Unhosted protocol: WebDAV!

[Daniel Gröber]

During one of my moments of doubt about the way our current protocol requires
you to set a signature for messages send with SEND i was drifting through my
thoughts. And suddenly it struck me. WEBDAV!

Why would we create another protocol that is basically just here to provide
read/write access to resources stored on a server addressed by a URI when there
are perfectly good protocols that do exactly that already available. Namely HTTP
and WebDAV.

WebDAV is basically just a extension to HTTP to support for resource properties,
directories (what they call 'collections'), COPY'ing of files and some other fs
isch operations. WebDAV is dead simple (the RFC has about 90 pages).

Now combining WebDAV as a storage protocol with a specification on how to extend
the capabilities of the server (like i described in "[proposal for UJ/0.2]
Extensions and capabilities") we could build an awesome framework that is based
on already existing protocols (WebDAV for storage, WebSockets for realtime
stuff, WebFinger for email based user identification and whatever else we can
think of ...).

The only thing that the 'Unhosted protocol' as such would then provide is a
specification on how to put all this together to make for a working system. And
how to discover what a server actually supports.

I really think this is the way to go.

Tell me what you think!

-Daniel

[pir...@gmail.com]

It's not bad idea just don't re-invent the wheel... :-)

[Curtis j Schofield]

On Tue, Jan 4, 2011 at 6:56 PM, Daniel Gröber <dark...@gmail.com> wrote:
>
> During one of my moments of doubt about the way our current protocol requires
> you to set a signature for messages send with SEND i was drifting through my
> thoughts. And suddenly it struck me. WEBDAV!
>

I'm very much for this. I think the work unhosted is doing with the
security and the formalizing
the way to use protocols to create unhosted clients is important work
and where possible
we are best enabled to utilize existing spec.

kudos

--
make haste slowly \
festina lente  \
-
mobile  +1_415_632_6001
cur...@robotarmyma.de
http://robotarmyma.de

[Thad Guidry]

CMIS is another protocol that comes to mind for documents and
querying. It's used in Drupal, Nuxeo, Alfresco, IBM, Microsoft Office
2010, OpenSocial
(http://wiki.opensocial.org/index.php?title=Align_CMIS_and_OpenSocial)
and a slew of others.

I think CMIS would be outside the unhosted protocol ? But curious how
exactly it would or could fit in the mix as well. But it's
interesting that those products above also do WEBDAV as well. Food
for thought and research.

-Thad
http://www.freebase.com/view/en/thad_guidry

[Michiel de Jong]

On Wed, Jan 5, 2011 at 10:36 PM, Thad Guidry <thadg...@gmail.com> wrote:

CMIS is another protocol that comes to mind for documents and
querying.  It's used in Drupal, Nuxeo, Alfresco, IBM, Microsoft Office
2010, OpenSocial
(http://wiki.opensocial.org/index.php?title=Align_CMIS_and_OpenSocial)
and a slew of others.

I think CMIS would be outside the unhosted protocol ?  But curious how
exactly it would or could fit in the mix as well.  But it's
interesting that those products above also do WEBDAV as well.  Food
for thought and research.

Definitely worth studying. The querying could be hard to get working repository-to-repository, because an unhosted storage node never knows what data it hosts. but at least we can provide a way to import CMIS-compatible document repositories into unhosted applications.

Relating this to the modules discussion -

When this sort of things come up (WebDAV, CMIS, and also TeleHash in another thread), I can really appreciate the need for a modular setup of unhosted. That way anybody can just get to work with something they need (say the need CMIS for a specific app), and don't need to get any form of central approval. We would just require all unhosted modules to be WebFinger-discoverable. i can see a possible distinction between gateway modules (to interact with existing hosted software) and persistence modules (ones that provide persistent storage). For modules that provide persistence, all we would have to do is make sure each module's data (if it involves any persistent data) is exportable/migratable via some common interface.

I'll continue to study WebFinger and to think about how we can further minimize the centralized part of the unhosted protocol. Is anybody aware of some application-independent data liberation standard that we could base the data migration mechanism on?

[pir...@gmail.com]

> approval. We would just require all unhosted modules to be
> WebFinger-discoverable. i can see a possible distinction between gateway
> modules (to interact with existing hosted software) and persistence modules
> (ones that provide persistent storage). For modules that provide
> persistence, all we would have to do is make sure each module's data (if it
> involves any persistent data) is exportable/migratable via some common
> interface.

