Adrian-Stefan Mares

I'm more in favor of option two (don't decreate data rate via ADR). I think that the margin approach basically is self-defeating: when the margin is big enough to avoid...

> @adriansmares are we still doing this? Yes. I've familiarized myself with the algorithm and I think we can do this safely, albeit not by default: it would be something...

I can confirm this bug - I'll take a look. @johanstokking what should we do in these cases where the frame is too big - drop the downlink or stall...

I've ran the tests again and this has been fixed in the meantime. Test procedure on AS923 device: - Erase `mac-settings.use-adr` and `mac-settings.adr-margin`. - Set `mac-settings.adr.mode.static.data-rate-index` to 2 and `mac-settings.adr.mode.static.nb-trans`...

> * `as.down.data.queue.invalid`/`as.down.data.queue.lost` (maybe, @adriansmares?) These events actually are no longer possible (:tada:) and we should deprecate them. Thanks for reminding me.

Since you're chaining PRs, you can make this one target `feature/remove-app-id-from-dr` in order to make things easier to review.

Hey @andreusoler8, Could you please provide the logs for the actual migration operations ? (`is-db migrate`, `ns-db migrate` etc.) We cannot really tell what caused your inventory to be lost...

Implementation wise this is not complex. On the NS/AS/IS side we just need to add it to the allowed field mask, and on the IS side we need to add...

Yes, it is still relevant. You can easily test this by adding an ABP device, sending some downlinks, then resetting the MAC state. The frame counter for newer downlinks won't...

Field masks are centered around the `protobuf` format used to actually pass these messages around inside The Things Stack. Fields such as the `data_rate` have `oneof` fields such as `modulation`...