/irc-logs / freenode / #whatwg / 2011-07-12 / end

Options:

# Session Start: Tue Jul 12 00:00:00 2011
# Session Ident: #whatwg
# [00:00] <zewt> but doing what?
# [00:00] <TabAtkins> zewt: Extracting image data from a tainted canvas.
# [00:00] <jamesr> revealing whether a pixel in a cross-origin image/canvas is fully transparent or not
# [00:00] <jamesr> not arbitrary data
# [00:00] <Hixie> jamesr: no, that's just the starting point. The up-side of the range should be a few trillion years more.
# [00:01] <zewt> webgl dropped support for non-same-origin/CORS images as textures
# [00:01] <Hixie> jamesr: and since it's relative, it just means that you can't have one browser run for more than a few trillion years.
# [00:01] <zewt> Hixie: er, not with a double; that would limit precision to milliseconds...
# [00:01] <TabAtkins> Darn, img-timing-2 doesn't work in chrome/linux. It's too noisy or something.
# [00:01] <jamesr> Hixie: you lose resolution a lot more quickly than that
# [00:01] <Hixie> scale my numbers accordingly
# [00:02] <Hixie> do i have to do everything here? :-P
# [00:02] <jamesr> TabAtkins: the attack depends heavily on implementation details of the gfx libraries involved
# [00:02] <Philip`> jamesr: I think it'd be easy to extend to the whole alpha channel - you can tell if the alpha is 0 or non-0, then just draw onto a temporary canvas with 50% globalAlpha then do timing to determine if it's 0-or-1/256 or higher, then repeat until you know all the bits
# [00:02] <jamesr> Hixie: well, your numbers scale into ranges that we have to care about :)
# [00:02] <TabAtkins> Hixie: Come now, at least be within orders of magnitude. You can't even reach a million years if you're using JS numbers and counting milliseconds.
# [00:02] <jamesr> Philip`: depending on roundoff on the intermediate canvases? that's a lot more interesting
# [00:02] <jamesr> Philip`: can you check if that really works?
# [00:03] <jamesr> i think we can can afford 1 year of noise and still keep 20 bits around for subpixel precision
# [00:03] <zewt> webgl never sets a canvas to non-clean
# [00:03] <jamesr> which is easily good enough for exactly representing nanoseconds
# [00:04] <zewt> so I'm not sure that it matters anymore whether you can read back a texture with it
# [00:04] <jamesr> zewt: it does
# [00:04] <jamesr> zewt: timing attacks
# [00:04] <jamesr> it's a lot easier to extract arbitrary image data from a webgl canvas than a 2d canvas
# [00:04] <zewt> they just changed it a couple months ago, dropping support for non-same-origin/CORS images entirely
# [00:04] <jamesr> right ,for this reason
# [00:05] <zewt> so ... it doesn't, then :)
# [00:05] <zewt> that was the point--this problem already seems fixed
# [00:05] <zewt> unless there's something I'm missing
# [00:06] <zewt> (it's not a very nice fix, it broke some of my code badly, which is very annoying)
# [00:06] * jernoble|afk is now known as jernoble
# [00:06] <zewt> early adopter woes o/~
# [00:08] <jamesr> broke angry birds too
# [00:08] <jamesr> it's necessary, tho
# [00:10] <zewt> making them even angrier
# [00:11] <zewt> i think having a really big noise factor for the base time doesn't matter: the case where that's a problem is where you can take a lot of samples and average out the noise, but in this case by definition you can only get 1 sample, ever
# [00:12] <zewt> (it does mean you can tell fairly reliably if the user restarted his browser)
# [00:13] <zewt> (or his OS, if that's what the time base uses)
# [00:13] * Quits: smaug____ (~chatzilla@YYYKMMMCMI.gprs.sl-laajakaista.fi) (Ping timeout: 246 seconds)
# [00:14] * jernoble is now known as jernoble|afk
# [00:14] <zewt> devil's advocating: it's also a user identifier (so long as he doesn't); even with cookies and everything else disabled, and using eg. Tor, you can probably still track an active user based on his timebase
# [00:14] <zewt> (... as he doesn't restart)
# [00:16] <jamesr> how could you detect the timebase? juts by the approx range?
# [00:16] <annevk-cloud> Is http://lists.w3.org/Archives/Public/public-webrtc/2011Jul/0010.html some indication Mozilla is not planning on working with the WHATWG on this? Kind of weird no email went to the WHATWG list...
# [00:17] <zewt> when a user views a page, read the time and send it to the server; when the user views another page do it again; then correllate values
# [00:17] <zewt> if it's the same user, the amount of time passed will be roughly the amount of time between the requests; for different users they'll be wildly different
# [00:17] * annevk-cloud is now known as annevk5
# [00:18] <jamesr> yeah
# [00:19] <zewt> using Window-specific (or perhaps tab-group-specific or whatever) at least alleviates that...
# [00:20] <Philip`> jamesr: http://philip.html5.org/demos/canvas/img-timing-3.html
# [00:20] <Philip`> jamesr: In (at least) Opera that reproduces the grey (alpha=0.5ish) and black (alpha=1) lines
# [00:21] <jamesr> Philip`: are you testing runs/second or seconds/run?
# [00:22] <Philip`> jamesr: It just does 100 drawImage()s and counts the milliseconds
# [00:23] <jamesr> yeah it would burn a lot less of my CPU and be more accurate if it tried to do drawImage()s until it hit some time threshhold, then checked how many millis expired
# [00:23] <Philip`> then repeats forever and does the shading based on the median timing value for each cell
# [00:23] <zewt> jamesr: could alleviate that by going in the other direction: set the timebase so that the timestamp is, say, roughly Epoch-based (according to whatever the system clock is when the browser starts up)
# [00:23] <jamesr> zewt: i think having it be unix-epochy is just going to be more confusing
# [00:23] <Philip`> It's a proof-of-concept, not a production-quality implementation ;-)
# [00:24] <zewt> that way, user timebases would tend to cluster around that timebase, making correllation not useful
# [00:24] <jamesr> Philip`: yeah, works pretty well
# [00:24] <jamesr> ok, so the whole alpha channel :)
# [00:25] <Philip`> You should pick an epoch far in the future, and be mysterious about the reasons for picking it, then people far in the future will think it must be counting down to some doomsday event
# [00:25] <jamesr> i doubt webdevs can properly do math with negative numbers
# [00:25] <zewt> burn
# [00:27] * Joins: shans (~shanestep@74.125.56.17)
# [00:27] <The_8472> negative margins already confuse them
# [00:27] <zewt> i suspect the problem of user trackability is a bigger one than the problem of detecting how long the browser has been running...
