/irc-logs / w3c / #webapps / 2009-04-20 / end
Options:
- # Session Start: Mon Apr 20 00:00:00 2009
- # Session Ident: #webapps
- # [00:17] * Joins: shepazu (schepers@128.30.52.30)
- # [00:46] * Joins: Marcos (Marcos@84.215.160.79)
- # [01:23] * Quits: Marcos (Marcos@84.215.160.79) (Quit: Marcos)
- # [01:26] * Quits: heycam (cam@124.168.17.176) (Quit: bye)
- # [03:26] * Joins: heycam (cam@130.194.73.110)
- # [04:19] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [06:48] * Quits: timeless (timeless@65.75.195.122) (Ping timeout)
- # [08:15] * Quits: heycam (cam@130.194.73.110) (Quit: bye)
- # [08:28] * Joins: heycam (cam@130.194.221.3)
- # [08:36] * Quits: heycam (cam@130.194.221.3) (Quit: bye)
- # [08:39] * Joins: heycam (cam@130.194.221.3)
- # [08:45] * Joins: timeless (timeless@65.75.195.122)
- # [09:09] * Joins: Marcos (Marcos@213.236.208.247)
- # [09:24] * Joins: tlr (tlr@128.30.52.28)
- # [09:28] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Tomorrow to fresh woods, and pastures new.)
- # [09:41] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [09:51] * Joins: anne (annevk@213.236.208.22)
- # [10:13] * Joins: Marcos_ (Marcos@213.236.208.247)
- # [10:13] * Quits: Marcos (Marcos@213.236.208.247) (No route to host)
- # [10:14] * Joins: arve (arve@213.236.208.22)
- # [10:18] * Quits: heycam (cam@130.194.221.3) (Quit: bye)
- # [11:09] * Joins: heycam (cam@124.168.17.176)
- # [11:11] * Quits: Marcos_ (Marcos@213.236.208.247) (Quit: Marcos_)
- # [11:16] * Joins: Marcos (Marcos@213.236.208.247)
- # [11:21] * Quits: Lachy (Lachlan@85.196.122.246) (Quit: This computer has gone to sleep)
- # [11:34] * Joins: Lachy (Lachlan@213.236.208.22)
- # [11:34] <Hixie> what is this discussion of <access> about?
- # [11:34] <Hixie> is it a widget thing?
- # [11:34] <Hixie> or is it something i care about?
- # [11:36] <anne> maybe both
- # [11:37] <anne> widgets has such an element and it clashes with http://www.w3.org/TR/xhtml-access/ which is something PF and others from WAI are advocating now and then
- # [11:38] <anne> though I suppose they're mostly silent about it until it reaches REC and will then try to get it into UAs and such similar to how RDFa happened, but maybe not, we'll see :)
- # [11:40] <Hixie> holy crap
- # [11:41] <Hixie> xhtml-access takes all the problems of accesskey, combines them with the problems of qnames-in-content, and introduces the ability to have platform-inconsistent UI behaviour
- # [11:41] <Hixie> all at the same time
- # [11:41] <Hixie> that's fantastic
- # [11:52] <tlr> hixie, the one thing you'll care about (which I think is only loosely related to access) is how widgets will eventually deal with origins and base uris
- # [11:52] <tlr> and no, the widget access thing *is* separate from xhtml-access
- # [11:53] <Hixie> widgets as far as i can tell are native apps, not web content, so i'm not really worried about any aspect of them
- # [11:53] <Hixie> not that i have anything against them
- # [11:54] <Hixie> i just don't have the bandwidth to worry about them
- # [11:54] <tlr> My suspicion is that they will blur with web content within relatively few years.
- # [11:54] <Hixie> i doubt it
- # [11:54] <Hixie> how do you see them doing so?
- # [11:55] <tlr> (a) being executed in the same runtime environment
- # [11:55] <tlr> (b) perhaps being executed off a web page
- # [11:55] <tlr> (c) perhaps ending up as the framework for web apps' access to things like address books and microphones
- # [11:56] <tlr> The minimum requirement I have for the widget model (in terms of origin and all that) is that I want to be able to think of a widget as a navigation context (or whatever you call it these days), and still have the html5 security model make sense.
- # [11:56] <Hixie> if you mean browsing context, i don't think we've ever called them anything but browsing context
- # [11:56] <tlr> yep, browsing context
- # [11:56] <Marcos> hixie: http://dev.w3.org/2006/waf/widgets/#the-access-element
- # [11:57] <tlr> (sorry, it's been a while that I looked into that part. )
- # [11:57] <Hixie> i thought widgets were native apps, security-wise?
