| 00:15 | <gsnedders> | Anyone here go to Imperial College London? |
| 00:44 | gsnedders | is somewhat tempted to write to Imperial to ask to change what he's applying for |
| 00:53 | <gsnedders> | (from theoretical physics MSci. to mathematics with computer science MSci.) |
| 01:06 | <gsnedders> | Or do I try and change to computer science? |
| 01:06 | <gsnedders> | Meh. |
| 01:11 | <Hixie> | do what you find interesting |
| 01:11 | <Hixie> | :-) |
| 01:12 | <gsnedders> | I'd find both interesting, it's just I'm probably less likely to get in for the joint degree :) |
| 01:13 | <gsnedders> | The main problem is it's far harder to change in England what course you are doing once you've started. |
| 01:36 | <BenMillard> | jgraham & annevk, I'm tried a more vibrant "web 2.0" pallette of colours for links: http://ben.local/!dev/whatwg-blog/ |
| 01:36 | <BenMillard> | I think it makes the links complement the runs of orange code a bit better than #00c |
| 01:37 | <BenMillard> | it is inconsistent with the spec now, of course |
| 06:42 | <takkaria> | gsnedders: not convinced it is that hard, you know |
| 06:42 | <takkaria> | gsnedders: I know a fair few people who transferred within the first few weeks at a few UK unis |
| 08:19 | <olliej> | yoyo dave_levin |
| 08:19 | <dave_levin> | olliej: hi |
| 08:19 | <dave_levin> | You found me in a different irc :) |
| 08:22 | <hsivonen> | Hixie: thanks for bringing the wiki back up. |
| 08:22 | <hsivonen> | Hixie: any pointers on filing a bug against Google Code operations? |
| 08:23 | <annevk> | BenMillard, ben.local won't work |
| 08:26 | <gpy> | annevk: Is the supporting of win95 + win98 of Opera something intended or just a feature? |
| 08:26 | <hsivonen> | does Opera 10 still support those? |
| 08:27 | <annevk> | we try not to break support |
| 08:27 | <annevk> | afaik |
| 08:27 | <gpy> | http://weblogs.mozillazine.org/asa/archives/2009/01/ie6_on_the_way.html |
| 08:27 | <gpy> | (Hey Opera, how about seriously targeting those users since you guys are the only modern browser left supporting ancient versions of Windows?) |
| 08:27 | <gpy> | </quote> |
| 08:28 | <hsivonen> | isn't it bad to support staying with OS releases that don't get security patches? |
| 08:33 | <dave_levin> | hsivonen: Do you want http://code.google.com/p/support/issues/list? |
| 08:33 | <annevk> | at least if they don't upgrade they can at least get a secure browser |
| 08:36 | jgraham | wonders why gsnedders has dismissed colleges not in the centre of Cambridge |
| 08:44 | <hsivonen> | dave_levin: thanks. (I wonder why that isn't where the "Report" link points to in the error message.) |
| 08:45 | <dave_levin> | hsivonen: Maybe you should file that issue too :) (I went thru their blog entries to find this.) |
| 08:45 | <hsivonen> | dave_levin: yeah, I'll file both. |
| 08:47 | <dave_levin> | hsivonen: fwiw this entry is the one that mentioned it: http://google-code-updates.blogspot.com/2008/11/improving-issue-tracker-for-larger.html |
| 08:48 | <hsivonen> | perhaps I googled for it using the wrong words |
| 08:57 | <hsivonen> | ok. bugs filed. http://code.google.com/p/support/issues/detail?id=2032 and http://code.google.com/p/support/issues/detail?id=2031 |
| 09:02 | hsivonen | wonders if http://code.google.com/p/support/issues/detail?id=873&colspec=ID%20Type%20Status%20Milestone%20Priority%20Stars%20Owner%20Summary#c54 is like http://weblogs.mozillazine.org/gerv/archives/2008/10/top_hackers.html |
| 09:29 | <jgraham> | BenMillard: Sorry about the table inspector UI regression |
| 09:30 | <jgraham> | It is fixed locally. The bug you found looks like stray debugging code so it should be pretty easy to fix too (but I don't know when I will be able to resync the live version) |
| 11:02 | Hixie | updates web forms 2 |
| 11:06 | <hsivonen> | is there a process for adding the note to the W3C copy? |
| 11:06 | <annevk> | you can update the shorthand to redirect to html5 |
| 11:06 | <Hixie> | hsivonen: the /TR copy? |
| 11:07 | <Hixie> | annevk: done |
| 11:08 | <hsivonen> | Hixie: yes |
| 11:09 | <annevk> | it was more in reply to hsivonen regarding the /TR/ shorthand, but for the WHATWG one it makes sense too :) |
| 11:09 | <Hixie> | hsivonen: dunno, whose charter does it fall under? |
| 11:09 | <Hixie> | annevk: oh :-) |
| 11:09 | <hsivonen> | that is, is there a process for making the /TR/shortname of redirect to a different place once the draft has fallen off charters? |
| 11:09 | <hsivonen> | s/ of / of a draft / |
| 11:09 | <Hixie> | the w3c process is so heavy for this kind of thing, i don't really want to get involved in working out what the process is |
| 11:10 | <Hixie> | but feel free to champion the cause if you want to |
| 11:10 | <annevk> | hsivonen, there's no process, you just tell sysreq, prolly easiest to do when we publish a new html5 draft and then list the WF2 draft as a previous version |
| 11:10 | <Philip`> | Why does the first bit of the HTML5 forms page list "Submittable elements" and "Resettable elements", but "Listed" and "Labelable" [not "... elements"]? |
| 11:11 | <Hixie> | Philip`: editorial oversight. file a bug / send an e-mail / ignore it, as desired :-) |
| 11:12 | <zcorpan> | Hixie: no, unless you intend to add similar prose everywhere else that invokes another algorithm :) |
| 11:12 | <Hixie> | zcorpan: bummer |
| 11:12 | <annevk> | (the CSS WG did the shorthand change thing with the borders and background drafts) |
| 11:29 | <zcorpan> | Hixie: could you link to Philip`'s static view of the whatwg issues from http://www.whatwg.org/issues/ ? it's pretty hard to find |
| 11:29 | <Hixie> | url? |
| 11:30 | <zcorpan> | haven't found it yet :) |
| 11:30 | <zcorpan> | http://canvex.lazyilluminati.com/misc/cgi/issues.cgi/ there |
| 11:31 | <Hixie> | done |
| 11:31 | <zcorpan> | thanks |
| 11:31 | <Philip`> | I could move it somewhere better if I could think of a nice domain name :-) |
| 11:32 | <Hixie> | i really should do something about the base64 e-mails |
| 11:32 | <Hixie> | oh well |
| 11:32 | <Philip`> | And about the stripping of lines that start with "*" |
| 11:33 | <zcorpan> | is the parsing folder empty? |
| 11:33 | <jgraham> | Philip`: How about phillip.html5.org/issues |
| 11:33 | <jgraham> | but maybe with your name splet right |
| 11:33 | <Hixie> | stripping of lines that start with "*"? |
| 11:33 | <Hixie> | zcorpan: there is no parsing folder |
| 11:33 | <Hixie> | zcorpan: parsing issues are in the processing-model folder at the moment |
| 11:33 | <zcorpan> | Hixie: ok |
| 11:34 | zcorpan | wonders whether http://lists.whatwg.org/htdig.cgi/whatwg-whatwg.org/2006-October/007528.html got a reply |
| 11:36 | <Hixie> | it doesn't appear to be in my folders anymore |
| 11:36 | <Hixie> | so i assume yes |
| 11:37 | <zcorpan> | we're having trouble with innerHTML roundtripping -- pages depend on the ie6 behavior that form controls don't lose their form after innerHTML assignment |
| 11:37 | <Hixie> | zcorpan: how do you mean? |
| 11:37 | <zcorpan> | Hixie: <table><form><tr><td><input> |
| 11:37 | <zcorpan> | table.innerHTML = table.innerHTML |
| 11:38 | <Hixie> | how do other browsers handle it? |
| 11:39 | <zcorpan> | in safari and firefox the control looses its form |
| 11:41 | <zcorpan> | Hixie: i can't find the reply if there is one |
| 11:42 | <zcorpan> | http://www.google.com/search?q=%22I+didn%27t+consider+that.+I+see+the+issue%22 should probably have 2 results if there was :) |
| 11:43 | <Hixie> | well, let me answer it now then: |
| 11:43 | <Hixie> | making <form> into a table body wouldn't solve your problem for tables that contained table bodies. |
| 11:43 | <Hixie> | and it would be _really_ weird behavior |
| 11:44 | <Hixie> | since all but one of the major browsers don't support this page, i recommend evangelisation. |
| 11:45 | <zcorpan> | i agree it would be weird |
| 11:45 | <Philip`> | Hixie: I meant that (when I last looked) lines starting with "*" were stripped :-) |
| 11:45 | <Hixie> | Philip`: that's odd |
| 11:45 | <Philip`> | but I can't find any examples because I don't know how to search for mails that have lines starting with "*" and that were turned into issues |
| 11:46 | <Hixie> | Philip`: in the body? |
| 11:46 | <Philip`> | Hixie: Yes - I wrote emails with list items starting with "*", and the first line of each list item (i.e. the line starting with "*") was nowhere to be seen in the version on the issues list |
| 11:47 | <zcorpan> | Hixie: though it would work for tables that have tbodies (even though it's rare) by making <form> close tbody |
| 11:48 | <hsivonen> | FYI http://lists.xiph.org/pipermail/accessibility/2009-January/000272.html |
| 11:48 | <Hixie> | zcorpan: only at the cost of breaking the styling |
| 11:49 | <Philip`> | jgraham: I don't really want to run it on the html5.org server because I don't want to bother with having to install Perl modules and stuff |
| 11:49 | <Hixie> | hsivonen: good lord, captions shouldn't use JS |
| 11:50 | <Philip`> | jgraham: but I suppose I could run it on my own server at issues.philip.html5.org/ or something |
| 11:50 | <hsivonen> | Hixie: I'm inclined to think that the captioning browsing context should have scripting disabled. |
| 11:50 | <Hixie> | Philip`: there don't appear to be any e-mails with "*" at the start of them in the issues list right now |
| 11:51 | <Philip`> | Hixie: Well, there wouldn't be since those lines all got stripped out ;-) |
| 11:51 | <Hixie> | Philip`: but it could just be that i never see the *s in the first place |
| 11:51 | <Hixie> | Philip`: i mean, my script doesn't see any *s in the raw data |
| 11:51 | <zcorpan> | Hixie: yeah but the combination of using tbody and form in a table *and* styling tbody in a way that would make a difference is so rare that it's just not there |
| 11:51 | <Philip`> | Hixie: Ah |
| 11:51 | <Hixie> | zcorpan: apparently the combination of <form> and innerHTML is rare enough that no browser other than IE supports it either :-) |
| 11:52 | <zcorpan> | Hixie: fair enough but we're still getting bugs about it |
| 11:52 | <hsivonen> | Hixie: other than JS, what's your take on overlaying a browsing context with a CSS formatter onto the video to implement captioning? |