# [00:27] <zewt> eg. adding noise to the base is causing a worse problem than it solves
# [00:28] <zewt> (tracking that way without noise still works sort-of, but much less reliably, I think, since tons of users will have fairly small values)
# [00:28] <jamesr> if they are defined as not comparable across contexts, then the tracking problem goes away
# [00:29] <Philip`> Could you make the API never return an absolute time, and just return an opaque object with a .getDifference(anotherOpaqueTimeObject) method?
# [00:29] <jamesr> thought of that too
# [00:29] <jamesr> pain in the butt to use
# [00:29] <The_8472> Philip`, btw... your is a nice DoS :P
# [00:30] <The_8472> just had to zap my browser
# [00:30] * Quits: David_Bradbury (~chatzilla@75-147-178-254-Washington.hfc.comcastbusiness.net) (Quit: ChatZilla 0.9.87 [Firefox 5.0/20110615151330])
# [00:30] <zewt> Philip`: that's also breaking cross-context use implicitly anyway
# [00:30] <zewt> assuming the object isn't structured clonable or anything
# [00:32] <Philip`> You could do a "window.timeBase = getCurrentTimeObject()" and then use getCurrentTimeObject().getDifference(timeBase) everywhere in your application and get just a number, which would be a bit verbose but no harder to use than a number-return API, unless you want to use it cross-window in which case you just pass the objects instead
# [00:32] <Philip`> (and extend the structured clone algorithm to work with it)
# [00:33] <zewt> i guess what that essentially does is give each page its own implicit time base, of whenever it initialized a base-time object
# [00:34] <jamesr> or define it such that the first call to getMonotonicTime() returns 0 and everything else is relative to that
# [00:34] <Philip`> The_8472: Yeah, some browers seem to deal quite inelegantly with it :-(
# [00:34] <zewt> a time base which, having an actual representation, can be passed around as needed to different contexts
# [00:34] <jamesr> so the time base is explicit and decided by the author
# [00:34] <zewt> jamesr: that has the context problem again though
# [00:35] <zewt> (if handling multiple contexts is given up on, then that's probably the simplest, though)
# [00:35] <jamesr> zewt: only if we want these comparable across contexts, which i don't think is valuable
# [00:36] * Quits: CvP (~CvP@123.49.23.90) (Quit: [ UPP ] > all)
# [00:39] * Quits: oknoway (~oknoway@72.11.82.226) (Quit: oknoway)
# [00:39] <zewt> messages across threads like "do action when the current time is >= value" can be useful (more precise than saying "do it in 60 seconds"), or "time out when at a specified time" sort of things...
# [00:40] <jamesr> that would have to be relative to when they receive a message, i think
# [00:40] <zewt> huh? i mean relative times are less precise than calculating a time in advance and passing it down the line, generally speaking
# [00:40] * Joins: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net)
# [00:40] <jamesr> yeah, but we don't have shared state between workers
# [00:41] <zewt> but if the monotonic timebase is universal across the browser then you can do that
# [00:41] <Philip`> I guess hardware is going to become more like a distributed system in the future, so maybe it's better to avoid assuming an architecture that has a universal clock across all windows/threads/processes/etc, given that universal clocks are fundamentally impossible
# [00:42] <roc> annevk: no, I think it's just an indication that Anant thought the RTC group was appropriate
# [00:43] <roc> I hope that Web apps never have to worry about clock skew
# [00:43] <jamesr> Philip`, zewt: yeah, the question really is "how much skew are you willing to put up with?"
# [00:43] * Quits: magistr (~magistr@92-127-122-186-xdsl-dynamic.kuzbass.net) (Ping timeout: 276 seconds)
# [00:43] <jamesr> so i'd rather just not have any promises
# [00:43] * Joins: magistr (~magistr@92-127-98-26-xdsl-dynamic.kuzbass.net)
# [00:45] <annevk5> roc, would prolly be good to send at least one email indicating a fork is in the works, since quite a few people gave feedback on the WHATWG list already
# [00:46] * Joins: beverloo (~beverloo@nat/google/x-wwnpbukymapkzeii)
# [00:47] * Joins: Transformer (~Transform@ool-4a59e397.dyn.optonline.net)
# [00:47] * Parts: Transformer (~Transform@ool-4a59e397.dyn.optonline.net)
# [00:48] * Joins: llrcombs (~llrcombs@64.130.210.214)
# [00:49] <roc> I'll tell him to provide feedback on the WHATWG list
# [00:50] <roc> I don't really think of it as a fork, since I don't think Ian's proposal is baked enough for that, but maybe it is
# [02:50] * Disconnected
# [02:51] * Attempting to rejoin channel #whatwg
# [02:51] * Rejoined channel #whatwg
# [02:51] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [02:51] * Set by annevk42 on Mon Oct 19 23:03:06
# [02:51] <MikeSmith> and Microdata and RDFa seem to be a case where both could exist and find widespread use in the same way that Relax NG and XML Schema (or whatever) have
# [02:54] * Quits: dbaron (~dbaron@nat/mozilla/x-vjjfcmvmqjxodoop) (Quit: 8403864 bytes have been tenured, next gc will be global.)
# [02:55] <TabAtkins> On that note, every time I look at RDFa I become more convinced that it's a technically worse solution in general.
