W3C home > Mailing lists > Public > public-rww@w3.org > April 2014

Re: writes from Cimba to data.fm

From: Andrei Sambra <andrei.sambra@gmail.com>
Date: Tue, 8 Apr 2014 15:47:27 +0900
Message-ID: <CAFG79eiQmzSfM5kUKMyiykzo+R5P2wKU634xLNGwaXnBSUMS-g@mail.gmail.com>
To: cr <_@whats-your.name>
Cc: "public-rww@w3.org" <public-rww@w3.org>

On Tue, Apr 8, 2014 at 3:29 PM, cr <_@whats-your.name> wrote:

> https://www.youtube.com/watch?v=z0_XaJ97rF0 is certainly evidence that it
> "works for deiu"..
> and probably any other user of rww.io.. of course, i want to know exactly
> what's required for all of this RWW thing to work. so had to write a
> webserver from scratch, and it's been a constant tweaking of things here
> and there since - ok, text/turtle posts weren't accepted.. ok OPTIONS
> wasn't saying this or that. ok a CORS header for * didnt' exist - but now
> we're in a credentialed scenario so it has to be the exact domain - or
> something - cue reading 12 more spec docs, gold source, mitmproxy of
> https:// sessions to data.fm with custom-imported snakeoil certificates..
> furthest i can get using data.fm 'Creating...' on a 'microblog space'
> just hangs, with either a 200 or 400:

I'm not sure data.fm is using the latest LDP spec implementation that I
have added to ldphp. You have to remember that neither Cimba nor rww.io are
production (commercial) software. Unfortunately, they are just proof of
work for a decentralized Web that is based on standards. You should only
use them if you want to be on the bleeding edge of the work that is being
done in this domain.

> creating mb w/ #storage set to https://user.data.fm/space
> POSTs a 0byte text/turtle -> 200 OK, blank "RDF file" appears in data.fm
> #storage set to https://user.data.fm/ throws the aforementioned 400:
> <h1>400 Bad Request</h1>
> <p>Cannot read from a directory '/mnt/md1/srv/src/data.fm/data/cr.data.fm/
> '</p>
> which looks like it's trying to read a file using a directory path

Yes, both data.fm and rww.io use the filesystem as storage medium, so you
end up with a 1-1 mapping to the filesystem.

> probably should just try rww.io, but it's no fun just duplicating the
> exact environment when interoperable implementations are the goal. but had
> to use chrome - seems like Angular.JS is broken in Firefox-28. getting
> uncompiled {{}} handlebar.js-looking templates mixed into a semiworking UI
> along with all sorts of exotic-sounding errors with strange-looking
> operators like //# or //@ http://mw.logbook.am/img/cimba.hmm.png

Where do those weird operators come from?

> unclear on exactly what's different between rww.io and data.fm - clearly
> they are different, eg:
> p ~ curl -I https://notfound.data.fm/space
> HTTP/1.1 200 OK
> p ~ curl -I https://notfound.rww.io/space
> HTTP/1.1 404 Not Found

Yes, that's because data.fm uses a "wiki" convention, in which a Web
resource always exists but it may be empty. Whether that resource exists or
not, it should not matter for the app, since you are interested in the
triples inside the document instead of the document itself.

> 'datawiki' scripts from 2008 also did 200 on every nonexistent but
> creatable resource
Soon we plan on releasing a detailed document on how to build such a
server, how the interaction model (LDP) works in this case, and a lot more
interesting stuff. Stay tuned! :)

-- Andrei
Received on Tuesday, 8 April 2014 06:48:17 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:10:45 UTC