Trying to gauge the level of interest in a library I have been developing, to see if it would be worth fleshing out and open sourcing/offering as a service.
One of my recent esp32 projects required me to build a cloud building and deployment system for many distributed ESP32-based systems.
For this use, I developed a cloud firmware building and hosting tool, and an esp-idf library to communicate with the cloud service so firmware could be upgraded automatically.
Along with this, the esp could also automatically send diagnostic information whenever it chose to, such as redirecting its ESP_LOGX calls to the server, or sending up debut information when something went wrong. Whenever an ESP core panic occurred, the core dump would be sent up on the next boot and debugged on the server end, and automatically analyse the reason for the core dump, a stack trace of the issue location, and the state of the system when the crash occured.
Full list of features it has/would be quick to implement include. - automatic firmware building and updating to esp-based devices in the field. - A/B firmware testing - each ESP can request a certain ‘branch’ of the firmware, depending on its role. - Automatic firmware testing with a dedicated test ESP before that firmware is deployed to the field. - Full core dump analysis after crashes - debugged against the ELF file produced with the firmware build that is running. - Simple diagnostic information sending through the library - Sentry integration so any errors within the firmware can be reported to a Sentry Error Tracking project
Here is an example of the sort of events that can are tracked in Sentry:
I am wondering what the interest would be to tidy up the code and make it more generic and ship an open source version - as well as possibly offering a paid service where hosting and analysis would be provided for a small per unit charge.
I have seen projects like: https://iotappstory.com which seem to be good in their own right, although seems to be tuned more towards hobbyists with multiple small projects, as opposed to a service more suitable for production-level applications, where it is currently deployed. My project would also be designed to be mostly diagnostic and error collection, as opposed to controlling GPIO pins from the cloud or anything.
I can port the library to work with vanilla installations of esp-idf and Arduino-Esp32. And most of the functionality could be ported to be esp8266 based if people wanted it.
Looking forward to hearing any thoughts.