# [02:58] <MikeSmith> well, I'm not personally taking any position on that, but I will say that was the view that James Clark and Murata-san expressed about XML Schema, and why they came up with Relax NG
# [02:58] <MikeSmith> and I realize I'm harping on about XML Schema and Relax NG and that those are somewhat esoteric
# [02:58] <MikeSmith> but to me at least that seems like the most relevant precedent
# [03:02] * Joins: davidb (~davidb@bas1-toronto06-2925210136.dsl.bell.ca)
# [03:03] * Quits: ojan (~ojan@nat/google/x-etrevrnfqtfyhqxh) (Quit: ojan)
# [03:03] * Joins: wakaba_ (~wakaba_@122x221x184x68.ap122.ftth.ucom.ne.jp)
# [03:03] * Joins: agektmr (~Adium@220.109.219.244)
# [03:04] * Quits: mpilgrim (~pilgrim@rrcs-24-206-36-125.midsouth.biz.rr.com) (Ping timeout: 260 seconds)
# [03:07] * Quits: karlcow (~karl@nerval.la-grange.net) (Quit: This computer has gone to sleep)
# [03:12] * Quits: davidb (~davidb@bas1-toronto06-2925210136.dsl.bell.ca) (Ping timeout: 250 seconds)
# [03:19] * Joins: mahir256 (~mahir256@50.40.186.8)
# [03:20] * Parts: mahir256 (~mahir256@50.40.186.8)
# [03:25] * Quits: ap (~ap@2620:149:4:401:226:4aff:fe14:aad6) (Quit: ap)
# [05:26] * Disconnected
# [05:27] * Attempting to rejoin channel #whatwg
# [05:27] * Rejoined channel #whatwg
# [05:27] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [05:27] * Set by annevk42 on Mon Oct 19 23:03:06
# [05:35] * Quits: pdr__ (~pdr@nat/google/x-tprcpoejnbixyfos) (Quit: pdr__)
# [05:40] <roc> did Richard Schwerdtfeger ever explain *why* his people are using canvas instead of SVG?
# [05:43] * Quits: aswetrave (~aswetrave@c-24-131-190-97.hsd1.mn.comcast.net) (Ping timeout: 252 seconds)
# [05:46] * Joins: aswetrave (~aswetrave@c-24-131-190-97.hsd1.mn.comcast.net)
# [05:46] * Quits: chriseppstein (~chris@99-34-231-235.lightspeed.sntcca.sbcglobal.net) (Read error: Operation timed out)
# [05:46] * Joins: chriseppstein (~chris@99-34-231-235.lightspeed.sntcca.sbcglobal.net)
# [05:48] * Quits: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net) (Remote host closed the connection)
# [05:52] * Joins: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net)
# [05:54] * Joins: nonge__ (~nonge@p5082B626.dip.t-dialin.net)
# [05:59] * Quits: shans (~shanestep@nat/google/x-arptzbvbbwjuiigc) (Quit: shans)
# [05:59] * Quits: nonge_ (~nonge@p5B326D5E.dip.t-dialin.net) (Ping timeout: 264 seconds)
# [08:02] * Disconnected
# [08:03] * Attempting to rejoin channel #whatwg
# [08:03] * Rejoined channel #whatwg
# [08:03] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [08:03] * Set by annevk42 on Mon Oct 19 23:03:06
# [08:03] -asimov.freenode.net:#whatwg- [freenode-info] channel trolls and no channel staff around to help? please check with freenode support: http://freenode.net/faq.shtml#gettinghelp
# [08:04] * Quits: shans (~shanestep@74.125.56.17) (Quit: shans)
# [08:08] * Joins: shans (~shanestep@nat/google/x-qzigmvsscskihxed)
# [08:08] * Joins: niftylettuce (~niftylett@c-71-61-47-144.hsd1.pa.comcast.net)
# [08:17] * Joins: mhausenblas (~mhausenbl@wg1-nat.fwgal01.deri.ie)
# [08:18] <hsivonen> hmm. http://www.xanthir.com/demos/collidingcircles/ doesn't work for me in the Firefox build I happen to have
# [08:18] <hsivonen> clicking the canvas does nothing
# [08:21] * Joins: shetech (~shetech@c-76-126-167-49.hsd1.ca.comcast.net)
# [08:22] * Quits: shetech (~shetech@c-76-126-167-49.hsd1.ca.comcast.net) (Client Quit)
# [08:29] * heycam|away is now known as heycam
# [08:35] * hsivonen is reading 10 days of public-html email
# [08:35] * Joins: sharksoda (~colby@c-67-168-136-89.hsd1.wa.comcast.net)
# [08:36] <hsivonen> the accessibility rhetoric does not seem to help with finding solutions
# [10:36] * Disconnected
# [10:37] * Attempting to rejoin channel #whatwg
# [10:37] * Rejoined channel #whatwg
# [10:37] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [10:37] * Set by annevk42 on Mon Oct 19 23:03:06
# [10:41] * Joins: jeremyselier (u2513@gateway/web/irccloud.com/x-fjicwxvqinkkfczu)
# [10:42] * Quits: niftylettuce (~niftylett@c-71-61-47-144.hsd1.pa.comcast.net) (Ping timeout: 255 seconds)
# [10:42] * Joins: manu-db (~msporny@digitalbazaar.com)
# [10:49] * Joins: danbri (~danbri@ip176-48-210-87.adsl2.static.versatel.nl)
# [10:49] <annevk5> XML Schema vs Relax NG is a nice one
# [10:50] <annevk5> manu-db, despite what you suggest, people have did think CSS should be removed in favor of XSL-FO
# [10:50] <annevk5> manu-db, same with Selectors and XPath
# [10:51] <annevk5> manu-db, and again with Media Queries vs some XPath-based syntax
# [10:58] * Quits: othermaciej (~mjs@c-24-6-209-6.hsd1.ca.comcast.net) (Quit: othermaciej)
# [11:00] * Joins: micheil (~micheil@109.231.193.164)
# [13:08] * Disconnected
# [13:09] * Attempting to rejoin channel #whatwg
# [13:09] * Rejoined channel #whatwg
# [13:09] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [13:09] * Set by annevk42 on Mon Oct 19 23:03:06
# [13:09] <Zeddy> when accessing the page using http://www.kokonniemi.fi/guide/index.php/view/map/1/1 the manifest file saves 0kb
# [13:09] * Joins: erlehmann (~erlehmann@89.204.153.64)
# [13:09] <Zeddy> when accessing using http://kokonniemi.fi/guide/index.php/view/map/1/1 it saves 602kb
# [13:12] <Philip`> The &lang= would probably be a validator error in the HTML, but shouldn't affect the behaviour of browsers
