Comments on: Down for everyone, but not me.

By: FETHR roadmap. (dsandler.org)

FETHR roadmap. (dsandler.org) — Fri, 01 May 2009 14:15:35 +0000

[...] twittergw, which acts as a gateway between the two networks. (I first described the gateway in a previous blog post.) Long-term, Twitter should support the FETHR API directly; twittergw was created to provide a [...]

By: Odin/Velmont

Odin/Velmont — Wed, 22 Apr 2009 17:27:17 +0000

Huh. Didn’t OpenMicroBlogging and Laconica exist back then? :-)

http://openmicroblogging.org/
http://laconi.ca/
http://identi.ca/

Why another federated microblogging-protocol? They are trying to make a 0.2, adressing problems with 0.1 — so please say if there is something to be improved upon.

By: dsandler

dsandler — Tue, 25 Nov 2008 15:54:37 +0000

This design shares the spirit of SMTP: It is decentralized and federated by design. In fact, you could probably hack together a passable micropublishing implementation based entirely upon email. But email is easily abused; the default-accept nature of SMTP combined with its lack of mandatory security features is what makes spam possible. FETHR an opt-in subscription model and security features to make this work in a decentralized environment. The design is tuned for the µpublishing/µblogging paradigm, however; I’m not simply trying to design a “better email.” That’s a problem for another day. :)

By: Matt

Matt — Mon, 24 Nov 2008 22:18:52 +0000

Shouldn’t we be concerned that on some level we’re doomed to re-implement SMTP.

By: dsandler

dsandler — Tue, 18 Nov 2008 19:55:25 +0000

@Dave No, it's a totally new framework based on RESTful HTTP requests between servers. (FETHR := “Featherweight Entangled Timelines via HTTP Requests.”) A FETHR server can be run from a vanilla webhost, or at home, or anywhere you can be reached on port 80. Servers interact by gossiping updates via POST. I really need to get that TR out.

By: Dave Peck

Dave Peck — Tue, 18 Nov 2008 19:25:20 +0000

Dan — is this built on top of the Pastry/Scribe work that FeedTree was originally built on top of?

By: dsandler

dsandler — Tue, 18 Nov 2008 19:17:58 +0000

Update: Today's outage identified a new failure mode that I wasn't properly catching in the twittergw. During the outage, Twitter responded 200 OK but with an error embedded in the JSON:

{'error': 'Twitter is down for database maintenance. It will return in about an hour.'}

It's implied in the API docs that twitter.com should have returned something other than 200 in this case; as it turns out, the twittergw code looked no further than the 200 and did not (as I hoped it would) batch my tweets for re-posting once the service returned. Bug fixed.