Re: Next steps for rfc6874bis

On Fri, 13 Aug 2021 at 03:24, Kerry Lynn
<kerlyn=40ieee.org@dmarc.ietf.org> wrote:
>
> If I could add my $0.02 (even though I'm not current on all the
> discussions and details):

As I'm not either, however another 2c.

Non-technical end users shouldn't be exposed or have to even enter
IPv6 addresses, including link-locals. While convenient to be able to
enter them into a URL field in a browser, it shouldn't be a common
thing to do.

For user friendly end user access to device web GUIs over link-local
addresses, the devices announcing themselves via MDNS would be better.
I have a laser printer that already uses MDNS to announce its IPv6
address.

Web browsers could provide a user-friendly selection list showing
devices discovered via MDNS. An extension for Firefox that looks to do
this already exists -
https://addons.mozilla.org/en-US/firefox/addon/mdns-discover/

For technical users who need to enter link-local addresses and
interface/zone IDs, having a technical dialog box, with separate IPv6
address and interface/zone ID fields, hidden away under more technical
menu options (like Firefox's "More Tools" menu) should be adequate to
resolve this thread's issue by avoiding it.

Regards,
Mark.

> - I believe http://[fe80::%eth0] _used to work_ in Mozilla browsers;
>   we're talking maybe early 2000s. The earliest version I can find online
>   is c. 2015. Pretty sure I might have old VMs that can demonstrate
>   the functionality. (Let's assume there's an existence proof for parsing
>   a naked '%' between square brackets.)
> - If this has to be approved by the browser community as a condition
>   of it becoming widely available, do we know if they will support link
>   local addresses in principle? (If not, then the utility of this for e.g.
>   consumer router setup is moot.)
> - If the browser community can support this in principal, then we can
>   discuss EBNF. I strongly favor the syntax above, but any single char
>   is better than "%25". Certain OSes (not naming any names) use
>   numeric ZoneIDs and I think the cognitive dissonance of "%25<nn>"
>   is problematic.
> - If the browser community cannot support this, then we could perhaps
>   develop a patch for cURL to make it available at least for the engineering
>   community (I haven't checked cURL to see if it's already supported).
>
> Cheers, Kerry
>
> --------------------------------------------------------------------
> IETF IPv6 working group mailing list
> ipv6@ietf.org
> Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
> --------------------------------------------------------------------