# [13:12] <Zeddy> ok, had the html tag defined twice
# [13:12] <moo-_-> Zeddy: I suggest you contact your CMS authors and file a bug ;)
# [13:12] <Zeddy> that might explain some problems x)
# [13:12] <Zeddy> im using codeigniter
# [13:14] <Zeddy> god damnit, now that i fixed my html the site isn't caching anything
# [13:14] <Zeddy> is it working for you?
# [13:14] <Zeddy> chrome again tells me the following: Application Cache Error event: Failed to parse manifest http://www.kokonniemi.fi/guide/index.php/manifest?map=1&lang=2
# [13:14] <Zeddy> x)
# [13:15] <Zeddy> oh, chrome doesn't like comments it seems
# [13:16] <Zeddy> haha, neither did firefox, works in all browsers now :)
# [13:16] <annevk> did you have a comment as first line?
# [13:16] <annevk> because that is bogus
# [13:17] <annevk> read http://www.whatwg.org/specs/web-apps/current-work/complete/offline.html#manifests
# [13:18] <Zeddy> hmm, yeah just above Cache manifest x)
# [13:18] <Zeddy> moved it beneath the line and it works
# [13:21] <annevk> well of course
# [13:21] <annevk> CACHE MANIFEST is the file signature
# [13:21] <annevk> if you replace that with something random it is no longer a cache manifest
# [13:21] <Zeddy> yeah, i realised that
# [13:21] <Zeddy> i didn't even notice i had that line beneath the comment
# [13:22] <Zeddy> i also noticed that i cannot output time() in the comments, because the manifest file changes during the download process and gets corrupt
# [13:23] <Zeddy> wont the browser re-cache everything if a single file is added or deleted to the manifest?
# [13:23] <annevk> it will
# [13:23] <Zeddy> because i could have some kind of version number in the database which would be updated everytime the admin makes changes to text data
# [13:24] <Zeddy> in most cases there will probably be changes in the list of files aswell, but sometimes there might be tiny corrections to the main page, and that won't automatically cause a re-cache, will it?
# [13:25] <annevk> no
# [13:25] <annevk> everything is loaded from cache
# [13:25] <annevk> the manifest is then fetched and compared to the current copy
# [13:25] <annevk> if it is changed new stuff is fetched
# [13:25] <Zeddy> so whenever an admin makes any changes i just increment a version number which is updated in the database and i should always have everything up to date as long as the user visits the page in online mode
# [13:25] <annevk> the next time a refresh happens the new app is loaded
# [13:26] <annevk> if the user just keeps a tab open you might have to indicate a new version is available and a refresh is required
# [13:26] <annevk> or some such
# [13:26] <Zeddy> echo "#version".$map_version."\n";
# [13:26] <Zeddy> yeah but thats a bit far fetched since the data wont be updated that often
# [13:28] <Zeddy> Another thing i was wondering, since my pages will be dynamic aswell as the manifest file, is it possible to clear everything cached by my domain even though the url is different? Or will it perhaps do that automatically
# [13:28] <Zeddy> Since the user might switch from http://www.kokonniemi.fi/guide/index.php/manifest?map=1&lang=2 to http://www.kokonniemi.fi/guide/index.php/manifest?map=1&lang=3
# [13:28] <Zeddy> and i don't want it to cache both
# [13:28] * Joins: virtuelv (~virtuelv_@20.74.9.46.customer.cdi.no)
# [13:29] * Quits: virtuelv (~virtuelv_@20.74.9.46.customer.cdi.no) (Remote host closed the connection)
# [13:30] * Quits: homata_ (~homata_@58x158x182x50.ap58.ftth.ucom.ne.jp) (Remote host closed the connection)
# [13:32] <hsivonen> now that Zakim supports SIP, what's a good way to get a provider?
# [13:34] <annevk> does it work with Skype?
# [13:36] <Zeddy> i don't get this, i visit my page with the manifest file, it caches the manifest, i visit it again and it also loads javascript and css not included in the manifest, but if i go offline once and visit the page again the css and javascript are obviously no there, the weird part is that when i go online again and refresh, it doesn't even attempt to load the css and javascript
# [13:36] * Quits: dirkpennings (~Vuurbal@240-216-164-193.fiber.trined.nl) (Ping timeout: 250 seconds)
# [13:36] * Joins: FireFly (~firefly@unaffiliated/firefly)
# [13:36] <Zeddy> why does it load the javascript and css when refreshing in online mode in the first place then?
# [13:36] <annevk> because it has not seen the manifest yet
# [13:37] <Zeddy> Huh, the first time i visit it sees the manifest and download everything, then i press F5, shouldn't it have seen the manifest at that point?
# [13:38] <annevk> yes
# [13:38] <annevk> oh wait
# [13:38] <annevk> not sure what is going on there
# [13:40] * Quits: yuuki (~kobayashi@58x158x182x50.ap58.ftth.ucom.ne.jp) (Quit: Leaving...)
# [13:40] * Quits: wakaba_ (~wakaba_@122x221x184x68.ap122.ftth.ucom.ne.jp) (Quit: Leaving...)
# [13:41] <Zeddy> you can try it yourself here: http://www.kokonniemi.fi/guide/index.php/view/map/1/1
# [13:41] <Zeddy> it seems to be the same with both chrome and firefox
# [15:41] * Disconnected
# [15:43] * Attempting to rejoin channel #whatwg
# [15:43] * Rejoined channel #whatwg
# [15:43] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [15:43] * Set by annevk42 on Mon Oct 19 23:03:06
# [15:43] <hsivonen> annevk: were we supposed to have an HTML-XML call today?
# [15:43] * Quits: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net) (Read error: Connection reset by peer)
# [15:43] * Joins: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net)
# [15:49] <manu-db> annevk5, annevk: re: CSS should be removed in favor of XSL-FO - I hope that 'removal' of Microdata isn't the intention with the Microdata/RDFa TF. I think that they could be merged successfully (with a number of changes to Microdata or RDFa). I think this discussion could positively impact Microformats as well. I want to see that dialog start. If all of us end up in a place that we can agree...