| 11:53 | <Hixie> | zcorpan: i'm very reluctant to special case <form> here, it's a hack that my intuition says will cause far more problems down the road than anything it might solve today |
| 11:53 | <hsivonen> | Hixie: it seems to me that if a shell document with UA-set innerHTML is used, the result would be massively simpler to implement in browsers than W3C TimedText |
| 11:53 | <zcorpan> | Hixie: ok |
| 11:59 | <Hixie> | hsivonen: it sounds like trying to dig a hole to plant a flower using one of these: http://tinyurl.com/yqodhg |
| 11:59 | <Hixie> | zcorpan: having said that i do wish we had a solution to the innerHTML problem |
| 11:59 | <hsivonen> | Hixie: I guess srt would be simple flower planting |
| 12:00 | <hsivonen> | Hixie: but going beyond srt, what would you suggest? implementing a parallel W3C TimedText formatter instead of reusing the CSS formatter would suck |
| 12:00 | <Hixie> | zcorpan: have you had any bugs on other innerHTML weirdness in IE? like with what it does with mis-nested tags? |
| 12:01 | <Hixie> | hsivonen: my dvd player should not have to support <iframe seamless>, and my web browser should support the same timed text standard as my dvd player. |
| 12:01 | <Hixie> | hsivonen: i don't understand where "formatting" comes into this. we're talking about lines of text overlayed at the bottom of the screen, right? |
| 12:01 | <hsivonen> | Hixie: your DVD player support overlaid bitmaps |
| 12:02 | <Hixie> | hsivonen: do we even need colour or italics? |
| 12:02 | <zcorpan> | Hixie: yeah we've had lots of innerHTML bugs |
| 12:02 | <Hixie> | zcorpan: would be interesting to have a list of cases that fail in safari/firefox also |
| 12:02 | <hsivonen> | Hixie: well, srt covers 99% of usage (text at bottom of screen without styling) |
| 12:02 | <Hixie> | zcorpan: i.e. cases the spec can't handle |
| 12:02 | <Hixie> | hsivonen: cool, that's 19% more than i care about :-) |
| 12:03 | <Hixie> | hsivonen: sounds perfect :-) |
| 12:03 | <hsivonen> | Hixie: the timed text WG has respecified the kitchen sink without CSS |
| 12:03 | <Hixie> | hsivonen: the timed text WG's work will likely be irrelevant then |
| 12:03 | <zcorpan> | Hixie: i think most of them would be solved by implementing an html5 parser and going with html5's innerHTML algorithm |
| 12:04 | <hsivonen> | Hixie: anyway, when one suggests srt, someone always has a use case that isn't satisfied. by induction, you end up with CSS or a poor replacement for CSS |
| 12:04 | <Hixie> | zcorpan: if there are any that wouldn't, like the table-form-input one, i'd love to get a list, maybe there's something we can do to fix a bunch at once |
| 12:05 | <Hixie> | hsivonen: there appears to be a problem in the process you describe. namely, there doesn't seem to be a step at which someone is told "no". |
| 12:05 | <hsivonen> | Hixie: is it your position that browser vendors should not implement native captioning capability beyond roughly the feature set of srt? |
| 12:05 | <hsivonen> | nessy: ping? see above |
| 12:07 | <annevk> | it seems sort of nice to have a subset of HTML in subtitle tracks |
| 12:07 | <Hixie> | hsivonen: do you have a reference for SRT so I could educate myself before answering that? |
| 12:07 | <annevk> | e.g. links |
| 12:07 | <annevk> | dvd players can just use a browser :) makes them also capable of running youtube and all |
| 12:08 | <annevk> | and following the links :p |
| 12:08 | <hsivonen> | Hixie: http://en.wikipedia.org/wiki/SubRip#Sample_file |
| 12:09 | <hsivonen> | Hixie: doom9 documents weird HTMLish extensions to srt. I don't mean those. |
| 12:10 | <annevk> | "Unicode and basic HTML tags such as <i>, <b>, <u> and <font> in subtitles are supported by most players, but VLC (the nightly builds are starting handling them) and some subtitle editors (SubResync, Subtitle Workshop) do not handle these tags well." |
| 12:11 | <hsivonen> | fwiw, if you start supporting <foo>-like tags in srt, you are going down a slippery slope fast, and I'd prefer a well-specified format in terms of innerHTML |
| 12:11 | <hsivonen> | looks like srt is already in the RSS <title> hell |
| 12:11 | <annevk> | agree |
| 12:12 | <deane> | hsivonen: Hi. Was there ever any talk about the W3C adopting v.nu as the front end for validator.w3.org? It's a bit of a mission getting the W3C's validator to validate (X)HTML5 stuff properly. And I haven't even looked at file upload yet. |
| 12:13 | <gsnedders> | jgraham: Because I'm a picky pedantic asshole :) |
| 12:13 | <hsivonen> | deane: no, not really |
| 12:14 | <gsnedders> | jgraham: Actually, it's more a case of I don't want to be to far out |
| 12:14 | <gsnedders> | *too |
| 12:14 | <Hixie> | hsivonen: i, a complete novice in terms of subtitling and timed text, would be surprised to see features beyond the following in a timed text specification: start/end time, full unicode support, device- and media-independent way of annotating non-overlapping, non-nested spans of text as being related to a particular class, _maybe_ the ability to indicate emphasis for spans of text, some mechanism to encode position, and some definition of the expected relationship |
| 12:15 | <hsivonen> | Hixie: it seems that subtitle formats that aren't HTML+CSS are doomed to reinvent HTML+CSS in an ad hoc way |
| 12:15 | <Hixie> | subtitle formats that are HTML+CSS are almost certainly doomed to collapse under their own weight |
| 12:16 | <jgraham> | gsnedders: Bad reason. Because 1) It's not really /that/ far out and 2) you'll likely be closer to Comp. Sci / Phy. buildings anyway (and and 3) because it hardly makes a difference when you're there) |
| 12:16 | <gsnedders> | takkaria: It's certainly true it's harder than in Scotland :P |
| 12:16 | <annevk> | Hixie, if videos are primarily created for the Web and primarily rendered in browsers, why? |
| 12:17 | <Hixie> | videos aren't primarily created for the web nor primarily rendered in browsers. |
| 12:17 | <annevk> | hsivonen, CSS would be optional, right? |
| 12:17 | <gsnedders> | jgraham: I gave up on good reasons a while ago. I mean, whatever criteria I try and apply now isn't going to be what I'd apply after having started. |
| 12:17 | <hsivonen> | Hixie: even in browsers? with my suggested changes, implementing HTML+CSS captions can be done in terms of existing DOM operations given the anonymous iframe |
| 12:17 | <hsivonen> | Hixie: youtube is evidence to the contrary |
| 12:18 | <Hixie> | youtube is a great example |
| 12:18 | <hsivonen> | annevk: as optional as it is for web browsers in general |
| 12:18 | <jgraham> | gsnedders: For example I lived roughly by New Hall/Fits for four years and it wasn't a problem. Plus most of the central colleges have no space so their actual accommodation isn't in the centre of twon anyway |
| 12:18 | <Hixie> | it's streamed to the iPhone, to Apple TV, and to various other devices |
| 12:18 | <jgraham> | *Fitz |
| 12:18 | <gsnedders> | Hixie: having "[html5] r46 - This draft is now obsolete." as a subject for a WF2 commit message is confusing |
| 12:18 | <hsivonen> | Hixie: Apple has a CSS formatter at their disposal |
| 12:18 | <Hixie> | and the idea of turning those devices into web browsers would likely horrify their developers |
| 12:18 | <hsivonen> | iPhone already has a browser |
| 12:19 | gsnedders | always puts Fitz at the wrong side of Cambridge |
| 12:19 | <gsnedders> | jgraham: Huntingdon Rd? |
| 12:19 | <jgraham> | gsnedders: Anyway my point is that if you get pooled to somewhere further out that should not be a sufficient reason to reject it |
| 12:19 | <jgraham> | gsnedders: Just off there, yes |
| 12:20 | <gsnedders> | jgraham: I always want it to be on the south side of Cambridge, for some non-obvious reason |
| 12:20 | <annevk> | hsivonen, e.g. I think it should be possible to extract the subtitles from the HTML and simply render them with default rendering for various elements (or no rendering at all) |
| 12:20 | <hsivonen> | Hixie: fwiw, my initial reaction was also against requiring non-browser players to render HTML, but the prospect of supporting W3C TimedText alongside CSS sucks pretty badly from a browser POV |
| 12:21 | <annevk> | hsivonen, so if e.g. mobile phones do not want to use a CSS formatter they do not have to |
| 12:21 | <hsivonen> | Hixie: and from non-browser POV, HTML sucks about as much as W3C TimedText |
| 12:21 | <jgraham> | Being able to trivially render the subtitles in a web browser does sound like a win |
| 12:21 | <annevk> | hsivonen, then the only cost would be parsing HTML which is negligible |
| 12:21 | <Hixie> | hsivonen: having the features of html and css without using html and css is even worse than using html and css, but that doesn't mean that using html and css is a good idea |
| 12:22 | <gsnedders> | jgraham: Oh well, I guess I'll see. |
| 12:22 | <Hixie> | hsivonen: a bad idea doesn't become a good idea just because it is compared to an even worse idea |
| 12:22 | <hsivonen> | Hixie: it becomes preferable over the worse idea, though :-) |
| 12:23 | <Hixie> | hsivonen: there are more than two options here |
| 12:23 | <jgraham> | Hixie: Given the assumption that the desired feature set of subtitles will expand over time, it seems like a bad idea not to start out with something that is a subset of HTML since otherwise we will end up reinventing HTML, badly |
| 12:23 | <hsivonen> | what jgraham said |
| 12:23 | annevk | agrees |
| 12:24 | <Hixie> | that assumption is a bad assumption |
| 12:24 | <annevk> | Hixie, why? it seems several subtitle formats are heading in that direction |
| 12:24 | <Hixie> | and that assumption assumes that the ultimate end point is html |
| 12:24 | <hsivonen> | but subsetting HTML in a way other than disabling scripting in the browsing context is more complex than just importing the whole thing |
| 12:24 | <Hixie> | why not assume that subtitles will end up being svg? |
| 12:24 | <Hixie> | then we should use an svg subset |
| 12:24 | <annevk> | Hixie, e.g. SubRip and W3C TimedText |
| 12:24 | <jgraham> | Hixie: It's not clear to me whether it is a good thing, but it doesn't seem like a bad assumption. People generally want new features from technology over time |
| 12:24 | <hsivonen> | Hixie: HTML5 innerHTML implicitly imports SVG, too |
