@domain.tld identifiers?

[elf Pavlik]

Hello,

I would see many interesting use cases for identifiers like @domain.tld (without local username part). One of them people could use them for personal ones like @franky.me (rather than fra...@franky.me) second for identifying groups like @it.example.com

Does anyone see significant drawbacks of having such identifiers available?

Cheers!
=)
~ elf Pavlik ~

[Michiel de Jong]

i like the idea, the drawback i see is that because mentioning a nick has been done by putting an @ symbol before the nick, @elfPavlik now means "this is directed "AT" elfPavlik", and on twitter this has even evolved to always putting the @ symbol before a nick. So if you tell people you're @franky.me, a lot of people will think that that's your twitter or identi.ca handle. even though a domain will always have a dot in it, and twitter/identi.ca handles never do.

also, i think '@' in the user@host syntax, which i think (but i don't know this) is even older than email (used in ssh for instance) refers to 'this user, at this host'. there is a syntax for email addresses which is never used, but is part of the spec afaik, which is {you,me}@domain.tld - although i don't think it goes much further than that - you could imagine things like wildcards, *@domain.tld to identify a group maybe?

just brainstorming :)
Michiel

[Markus Sabadello]

I think we've talked about these identifiers quite a bit during the Federated Social Web SWAT0 effort..

If you have to type e.g. @dan...@danube.status.net in order to mention someone in a post, then that is obviously not too user-friendly :)

If you want to allow something like @franky.me, then everyone needs their own domain, or sub-domain, hmm why not, I guess it could work, but it would need an updated Webfinger spec, no?

Markus

[Blaine Cook]

I'd encourage folks to *not* adopt that kind of syntax, since the
point of webfinger was to take well-established patterns and update
them to work nicely with the web.

As evidence, I can only offer that in ~30 years, email has never
adopted that form of addressing, even though nothing ever stopped
anyone from trying.

b.

[Paul E. Jones]

Elf,

I would not recommend this. Of course, one can always query any URI, so querying http://franky.me is a valid query. If you wish to get a person or group, though, I think there is far less confusion with user@domain notation. Groups are identified similarly.

I think it's important that there is a clear delineation between user addresses and hostnames, presented in a form humans can quickly recognize.

Paul

[elf Pavlik]

Thank you for all for sharing your opinions on this. I agree it could bring some confusion when people think webfinger == email address. Anyhow hearing 2 strong opinions against from Blaine and Paul, I feel like dropping this idea =)

~ elf pavlik ~

[Bob Wyman]

I agree with Blaine and Paul that the proposed syntax should be discouraged. The proposed benefits do not rise to a level that overcomes the confusion that would probably arise.

bob wyman

[Daniel E. Renfer]

On 02/23/2012 10:34 AM, elf Pavlik wrote:
> Thank you for all for sharing your opinions on this. I agree it could bring some confusion when people think webfinger == email address. Anyhow hearing 2 strong opinions against from Blaine and Paul, I feel like dropping this idea =)
>
> ~ elf pavlik ~

Dealing with single user domains is always difficult, but I think it's
best to always consider that a user has a username and reserve the bare
domain id for addressing the server itself.

If you take into account XMPP, where the bare domain is a valid JID and
is used, you don't want to get that mixed up with the default user.

[Chuck Houpt]

The traditional Finger protocol and command treated @domain.tld as a request to list all users logged onto the remote computer. A fake example:

% finger @vax1.example.com
[vax1.example.com]
Login Name Tty Idle Login Time Office Office Phone
john John Doe pts/2 4 Feb 21 09:40 (76.139.98.235:S.0)
jane Jane Doe pts/7 Feb 23 10:57 (57.129.122.14)
...

Note that modern finger servers typically no longer allow this type of query for security reasons.

For wfinger (the finger command with WebFinger support: http://extechops.net/2010/08/28/wfinger/), I thought the closest matching modern protocol would be humans.txt (http://humanstxt.org/). So wfinger displays the sites humans.txt file when given @domain.tld. For example:

% wfinger @google.com
[google.com - humans]

Google is built by a large team of engineers, designers, researchers, robots, and others ...

Other examples via the wfinger gateway:

http://wfinger.habilis.net/@google.com
http://wfinger.habilis.net/@movethewebforward.org
http://wfinger.habilis.net/@technologyreview.com

Anyway, an example use of @domain.tld "in the wild."

Cheers - Chuck
http://chuck.habilis.net/

[elf Pavlik]

Interesting!

I really would like to have something like @domain.tld available to refer to projects. For example @unhosted.org or @ostatus.org, I wonder if LRDD could express something similar as DOAP does? Short description, description, homepage, participants list to get started.Or maybe just link to DOAP file...

Also specifying group/project salomon endpoint could come useful for some OStatus related features...
Hearing that xmpp has bare domain jid available also may gave some foundation for compatibility in this direction?

Thanks everyone for sharing your creative ideas!

=)
~ elf Pavlik ~

Excerpts from Chuck Houpt's message of 2012-02-24 17:21:09 +0000:

[Melvin Carvalho]

According to the IETF the local identifier before the @ is 'dot-atom-text'

http://www.ietf.org/rfc/rfc2822.txt

From the doc it seems that must be at least 1 character?