- # [11:57] <Hixie> has that changed?
- # [11:57] <Marcos> no
- # [11:57] <Marcos> yes
- # [11:57] <Hixie> i thought they had, like, access to the filesystem and so forth
- # [11:57] <Marcos> widgets are both
- # [11:57] <Marcos> hixie, no access is access to resources on the interwebz
- # [11:57] <Marcos> or any URI
- # [11:58] <Hixie> i'm confused
- # [11:58] <Hixie> how is a widget different from a web page then?
- # [11:58] <Marcos> widget signed, packaged
- # [11:58] <tlr> (a) different notions of identity
- # [11:58] <Marcos> and that
- # [11:59] <tlr> (b) separate instances are separate trust domains
- # [11:59] <tlr> therefore the proposal to use synthetic origins
- # [11:59] <Marcos> oh, we are talking about that
- # [11:59] <Hixie> why not just do what safari does with dashboard widgets now, and allow any web page to be turnd into a widget? that has the same end-user result, no?
- # [12:00] <Hixie> without any new specs needed
- # [12:00] * tlr chuckles. I'd actually like to get there.
- # [12:00] <Hixie> we are there
- # [12:00] <Hixie> safari's alread shipped that for years
- # [12:00] * Hixie slides a 'y' in there
- # [12:01] <tlr> safari is there, but the rest of the world isn't.
- # [12:01] <Marcos> I'm sure they are just adding an iframe or something to do that
- # [12:01] <tlr> Plus, people indeed see widgets as native apps, and want to build app stores around that.
- # [12:02] <tlr> therefore, what's interesting here is the tension between that kind of business model (with requisit requirements) and the "let's take a web page and run it without chrome" approach.
- # [12:02] <Hixie> Marcos: they just crop a browser window basically
- # [12:02] <Marcos> Hixie: yep
- # [12:02] <Hixie> no widget spec needed
- # [12:02] * Marcos deletes widget spec, resigns as editor
- # [12:02] <hendry> :)
- # [12:02] * tlr grins, takes spec, renames as "web application packaging"
- # [12:03] <Hixie> Marcos: well my point is i don't understand what the widget spec does that's useful, given what you've said about the security model. what's the point? i assume there is one!
- # [12:03] <hendry> Hixie: have you seen my blog? http://dabase.com/blog/Widgets_are_simple_offline_packages/
- # [12:03] <Marcos> Hixie, ok, lets say. Widgets are native apps.
- # [12:04] <Marcos> They provide access to APIs beyond HTML5
- # [12:04] <Marcos> APIs are the ones developed by BONDI and others
- # [12:04] <tlr> precisely, re additional APIs. That's where the security model is likely to diverge (or be done first for widgets, and make it to web apps later)
- # [12:05] <Hixie> hendry: offline web apps are no more immature than widgets; no less fully defined; your third bullet point doesn't seem to be a reason either way; and synchronisation of user data is hard with widgets too
- # [12:05] <Hixie> gaaah, BONDI
- # [12:05] <Hixie> run away!!!
- # [12:05] <Marcos> you know you love it :D
- # [12:05] <hendry> Hixie: dude I work on BONDI
- # [12:05] <Hixie> sorry to hear that
- # [12:05] <Marcos> eheh
- # [12:06] <hendry> Hixie: i don't widgets are met to by synced ?
- # [12:06] * hendry puts in a 'think' in there somewhere
- # [12:06] <Hixie> hendry: well if you don't need syncing, then offline web apps aren't any harder
- # [12:07] <hendry> Hixie: offline web apps implies sync to me. but i guess you're right
- # [12:07] <Hixie> offline web apps is just a packaging mechanism
- # [12:07] <Marcos> Hixie: we provide updates. And I guess that would work with HTML5 sync
- # [12:07] <Hixie> the manifest autoupdates resources from the web
- # [12:08] <hendry> Hixie: Don't understand ... what is the package?
- # [12:08] <tlr> hendry, a collection of resources from the web
- # [12:08] <Hixie> right
- # [12:08] <Marcos> hendry: is a "logical" package
- # [12:08] <hendry> tlr: is it a zip or a tar ball that i could archive offline?
- # [12:08] <tlr> it is a collection of resources that you could archive offline
- # [12:08] <Hixie> right
- # [12:08] <hendry> as a tar ball or zip or anything? what what?