# [15:50] <manu-db> ...on, all the better. I think we're not there today because the Microdata, the Microformats and the RDFa camp haven't been communicating as much as they should have been.
# [15:51] <manu-db> So, yes - I agree that there was a contingent of people that wanted a black and white decision on CSS and XSL-FO, Selectors and X-Path, etc. - Personally, I'm not looking for that to happen. I want to know if the changes we're proposing to RDFa could find common ground, or if changes to Microdata could find common ground.
# [15:52] <AryehGregor> manu-db, I don't think the syntaxes can be merged at this point, realistically. The data models, maybe, and that would be productive.
# [15:52] <manu-db> If the use cases are supported, I don't think that it matters which one ends up "winning".
# [15:52] <AryehGregor> I don't see why a committee is going to be remotely useful, though. It seems like a case of "when in doubt, form a committee".
# [15:53] <AryehGregor> It would seem more productive for people to propose specific changes in the form of bugs.
# [15:53] <manu-db> AryehGregor: People don't really care that much about the syntaxes - both are equally confusing to some, both are reasonable to others. The data models are important and yes, I do think a discussion about that would be productive.
# [15:54] * Philip` imagines the main danger is that merging two different systems will result in greater complexity than either of the systems by itself, since it's easier to argue for features than for simplicity
# [15:54] <AryehGregor> I think the syntaxes aren't going to change substantially because both have too much existing content.
# [15:54] <manu-db> AryehGregor: I don't see this as a "design by committee" exercise - maybe it'll turn into that, but I want this to primarily be a meeting between people that are in the same area (structured data), but who have not had the chance to talk to each other and express their pain points.
# [15:54] <Philip`> (and the complexity of RDFa was one of the main reasons for designing Microdata as a replacement)
# [15:55] <hsivonen> manu-db: is there a Web reference for removing CSS to pave way for XSL-FO?
# [15:55] <manu-db> Philip`: yes, more complexity is one of the concerns. Now, I'm talking about stripping stuff out of RDFa (just me, not the WG) to see if that could create consensus.
# [15:55] <AryehGregor> manu-db, well, good luck.
# [15:55] <manu-db> hsivonen: annevk said that, not I. I just trusted that he was correct. :)
# [15:56] <hsivonen> manu-db: It sounds plausible, sure.
# [15:56] <AryehGregor> I don't think there are going to be strong enough incentives for compromise here. The W3C can't exert much pressure on RDFa because it has too much entrenched support, and can't exert much pressure on microdata because the microdata people would be happy to tell the W3C to drop dead and only publish at the WHATWG.
# [15:56] <karlcow> "do not fear death"
# [15:56] <hsivonen> manu-db: in fact, I believe the CSS vs. XSL situation was more combative initially, but I haven't read up on that story enough
# [15:57] <manu-db> AryehGregor: I don't think that logging bugs will do much at this point, because there are pockets of people that don't think there is a problem - or that two syntaxes isn't causing confusion. Those folks need to speak to implementers - implementers are getting increasingly agitated that they're having to put both types of markup into their pages.
# [15:57] <AryehGregor> Theoretically everyone agrees that having only one solution is better, but in practice both sides think that solution should be theirs.
# [15:57] * Joins: eric_carlson_ (~eric_carl@2620:149:4:401:217:f2ff:fe03:a2e)
# [15:57] <hsivonen> manu-db: anyway, I suspect it's not correct that XSL-FO has always been an XML print thing
# [15:57] <karlcow> hsivonen: if I remember there were a few contentious
# [15:57] <AryehGregor> manu-db, you mean authors, I guess, not implementers? One way to solve that problem is to just decide on one or the other, as schema.org has done.
# [15:57] <karlcow> xpath vs css selectors
# [15:57] <karlcow> was one
# [15:57] <hsivonen> manu-db: why would howcome have written the Considered Harmful essay if there wasn't a threat of FO leaking to the Web?
# [15:58] <hsivonen> manu-db: which people don't care about the syntaxes?
# [15:58] <manu-db> hsivonen: I didn't mean to say that XSL-FO has always been a "print" thing - I meant that it was /one/ of the targets - which made it a bit different than CSS. If you guys think XSL vs. CSS was a better example of something like this happening in the past, I'd be happy to use that as an example - especially if it resulted in some sort of consensus.
# [15:58] * Quits: eric_carlson (~ericc@adsl-67-112-12-110.dsl.anhm01.pacbell.net) (Disconnected by services)
# [15:58] * eric_carlson_ is now known as eric_carlson
# [15:58] <AryehGregor> So I don't think there's a huge problem right now. Compromising on the syntax seems like a bad idea -- it will just result in something that's worse than either of the existing ones.
# [15:58] <hsivonen> manu-db: do we get to kill CURIEs now if syntax isn't cared about? :-)
# [15:59] <AryehGregor> I think the only productive thing to do at this point is to make sure you can mechanically transform microdata to RDFa and vice versa to the greatest extent possible.
# [15:59] <manu-db> hsivonen: killing CURIEs - it's a possibility (in my mind, at least)
# [16:00] <AryehGregor> The microdata spec has a conversion algorithm, but it's not targeted at allowing you to reuse existing RDFa vocabularies, without some translation layer -- e.g., it can only create a very specific sort of vocabulary URL.
