chrysn

I'm a bit conflicted about this requiring keyboard interaction. On one hand, this is excluding the feature from keyboardless use cases -- on the other hand, the predominant keyboardless platforms...

Just a note to self: AI_V4MAPPED seems not to be available on Android (Termux' build) and is used on both udp6 and simple6. The only thing that's not using V4MAPPED...

aiocoap (as asyncio in general) tries to fully use a single core. an asyncio *should* be runnable in multiple instances in parallel to use more than that, but to be...

Message IDs should only be limiting if all connections stem from the same remote address (alas, aiocoap currently uses global message IDs, see #80; shouldn't be too bad for a...

This value would need to be agreed between software to be upgradable. riotboot does not concern itself with this, but SUIT does. Should some form of this value (say, `-aux32768`)...

It adds a parametrization to boards that was previously dependent only on RIOTBOOT_LEN / NUM_SLOTS [edit: (or more precisely, `(flash - RIOTBOOT_LEN) / NUM_SLOTS)`]. Before this PR, one could use...

That sounds good to me -- a good default might be "two erasure units". Do you plan to add the AUX in as this PR matures? (I'd welcome it --...

> I'm not sure what you mean - specify it for a select few boards? Yes, or even all boards that have no EEPROM or other dedicated storage. As this...

But this will only cut down sizes when one is already using a bootloader, and in that case the usable space is halved already (and often needlessly -- once a...

OK, then I'm seriously misunderstanding this. For storing data within a single firmware version, FLASH_WRITABLE_INIT works well already -- only in connection with bootloaders do I see the need a...