diff: https://ci.embassy.dev/jobs/1b945c6febda/artifacts/diff.html

diff: https://ci.embassy.dev/jobs/7b09db6e4a01/artifacts/diff.html

Putting a single bool in the generated data is too "ad hoc". I think the data should contain the full info of which peripheral+register+field to set to enable the remap.

Reasons:

• The bool is specifying the value, but it's not specifying which field. This makes users (like the embassy-stm32 PR you sent) to search it by name themselves. It's nicer if the data is "as cooked as possible" so using it is easier.
• There's more chips with "remaps" like this, for example F1 for GPIO. These are on different peripherals (AFIO) and registers (MAPR, MAPR2). The data model should ideally be the same for all chips, so remaps can be handled the same way.

For example it could be this:

{
  "signal": "ADC2",
  "channel": "DMA2_CH1",
  "remap": {
    "peripheral": "SYSCFG",
    "register": "CFGR1",
    "field": "TIM17_DMA_RMP",
    "value": 0
  }
}

Thank you very much for your feedback! I will incorporate it. I wasn't aware there were even more remapping bits.