[tyeth]

Closes #920

• Vendored nanopb runtime (src/nanopb/*, src/pb.h, src/nanopb/nanopb.pb.h) updated from a pristine 0.4.6 to the nanopb-0.4.9.2 release (includes the GHSA-p24j-vqcp-x988 and GHSA-9w99-4pfq-6396 decoder fixes).
• src/wippersnapper/** regenerated with the 0.4.9.2 generator from current https://github.com/adafruit/Wippersnapper_Protobuf master protos. Compared with the previous 0.4.5-dev output: tags, sizes, init macros and field lists are identical; only additions are *_ENUMTYPE convenience macros. Cosmetic: no timestamp line, shorter include guards, proto comments emitted, .pb.c includes its header by bare name.
• CI workflow now downloads nanopb 0.4.9.2 instead of 0.4.8. Since the vendored files match, that step is now a no-op and could be removed in a follow-up.

Companion: adafruit/Wippersnapper_Protobuf#218 bumps the generator submodule so future CI-generated wrappers match these files.

Local builds (PlatformIO): adafruit_feather_esp32s3 and raspberypi_picow succeed with no nanopb-related warnings; adafruit_pyportal_m4 result to follow in a comment if it differs.

[tyeth]

LGTM, no real changes except we now get enum docstrings for free 🎉

[brentru]

@tyeth Was this tested against IO Prod?

[tyeth]

Linked evidence here:

Details

All three jobs completed with successful esptool read-back verification, though my regex checks missed the version and IO-connection details. I'll now directly inspect the serial logs, boot_out dumps, and B0 verify transcript to confirm those.

Version verification checks out—beta.130 on baseline and beta.132 on the #993 build. Now I'll dig into the end of each serial log past "Registering hardware" to confirm IO registration and MQTT connection succeeded, along with the flash-verify transcript tails.

Both builds successfully connected to production Adafruit IO as tyeth_demo, registered, decoded the pin-configuration signal, and maintained a live MQTT session throughout. I'll now write up this evidence as a report, save it to memory, and summarize the findings.

Yes, the v1 firmware is now tested on real hardware against production IO, and it passes.

Result on the QT Py ESP32-S3 N4R2 (rpi-hil006), account tyeth_demo, io.adafruit.com:8883

job	what	esptool read-back	version reported by the running board	IO
A 96334361	flash beta.130 baseline	full image, digest matched	1.0.0-beta.130	registered, configured, MQTT pings OK through the hold
B0 eda2375e	read-back before upgrade	app region at 0x10000 matches the beta.130 app image	n/a	n/a
B c162888b	flash the #993 build	full image, digest matched	1.0.0-beta.132	registered, configured, MQTT pings OK through the hold

The board provably held beta.130 right before the erase and provably holds the #993 image after it, and the running firmware's self-reported version moved from beta.130 to beta.132. On the #993 build the serial log shows WiFi join, TLS MQTT connect, the nanopb-encoded registration request published, the registration response decoded, the broker's pin-configuration signal message decoded through the oneof callback path, "Hardware configured successfully", and pings every 5 s until the window ended. Behaviour matches the baseline line for line.

hil-report-pr993.md