# [16:00] <manu-db> hsivonen: "which people don't care about syntaxes": Regular web devs (theory). In practice those that I've been able to talk to at the International Press Telecommunications Council, Good Relations, etc.
# [16:00] <AryehGregor> So I think the best thing to do is acknowledge that microdata and RDFa will both continue to exist for now, and try to make sure they can interoperate as much as possible.
# [16:00] <karlcow> xpath vs css selectors, xml query vs xslt, cassoulet vs choucroute, pepsi vs coke, css and fo, xml vs rdf, xhtml vs html,
# [16:01] <manu-db> AryehGregor: The problem isn't interoperability - that's fairly easy *wild hand waving*
# [16:01] <AryehGregor> Microformats is a separate thing, that should just be ignored. It's a legacy hack designed to embed data in pages while still being valid HTML 4.
# [16:01] <scor> AryehGregor: that assumes that those consuming this structured data can parse both syntaxes
# [16:01] <manu-db> AryehGregor: The problem is that implementers are having to put /both/ Microdata /and/ RDFa into their markup to placate different folks: Facebook OGP and Google schema.org, for example.
# [16:01] <AryehGregor> scor, yes, exactly. Put work into making sure that the same libraries support both interchangeably without extra work, as much as possible.
# [16:02] <manu-db> AryehGregor: We're working on that now, with the Structured Data API in the RDF Web Apps WG - it's meant to be syntax agnostic.
# [16:02] <AryehGregor> manu-db, yeah, that's a pain. But a compromise syntax is no more likely to resolve it than just trying to get everyone to converge on one or the other.
# [16:02] * Philip` thought every big company made up their own 'standard' vocabularies, so you'd have to insert two separate sets of metadata anyway
# [16:02] <AryehGregor> I mean, if you announce a great new compromise tomorrow, Facebook probably isn't going to instantly support it for OGP.
# [16:02] <manu-db> AryehGregor, but again - that's not the real problem. The real problem is that Web devs are having to put /both/ types of markup in their pages.
# [16:03] <AryehGregor> It would probably be just as easy to get them to support microdata.
# [16:03] <AryehGregor> The reason people have to use both isn't because there's no compromise, it's because some organizations have gone with one and some have gone with another.
# [16:03] <manu-db> Philip` - yes, the whole NIH vocabularies are a pain. We're trying to get the large companies to converge on vocabularies, there is some progress on that front.
# [16:03] <AryehGregor> That's not something that can be resolved in standards-land, we have to talk to the organizations.
# [16:03] * Quits: Ankheg (~Ankheg@91.224.77.4) (Quit: Leaving.)
# [16:04] <manu-db> AryehGregor: I think you would find Google, Facebook, etc. more willing to support a unified way forward than you imagine. At least, that's my read on them.
# [16:04] <AryehGregor> Yes, but that unified way forward may as well just be microdata. It's not like Facebook is using RDFa because they think it's better, they're using it because nobody had heard of microdata when they introduced OGP.
# [16:05] <linclark> AryehGregor: just catching up on this... there are some implementers of tools for authoring RDFa/microdata. we're probably a small bunch, but there are some none-the-less ;)
# [16:05] <karlcow> hmm the convergence of vocabularies will happen through business practices. I'm pretty sure. Where and when it matters. Community of interests with business needs, requirements aka profitability
# [16:05] <manu-db> AryehGregor: That's what I meant - we /are/ talking to the organizations and there seems to be a desire to work together on vocabs.
# [16:06] * Quits: eric_carlson (~eric_carl@2620:149:4:401:217:f2ff:fe03:a2e) (Quit: eric_carlson)
# [16:06] * Joins: eric_carlson (~eric_carl@2620:149:4:401:217:f2ff:fe03:a2e)
# [16:07] <manu-db> AryehGregor: I think that's a fairly wild claim, to say that you understand why Facebook picked RDFa. I'm just pointing that out - I don't think any one of us can clearly understand why they picked one or the other.
# [16:07] <AryehGregor> When was OGP introduced, and when was microdata first published?
# [16:07] <manu-db> I think something that does hold true is that all of the companies that have picked one or the other, have not clearly understood /both/ RDFa or Microdata when they adopted it.
# [16:08] <manu-db> OGP was announced April 21st 2010
# [16:08] <AryehGregor> Oh, sure. But that's true for basically every web technology. Almost nobody gets anything except the very most concrete aspects.
# [16:08] <manu-db> Yes, exactly
# [16:09] <manu-db> Which is why I think that people making the argument from the standpoint of the adopting organizations /completely/ understanding the options is a very difficult argument to make.
# [16:09] <AryehGregor> So microdata had been around for a while, but nobody actually used it until schema.org came around, and that might have been related to internal decisions within Google. (I have no idea, I'm an outside contractor.)
# [16:09] * Quits: maikmerten (~merten@ls5dhcp197.cs.uni-dortmund.de) (Ping timeout: 252 seconds)
# [16:09] <AryehGregor> Whereas RDFa had been gaining traction for years.
# [16:10] <AryehGregor> Unless you were a standards aficionado, if you heard of any technology in this area it would be RDF.
# [16:11] <AryehGregor> Also, didn't Facebook already use RDFa in the form FBML before OGP?
# [16:12] <AryehGregor> Anyway, this really isn't a subject I'm particularly knowledgeable about or interested in. All I'll say is that no one seems to prefer RDFa to microdata unless they were committed to RDFa before microdata was developed, and at least one major organization (Google) that previously used RDFa is transitioning to microdata, so I'm going to bet the best way to get standardization is just to standardize on microdata. I think the worst way to proc
# [16:12] <AryehGregor> eed is to try introducing yet another markup language that will complicate matters still further. Regardless, good luck.
