Why does mbed support COTS dev boards only, and not generic microcontrollers themselves? This has stopped me using this platform at work and at home for years. I've got an STM32F207VG-based board I built and was writing firmware from scratch on. I'd much rather use mbed. I decide to come back and give it another go, and start off with the STM32F207*Z*G NUCLEO target; almost the same board. Nucleo put an LED on PB_0. I'm using that pin for an ADC. Now I've got to export the project and use an offline compiler just so I can uncomment a single line in PeripheralPins.c. Jeez, that's a lot of work instead of just sending other devs who might want to use this project to my repo and having them import it into their own mbed compiler, right? Maybe I can import the whole mbed library into my project? Turns out I can, but then the compiler doesn't work right. So I remove the target from the IDE and hope I can just use the library, but then the IDE won't build the project for me because I haven't picked a valid target. So I'm stuck downloading this and working offline, and other devs have to set up a toolchain, which is 75% of why I wanted to use this in the first place. All just to uncomment one line of code that was commented so that somebody could blink an LED real quick without having to look up the pinout and type "PB_0".