| 12:25 | <hsivonen> | Hixie: once the relevant bits of the spec are uncommented |
| 12:25 | <Hixie> | why not assume it shouldn't become OOXML? |
| 12:25 | <annevk> | Hixie, because currently they use HTML like tags and rendering rather than SVG |
| 12:26 | <hsivonen> | Hixie: because current evidence suggests ad hoc feature copying from HTML and CSS--not OOXML |
| 12:26 | <Hixie> | when building technology one needs to start from the problem, not the solution. it doesn't matter what timed text does when considering what timed text _should_ do |
| 12:26 | <annevk> | OOXML is also not really an option |
| 12:26 | <Hixie> | i think you are seeing HTML because that's what you're used to |
| 12:27 | <hsivonen> | Hixie: fwiw, the HTML proposal flowed from trying to address problems where the problem space was defined more broadly than in European TV subtitling |
| 12:27 | <jgraham> | Hixie: So, for example, people have suggested putting hyperlinks in timed text to link to contextual information about the video content |
| 12:27 | <annevk> | Hixie, http://en.wikipedia.org/wiki/SubRip#Fields simply says "HTML" |
| 12:27 | <annevk> | Hixie, and has tags that are from HTML, literally |
| 12:27 | <Hixie> | jgraham: people have been talking about hyperlinks in video for decades, but i've yet to see anyone actually want to click a hyperlink while watching video |
| 12:28 | <hsivonen> | Hixie: Ogg Kate is different, but its implementation isn't suitable for importing into Gecko |
| 12:28 | <Hixie> | annevk: yeah, and imho that's a terrible design :-) |
| 12:28 | <deane> | hsivonen: OK, I guess it's just a case of persevering with the present system then, thanks. |
| 12:28 | <annevk> | Hixie, http://www.w3.org/TR/2006/CR-ttaf1-dfxp-20061116/#content-element-vocabulary also has tags copied from HTML |
| 12:28 | <Hixie> | hsivonen: why would Gecko need to do any of this? Surely the video subsystem should take care of this so it'll work in all video players on the system. |
| 12:29 | <annevk> | Hixie, whatever, but we're not seeing HTML because that's we're used too |
| 12:29 | <annevk> | Hixie, apparently the people working on subtitles see a need for HTML features |
| 12:29 | <jgraham> | Hixie: Well it is a problem statement "link to contextual information in video". And if it is hard to do it's not so surprising that people haven't done it so much |
| 12:29 | <Hixie> | annevk: they have similar names, but that's about all that seems to be like html in that list |
| 12:29 | <hsivonen> | Hixie: I'm pretty much in agreement with you about what people need. However, I also recognize that creeping featurism looks inevitable and needs to be guided to a path that leads to least code duplication. |
| 12:30 | <annevk> | Hixie, they have identical functions, for one |
| 12:30 | <Hixie> | you don't combat creeping featuritism by throwing more features at the solution than will ever be needed in the history of mankind |
| 12:30 | <nessy> | ups, looks like I missed a discussion |
| 12:30 | nessy | reads up |
| 12:30 | <hsivonen> | Hixie: when looking for a solution for Web video, it's pretty relevant how it suits Gecko, since Gecko is a significant browser engine |
| 12:31 | <annevk> | Hixie, http://www.w3.org/TR/2006/CR-ttaf1-dfxp-20061116/#styling-attribute-vocabulary is identical to CSS/XSL |
| 12:31 | <Hixie> | annevk: your definition of identical is interesting :-) |
| 12:31 | <Hixie> | hsivonen: i don't understand the adjective "web" in "web video". |
| 12:31 | <hsivonen> | Hixie: video that the user primarily encounters through a Web browser |
| 12:32 | <Hixie> | hsivonen: i don't know of any content that falls into that category. |
| 12:32 | <Hixie> | (and i work for youtube) |
| 12:33 | <jgraham> | Hixie: How is Youtube primarilly encountered then? |
| 12:34 | <nessy> | I think you guys are vastly underestimating the use of time-aligned text that goes beyond simple subtitles |
| 12:34 | <nessy> | srt may solve 50% of the needs |
| 12:34 | <Hixie> | it is encountered in a variety of media and i think any current predominance of a particular outlet is more a function of ease of deployment than anything to do with that outlet itself. |
| 12:34 | <nessy> | nowhere near 99% |
| 12:34 | <nessy> | all of the video sites are starting to have annotation tags and speech bubbles and the like |
| 12:34 | <hsivonen> | nessy: depends on whether you count use cases or use cases weighed by frequency of occurrence |
| 12:35 | <Hixie> | for example, if youtube was widely available on TVs, then people would view it there at least as much as on the Web, maybe even more. |
| 12:35 | <Hixie> | same with small devices. |
| 12:35 | <nessy> | since nobody has done an actual count, I don't think your 99% stands up |
| 12:35 | <hsivonen> | Hixie: how do you follow links from IRC to your TV to view youtube? |
| 12:36 | <Philip`> | jgraham: I've seen some TV programmes include clips from YouTube, which means millions of viewers each time |
| 12:36 | <Hixie> | hsivonen: most users aren't finding videos from IRC |
| 12:36 | <nessy> | if your tv is just a mythtv box like mine, easy :) |
| 12:36 | <hsivonen> | nessy: ok. note that I quite like the HTML idea as long as it's defined to use the same parsing as innerHTML instead of something else |
| 12:37 | <jgraham> | Hixie: The obvious way to deploy Youtube on TV would be via something like an embedded web browser |
| 12:37 | nessy | goes to read up on innerHTML |
| 12:37 | <Hixie> | also, i have to say that the idea of reusing the html syntax in another specification strikes me as a terribly bad idea, given how screwed up html's syntax is. |
| 12:37 | <hsivonen> | nessy: for the ogg-muxed packets, that is |
| 12:37 | <takkaria> | let's use XML instead |
| 12:38 | <Hixie> | jgraham: appletv isn't an embedded web browser. |
| 12:38 | <hsivonen> | using XML sucks more due to the namespace mapping context when parsing fragments |
| 12:38 | <hsivonen> | and due to XML spec itself not specifying fragment parsing |
| 12:38 | <Philip`> | Why does it have to use fragments? |
| 12:38 | <nessy> | Hixie: your alternative is TimedText |
| 12:38 | <annevk> | hsivonen, why fragments? |
| 12:38 | <Hixie> | nessy: alternative to what? |
| 12:38 | <jgraham> | Hixie: For all its faults, HTML is one of the most widely authored languages in the world, so "too hard to author" seems like a weak argument |
| 12:39 | <hsivonen> | annevk: for seeking |
| 12:39 | <Hixie> | jgraham: i didn't say too hard to author :-) |
| 12:39 | <hsivonen> | annevk: in the Ogg-multiplexed case |
| 12:39 | <nessy> | alternative to using plain-old html to specify time-aligned text |
| 12:39 | <hsivonen> | annevk: for separate HTML files, the normal parse mode is fine |
| 12:39 | <jgraham> | Hixie: Well what is the problem with the syntax if it is easy for authors and easy to parse? |
| 12:39 | <Hixie> | jgraham: but propagating design errors into more specs just because "well one day subtitles might need <iframe> and <eventsource>" seems like a terrible way of doing things |
| 12:40 | <Hixie> | jgraham: it's clearly not easy for authors, given how widespread syntax errors are. |
| 12:40 | <Hixie> | jgraham: but that isn't the problem |
| 12:40 | <nessy> | so you prefer to create a new spec that just picks the stuff that we currently deem as relevant out of HTML |
| 12:40 | <nessy> | ? |
| 12:40 | <Hixie> | jgraham: the problem is that it is an inconsistent disaster :-) |
| 12:41 | <nessy> | works fairly well for the web :) |
| 12:41 | <Hixie> | nessy: i don't think there's anything in HTML that really makes sense for timed text tracks |
| 12:41 | <nessy> | what about YouTube's annotation tracks? |
| 12:41 | <nessy> | what about the speech bubbles you can overlay onto videos? |
| 12:42 | <jgraham> | (FWIW I would be fine with having a document conformance class for this use case without <eventsource> and other stuff that could be added later if needed. Of course desktop UAs would just use their ordinary engine so people might use those things anyway, but at least that would demonstrate a use case for them) |
| 12:42 | <nessy> | what about music video lyrics, that are displayed alongside the video and scrolled down? |
| 12:42 | <jgraham> | Hixie: As opposed to XML which is a consistent disaster :) |
| 12:42 | <Hixie> | nessy: what about them? |
| 12:42 | <Hixie> | jgraham: i never suggested XML :-) |
| 12:43 | <nessy> | and karaoke |
| 12:43 | <nessy> | they all need some kind of formatting |
| 12:43 | <nessy> | and hyperlinks inside them |
| 12:43 | <Philip`> | They need a little bouncing ball along the words too |
| 12:43 | <Hixie> | nessy: HTML doesn't have formatting. |
| 12:43 | <nessy> | and background images |
| 12:43 | <Hixie> | nessy: and i'm not at all convinced of the need for hyperlinks. |
| 12:43 | <Hixie> | nessy: nor background images. |
| 12:43 | <nessy> | ask any advertiser |
| 12:43 | <jgraham> | Hixie: Sure but many people would |
| 12:43 | <nessy> | speech bubbles are background images on text |
| 12:43 | <Hixie> | show me a successful advertisement on video and i'll reconsider. |
| 12:44 | <Hixie> | jgraham: then argue against them :-) |
| 12:44 | <nessy> | almost all video hosting sites live off overlay advertisements |
| 12:44 | Philip` | has seen an overlay advert popping up on a video on YouTube |
| 12:44 | <nessy> | as much as banner ads and the like |
| 12:44 | <Hixie> | nessy: we'll see how many survive the depression :-) |
| 12:44 | <Philip`> | (It was really annoying and distracted me from the video for a couple of seconds while I looked for the 'x' button) |
| 12:44 | <takkaria> | how did we got from subtitling to overlaying ads? |
| 12:44 | <jgraham> | Hixie: Most films and TV shows are *full* of successful advertisments on video. Not ones you could obviously click on admittedly |
| 12:45 | <nessy> | overlaying ads are also time-aligned text |
| 12:45 | <Hixie> | jgraham: i rest my case |
| 12:45 | <takkaria> | right, but there are different use-cases here |
| 12:45 | <nessy> | since they are shown on TV, clicking would be a problem :) |
| 12:45 | <Philip`> | Movie teaser trailers often show a URL at the end, and it'd be nice if you could visit that without manually typing it into your browser |
