Is incus port-forward a simpler/reduced proxy device in NAT mode?

As per the title, I’m trying to understand where the new since incus 7.3 incus port-forward fits within Incus network capabilities.

In my understanding, so far we had two main avenues to transfer connections reaching the host to instances on a managed bridge:

(I’ve found Difference between network forward and proxy device (this reply especially) and Forward and map port to container - #2 by stgraber useful to understand the differences and respective use cases.)

Both options always felt a bit heavy/cumbersome to use to me due to their syntax, and that’s why I’m delighted we now get the simpler port-forward.

Am I right in thinking it offers a simpler, TCP-only proxy device (because it is per instance), in NAT mode (because it uses nftables instead of a separate connection/virtual device) but able to listen on the wildcard address, which didn’t use to be the case (at least for VMs)?

Trying it just now highlighted something that had completely escaped me: incus port-forward only does temporary/transient forwarding, à la kubectl port-forward.

So it really is on a different level than the other options, and is only suitable for quick debug/check.

I feel that neither the announcement nor the docs made this clear.

Case closed!

No. It’s a temporary port forward done through the Incus API by the CLI tool itself.

It’s equivalent to the same feature in SSH.