Re: Comparison between Mozilla & Chrome App Manifests

Hi Scott (cc'd both lists because I'm not sure where this discussion should
take place),

You said:
"I think Mozilla had in mind self-hosted-apps (similar to Facebook
apps) rather than a portable/mobile/desktop type of model, which is perhaps
why they went off in a different direction - although they ended up with
very similar manifest content to W3C."

The important differentiation in my mind is "hosted" vs. "packaged" apps.
W3C Widgets are for "packaged" apps where the resources are packaged in a
zip file and downloaded, Mozilla's Open Web Apps and Chrome's hosted apps
are "hosted" web apps where each of the resources are hosted on a web
server (though can be cached locally using HTML5 APIs) and can be updated
without downloading a whole new package.

Incidentally Open Web Apps are very different to Facebook apps, in fact
Facebook/Apple App Store/Android Marketplace style apps all seem part of
the "problem" they are trying to solve.

IMHO packaged apps are not "web" apps because for something to be part of
the web it must have a URL, and the resources inside a package exist
locally but don't correspond to a unique URL. I explain this in more detail
in a blog post entitled "Keeping Web Apps Open"
http://webian.org/blog/2011/10/10/keeping-web-apps-open/

In comparing the manifest formats for Mozilla Open Web Apps and Chrome
hosted apps I hope to converge on a common standard for *hosted* web apps,
I have less interest in packaged apps.

Ben

On Thu, Dec 22, 2011 at 10:59 AM, Scott Wilson <
scott.bradley.wilson@gmail.com> wrote:

>
> On 22 Dec 2011, at 00:34, Michiel de Jong wrote:
>
> Hi!
>
> CC Ben, it's posts like yours that constitute the hope of the open web. A
> truly great initiative!
>
> Everybody else who reads this, let's put our weight behind Ben, and help
> him with this. Let's team together with mozilla, google and w3c, and be one
> web with one manifest format. Please read this thread as a starting point:
>
>
> http://groups.google.com/group/mozilla-labs/browse_thread/thread/29d186bd2f3580e4
>
> If the json scares you, just imagine the xml syntax with your mind. I'm
> not joking, it really is what I do the other way around when i read xml,
> and it really helps to bridge gaps between the author of a spec and you as
> a reader. Even if this group will use xml, then we can still make sure
> there is a one-to-one mapping with the json manifests used by mozilla and
> google. (see xrd and jrd, or linked data and json-ld, for examples of such
> mappings). Marcos already hinted at this possibility in one of his blog
> posts.
>
>
> It makes sense for an author to create a manifest in XML (W3C Widgets:
> C&P) using any editor, and for a user agent to generate JSON for it if it
> has an API. E.g. Apache Wookie consumes W3C Widgets, but exposes various
> REST APIs that provide the metadata in XML, Atom, and JSON formats.
>
> See also:
>
>
> http://www.brucelawson.co.uk/2011/installable-web-apps-and-interoperability/
>
> also:
>
>
> http://scottbw.wordpress.com/2011/02/17/converting-chrome-installed-web-apps-into-w3c-widgets/
>
> Let's do it! Let's join forces and be 'the one web'.
>
>
> We keep trying :-)
>
> I think Mozilla had in mind self-hosted-apps (similar to Facebook
> apps) rather than a portable/mobile/desktop type of model, which is perhaps
> why they went off in a different direction - although they ended up with
> very similar manifest content to W3C. In which case its less important to
> make the manifest easy to author in a text editor (for example), as the app
> is principally server-side rather than client-side anyway.
>
> Cheers,
> Michiel
>
>
>


-- 
Ben Francis
http://tola.me.uk