- # [12:09] <Marcos> Lets get back to talking about access? :)
- # [12:09] <Marcos> (now that Hixie is convinced that widgets are awesome)
- # [12:10] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Ping timeout)
- # [12:11] <Marcos> the <access uri=""> element lets you basically do cross domain
- # [12:13] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [12:17] <tlr> it lets a widget declare that it wants to do cross domain
- # [12:19] * Marcos pictures Hixie reading the widget spec and running around his house shouting "this is great!" :P
- # [12:20] <Hixie> why can't it just do cross-domain anyway
- # [12:20] <Hixie> i'm confused
- # [12:20] <Marcos> because we don't allow it.
- # [12:20] <Marcos> You have to declare every domain you want to access
- # [12:22] <Hixie> why?
- # [12:22] <Hixie> that seems... odd
- # [12:22] <Hixie> can't these widgets run arbitrary binary code?
- # [12:22] <Marcos> Hixie, that's like asking why can't web pages just do cross domain.
- # [12:23] * tlr notes that that's the actual oddity for dashboard widgets; however, TR/widget widgets probably aren't supposed to.
- # [12:23] <Marcos> (well, they can, thanks to script and img)
- # [12:23] <Marcos> but anyway, that, we are plugging that hole in widgets
- # [12:23] <Marcos> hixie, that's why you need to declare
- # [12:23] <Marcos> what domain you want to access
- # [12:23] <tlr> marcos, Hixie's point is that *if* (like the dashboard) you permit binary code plugins in a widget, *then* all bets are off.
- # [12:24] <Hixie> Marcos: web pages can just do cross-domain
- # [12:24] <Hixie> i'm very confused
- # [12:24] <Marcos> widgets can't do cross domain
- # [12:24] <Hixie> what's the point of making this more secure than web pages, if the user has to install them anyway
- # [12:25] * tlr notes that you're talking cross-purpose as as what you mean by "do cross-domain" is concerned.
- # [12:25] <Hixie> the user has to go to a web page to do it, right? so they've already been exposed to cross-domain...
- # [12:25] <Marcos> no, they got it over bluetooth
- # [12:25] * tlr drops a "far" between "as as"
- # [12:26] <Hixie> people still use bluetooth?
- # [12:26] <Hixie> what kind of device is this meant for?
- # [12:26] <Marcos> lets say, as series 60 device
- # [12:28] <Hixie> series 60 devices can't browse the web??
- # [12:28] <Marcos> ha! you've not used Opera?
- # [12:28] <Hixie> yes but opera didn't use bluetooth to visit web pages!
- # [12:28] <Hixie> i'm even more confused now
- # [12:29] <Marcos> Hixie, I get the widget over bluetooth
- # [12:29] <Marcos> I browse the web over GPRS
- # [12:29] <Marcos> Widget uses GPRS to access network
- # [12:30] <Marcos> get it?
- # [12:30] <Marcos> (or I got the widget over BitTorrent)
- # [12:32] <Hixie> why don't you just get the widget from GPRS?
- # [12:32] <tlr> hixie, app store use case
- # [12:33] <tlr> domain from which you got widget is separate from anything that would identify the widget's author usefully
- # [12:34] <Marcos> hixie, the point is that we can't say where widgets will come from. We want them to come from anywhere, not just HTTP
- # [12:34] <Marcos> Hence we package them using Zip
- # [12:35] <Marcos> If they are sent over HTTP, they are labelled with the appropriate content type ("application/widget")
- # [12:35] <Marcos> otherwise, anything goes and it need to work
- # [12:35] <Marcos> so zip is the easiest, most interop. thing to use
- # [12:36] <Hixie> tlr: when i buy an app from the apple or android app stores, the app comes over the same connection as the web...
- # [12:36] <Marcos> Hixie, over HTTP?
- # [12:36] <tlr> hixie, for all intents and purposes, these app stores are transparent proxies for arbitrary crap
- # [12:36] <Marcos> by crap he means widgets :D
- # [12:36] <Hixie> Marcos: i still don't understand why these apps have to be more secure than a web page
- # [12:37] <tlr> and I *really* don't want to give a widget full access to an app store where CSRF means loss of money.
