 0.6.1 RELEASE BUMP. The three prior notes above all record the same recurring
+# defect -- a development tree carrying a PUBLISHED tag's exact version -- and
+# are about moving OFF a released number. This one is the other direction: the
+# `.dev0` tail comes off because this tree is what `v0.6.1` will name. Per
+# `release_target()` a releasable version is its own CHANGELOG section verbatim,
+# and `## [0.6.1]` already exists (the 204 `changelog.d/` fragments were folded
+# into it in this same change), so no heading moves.
+#
+# The tag is NOT pushed by this change, deliberately. `release.yml`'s `build`
+# job is `needs: [verify-version, gates, python-gates]`, and `python-gates`
+# calls `parity.yml` with `python_only: true`, whose `release-sdk-parity` job
+# byte-compares tan's renders against alp-sdk's `releases/latest`. That is
+# `v0.16.0` today, against which `tests/parity/scaffold_byte_parity.py` exits 1
+# (8 of 10 trees, 23 files -- measured 2026-09-17). A tag pushed now would fail
+# `python-gates`, never reach `build`, and publish ZERO assets under an
+# immutable tag -- the exact shape `release.yml:300`'s own comment warns about.
+# tan-cli#1258 tracks the alp-sdk release that clears it; tag after that lands.
+#
+# RENUMBERED 0.6.1 -> 0.7.0 (maintainer decision). Everything above this line
+# is left as written -- it is the accurate record of how this tree came to
+# carry a plain, non-`.dev0` release number -- but the number itself was
+# wrong: this tree's content breaks v0.6.0 consumers, and pre-1.0 SemVer puts
+# a break in the MINOR, not the patch. `## [0.6.1]` becomes `## [0.7.0]` in
+# CHANGELOG.md (same section, same 204 folded fragments, renamed, then
+# extended with hand-written entries -- the floor, the doctor sub-bullet,
+# `debug-config.sdk-identity-appended`, the model-code removal -- and with
+# the preamble + BREAKING leads a follow-up review added), and
+# `python/pyproject.toml` moves with it. Nothing else about the readiness
+# picture above changes: the alp-sdk-floor blocker this file already
+# documented (tan-cli#1258) still gates the tag push, under the new number.
+#
+# v0.7.0-rc1 published 2026-10-10 (tan-cli#770's automated post-release bump):
+# the development line still carried the published tag's exact version, the
+# state `version_check.py --not-released` exists to refuse. TAN_VERSION moves
+# to `0.7.0-rc2.dev0` -- a `.devN` tail on the next pre-release -- which
+# promises neither that release nor that pre-release number, only that both
+# sort above it in SemVer and PEP 440 alike. CHANGELOG home is `## [0.7.0] —
+# Unreleased`, per `release_target()`.
+#
+# 0.7.0 RELEASE BUMP (2026-10-13). alp-sdk `v0.17.0` is the first stable
+# release containing the floor (tan-cli#1258), so the `.dev0` tail comes off:
+# this tree is what `v0.7.0` will name, and `## [0.7.0]` is dated in the same
+# change, with the 20 post-rc1 `changelog.d/` fragments folded into it.
+TAN_VERSION = "0.7.0"