| 12:45 | <takkaria> | one is ads, one is subtitles |
| 12:45 | <jgraham> | but I can see advertisers _killing_ for the ability to click on their product when it appears on the screen and be taken to their website |
| 12:45 | <nessy> | takkaria: they can all be solved with the same technology |
| 12:45 | <hsivonen> | Hixie: Revver's business is based on putting hyperlinks into videos |
| 12:46 | <takkaria> | it doesn't necessarily make sense to use the same technology to solve them, though |
| 12:46 | <takkaria> | for example, subtitling is a fairly well-defined problem |
| 12:46 | <Hixie> | jgraham: if 80% of people watching videos would at some point or other be actually interested in clicking on one of those links, then fair enough. i'm not at all convinced that even 5% of people would ever be so interested. |
| 12:46 | <nessy> | takkaria: if you analyse the cases thoroughly, you will notice that they have more incommon than different and that it makes sense to use the same technology to solve them |
| 12:46 | <takkaria> | and a format that tried to only cater for subtitles, it could do quite well at it |
| 12:47 | <nessy> | Hixie: many people click on these ads |
| 12:47 | <takkaria> | but if you want ads with speech bubbles and scrolling text, it seems like the kind of thing you just want to do using JS |
| 12:47 | <nessy> | probably more than on banner ads |
| 12:47 | <Hixie> | hsivonen: i'm not working on standards to support business models, i'm working on standards to help users |
| 12:47 | <nessy> | takkaria: javascript together with a time-aligned text spec, yes |
| 12:47 | <Hixie> | Philip`: the entire movie teaser trailer should be a link, not just the url at the end. Put the <video> in an <a>. |
| 12:48 | <nessy> | Hixie: it's not about business models, it's about use cases |
| 12:48 | <nessy> | and we are talking about more than just ads here |
| 12:48 | <jgraham> | Hixie: I'm not sure. It seems like Philip`'s example of a movie trailer with a URL at the end is quite reasonable and >5% of people would click on one of those at some point |
| 12:49 | <annevk> | we might use this for e.g. standards suck to provide URLs to the stuff we're talking about |
| 12:49 | <hsivonen> | Hixie: well, if HTML5/Ogg videos don't support insterting a linked ad, content providers will stick to Flash and QuickTime, which means users still have to deal with the badness of plug-ins |
| 12:49 | <jgraham> | (and on something like YouTube you don't necessarily have access to the surrounding page) |
| 12:49 | <nessy> | using innerHTML, CSS and javascript to solve the time-aligned text problem seems simple and elegant to me |
| 12:50 | <takkaria> | annevk: I would have thought it would be easier just to include the links in the post containing the video, no? |
| 12:51 | <nessy> | almost like a recursive function (from an elegance view-point) :) |
| 12:51 | <annevk> | takkaria, that would give away spoilers |
| 12:51 | <nessy> | takkaria: then you cannot have the links turn up at the time that they are relevant |
| 12:51 | <Hixie> | hsivonen: i'd be willing to take that risk, frankly, because i don't believe that enough people care to make a difference. |
| 12:52 | <nessy> | it's not just about Ogg, btw |
| 12:52 | <nessy> | it's also about the external files that contain time-aligned text |
| 12:52 | <nessy> | if we get to a format that is web-friendly, web developers will easily be able to author them |
| 12:53 | <Hixie> | anyway. Both the Timed Text W3C draft and HTML are suggestions that go completely overboard. One can provide much more than even SRT without going all the way to either of those. |
| 12:53 | <nessy> | so far, srt seems the only sensible format, but it is far too incapable for any richer need |
| 12:53 | <Hixie> | specs have to be built in baby steps, well-defined each step of hte way, only trying to address the 80% cases each time, if they are to be truly successful. |
| 12:54 | <nessy> | yes, the first web page was also far simpler than any current web page |
| 12:54 | <nessy> | that's why there are so many time-aligend text formats |
| 12:54 | <nessy> | each one with a bit more complexity |
| 12:54 | <hsivonen> | Hixie: so from a POV of a browser vendor, the options are roughly: 1) use timed plain strings (original srt; no extension path without the RSS <title> hell), 2) Use W3C TimedTimed or derivative, 3) Use the CSS formatter you already have, 4) Use an unproven single-implementation format like Kate or 5) design a new non-CSS/HTML format |
| 12:54 | <nessy> | and with TimedText we have reached a very complex level for a reason |
| 12:55 | <Hixie> | hsivonen: i don't understand why a browser vendor would ever have to even remotely be looking at implementing any of these proposals. |
| 12:55 | <Hixie> | hsivonen: it should be implemented at the same layer as the video codec. |
| 12:55 | <webben> | IIRC the main problem with TimedText I saw was that it attached categories to text without enforcing any meaning to those categories. |
| 12:55 | <webben> | which was just bonkers. |
| 12:55 | <webben> | maybe they've fixed that now |
| 12:56 | <annevk> | nessy, btw http://wiki.xiph.org/index.php/Timed_Divs_HTML#2._The_div_element says it matches HTML5 but HTML5 uses a point as a microsecond separator and not a comma so the examples are wrong |
| 12:56 | <nessy> | Hixie: almost nobody wants timed-aligned text inside video files |
| 12:56 | <webben> | nope, not fixed: http://www.w3.org/TR/2006/CR-ttaf1-dfxp-20061116/#metadata-attribute-role |
| 12:56 | <hsivonen> | Hixie: you can't pick stuff that doesn't exist off the shelf |
| 12:56 | <nessy> | annevk: I created that page today! nobody apart from hsivonen has even seen it - wow! |
| 12:56 | <nessy> | will fix it :) |
| 12:57 | webben | would love to know what the difference between 'dialog', 'transcription', and 'caption' is supposed to be. |
| 12:57 | <Hixie> | nessy: what do people suggest happens to the subtitles when the user saves the video? |
| 12:57 | <nessy> | same as what happens to images and html pages |
| 12:57 | <nessy> | all go into a directory |
| 12:58 | <Hixie> | annevk: HTML5 has a microsecond separator? i thought we dropped that. |
| 12:58 | <Hixie> | nessy: and how do those subtitles get played when i load quicktime, then? |
| 12:58 | <annevk> | Hixie, http://www.whatwg.org/specs/web-apps/current-work/multipage/infrastructure.html#times suggests seconds can have a fractional part |
| 12:59 | <Hixie> | annevk: oh, fractional seconds, ok. those aren't necessarily microseconds. |
| 12:59 | <nessy> | it depens on your player |
| 12:59 | <Hixie> | quicktime |
| 12:59 | <Philip`> | (Hmm, 00:99:00 is a valid time string?) |
| 12:59 | <hsivonen> | Hixie: basically browsers are now driving the need for timed text for Ogg, so browser vendors don't have existing Ogg stack code to reuse |
| 12:59 | <annevk> | Hixie, true, the conversion is indeed slightly more involed |
| 13:00 | <annevk> | nessy, ^^ |
| 13:00 | <hsivonen> | Hixie: also, wouldn't you want to have the full Unicode goodness of Gecko for line layout instead of shipping duplicate code for it? |
| 13:00 | <Hixie> | hsivonen: so stop considering yourself as a browser implementor and just add the code to ogg :-) |
| 13:00 | <nessy> | the quicktime player is not as capable as for example vlc or mplayer, who load external srt and similar files to sync upon playback |
| 13:01 | <Hixie> | hsivonen: the OS already has full Unicode goodness for line layout. |
| 13:01 | <hsivonen> | Hixie: so would the Ogg stack merely blit pixels as far as Gecko is concerned? what about exposing the textual data to JS and AT? |
| 13:01 | <annevk> | Philip`, seems like a bug in the spec |
| 13:01 | <Hixie> | hsivonen: we don't need pixel-perfect precision for subtitles. Heck we don't even need the same font or color to be used. |
| 13:02 | <Hixie> | Philip`: oops. i'll file a bug. |
| 13:02 | <nessy> | that's like saying you don't care if a web page author created a red heading - you display the heading in whichever colour you like |
| 13:03 | <nessy> | also, we really want to expose the time-aligend text so crawlers and other engines can deal with it |
| 13:03 | <Hixie> | crawlers can read movie files. |
| 13:04 | <nessy> | a web crawler should not have to |
| 13:04 | <Hixie> | i don't see why not |
| 13:04 | <Hixie> | my point with the subtitles is that subtitles shouldn't support formatting to that level of detail at all |
| 13:04 | <webben> | Hmm. |
| 13:04 | <nessy> | there is a need for it |
| 13:04 | <webben> | Captioning positioning may be important however. |
| 13:04 | <hsivonen> | Hixie: doing too much on the Ogg stack side is actually an argument *against* Kate |
| 13:04 | <Hixie> | no there isn't :-) |
| 13:04 | <Hixie> | nessy: what is the need? |
| 13:05 | <nessy> | and we are talking about more than subtitles - all other time-aligned text types that are similar |
| 13:05 | <hsivonen> | Hixie: partially because it doesn't do it on all platforms that Gecko does text on |
| 13:05 | <webben> | e.g. TV captions are sometimes in the video, not beneath or above it. |
| 13:05 | <Hixie> | webben: position was on my list of the few things i'd expect |
| 13:05 | <nessy> | Hixie: now you're just playing devil's advocate |
| 13:06 | <webben> | Hixie: Ah okay, by pixel perfect you mean what? text smoothing/hinting type things? |
| 13:06 | <nessy> | so to add to my above list: |
| 13:06 | <webben> | isn't that important to legibility? |
| 13:06 | <Hixie> | nessy: i honestly don't think that there are valid use cases for formatting information in timed text, but would welcome contrary evidence. |
| 13:06 | <Philip`> | Hixie: Crawlers can't read movie files, because http://youtube.com/robots.txt says "Disallow: /get_video" |
| 13:06 | <Hixie> | webben: i mean, i don't think it matters what the line height is or what font is used, so long as the text is roughly in the right place. |
| 13:07 | <nessy> | it has been shown that captions that are formatted in particular ways (and there are guidelines by captioning organisations) will help deaf people understand things better faster and more in-depth |
