One example: I might have a marginal power supply design that has limited or no ability to reach 400mA. In my application I have tested all worst-case scenarios and find that I never need more than 200mA (this is just a hypothetical for now)
I am concerned about releasing such a design due to the fact new firmware may be pushed to the imp one day and do untold things if it is starved for power.
do I have the correct line of reasoning? Do I have no choice but to build up and beef up the power supply?
What if I have another reason for not desiring an upgrade or update? (Sometimes in the wide wide world, upgrades break things that are working just fine though that has not happened to me with an imp)
The imp specifies up to 400mA power drain, so you better build your power supply capable of delivering this, or else expect problems. If you use any product under out-of-spec conditions, you can’t expect reliable operation.
The imp during firmware update actually only takes about 100mA average; we spec 400mA for future expansion (because it’s a slot, different things could go in there) and because often people under-design…
There are facilities coming in the API which allow you to be informed of when a firmware update is pending, allowing you to choose when you accept the update. Commercial customers get to run one API revision behind for a limited period (to allow them to test further) but for many reasons including security, upgrades are generally a good thing.
We do our best to ensure that API changes are backwards compatible - eg the new SPI API was backwards compatible with old code.