Comparison between Mozilla & Chrome App Manifests

[Ben Francis]

Hi Mozilla Labs (and a few people I hope might be interested),

Just over a year ago I emailed this list with a comparison between Open Web Apps and Chrome Hosted Apps manifest formats, with a hope of identifying areas where they could converge towards a common standard.

Things have moved on since then and with the launch of the developer preview of Mozilla Apps (https://apps-preview.mozilla.org/) and the work I'm involved in on web apps for Boot to Gecko (https://wiki.mozilla.org/B2G) I thought I would re-visit this.

I've created a comparison on the Mozilla wiki based on my current understanding of the two manfiest formats and I'd be very grateful if you could have a look over it to correct any inaccuracies and hopefully spark off a discussion on a path towards converging the two formats.

https://wiki.mozilla.org/Apps/ManfiestFormatComparison

If this discussion is already going on elsewhere I'd be delighted to hear about it!

Thanks

Ben

Disclaimer: I am not an employee or representative of Mozilla, but I am currently contracting with them to work on B2G and I also have an interest in this topic in relation to the Webian project.

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

[Ben Francis]

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.brad...@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

[Michiel de Jong]

Hi Ben,

On Thu, Dec 22, 2011 at 7:34 PM, Ben Francis <b...@tola.me.uk> wrote:

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.

That's fair enough, and it's very admirable what you're doing. I myself am interested in both hosted and unhosted (as i call them) apps, and after reading http://www.brucelawson.co.uk/2011/installable-web-apps-and-interoperability/ (including the comments) i think we need converters like https://github.com/scottbw/crx2widget for going back and forth between all three spec branches (mozilla, google, w3c). Such a set of converters, as open source reusable tools, would have a lot of the benefits that real convergence would have, and might be a good start towards achieving actual convergence. Just like jQuery is a good way to hide differences in DOM implementations, and allows us in practice to "pretend" that these differences don't exist.

To start with, is there a converter to go from mozilla format to google format and back? Should we make one?