| 13:07 | <nessy> | karaoke text needs specific formatting |
| 13:07 | <webben> | Hixie: Well. There either needs to be semantic information (different speakers) or presentational info. |
| 13:07 | <webben> | However, semantic info would be better, since then it can be formatted by the UA. |
| 13:07 | <webben> | @nessy's point. |
| 13:07 | <nessy> | speech bubbles need positioning and a background image, as well as appropriately formatted text |
| 13:07 | <Hixie> | nessy: sure, that's an argument for including particular classes of spans, as i mentioned; that formatting should certainly _not_ be hardcoded in the file. |
| 13:08 | <nessy> | in which file? |
| 13:08 | <Hixie> | the timed text data |
| 13:08 | <nessy> | so it should be in a css file and javascript accompanying this file? |
| 13:09 | <webben> | http://www.w3.org/TR/2006/CR-ttaf1-dfxp-20061116/#metadata-vocabulary-agent ... timed text can differentiate speakers at least. |
| 13:09 | <Hixie> | nessy: why does karaoke text need formatting? (and why do we need to handle karaoke? that seems like a rather specific case.) |
| 13:09 | <nessy> | it's one of the use cases, and not a small one in certain parts of the world |
| 13:09 | <Hixie> | nessy: what's the use case for speech bubbles? |
| 13:09 | <Hixie> | nessy: what? no css and js, please! |
| 13:09 | <webben> | Hixie: Showing who is speaking, I guess. |
| 13:10 | <nessy> | and singing along to karaoke, colouring words as their time comes is the best way |
| 13:10 | <nessy> | and speech bubbles are annotations - youtube has them and most other hosting sites have them - they add to the message that an author can provide about their videos |
| 13:10 | <webben> | Hixie: see also http://googlesystem.blogspot.com/2008/06/youtube-annotations.html |
| 13:10 | <nessy> | people use them, so there is the use case |
| 13:10 | <Hixie> | nessy: i have yet to see an karaoke video that colours the words in a way that actually leads to even half-decent singing, so if it's the best way, we're doomed. |
| 13:11 | <nessy> | Hixie: where do you put the formatting information if not into the text file hard-coded and not into css? |
| 13:11 | <Hixie> | nessy: people use lots of things, that doesn't mean we have to support them |
| 13:11 | <webben> | Hixie: They may be an improvement over just positioning the text in different places. |
| 13:11 | <hsivonen> | nessy: btw, what's your plan for timed coloring? |
| 13:11 | <Hixie> | nessy: in particular, i would not expect youtube-style annotations to ever be implemented using timed text features |
| 13:11 | <nessy> | hsivonen: javascript? :-) |
| 13:12 | <Philip`> | Subtitled foreign films often have positioned English subtitleish text overlaying the original foreign text that's encoded in the video, and those require reasonable control over position and size to avoid the bits of English text overlapping each other |
| 13:12 | <Hixie> | nessy: no formatting information. let the UA do it. |
| 13:12 | <nessy> | and the UA should get it from a CSS |
| 13:12 | <nessy> | it should definitely not fall from the sky |
| 13:12 | <Hixie> | good lord no |
| 13:12 | <Hixie> | the whole point is taht the formatting has to follow the guidelines |
| 13:12 | <Hixie> | don't let the author style it |
| 13:13 | <hsivonen> | nessy: wouldn't you then need time-based callbacks inside the captioning browsing context, too |
| 13:13 | <Hixie> | that's a recipe for disaster |
| 13:13 | <nessy> | ok, let's stop all css and not style any web pages any more either - the guidelines we have for styling web pages are better than any author will ever do it, right? |
| 13:13 | <Hixie> | what guidelines? |
| 13:13 | <nessy> | the ones that existed when the web was created |
| 13:14 | <nessy> | ok, let's take a step back |
| 13:14 | <Hixie> | i wasn't aware that there were any |
| 13:14 | <webben> | I wonder if one could cater to UAs either positioning captions or providing speech bubbles by allowing content creators to specify origin points for text. |
| 13:15 | <webben> | I guess it depends on whether TV captions are positioned out of the way of the action or near to the speaker. |
| 13:15 | <Hixie> | regarding the youtube annotations -- it is far too early to standardise them. the youtube folk are inventing new things every week, you'd never be able to keep up, and they'd end up never using what the timed text spec supported. |
| 13:16 | <webben> | I also wonder whether it's possible to detect where the action is in a set of frames and position captions appropriately. I suspect not. |
| 13:16 | <Hixie> | they'll just overlap actual html or svg or canvas or flash or whatever on top of the video, with the data in a custom format fetched by xhr or some such. |
| 13:16 | <hsivonen> | jgraham: I've now verified that Gecko with HTML5 parsing does not build on Linux at this time |
| 13:16 | <webben> | (e.g. superman flying - superman is still, the sky is moving, but you probably want to see superman) |
| 13:16 | <hsivonen> | (with the default build settings) |
| 13:16 | <nessy> | Hixie: I have spoken to the YouTube guys and they would love to have a standard available to provide them with the functionality |
| 13:16 | <Hixie> | who'd you speak to? |
| 13:17 | <Hixie> | that's not what i've heard |
| 13:17 | <nessy> | Ken and the guys around him |
| 13:18 | <nessy> | when they implemented time-aligned text functionality, it was a big question to use which format |
| 13:18 | <nessy> | when they implemented time offset addressing, they didn't know what to use |
| 13:18 | <nessy> | they are grateful that somebody is looking into standards |
| 13:18 | <nessy> | that's what I heard |
| 13:19 | <Hixie> | Ken Harrenstien? |
| 13:19 | <nessy> | also, speech bubbles for example are not just a youtube phenomenon - many other sites use them |
| 13:19 | <nessy> | yup, that Ken |
| 13:21 | <nessy> | anyway - we'll implement srt and experiment with other stuff |
| 13:21 | <hsivonen> | Hixie: why would it be less disastrous for sites to overlay HTML+CSS onto the frame from a script in the containing document than browser C++ code driving innerHTML setting on the overlay? |
| 13:22 | <hsivonen> | nessy: btw, is there a plan on what line layout code will be used for srt and what glyph rasterization code? |
| 13:22 | <Hixie> | hsivonen: because with the latter you are importing all of HTML and CSS into a format that will be reused outside browsers. |
| 13:23 | <hsivonen> | Hixie: in the former, you just can't take the stuff out of a browser. how's that better? |
| 13:24 | <Hixie> | hsivonen: it allows for conventions to arise and implemenetation experience to be collected so that the eventual standard can be designed sanely and with a suitably limited scope that still handles the important use cases |
| 13:24 | <Hixie> | hsivonen: it's the same reason we often don't add features to html5 until there are cowpaths to pave |
| 13:25 | <Hixie> | video on the web is very young |
| 13:25 | <Hixie> | video annotation on the web is in its infancy |
| 13:25 | <Hixie> | it's one thing to be writing standards for things like subtitles, which are well understood and mature |
| 13:25 | <Hixie> | but things like timed annotations are far too new to be standardised successfully |
| 13:26 | <hsivonen> | Hixie: some use cases (captioning for the deaf) aren't waiting. if browsers don't deliver native support soon, we'll be stuck with JS-based captioning, in which case your goals about moving the video with captions outside the browser fail |
| 13:26 | <nessy> | hsivonen: the srt layout question is unsolved at this stage - I hope to solve this with your and doublec's help |
| 13:27 | <Hixie> | hsivonen: right, that's why we should develop standards today for subtitles, which are well understood and mature, and which are incredibly simple to do compared to animated, interactive, formatted, web pages. |
| 13:27 | <webben> | Hixie: I guess one way of harmonizing this stuff would be to allow extraction of the data with the HTML5 video API, so you don't need to make an XHR call if you want to pull some sort of SVG/Canvas stunt. |
| 13:28 | <webben> | Hixie: where subtitles includes captions? |
| 13:28 | <Hixie> | Most of the world does not distinguish captions from subtitles. |
| 13:29 | <hsivonen> | webben: what would the API return when the data isn't a naked DOMString? |
| 13:29 | <nessy> | Hixie: there is a huge gap between what the subbing community considers as good subtitles and what professional captioners regard as good subtitles - and it has a lot to do with formatting |
| 13:29 | <Hixie> | nessy: could you give an example of something where that formatting choice is not a choice the user agent could make? |
| 13:30 | <nessy> | Hixie: in fact, the part of the world that cares about captions, very much distinguishes between subtitles and captions, but let's not go down that path, too :) |
| 13:30 | <nessy> | in professional captions, you will find that different speakers are given different colours |
| 13:30 | <hsivonen> | Hixie: George Lucas uses yellow subtitles with black outline when the TV convention over here is to use white text with black outline |
| 13:31 | <hsivonen> | I guess Lucas is special and gets to master the subtitles where others don't care |
| 13:31 | <gsnedders> | That rebel! |
| 13:31 | <nessy> | so if a speaker is off-screen, you don't have to provide the name of the person (which incidentally may not be known) to refer to the speaker |
| 13:31 | <Hixie> | hsivonen: right, so not providing formatting choice would have improved matters. |
| 13:32 | <webben> | hsivonen: Depends what format is chosen, I guess. e.g. Timed Text DOM, or a DOM of whatever format Hixie has in mind. |
| 13:32 | <nessy> | also, if there are two speakers on screen talking fast, professional captions often create two different boxes with text and position them close to the person that actually said that text |
| 13:32 | <Hixie> | nessy: yes, my list of features did indicate that the text should be able to specify classes for each span of text, which the UA can then style accordingly (e.g. picked colours that the viewer can distinguish) |
| 13:32 | <hsivonen> | <devilsadvocate>Should all Web documents be rendered in Computer Modern for consistency?</devilsadvocate> |