- # [12:37] <Marcos> Hixie, because we have an opportunity to patch a few holes. We should take them
- # [12:37] <tlr> hixie, see above, I think you're largely talking cross purposes when you say "do cross-domain"
- # [12:38] <Hixie> tlr: quite possible
- # [12:38] <tlr> my understanding is that <access> does *not* affect inline elements, for example
- # [12:38] <tlr> ... but rather replaces the same origin policy for something like XHR
- # [12:38] <Hixie> anyway
- # [12:38] <Hixie> as i said, this isn't the web, so i really don't have the bandwidth to worry
- # [12:38] <Marcos> Hixie, sorry, it is the web
- # [12:38] <Hixie> if it's the web, stop plugging holes and just use the web
- # [12:38] * tlr mumbles "film at 11" and waits for hixie and Marcos to throw spiders at each other
- # [12:39] * Marcos notes his is rubber and Hixie is glue :)
- # [12:39] <Marcos> s/his/he's
- # [12:40] <Marcos> Hixie, point taken. Problem is that widgets don't have an origin
- # [12:40] <Hixie> it seems to me you want it both ways -- but you can't be the web if you're not actually using the web's concepts, security model, apis, etc
- # [12:41] <Hixie> bondi isn't the web
- # [12:41] * Joins: darobin (robinb@82.233.247.234)
- # [12:41] <Hixie> not having an origin isn't the web
- # [12:41] <Hixie> having a different security model isn't the web
- # [12:41] <Marcos> ok, we are not the Web. You are right.
- # [12:41] <Hixie> :-)
- # [12:41] * Marcos allows Hixie to go free and not worry about widgets :)
- # [12:41] <Hixie> :-)
- # [12:42] <tlr> for the record, I dare a bet that "offline webapps 2.0" and "widgets 2.0" will be the same thing.
- # [12:43] <Marcos> but be warned, Hixie, widgets will crop up on the Web so it might be in your interest to take a peak a the spec and send some comments.
- # [12:43] <Hixie> i don't understand how to give comments, because i don't understand what widgets do that the web doesn't already do
- # [12:44] * Marcos starts to worry :(
- # [12:44] <darobin> widgets *are* native apps
- # [12:44] <darobin> they're just native apps for which people want to control some aspects of what they have access to
- # [12:44] * tlr starts to stop worrying and love the bomb
- # [12:44] <Hixie> well then they won't crop up on the web :-)
- # [12:45] <Hixie> native apps are fundamentally incompatible with the web security model
- # [12:45] <darobin> agreed
- # [12:45] <Hixie> Marcos doesn't seem to :-)
- # [12:45] * Marcos agrees
- # [12:45] <Marcos> now
- # [12:45] <darobin> if they do crop up on the web, it'll be as something like ActiveX controls (heh heh) except that you're supposed to get a say in what they can access
- # [12:45] <Marcos> you've convinced me that I'm crazy thinking otherwise
- # [12:46] <darobin> I think that trying to see widgets in a web context at this stage isn't helpful
- # [12:46] <Marcos> agreed
- # [12:46] <darobin> it could happen, and it probably won't need a spec actually
- # [12:46] <tlr> disagree
- # [12:46] <Marcos> widgets are native apps that use HTML and JS and stuff
- # [12:46] <darobin> but in the meantime, they're just native apps that use web-based technology, and for which we want a security model that allows people to be selective about what they grant access to
- # [12:47] <tlr> perhaps one way to think about them is as web runtimes (*shudder*) with mildly different feature sets and access to local code.
- # [12:47] <darobin> what falls out of that is that, indeed, it does make a potential convergence between apps and the web a fair bit easier — but we'll cross that bridge later
- # [12:47] <tlr> These beasts may still, e.g., have frames that came directly from the web, and behave just like every frame that comes from a web application willb ehave.
- # [12:47] <darobin> tlr: right, but you can already do that with native apps
- # [12:48] <tlr> darobin, you can do *everything* with native apps
- # [12:49] <tlr> the value proposition is to have a cross-platform development environment
- # [12:49] <tlr> which only works because people have to deploy web browsers anyway, and have incentives to reuse code, massively
- # [12:49] <tlr> i.e., the value proposition goes downhill in the moment in which you deviate from the web model
- # [12:50] <darobin> fair enough
- # [12:50] <darobin> but I'm not sure that influences how the spec is crafted
- # [12:50] <darobin> as opposed to how we might think about the ecosystem
- # [12:51] <Hixie> well having a different security system is a deviation
- # [12:51] <Hixie> for what it's worth :-)
- # [12:53] <tlr> precisely
- # [12:53] <tlr> therefore, the minimum requirement is that the beast must be consistent with the html security model to the extent possible
- # [12:54] <tlr> whether it behaves like a sandboxed iframe in that system, or like a web page, is a different question
- # [12:57] <darobin> I'd assume same security model with extensions to allow some things to be loosened up
- # [12:58] <darobin> what with it being a native app and all :)
- # [13:20] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Tomorrow to fresh woods, and pastures new.)
