[OT] Re: User @ domain.tld as ID (Once again)

[Ernst Johannes]

On Nov 3, 2005, at 8:01, Martin Atkins wrote:

> So my proposal is this: YADIS will go ahead and use URLs, because  
> it has to as a fundamental requirement. Another spec can try to  
> standardize mapping an email address onto a URL, which will then  
> automatically enable YADIS to use email addresses. There's no  
> reason why these two things have to be bound together into one spec.

I support this idea. Anybody want to create a "future potential work"  
section on the yadis.org wiki, and start putting some of this  
discussion there?

I agree that using identifiers that look like xxx at yyy.tld would be  
great for people, but for my part, there are too many issues that are  
unresolved, and maybe unresolvable, to make this possible in  
practice. This thread illustrates this very well.

There are few more problems:
  - you can type xxx at yyy.tld into a browser, but it does not do what  
you'd expect (in fact, browsers are beginning to be really suspicious  
about @ signs in http URLs because it's a common phishing technique).
  - you can't use Google and friends to easily "look up" the  
identifier of a person (the identifier is not REST-ful)
  - you can't bookmark, tag etc. the identifier (at least not very well)
  - it's hard to build a protocol around because unlike for http[s]  
URLs, it does not identify an end point that can do something in  
response to a request (there are no CGI scripts behind e-mail addresses)

So let's declare this a separate area of potential future work, as  
Martin suggests... regardless how it is resolved (nor not resolved),  
it doesn't impact anything else in the architecture.



Johannes Ernst
-------------- next part --------------
A non-text attachment was scrubbed...
Name: lid.gif
Type: image/gif
Size: 973 bytes
Desc: not available
Url : http://lists.danga.com/pipermail/yadis/attachments/20051103/ff482f42/lid.gif
-------------- next part --------------
  http://netmesh.info/jernst