| 13:32 | <Hixie> | nessy: positioning was also on my list |
| 13:32 | <nessy> | also, when there is already on-screen text at the bottom of a screen, professional captions end up at the top of the screen or some place where they don't obstruct important information |
| 13:33 | <webben> | hsivonen: If you prefer Computer Modern, maybe they should! |
| 13:33 | <Hixie> | hsivonen: web pages and subtitles are totally different. subtitles are more akin to the <title> or title="" text, and we don't provide any control over those (and rightly so). |
| 13:33 | <nessy> | lol |
| 13:33 | <nessy> | Hixie: I disagree |
| 13:33 | <nessy> | 10 years ago I would have said the same |
| 13:34 | <nessy> | but I have worked with professional captioners, I have created time-aligned text formats, and I have watched the Web2.0 space pick up on different types of time-aligend text |
| 13:34 | <nessy> | time-aligned text has almost as many requirements as web pages |
| 13:35 | <Hixie> | you have yet to give an example that shows that :-) |
| 13:35 | <nessy> | even picture-in-picture could be solved by having video inside text |
| 13:35 | <nessy> | video elements that is |
| 13:35 | <nessy> | but it's not something I'd advocate easily, since the synchronisation is a different matter |
| 13:36 | <nessy> | ok, let's take ticker text |
| 13:36 | <nessy> | ticker text often has images inside it |
| 13:36 | <nessy> | it has text in different fonts |
| 13:36 | <nessy> | text that may be fixed or run across the screen |
| 13:37 | <Hixie> | why would you put ticker text into the same solution as captions? |
| 13:37 | <Hixie> | ticker text has nothing to do with the video |
| 13:37 | <nessy> | because it is time-aligned text |
| 13:37 | <Hixie> | ticker text is pretty much exactly the opposite of time aligned text |
| 13:37 | <nessy> | yes it does - if the program changes, the text changes |
| 13:37 | <nessy> | if an ad comes, ticker text stops |
| 13:37 | <Hixie> | if the program changes, the video changes :-) |
| 13:37 | <nessy> | it is very much aligned to a segment of programming |
| 13:38 | <Hixie> | well in any case if we could kill ticker text forever, i think we'd have done the world a great favour |
| 13:38 | <Hixie> | so let's move on to the next use case |
| 13:38 | <nessy> | I have a whole bunch of other stuff that I'd like to put in that bucket - many html and css features |
| 13:39 | <nessy> | doesn't mean because we don't like it, the world doesn't need it |
| 13:39 | <Hixie> | please do send such feedback to the list, i'm trying to drop as much as possible in html5 |
| 13:39 | <nessy> | blinking text!? |
| 13:40 | <nessy> | javscript that doesn't let you leave a webpage? |
| 13:40 | <nessy> | anyway ... I digress |
| 13:40 | <hsivonen> | ticker plus blink is the new blinquee |
| 13:40 | nessy | shudders |
| 13:40 | <Hixie> | <blink> is dead in html5 and text-decoration:blink has always been optional |
| 13:41 | <nessy> | good |
| 13:41 | <Hixie> | javascript that doesn't let you leave a webpage is as far as i know impossible, and html5 doesn't allow it (in fact it goes to some lengths to prevent it) |
| 13:41 | <Hixie> | (in fact <blink> has never been in any html standard) |
| 13:42 | <annevk> | <marquee style=text-decoration:blink> |
| 13:44 | <nessy> | I came across a spam web page recently that always brought up this dialog box upon browsing away from it which you were only able to say "no" on - had to do some firebug magic to make it go away |
| 13:45 | <Hixie> | might have been a browser bug |
| 13:45 | <Hixie> | if you come across it again, please do let us know the url |
| 13:47 | <nessy> | ok, will |
| 13:47 | <nessy> | thought I still had it, but can't find it any more |
| 13:49 | <nessy> | ah, here it is - http://scan4plus.com/22/?uid=165 |
| 13:50 | <Hixie> | what browser? |
| 13:51 | <nessy> | had it in firefox and safari on the mac |
| 13:51 | <nessy> | it forces you to download an install.exe file, bascially |
| 13:51 | <Hixie> | i have no trouble closing it on safari |
| 13:51 | Hixie | tries firefox |
| 13:51 | <nessy> | I think it has a little delay effect |
| 13:52 | <nessy> | if closed immediately, it's fine - but after a while it becomes a nuisance |
| 13:52 | <nessy> | but I haven't investigated much |
| 13:53 | <Hixie> | i waited and in both cases it eventually prompted with a dialog that i hit cancel on and then could close without issue |
| 13:53 | <Hixie> | but anyway |
| 13:53 | <nessy> | my cancels were ignored |
| 13:55 | <Hixie> | odd |
| 13:55 | <Hixie> | oh well |
| 13:55 | <Hixie> | bed time |
| 13:55 | <Hixie> | nn |
| 13:56 | <webben> | nn Hixie |
| 13:57 | <nessy> | yeah - long past my bedtime |
| 13:57 | <nessy> | but at still 32 degrees, I don't really feel like sleep yet |
| 13:58 | <nessy> | nn Hixie :) |
| 14:02 | <annevk> | what do you call window.XMLHttpRequest ? |
| 14:02 | <gsnedders> | window.XMLHttpRequest? :P |
| 14:03 | <annevk> | you can go stand in the corner |
| 14:03 | <gsnedders> | Seeming I have two working legs, I'd tend to agree with that statement. |
| 14:05 | <Philip`> | But you might be standing in a circular room |
| 14:07 | <gsnedders> | I, however, am not. |
| 14:08 | <jgraham> | There might be more than one corner |
| 14:09 | <Philip`> | You might be outdoors |
| 14:09 | <Philip`> | Or you might be indoors in a room shaped like a Klein bottle, in which case you are still outdoors too |
| 14:10 | <annevk> | fail |
| 14:11 | jgraham | imagines gsnedders traped inside an Escher print |
| 14:12 | gsnedders | doesn't know which way gravity is acting on him now |
| 14:13 | <Philip`> | Gravity is down |
| 14:14 | <jgraham> | Unless you are observing youself from a distance |
| 14:15 | <gsnedders> | Philip`: No, there is a gravitational force acting on me in every possible direction, silly |
| 14:17 | Philip` | finds a link to a Wikipedia image in a (dead-tree) book |
| 14:17 | <Philip`> | which is quite user-unfriendly, and also seems rather a fragile thing to put onto paper |
| 14:24 | <gpy> | please whatwg, dont remove javascripts method .blink() |
| 14:24 | <annevk> | there's such a thing? |
| 14:25 | <annevk> | anyone here willing to run http://xmlhttprequest.testsuite.org/dump/2009/001.htm and http://xmlhttprequest.testsuite.org/dump/2009/002.htm in IE? |
| 14:25 | <annevk> | please give IE version number and output string |
| 14:25 | <annevk> | and the last 3 digits of the URI tested |
| 14:27 | <hsivonen> | rubys: I think I'm going to try to relax GCC errors overall first before trying to do it on a per directory or file basis |
| 14:28 | <rubys> | relax errors? Why not simply fix them? |
| 14:28 | <rubys> | My compile has started, I should be able to help shortly |
| 14:29 | <rubys> | annevk: ie8 beta2 shows ./support for both pages |
| 14:29 | <annevk> | cheers |
| 14:29 | <hsivonen> | rubys: the error of jumping past variable initialization should actually be harmless and fixing it would mean hoisting all variable declarations to the start of the method and initializing them to placeholder values |
| 14:29 | <annevk> | that at least works the same everywhere then :) |
| 14:30 | <Philip`> | hsivonen: I think the warning goes away if you just move the declarations up, and leave the variables uninitialised until the point where they're needed |
| 14:31 | <Philip`> | (i.e. you don't need placeholder values) |
| 14:31 | <rubys> | I'm getting what appears to be an unrelated error: /usr/include/gtk-2.0/gtk/gtkmenushell.h:117: error: expected ‘,’ or ‘...’ before ‘<expression error>’ |
| 14:31 | <Philip`> | (except to the extend that default constructors would provide placeholder values) |
| 14:32 | <hsivonen> | rubys: did you run the Ubuntu-specific build dependency apt-get incantations? |
| 14:32 | <rubys> | yup |
| 14:33 | <hsivonen> | then I have no idea why that happens :-( |
| 14:33 | <rubys> | weird... restarted the build and it picked up where it left off and continued on. |
| 14:34 | <hsivonen> | Philip`: but if I move the declarations up in Java, I'll waste cycles zeroing the variables needlessly, right? |
| 14:34 | <hsivonen> | not a big deal, probably |
| 14:35 | <rubys> | next error: /home/rubys/hg/html5parsing/content/html/parser/src/nsHtml5Parser.h:73: error: comma at end of enumerator list |
| 14:35 | <Philip`> | hsivonen: In Java I thought you could just leave stack variables uninitialised, and it's fine as long as the JVM can statically verify that you're not going to use it until it's been assigned |
| 14:35 | <hsivonen> | in any case, I'm reluctant to move the declarations in Java, because the variables would be declared totally out of context of use |
| 14:35 | <hsivonen> | rubys: already fixed in hg |
| 14:35 | <Philip`> | (and so it will just allocate stack space, which has zero cost) |
| 14:36 | <Philip`> | (though actually I suppose it'll all be JITted into unrecognisability) |
| 14:36 | <hsivonen> | Philip`: ok. well, then there's only the argument of losing context |
| 14:36 | <rubys> | hsivonen: is that a file that is produced by cpptranslate? |
| 14:36 | <hsivonen> | that is, the variables will appear less local to the reader than they are |
| 14:37 | <hsivonen> | rubys: no |
| 14:37 | <rubys> | ok, pulled and updated, restarting compile |
| 14:38 | <Philip`> | hsivonen: The warning also seems to go away if the variable is no longer in scope at the goto target, i.e. "goto x; { int n=1; }; x: ;" is fine, so is it easy to make your code do that? |
| 14:39 | <rubys> | ok, I think I'm now where you are: |
| 14:39 | <rubys> | /home/rubys/hg/html5parsing/content/html/parser/src/nsHtml5TreeBuilder.cpp:538: error: jump to case label |
| 14:39 | <rubys> | /home/rubys/hg/html5parsing/content/html/parser/src/nsHtml5TreeBuilder.cpp:488: error: crosses initialization of ‘PRInt32 currGroup’ |
| 14:40 | <hsivonen> | rubys: would you prefer moving the declarations or telling GCC not to treat it as an error? |
| 14:42 | <rubys> | I'm still trying to understand the error. Line 538 to me is not a goto but rather a label "default:" |
| 14:43 | <Philip`> | That looks like the label being goneto |
| 14:43 | <hsivonen> | rubys: I'm seeing defaults, too, on the relevant lines |