- # [13:21] * Quits: arve (arve@213.236.208.22) (Ping timeout)
- # [13:28] * Joins: taf2 (taf2@68.49.245.59)
- # [13:29] * Joins: ArtB (d0309a43@128.30.52.43)
- # [13:46] * Quits: taf2 (taf2@68.49.245.59) (Quit: taf2)
- # [13:52] * Quits: Marcos (Marcos@213.236.208.247) (Ping timeout)
- # [13:55] * Joins: arve (arve@213.236.208.22)
- # [13:57] * tlr is now known as tlr-bbl
- # [14:03] * Quits: arve (arve@213.236.208.22) (Ping timeout)
- # [14:04] * Joins: Marcos (Marcos@213.236.208.247)
- # [14:07] * Quits: Marcos (Marcos@213.236.208.247) (Ping timeout)
- # [14:18] * Joins: taf2 (taf2@65.210.82.235)
- # [14:20] * Joins: arve (arve@213.236.208.247)
- # [14:23] * Joins: Marcos (Marcos@213.236.208.22)
- # [14:48] * Quits: gsnedders (gsnedders@86.136.52.180) (Client exited)
- # [14:59] * Joins: chaals (chaals@89.130.83.193)
- # [14:59] * Joins: aroben (aroben@71.58.77.15)
- # [15:05] * Quits: arve (arve@213.236.208.247) (Ping timeout)
- # [15:09] * Joins: arve (arve@213.236.208.247)
- # [15:18] * tlr-bbl is now known as tlr
- # [15:21] * Quits: chaals (chaals@89.130.83.193) (Client exited)
- # [15:28] <ArtB> Marcos, is there a place for Versioning in the Widgets V2 wiki (http://www.w3.org/2008/webapps/wiki/Widgets2_UC%26R)? Or do see it as No/Never ?
- # [15:29] <Marcos> ArtB: never
- # [15:29] <Marcos> not for the lang
- # [15:30] * Quits: arve (arve@213.236.208.247) (Ping timeout)
- # [15:36] * Joins: arve (arve@213.236.208.247)
- # [15:36] * Quits: arve (arve@213.236.208.247) (Quit: Ex-Chat)
- # [15:41] * tlr is now known as tlr-brb
- # [16:28] * Quits: Marcos (Marcos@213.236.208.22) (Ping timeout)
- # [16:34] * Joins: Marcos (Marcos@213.236.208.247)
- # [17:21] * Quits: Lachy (Lachlan@213.236.208.22) (Quit: This computer has gone to sleep)
- # [17:48] * Joins: gsnedders (gsnedders@86.136.52.180)
- # [17:54] * Quits: darobin (robinb@82.233.247.234) (Ping timeout)
- # [18:21] * tlr-brb is now known as tlr
- # [18:29] * Quits: Marcos (Marcos@213.236.208.247) (Quit: Marcos)
- # [18:35] * Joins: Marcos (Marcos@213.236.208.22)
- # [19:08] * Quits: taf2 (taf2@65.210.82.235) (Quit: taf2)
- # [19:30] * Joins: taf2 (taf2@65.210.82.235)
- # [19:57] * Quits: Marcos (Marcos@213.236.208.22) (Quit: Marcos)
- # [20:12] * Joins: aroben_ (aroben@71.58.77.15)
- # [20:14] * Quits: aroben (aroben@71.58.77.15) (Ping timeout)
- # [20:23] * Joins: Lachy (Lachlan@85.196.122.246)
- # [20:31] * Quits: Lachy (Lachlan@85.196.122.246) (Connection reset by peer)
- # [20:31] * Joins: Lachy (Lachlan@85.196.122.246)
- # [20:52] * Quits: Lachy (Lachlan@85.196.122.246) (Quit: This computer has gone to sleep)
- # [21:24] * Quits: ArtB (d0309a43@128.30.52.43) (Quit: CGI:IRC)
- # [22:42] * Joins: arve (arve@95.34.170.26)
- # [23:02] * Joins: Lachy (Lachlan@85.196.122.246)
- # [23:49] * Quits: arve (arve@95.34.170.26) (Ping timeout)
- # Session Close: Tue Apr 21 00:00:00 2009
The end :)