Re: Finding user profiles on a Social Net

On 6 June 2013 16:41, Simon Tennant <simon@buddycloud.com> wrote:

> I generally dislike /.well-known because it makes lots of assumptions
> about the web-root being available.
>
> Three problems with this:
> 1.  Others might run hosted personal pages like those hosted on about.me.
> For example my sister runs a hosted store on her domain. Short of getting
> the eCommerce provider to change their code, she would never be able to
> implement anything social.
>

Well perhaps if she doesnt have access to .well-known, she probably could
not deploy that pattern, but neither could she deploy webfinger, or any
other number of patterns.  However, she could deploy an indieweb style
pattern which utilizes follow your nose, in the space she has access to.


>
> 2.Often times an organization will have their web-root maintained by
> another company. Page updates could easily overwrite a nice /.well-known
> hierachy.
>

As above, however, orgs that cannot access their web hosts probably are not
going to be able to deploy many social web patterns.


>
> 3. I don't know the answer to this, but how long should /.well-known be
> considered authoritative? What kind of refresh interval?
>
> When you start thinking about it, this is all a hack to accomplish what
> DNS already does. DNS-SD has already solved this, and has caching, and with
> zone signing, authority.
>

I would suspect that DNS is as hard to configure, if not harder, than your
web space.  However, it's an interesting idea.  I did ask stuart cheshire
about this, and he said it was an interesting idea "could be made to work",
so maybe.  I'll try and follow up on that.

Mark nottingham had another idea along these lines:
https://datatracker.ietf.org/doc/draft-nottingham-json-home/?include_text=1

You may be correct, but the .well-known pattern does seem to be gaining in
popularity (although I agree and share your concerns!)


>
> S.
>
>
>
>
> On 6 June 2013 16:22, Melvin Carvalho <melvincarvalho@gmail.com> wrote:
>
>> I was thinking about the issue of finding user profiles on a social net,
>> and it's not always easy to know where a user's data will be located.
>> There seems to be no well known place to get user information from a
>> profile.  Which means it's harder for HTTP based social web users to talk
>> to each other.
>>
>> One increasingly popular method is to use the /.well-known/ directory.
>> The disadvantage of this approach is that is it pretty rigid and people say
>> it amounts out of band hard coding.  However one advantage is that it can
>> save a round trip, compared with follow your nose, and it can client
>> implementations more straight forward.
>>
>> Taking the well known directory a logical pattern might be to register:
>> *
>> *
>> */.well-known/user/bob*
>>
>> For the FSW?
>>
>>
>> *Would it allow redirects* -- I would say yes.
>>
>> *What would it return* -- I would suggest linked data.  Ideally a
>> browser would see html and an ajax request would see JSON, but you could
>> start with just one of the two, say JSON only.
>>
>>
>> Good idea / bad idea / too hard to implement ... thoughts?
>>
>
>
>
> --
> Simon Tennant | buddycloud.com | +49 17 8545 0880 | office hours:
> goo.gl/tQgxP
>