| 14:44 | <rubys> | I'm assuming that the real error is "starttagloop" |
| 14:44 | <hsivonen> | that's possible |
| 14:48 | <hsivonen> | this can be addressed by putting braces around the affected statements in a switch-case |
| 14:49 | <hsivonen> | eww. that looks ugly with my Eclipse settings |
| 14:53 | <hsivonen> | I'm trying to locate the relevant cases in Java and wrap the statement in cases in braces |
| 14:54 | <rubys> | cool |
| 14:54 | <rubys> | if there is any way I can help, let me know. |
| 14:55 | <rubys> | I've reproduced building the cpp from java (diffs from hg were nil at the time) and compiling firefox to the point of getting the same error you are seeing. |
| 14:56 | <hsivonen> | hmm. I wonder if I should just generate the braces around all switch-cases in C++ instead of using braces case-by-case in Java... |
| 14:59 | <rubys> | i see you've done a checkin... rebuilding |
| 15:00 | <rubys> | still see the errors |
| 15:10 | <hsivonen> | committed to hg & svn. partial build succeeded already |
| 15:11 | <hsivonen> | I opted not to change the parser Java source |
| 15:11 | <hsivonen> | instead, I changed the translator to insert braces |
| 15:11 | <rubys> | reasonable for now (at least until it causes a problem that needs to be addressed) |
| 15:12 | <rubys> | pulled, updated, and rebuilding now... |
| 15:12 | <rubys> | unrelated question: is there a high res version of http://about.validator.nu/icon.png ? |
| 15:12 | <hsivonen> | rubys: I don't have one |
| 15:12 | <rubys> | alternately, how was it created? |
| 15:12 | <hsivonen> | zcorpan: do you have a high-res version? |
| 15:13 | <hsivonen> | rubys: zcorpan contributed it. I don't know. |
| 15:13 | <rubys> | if I blow it up, it looks like 8 line segments, but the underlying intent may have been different. |
| 15:15 | <rubys> | I guess putting it another away: if the image were to be scaled up, how would you expect it to look? |
| 15:16 | <hsivonen> | I would expect the top of the V to be horizontally straight but the tip at the bottom to be rounded |
| 15:16 | <zcorpan> | hsivonen: no sorry |
| 15:16 | <hsivonen> | I'd expect the top left of n to be angled and top right round |
| 15:17 | <hsivonen> | and u to be like n rotated by 180 degrees |
| 15:18 | <hsivonen> | zcorpan: ok. does my understanding of rounding match yours? |
| 15:23 | <zcorpan> | hsivonen: yeah. the V is dark blue and the "nu" has a black blurry shadow with no offset |
| 15:23 | <hsivonen> | zcorpan: ok |
| 15:24 | <rubys> | build is progressing, when done, I'll post a howto on my blog, complete with an svg rendering of the icon. Feel free to critique and/or steal the icon when done. |
| 15:24 | <zcorpan> | rubys: cool |
| 15:24 | <hsivonen> | rubys: cool |
| 15:29 | <annevk> | Philip`, why is it hard btw to find start tags with (significant) trailing slashes in them? |
| 15:30 | <Philip`> | annevk: Because I'd need to write code |
| 15:31 | <Philip`> | annevk: and that code might have to be embedded in the tokeniser, which is not trivial to modify |
| 15:31 | <annevk> | Philip`, can you put it on a wishlist? |
| 15:31 | <Lachy> | unfortunately, I'm now back in Oslo, after a fantasic holiday in England |
| 15:34 | <zcorpan> | rubys: btw, i created it in photoshop, drawing the "V" with a line tool and pixel-size pen for the edges, and the "nu" with the text tool with some font i don't remember and then shadow on the "nu"-layer |
| 15:52 | <Philip`> | annevk: I don't really do lists, since whenever I've written a list in the past I've never read it again :-) |
| 16:06 | <annevk> | Philip`, guess I'll just keep bugging you with it then :) |
| 16:22 | <hsivonen> | jgraham: now I've confirmed that Gecko+HTML5 parsing builds on Linux :-) |
| 16:25 | <jgraham> | hsivonen: Nice :) |
| 16:26 | <gsnedders> | hmmm |
| 16:26 | gsnedders | wonders if there are any blog packages which store their data as Atom |
| 16:27 | <hsivonen> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=6357 ouch. that's not nice |
| 16:27 | <annevk> | toally agreed |
| 16:29 | <jgraham> | It seems like a different survey of web pages (with different sample selection) would be a generally good thing |
| 16:30 | <jgraham> | (also to enable different data to be collected) |
| 16:31 | <Philip`> | annevk: If you only want to look for one specific tag (like script), and don't mind missing any tag split onto multiple lines, then I could just grep for it fairly easily |
| 16:32 | <annevk> | if you could do just script that would help |
| 16:32 | <annevk> | at least for that bug |
| 16:33 | <olliej_> | morning annevk Philip` |
| 16:33 | <annevk> | more like dinner time :p |
| 16:33 | <olliej> | annevk: details :D |
| 16:35 | gsnedders | guesses even with Javascript on file:// you couldn't get a directory listing |
| 16:36 | <rubys> | http://intertwingly.net/blog/2009/01/05/Building-Firefox-with-HTML5-parser-support |
| 16:37 | <annevk> | nice logo |
| 16:38 | <annevk> | btw, given that you write Python, JDK and such, maybe you should also have HTML5 and Firefox rather than html5 and firefox? |
| 16:38 | <Philip`> | rubys: s/prerequites/prerequisites/ |
| 16:39 | <rubys> | fixed and fixed |
| 16:39 | <gsnedders> | Hmm. I need some language and UI that I can expect to run on any computer I am on, without compiling or installing anything |
| 16:39 | gsnedders | guesses Java would probably work |
| 16:40 | <Dashiva> | Java with bundled jre? |
| 16:40 | <rubys> | I could have done better with the logo if I wanted to go for fancy filter effects |
| 16:40 | <Philip`> | rubys: You don't have to exit all Firefox windows - you can run "firefox -no-remote" and it'll start a new version even if you're already running Firefox |
| 16:40 | <Philip`> | ...though you need to choose a different profile |
| 16:40 | <gsnedders> | Issue: I don't know Java. |
| 16:41 | <rubys> | how do you chose a profile from the command line |
| 16:41 | <Philip`> | gsnedders: I imagine you're kind of stuck if you're ever on a Windows computer, because they don't have anything if you don't compile or install it |
| 16:41 | <hsivonen> | rubys: -P profilename |
| 16:41 | <Philip`> | rubys: Or -profilemanager |
| 16:41 | <hsivonen> | -P on its own starts the profile manager |
| 16:42 | <Philip`> | Oh, that's easier |
| 16:42 | <Dashiva> | gsnedders: That's not an issue, it's an opportunity :) |
| 16:42 | <gsnedders> | Philip`: But what is the probability of Java not being installed? |
| 16:42 | <Philip`> | gsnedders: 100% if it's a recent version of Windows and nobody has installed Java onto it :-) |
| 16:42 | <Dashiva> | Bundle the runtime |
| 16:43 | <Philip`> | gsnedders: I suggest using HTML and JS and CGI, then you can access it from anywhere |
| 16:43 | gsnedders | could almost do this with HTML and JS |
| 16:43 | gsnedders | realizes other more major issues with making it cross-platform |
| 16:44 | <hsivonen> | rubys: linked from http://hsivonen.iki.fi/html5-gecko-build/ ; thanks |
| 16:45 | <rubys> | when I try with -P and -no-remote, I get the profile manager. I can create a profile, but when I select it I get an immediate segmentation fault. |
| 16:45 | <rubys> | for now, I'll keep with the instructions I've got :-) |
| 16:46 | <rubys> | being able to do this is just a sanity test for what I really want to work on... integrating with Python and/or Ruby. |
| 16:47 | <gsnedders> | What supports XML Encryption? |
| 16:47 | <hsivonen> | gsnedders: things like SAML identity brokers, IIRC |
| 16:47 | <annevk> | rubys, sweet |
| 16:48 | gsnedders | wonders whether he could use Flash for what he wants |
| 16:48 | gsnedders | notes Flash has the issue of being wholly closed |
| 16:48 | <rubys> | http://santuario.apache.org/ |
| 16:49 | <gsnedders> | rubys: So that's a library for dealing with encrypted XML? |
| 16:49 | <rubys> | http://santuario.apache.org/c/prog_enc.html |
| 16:51 | <hsivonen> | gsnedders: from what I've heard, dealing with XML encryption and signing is painful |
| 16:51 | <gsnedders> | hsivonen: Without making up my own file-format it seems the only way to do this |
| 16:51 | <hsivonen> | probably the runner up in # of over beer complaints after Namespaces |
| 16:52 | <gsnedders> | rubys: Thanks for those, links, though |
| 16:53 | <Philip`> | gsnedders: Making up your own file format might be sensible, since you only have to support 1 use case and not 80% |
| 16:54 | <gsnedders> | Philip`: But if there are already libraries, like what rubys linked to, then it may be less effort to use them |
| 16:55 | <hsivonen> | gsnedders: what's your use case? |
| 16:55 | <hsivonen> | gsnedders: that is, do you really need to encrypt an XML subtree instead of encrypting an entire XML file? |
| 16:55 | <gsnedders> | hsivonen: A diary on a memory-stick, which should be editable anywhere while simultaneously being encrypted. |
| 16:55 | <Philip`> | gsnedders: That depends on whether the cost of learning the library, and the language restrictions it imposes on you, are less than the cost of simply writing what you want |
| 16:55 | <gsnedders> | hsivonen: No, I can just encrypt entire Atom entry documents, so entire files. |
| 16:56 | <hsivonen> | gsnedders: seems like a use case for an encrypted sparse disk image |
| 16:56 | <hsivonen> | gsnedders: then you can use general-purpose crypto |
| 16:57 | <gsnedders> | hsivonen: I guess I'd avoid the issue of platform-specificy by having my own encrypting/decrypting stuff for it, but it'd probably be easier just to use something like tar encrypted as then I don't need to worry about OS differences about mounting images |
| 16:58 | <Philip`> | Maybe something like http://www.truecrypt.org/faq.php ("How can I use TrueCrypt on a USB flash drive?") |
| 16:59 | <gsnedders> | Philip`: That seemingly requires to be installed, though |
| 16:59 | <Philip`> | gsnedders: http://www.truecrypt.org/docs/?s=traveler-mode says it doesn't quite |
| 16:59 | <gsnedders> | Philip`: That says on Windows you need to be an admin |
| 16:59 | <Philip`> | gsnedders: Everyone on Windows is an admin :-) |
| 17:00 | <gsnedders> | Philip`: At school I'm not :P |
| 17:00 | gsnedders | wonders whatever happened to that idea of doing schoolwork today |
