social
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Social] Fwd: GNU/social legacy


From: Melvin Carvalho
Subject: Re: [Social] Fwd: GNU/social legacy
Date: Wed, 12 Dec 2012 23:46:02 +0100



On 12 December 2012 23:08, hellekin <address@hidden> wrote:
On 12/12/2012 06:46 AM, Melvin Carvalho wrote:
>
> HTTP URIs
>
*** Too bad they are unidirectional. What's the status of XLINK?

Also, they tend to break. I had to correct a 404 today. That should be
addressed seriously.

I dont have a problem with 404s , they show that the web is constantly changing.  It is hard for a system as big as the web could survive without something like a 404.  I like a degree of fault tolerance.
 

As Evan suggested, and address@hidden does have it too, as that it
depends on the centralized Domain Name System: that is the underlying
protocol in both cases, is centralized, and utterly broken. DNSsec will
certainly solved part of the issues with DNS, but won't change its
centralized nature. Isn't using an HTTP URI or an email-like address an
implementation detail?

That string seems a good one for the UI interface.  Ostatus folks currently NOT considered an email address, but rather, a new URI scheme acct:.  It uses 'spectacularly bad' patterns, as was commented by Mark Nottingham on the HTTP list.  My view is that it's trying to reinvent too much.

I would prefer:

address@hidden at the User Interface level

mailto:address@hidden at the data level

An HTTP profile page listing data about the user using web standards.

I would find this a scalable solution accommodating the best of all worlds, allowing interop, and not having to create a new URI scheme.  I prefer reuse to reinvention.
 

In the current state of affairs, you have to pay a tribute to companies
for the service (contrary to, for example: phonebooks, where companies
actually pay to have you listed, and that's why the books come gratis),
and you cannot easily move from a domain to another.

As an unrelated, but analogical example, consider Google Reader, the RSS
service. It states you're free to move away, and can indeed export your
feeds as an OPML file. But when you do so, you actually lose all the
reputation you built over time using this service, as all links that
others made to your own instance of that service will not follow you.
And I don't see Google, or any other commercial company, maintaining
links for people who stopped using their services.

Yes.  The trust third party model and lock in are often deployed by businesses.  I actually think businesses would do better in the long term by allowing more user freedom.
 

In comparison, GNUnet offers a distributed DNS dubbed GADS[1], that
potentially brings an unlimited name space. Think categories vs. taxonomies.

You may also want to read lynX's thoughts about why using address@hidden is
not necessarily a good idea when it comes to interoperability. [2]


I like what lynX writes here

Opacity in identifiers seems a great feature.  One reason why why URIs hold so much value.  You can have psyc: xmpp: mailto: http: and they are all unambiguous. 


>
> I appreciate your world view.  I would be very happy if there were
> efforts from OStatus based systems to talk to web based systems based
> off HTTP URIs.  If efforts were made in the OStatus ecosystem to talk to
> heterogeneous systems outside of itself, with evidence efforts made to
> of interop, I would happily change my view.  But in the years that I've
> followed, this has not happened, and I dont expect that to change.
>
*** That's part of the issue we're trying to solve with the proposal I
made to this list (which, obviously, will be pursued somewhere else.) By
coordinating projects, I simply mean bring that kind of issues to the
table, make sure they're solved collectively, and that a maximum number
of developers are aware of them, and participate in solving them.

+1000

I dont believe there is 'one social protocol to rule them all'.

Trying to foster interop is something I'm passionate about.  However, it's easy to say, but harder to test.
 

If you didn't yet, you can read the proposal at [3].

I do like it.  Lorea is one of my favourite projects, is caedes still on the team?  I like freindica too.  You'll have to put me down as a heavy skeptic of OStatus.  If it gets approved by any recognized standards body or demonstrates good degrees of interop with systems other than itself, I'll be very happy to change opinion. 
 

==
hk

[1] GNU Alternative Domain System (GADS)
https://gnunet.org/sites/default/files/schanzen_defense_slides.pdf
[2] http://about.psyc.eu/Jabber#JIDs_aren.27t_flexible_enough
[3] http://cepheide.org/consensus/manifesto.html



reply via email to

[Prev in Thread] Current Thread [Next in Thread]