# [16:13] <karlcow> the walmart effect
# [16:13] <karlcow> the elephant in the room changing the business practice for everyone else
# [18:13] * Disconnected
# [18:15] * Attempting to rejoin channel #whatwg
# [18:15] * Rejoined channel #whatwg
# [18:15] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [18:15] * Set by annevk42 on Mon Oct 19 23:03:06
# [18:17] * Quits: Onderhond (~woef@79.232-136-217.adsl-static.isp.belgacom.be) (Ping timeout: 252 seconds)
# [18:21] <annevk> AryehGregor, I once worked on that too btw, in my first summer here in Norway: http://annevankesteren.nl/projects/whatwg/spec
# [18:21] <annevk> AryehGregor, I never got really far
# [18:22] <AryehGregor> annevk, interesting.
# [18:22] <AryehGregor> Never saw that.
# [18:22] * AryehGregor will take a look
# [18:22] <annevk> be careful, six years ago, I didn't really know how stuff worked
# [18:23] <AryehGregor> You'll probably have at least one angle on something that will be interesting to me.
# [18:24] <AryehGregor> I've gotten next to no feedback on the spec so far from anyone.
# [18:24] <annevk> that always sucks :/
# [18:24] <annevk> you basically need to find someone that either implements or writes tests
# [18:24] * Quits: Lachy (~Lachy@pat-tdc.opera.com) (Ping timeout: 260 seconds)
# [18:24] <annevk> for good feedback
# [18:24] <annevk> otherwise it just gets stuck until reviewed
# [18:25] <annevk> the same is true with lots of the HTML specification
# [18:25] <annevk> the parts that were already implemented have had way less review
# [18:25] <AryehGregor> Once I'm done polishing up the spec, hopefully within a couple of weeks, I'll publicize it and try to encourage implementers to respond speedily.
# [18:25] * Joins: Maurice` (copyman@5ED573FA.cm-7-6b.dynamic.ziggo.nl)
# [18:25] <annevk> let me know how that goes :)
# [18:28] * Joins: hij1nx (~hij1nx@12.49.89.159)
# [18:29] * Quits: seventh (seventh@64.9.130.68) (Ping timeout: 252 seconds)
# [18:30] <TabAtkins> AryehGregor: Make sure you pull in Ryosuke (rniwa@) as soon as you're ready. He's the "editting guy" on Chrome.
# [18:30] <AryehGregor> TabAtkins, I know, he's been my WebKit contact.
# [18:30] <TabAtkins> Excellent.
# [18:31] <AryehGregor> I always CC him when I have specific questions that I ask on whatwg.
# [18:31] <AryehGregor> Ehsan is my Mozilla person, and Hallvord is my Opera person, although the latter hasn't given much feedback.
# [18:32] <AryehGregor> When I'm ready, I'll post to public-html and CC Adrian Bateman also.
# [18:32] * Quits: jwalden (~waldo@2620:101:8003:200:221:6aff:fe6e:d10) (Quit: ChatZilla 0.9.86-rdmsoft [XULRunner 1.9.2.18/20110621100037])
# [18:32] * Joins: smaug____ (~chatzilla@GGYMKDIX.gprs.sl-laajakaista.fi)
# [18:33] <TabAtkins> annevk: When can the Origin header contain multiple origins?
# [18:34] * Quits: erlehmann (~erlehmann@89.204.153.64) (Quit: Ex-Chat)
# [18:34] * Quits: danbri (~danbri@ip176-48-210-87.adsl2.static.versatel.nl) (Remote host closed the connection)
# [18:36] * Quits: mpt (~mpt@canonical/mpt) (Ping timeout: 258 seconds)
# [18:37] * Quits: teear (teear@84-231-25-80.elisa-mobile.fi) (Remote host closed the connection)
# [18:40] * Quits: smaug____ (~chatzilla@GGYMKDIX.gprs.sl-laajakaista.fi) (Ping timeout: 258 seconds)
# [18:40] * Joins: teear (teear@84-231-25-80.elisa-mobile.fi)
# [18:42] <AryehGregor> This is what I love about functional programming:
# [18:42] <AryehGregor> / "Fix disallowed ancestors of each member of sublist."
# [18:42] <AryehGregor> sublist.forEach(fixDisallowedAncestors);
# [18:42] <AryehGregor> s!^/!//!
# [18:43] <Philip`> That looks like you're mutating the list
# [18:43] <Philip`> That's not functional :-p
# [18:44] <AryehGregor> It's not mutating the list.
# [18:44] * Joins: jwalden (~waldo@2620:101:8003:200:222:68ff:fe15:af5c)
# [18:44] <Philip`> What's it doing, then?
# [18:45] <AryehGregor> It's calling fixDisallowedAncestors on each member of sublist.
# [18:45] <AryehGregor> That mutates the DOM, but not the list.
# [18:46] * Philip` counts mutating things reachable from the list as mutating the list, in order to keep his statement valid
# [18:49] <AryehGregor> In that case, mutating any global variable or anything accessible from a global variable counts as a mutation, because of ownerDocument.defaultView.
# [18:49] * Quits: magistr (~magistr@92-127-98-26-xdsl-dynamic.kuzbass.net) (Ping timeout: 264 seconds)
# [18:49] * Joins: ap (~ap@2620:149:4:401:226:4aff:fe14:aad6)
# [18:50] * Joins: magistr (~magistr@92-127-98-26-xdsl-dynamic.kuzbass.net)
# [18:50] <AryehGregor> Hey, wait. Why can't we use instanceof like if (foo instanceof foo.ownerDocument.defaultView.Node) to avoid multiple-global problems, at least for Nodes?
# [18:50] <AryehGregor> Our use-case for testharness.js was mostly DOMException, admittedly.
# [18:51] * Joins: mpt (~mpt@91.189.88.12)
# [18:51] * Quits: mpt (~mpt@91.189.88.12) (Changing host)
# [18:51] * Joins: mpt (~mpt@canonical/mpt)
# [18:52] * Quits: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net) (Remote host closed the connection)
# [18:54] <TabAtkins> Who's "we" here? Because that's pretty verbose and horrible for authors.
# [18:54] <Philip`> If you really like functional programming, you should replace all global variables and DOM mutation functions with a monad system
# [18:54] * Quits: pdr__ (~pdr@nat/google/x-ggafeobqxexmdnul) (Quit: pdr__)
# [18:54] <AryehGregor> TabAtkins, I was specifically thinking of testharness.js, where we want to test things like "foo is a Node".
