OTA question - should mg_ota_end() be split into two functions?

[rossack]

mg_ota_end() returns false if the CRC fails. If the CRC is ok the function never returns, which means it's difficult to let the client know that the firmware upload is all good. Instead of proceeding with the bank swap and reset, wouldn't it be better for mg_ota_end() to just do the firmware validation so the server can send the appropriate reply? Then have another function void mg_ota_commit(void *) to do the final reset put on a timer to be called later. I have done this for my PicoW build and it works well, not sure about this logic for all the other architectures.

Go easy on me, I'm still learning. Thankyou for this wonderful body of work.

[scaprile]

There are as many possible OTA architectures as people designing them.
This is the one we came up with after simplifying a former one with all the bells and whistles you'd probably come up with after you sit down and think about all that can possibly happen if.
Once you start thinking "what happens if", things are no longer simple.
"Better" implies a judgment, a comparison.
Our aim is to have something simple, you can always define MG_OTA_CUSTOM and do whatever you want. BTW, our Wizard calls mg_ota_end on a timer, to leave some time.