Re: [autonomo.us] Web app stores

[l.m. orchard]

Hi there, I work as a webdev [0] at Mozilla. I'm not the lead of any
of Apps or Addons, but I know a bit about both...

[0]: http://blog.mozilla.com/webdev/

On Wed, Apr 27, 2011 at 4:08 PM, Christopher Allan Webber
<address@hidden> wrote:

> Both Google and Mozilla are talking about / moving forward with "web app
> stores":

Mozilla isn't launching an Official App Store so far, though the Apps
team has been working on the tech enable anyone to launch such a
store. I'm not the guy who's make this decision, but it's not
impossible that we could launch an Official store - if it looks like
the Mozilla brand would help advance the cause.

The main goal, though, is to get away from monolithic gatekeepers, and
enable The Web to be The Store.

>  'of course, but it seems "app store" is an incantation making people
>  forget "free software", hints of that in interview above...'

I'm not really sure what this means... I think the Mozilla plans for
web apps would accommodate both paid and free as in {beer,speech}
apps.

> On the other hand, maybe this could be not all terrible, even good for
> free software.  Honestly, in comparison to normal "app stores", free
> software already has great packaging solutions between apt, yum, etc,
> and I think that app stores really aren't much more than DRM'ed package
> managers that don't handle library dependencies in an efficient way. :)

So, the Mozilla paradigm of "installing" an "Open Web Application" is
basically a super-bookmark. You don't actually download and install
any packaged software in the traditional way.

"Installing" an app consists of your browser storing an URL paired
with a title, icon image, description, etc. [1] An app vendor can also
send along some book-keeping info at install time [2] that will be
retained by the browser. That is, say, a proof-of-purchase token, auth
credentials, etc.

There's no DRM as such, though app vendors could employ
purchaser-specific installation details (eg. a reg-code) to control
access to services while you're online. If you don't want to operate
that way, you don't have to.

[1]: https://developer.mozilla.org/en/OpenWebApps/The_Manifest
[2]: 
https://developer.mozilla.org/en/OpenWebApps/The_JavaScript_API#Installation_API_%28navigator.apps.*%29

> However, on the web front we really don't have anything for client-side
> web applications.  ... Would our existing packaging
> solutions really be a good solution for that?  Maybe not, so maybe if
> this is done correctly this could be a good way to have client-side web
> applications packaging for free software.

So, "installation" is just storing a pointer to a web page in the
browser. Then, Mozilla-style web apps rely on "HTML5" features like
cache manifests [3] to manage offline access and online updates to
assets that make up an app (ie. HTML, JS, CSS, images, audio, etc).
So, no software packages, just conventions of HTML5 and HTTP.

[3]: http://www.w3.org/TR/html5/offline.html

> However I'm doubtful, if for no other reason that *at minimum* there'd
> have to be a good way to tell what the licensing is of applications,

Hmm, yeah, it doesn't look like "license" is a field in the current
web app manifest [1]. That seems like a bug, though hopefully the app
itself would identify what license applies.

> ...and
> right now it's an absolute nightmare for me to figure out what license
> things are under when I'm installing things from the firefox extensions
> site.  But maybe we could be involved and guide things in a positive
> direction.

As for the addons site (AMO), I used to work on that. It's been
awhile, but I can say the team is pretty open to contributors and
discussion. The software is admittedly a bit high-bar to get running
and hacking.

But, you can find more details here: https://wiki.mozilla.org/AMO

Also, the guys behind Apps at Mozilla hang out on mailing lists and
IRC, if you're interested:

http://groups.google.com/group/mozilla-labs
irc://irc.mozilla.org/labs
irc://irc.mozilla.org/openwebapps

> PS: Also, this is being done under the banner of "open web", which
> seems misleading considering the "open web gaming" stuff:
>
> https://mozillalabs.com/gaming
>
> ...had a competition in which iirc the majority of winning finalists
> were proprietary games.  Yes yes, "open" has long had problems as a
> term, but maybe the term "standardized web" is less misleading.
>
> (This isn't to hate on Mozilla, which I think is massively an ally of
> ours generally.)

I wasn't involved in the Game On project, myself. But, I do know
everyone involved, and I can say there were no nefarious motives at
work. I'm not sure what the whole story was, but I think it was
intended as an honestly open effort.

-- 
Leslie Michael Orchard
address@hidden
http://decafbad.com
{web,mad,computer} scientist