# [18:55] <AryehGregor> But can't because of the multiple-global problem.
# [18:55] <AryehGregor> But actually the major thing we wanted to test is "foo is a DOMException", and this idea doesn't help for that.
# [18:55] <AryehGregor> Philip`, I never said *pure* functional.
# [18:55] <TabAtkins> Ah, kk. Then I dunno, because the DOM is confusing. But sure, sounds reasonable.
# [18:56] <Philip`> Impure functional isn't really functional :-p
# [18:56] <TabAtkins> Sure it is.
# [18:56] <TabAtkins> Otherwise Lisp isn't functional, and then "functional" isn't a very useful term.
# [18:58] <AryehGregor> Sure it is, if you only care about Haskell.
# [18:58] <AryehGregor> Anyone who cares about non-Haskell languages doesn't matter.
# [18:58] <TabAtkins> Still, any definition of "functional" that excludes Lisp is a silly definition.
# [18:59] <Philip`> It's a useful term for computations written in a subset of the language that don't do any mutation, since it means you can happily refactor things or do memoisation etc without worrying about changing the semantics of the program
# [19:00] <TabAtkins> That's what the term "pure functional" is for.
# [19:01] <Philip`> It doesn't seem a very useful term for referring to any syntax that happens to use functions as first-class objects, like forEach, since pretty much any language can do that
# [19:02] <TabAtkins> Many languages still can't do so easily. Passing around function pointers doesn't really make functions "first-class objects".
# [19:03] * Joins: smaug____ (~chatzilla@ZYYYMMDCCCLXVIII.gprs.sl-laajakaista.fi)
# [19:04] * Joins: david_carlisle (~chatzilla@dcarlisle.demon.co.uk)
# [19:05] * Quits: micheil (~micheil@109.231.193.164) (Quit: http://brandedcode.com | http://github.com/miksago)
# [19:07] <AryehGregor> "One of the problems with block-break is that its exact semantics depend on the markup language being used and the nodeName of the editing host. For example, in HTML you want to disable the ability to insert p element nodes inside a h1 element node. For generic XML languages this is almost impossible with using a schema language of some sort."
# [19:07] <AryehGregor> annevk, I can tell this is old. :)
# [19:08] * Joins: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net)
# [19:10] * Quits: davidwalsh (~davidwals@75-135-74-55.dhcp.mdsn.wi.charter.com) (Quit: Reading http://davidwalsh.name)
# [21:12] * Disconnected
# [21:13] * Attempting to rejoin channel #whatwg
# [21:13] * Rejoined channel #whatwg
# [21:13] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [21:13] * Set by annevk42 on Mon Oct 19 23:03:06
# [21:15] * Joins: danbri (~danbri@ip176-48-210-87.adsl2.static.versatel.nl)
# [21:16] * Joins: jamesr (~jamesr@216.239.45.98)
# [21:20] * Joins: Lachy (~Lachy@cm-84.215.59.50.getinternet.no)
# [21:21] * Quits: zdobersek (~zan@cpe-46-164-14-148.dynamic.amis.net) (Quit: Leaving.)
# [21:22] * Joins: The_8472 (~stardive@azureus/The8472)
# [21:29] * Joins: othermaciej_ (~mjs@17.246.18.226)
# [21:33] * Quits: othermaciej (~mjs@17.203.114.102) (Ping timeout: 250 seconds)
# [21:33] * othermaciej_ is now known as othermaciej
# [21:34] * Quits: danbri (~danbri@ip176-48-210-87.adsl2.static.versatel.nl) (Remote host closed the connection)
# [21:35] * Quits: smaug____ (~chatzilla@GZMCCXC.gprs.sl-laajakaista.fi) (Ping timeout: 258 seconds)
# [21:41] * Quits: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net) (Read error: Connection reset by peer)
# [21:41] * Joins: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net)
# [21:42] * Quits: Martijnc (~Martijn@d54C02C64.access.telenet.be) (Ping timeout: 276 seconds)
# [21:43] * Quits: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net) (Read error: Connection reset by peer)
# [21:43] * Joins: karlcow (~karl@nerval.la-grange.net)
# [21:44] * Joins: jacobolus (~jacobolus@208-90-212-189.PUBLIC.monkeybrains.net)
# [23:44] * Disconnected
# [23:45] * Attempting to rejoin channel #whatwg
# [23:45] * Rejoined channel #whatwg
# [23:45] * Topic is 'WHATWG: http://www.whatwg.org/ -- logs: http://krijnhoetmer.nl/irc-logs/ -- stats: http://gavinsharp.com/irc/whatwg.html -- Please leave your sense of logic at the door, thanks!'
# [23:45] * Set by annevk42 on Mon Oct 19 23:03:06
# [23:48] * Joins: niftylettuce_ (~niftylett@c-71-61-47-144.hsd1.pa.comcast.net)
# [23:48] * Joins: david_carlisle (~chatzilla@dcarlisle.demon.co.uk)
# [23:48] * Quits: niftylettuce (~niftylett@32.167.95.154) (Ping timeout: 255 seconds)
# [23:49] * Quits: beverloo (~beverloo@nat/google/x-jihkvglnjpyychub) (Quit: beverloo)
# [23:50] * Joins: beverloo (~beverloo@nat/google/x-hkeggldoxvtxzkau)
# [23:55] * Quits: othermaciej (~mjs@17.246.18.226) (Quit: othermaciej)
# [23:56] * Joins: davidb (~davidb@bas1-toronto06-2925210136.dsl.bell.ca)
# [23:59] * Quits: simplicity- (~simpli@unaffiliated/simplicity-) (Quit: simplicity-)
# [23:59] * Joins: cpearce (~chatzilla@203-97-204-82.dsl.clear.net.nz)
# Session Close: Wed Jul 13 00:00:00 2011

The end :)