We would need a roadmap, or at least some organization, this start to
become to huge and to big... :-$

> I'll continue to study WebFinger and to think about how we can further
> minimize the centralized part of the unhosted protocol. Is anybody aware of
> some application-independent data liberation standard that we could base the
> data migration mechanism on?

Just to my mind, Google Liberation Front:

http://www.dataliberation.org/
http://code.google.com/p/google-sites-liberation/
http://dataliberation.blogspot.com/

--
"Si quieres viajar alrededor del mundo y ser invitado a hablar en un
monton de sitios diferentes, simplemente escribe un sistema operativo
Unix."
– Linus Tordvals, creador del sistema operativo Linux

[Michiel de Jong]

On Wed, Jan 5, 2011 at 11:14 PM, pir...@gmail.com <pir...@gmail.com> wrote:

We would need a roadmap, or at least some organization, this start to
become to huge and to big... :-$

i think for now a wiki is good enough to let this self-organize. if we see conflicts or overlaps at some point, we can always consider introducing more organization later.

 

Just to my mind, Google Liberation Front:

yeah, i know, they should be implementing CMIS for Google Docs. I will ask them to make sure they don't forget about this ;)

they're a team within google, though, not a standardized protocol or mechanism, right? isn't google's data liberation mainly based on http download of xml? i don't know, never used it. sadly, there is rarely any product as good (yet) as the corresponding google product. (that's where we come in, of course) ;)

[Jonathan A. Marshall]

Unhosted can use the "AXIS" JavaScript WebDAV library, open source
from <a href="http://www.limebits.com/">LimeBits</a>.
http://groups.google.com/group/unhosted/browse_thread/thread/de3878ee55dc145f

--Jonathan

[pir...@gmail.com]

>> Just to my mind, Google Liberation Front:
>>
>
> yeah, i know, they should be implementing CMIS for Google Docs. I will ask
> them to make sure they don't forget about this ;)
> they're a team within google, though, not a standardized protocol or
> mechanism, right? isn't google's data liberation mainly based on http
> download of xml? i don't know, never used it. sadly, there is rarely any
> product as good (yet) as the corresponding google product. (that's where we
> come in, of course) ;)

It's true, it's just a group inside Google. Really i don't see any
format apart from the xml data from Google Code link, but it seems
they promove the use of standard formats, so it would be useful as
knowledge database.

[Daniel Gröber]

From: "Jonathan A. Marshall" <jonathan....@gmail.com>
Subject: [unhosted] Re: Unhosted protocol: WebDAV!
Date: Wed, 5 Jan 2011 07:55:27 -0800 (PST)

> Unhosted can use the "AXIS" JavaScript WebDAV library, open source
> from <a href="http://www.limebits.com/">LimeBits</a>.
> http://groups.google.com/group/unhosted/browse_thread/thread/de3878ee55dc145f

Thanks for the suggestion but I can't find any javascript library under the name 'AXIS'..?

[Jonathan A. Marshall]

Here's the AXIS library: http://www.limebits.com/!lime/root/library/limebits/latest/axis/
If you look around the site and documentation at www.limebits.com or
the source code at http://www.limebits.com/!lime/root/ you can see how
it's used.
The WebDAV-specific components are in WebDAV.js.

On Jan 5, 11:59 pm, Daniel Gröber <darkbo...@gmail.com> wrote:
> From: "Jonathan A. Marshall" <jonathan.a.marsh...@gmail.com>

> Subject: [unhosted] Re: Unhosted protocol: WebDAV!
> Date: Wed, 5 Jan 2011 07:55:27 -0800 (PST)
>
> > Unhosted can use the "AXIS" JavaScript WebDAV library, open source
> > from <a href="http://www.limebits.com/">LimeBits</a>.

> >http://groups.google.com/group/unhosted/browse_thread/thread/de3878ee...

[Daniel Gröber]

From: "Jonathan A. Marshall" <jonathan....@gmail.com>

Subject: Re: Unhosted protocol: WebDAV!
Date: Wed, 5 Jan 2011 21:27:07 -0800 (PST)

> Here's the AXIS library: http://www.limebits.com/!lime/root/library/limebits/latest/axis/
> If you look around the site and documentation at www.limebits.com or
> the source code at http://www.limebits.com/!lime/root/ you can see how
> it's used.
> The WebDAV-specific components are in WebDAV.js.

Thanks. I will have a look at it as soon as I am done with getting unhosted.js
working.