Jan Lübbe

You're right, MX isn't that useful. On the other hand [SRV records](https://en.wikipedia.org/wiki/SRV_record) are very useful in the DynDNS case, as you often have to use port forwards to make services...

@poettering Is there something we could do to make this easier to merge? Is the general approach acceptable?

@zsquareplusc Is there anything we can do to help with getting this merged?

What can we do to get this unblocked?

The boot-emmc type should only touch the mmcblk2boot0/1 partitions, which are exposed as block devices separate from the normal mmcblk2 ("user data area"). In old kernels, there was a corruption...

> There is no case for a `"*.img.caibx"` as the source and `"boot-emmc"` as the destination. So `img_to_raw_handler` is used. The issue with this is that `img_to_raw_handler` does not account...

> Not sure why you would want to pipe the output through a file first? At least for "boot-emmc" it would be easier to just use `write_image_to_dev()` which can distinguish...

The ``MountFlags=slave`` is intended to avoid other processes observing partially installed filesystems. It sounds like a [post-install hook](https://rauc.readthedocs.io/en/latest/using.html#bundle-based-customization-hooks) might be a better approach. That will run at the correct time...

The resulting signature is checked on the host if you pass a keyring during bundle creation. In that case, you also need the matching system.conf on the host. The benefit...

It might be useful to allow setting these options on the command line as well, avoiding the need for a system.conf on the development host.