| 17:00 | <Philip`> | gsnedders: If you're not, that just means you haven't tried hard enough |
| 17:01 | <gsnedders> | Philip`: Laziness wins. |
| 17:01 | <Philip`> | You're not meant to actually respect the security restrictions imposed by schools - they're just there as a fun challenge |
| 17:01 | <Philip`> | gsnedders: Does the thing really need to be on a memory stick rather than an online service? |
| 17:02 | <gsnedders> | Philip`: It being online means trusting someone else with my data :P |
| 17:03 | <hsivonen> | gsnedders: not if the server is in your apartment and is booted with a crypto key that you remove after booting so that shutting it down makes it unbootable without key |
| 17:03 | <gsnedders> | hsivonen: That sounds like too much effort. |
| 17:04 | <Philip`> | gsnedders: It being offline means trusting everybody who has access to any of the hardware that you will use it with |
| 17:04 | <Philip`> | and it's probably difficult to have that much trust when you're in somewhere like a school |
| 17:05 | <gsnedders> | Philip`: I hope the computer doesn't keep it unencrypted in memory for too long |
| 17:05 | <Philip`> | so, in comparison, a hosting provider is much more trustworthy :-) |
| 17:05 | <Philip`> | gsnedders: All they need is a keylogger |
| 17:05 | <gsnedders> | Another advantage of per-file encryption: if there is data leakage, it is minimalized |
| 17:05 | <gsnedders> | Philip`: A key logger will still get around online stuff |
| 17:06 | <Philip`> | gsnedders: Set up a one-time password system |
| 17:06 | <gsnedders> | Philip`: And if I have the physical hardware on which it is on, then they can't get at it even with a keylogger |
| 17:06 | <Philip`> | gsnedders: If you have the physical hardware, you can install stuff like TrueCrypt on it |
| 17:07 | <gsnedders> | Philip`: I mean a "memory stick" by physical hardware |
| 17:07 | <Philip`> | gsnedders: Ah |
| 17:07 | <Philip`> | gsnedders: I don't see how that prevents keyloggers from working... |
| 17:07 | <gsnedders> | Philip`: It doesn't, but it means they need to get the memory stick to get at the data even with a key logger |
| 17:08 | <gsnedders> | Philip`: Instead of just going to a URI and typing in what I did |
| 17:08 | <Philip`> | gsnedders: They can just copy the (encrypted) data off the memory stick when you've got it plugged in to the compromised machine |
| 17:08 | <gsnedders> | Philip`: I do realize it can always be circumvented :P |
| 17:09 | <Philip`> | gsnedders: HTTPS + one-time password would mostly prevent them from going to a URI and typing in what you did :-) |
| 17:09 | <gsnedders> | Philip`: HTTPS + one time password where? :P |
| 17:09 | <gsnedders> | s/where/how/ |
| 17:11 | <gsnedders> | Other problem with online: it requires internet access. |
| 17:12 | <Philip`> | gsnedders: Just set up a web server with HTTPS, and generate n random passwords, and have the server only let you use each one once, and then print out the list of passwords and, uh, be very careful not to lose it |
| 17:12 | <gsnedders> | :P |
| 17:12 | <Philip`> | I suppose the last step is a problem |
| 17:12 | <Philip`> | because if it wasn't, you could just make sure you don't lose your memory stick |
| 17:12 | <Philip`> | and then you wouldn't need to bother with encryption at all |
| 17:12 | <gsnedders> | Exactly. |
| 17:12 | <gsnedders> | I mean, I'm unlikely to lose it, but I may as well be an asshole :P |
| 17:14 | <Philip`> | Or: Find/write a Java application for your mobile phone which lets you read/write the data (with a password), because nobody's going to keylog a phone and you're probably not going to lose it very often |
| 17:15 | <gsnedders> | But number pads suck for typing on |
| 17:16 | <Philip`> | You'll get better with practice :-) |
| 17:19 | Philip` | sees that a compatibility bug in Wine he reported yesterday got fixed today, which is rather nice responsiveness |
| 17:21 | <Dashiva> | And two years later, it appears in package repos |
| 17:27 | <Philip`> | Dashiva: The lag on Gentoo is currently about one month (and three days if you don't mind using not-yet-stable versions), so it's nowhere near that bad for me :-) |
| 17:59 | gsnedders | can't remember how to simplify logs |
| 18:00 | <gsnedders> | does ln|x-2| - ln|x+2| = ln|(x-2)(x+2)| or not? |
| 18:02 | gsnedders | is just being stupid |
| 18:03 | <Philip`> | No |
| 18:03 | <Philip`> | ln|(x-2)/(x+2)|, I think |
| 18:03 | gsnedders | looks at marking scheme and sees he's made another mistake |
| 18:03 | <gsnedders> | Philip`: That's still wrong because it's a modulo function, IIRC |
| 18:04 | <Philip`> | gsnedders: Oh, you mean |x-2| like abs(x-2) and not just fancy brackets? |
| 18:04 | <Philip`> | In that case you'd want to calculate separate solutions for x <= -2, -2 < x <= 2, and 2 < x |
| 18:09 | <gsnedders> | I also realized I didn't actually need to simplify it |
| 18:09 | <gsnedders> | Philip`: yes |
| 18:10 | <Philip`> | annevk: http://philip.html5.org/data/script-with-trailing-slash.txt |
| 18:12 | gsnedders | fails to see how \frac{x^2}{x^2-4} = 1 + \frac{4}{x^2-4} |
| 18:13 | gsnedders | sees now |
| 18:13 | gsnedders | wonders how we're meant to realize that |
| 18:13 | <Philip`> | (x^2)/(x^2-4) = (x^2-4+4)/(x^2-4) = 1 + 4/(x^2-4) |
| 18:14 | <Philip`> | It's just a standard way of rearranging fractions :-) |
| 18:14 | <Philip`> | Try to add stuff to the top so it looks like the bottom, and then things cancel out nicely |
| 18:15 | <gsnedders> | Yeah, I saw that |
| 18:15 | <gsnedders> | It's just a question of how we are meant to realize that |
| 18:15 | <gsnedders> | I guess we're meant to think that x = y + 4 and work out that y = x^2-4 |
| 18:16 | <Philip`> | You're meant to have done some maths in the past, so you can recognise that in this kind of case you can simplify by adding stuff to the top so it looks like the bottom :-) |
| 18:17 | <Dashiva> | You see there's a 1, so you try to make a 1 on the left side too, and it just works out. :) |
| 18:17 | <Philip`> | It's just an instance of a/(a+b) = (a+b-b)/(a+b) = 1+b/(a+b), which is a simplification if b is simpler than a |
| 18:18 | <Philip`> | Uh, 1-b/(a+b) |
| 18:24 | <gsnedders> | Philip`: Yeah, sure, but I haven't done maths in weeks :P |
| 18:39 | Philip` | finds it strange that the BBC iPlayer appears to have the rights to stream some movies but [this is wild speculation] not to display screenshots of them on the web site, so the poster frames are unrelated images of the principal actor |
| 18:40 | <Philip`> | (e.g. http://www.bbc.co.uk/iplayer/episode/b009lt62/The_Magic_Roundabout/ and http://www.bbc.co.uk/iplayer/episode/b00gbjh8/Dances_with_Wolves/) |
| 18:43 | <BenMillard> | annevk, oops, I meant: http://projectcerbera.com/!dev/whatwg-blog/ |
| 18:44 | Philip` | unrelatedly notices that rubys's latest blog post breaks Opera's feed reader, because Opera scrolls horizontally to the end of the "java -classpath ..." line and does not display horizontal scrollbars and so the rest of the post disappears off the left edge of the screen |
| 18:46 | gsnedders | really threw himself a the hard end of stuff going straight into previous exam qustions |
| 18:47 | <BenMillard> | jgraham, good to know. :) |
| 18:50 | <rubys> | You mean you can break Opera with something as simple as a <pre>? |
| 18:51 | <Philip`> | rubys: It appears so |
| 18:52 | <Philip`> | rubys: (though it's only showing a single feed entry at a time, so you can only break your own and can't interfere with anyone else's) |
| 18:55 | <BenMillard> | it seems Opera 9 uses ul { margin: 1em 0; } when a <ul> is inside a <dd> whereas Firefox and IE have a rule like dl ul { margin-top: 0; margin-bottom: 0; } |
| 18:56 | <BenMillard> | (Fx and IE are what I expected) |
| 19:03 | <Philip`> | BenMillard: Would you mind if I said I thought that blog design looked quite ugly? :-) |
| 19:04 | <Philip`> | Actually I think it's just the green that I don't like much, mainly the blockquotes |
| 19:05 | <Philip`> | If I remove the colours and indent them more then it looks better to me - the green box is just a bit weird |
| 19:08 | <Philip`> | Also the green horizontal lines are a kind of sickly colour - if I make them something like 2px #ddf then it makes me happier |
| 19:09 | <Philip`> | Other than that, it looks fine to me :-) |
| 19:10 | <BenMillard> | Philip`, I just copied the colours from the spec. feel free to put your own demos online |
| 19:30 | <BenMillard> | jgraham, updated the table inspector UI demo: http://projectcerbera.com/!dev/table-inspector/new |
| 20:09 | gsnedders | is realizing how much maths he's missed |
| 21:00 | <gsnedders> | How do I start Firefox up without my profile? |
| 21:01 | <Philip`> | gsnedders: "firefox -P" and then create a new one |
| 21:01 | <Philip`> | At least, that's what hsivonen said, but it doesn't seem to work |
| 21:01 | <Philip`> | at least in FF2 |
| 21:02 | <Philip`> | but "firefox --profilemanager" does work |
| 21:08 | <jwalden> | add -no-remote to the commandline prior to the -P |
| 21:08 | <jwalden> | that makes it work if you already have a Firefox open |
| 21:08 | gsnedders | rarely has FF open |
| 21:44 | <gsnedders> | Peh. This spec is badly organized already. |
| 22:16 | <gsnedders> | "Implementations MUST support GZIP [RFC1952] compressed ustar [IEEE.1003‑1.2008] archives. Exporters SHOULD create such archives." — is that clear enough, where "exporters" are defined above? |
| 22:18 | <dglazkov> | Hixie: so what would be a better way to modify render tree directly? |
| 22:18 | <dglazkov> | it's not like I am a huge fan of the spec-that-must-not-be-mentioned, but it appears to be at least going in that direction |
| 22:19 | <Hixie> | well it's pointed in that direction |
| 22:19 | <Hixie> | but it's not going anywhere :-) |
| 22:19 | <Hixie> | at least not currently |
| 22:19 | <dglazkov> | great distinction! |
| 22:19 | <dglazkov> | :) |
| 22:19 | <Hixie> | i told the csswg i'd be able to resume editing it in 2017 or so |
| 22:20 | <dglazkov> | Hixie: I can't wait that long. It's like 5 years or something :) |
| 22:20 | <Hixie> | xbl2 is the other way to have diverging trees |
| 22:20 | <Hixie> | that one has a stable spec, at least |
| 22:20 | <dglazkov> | ptr? |
| 22:20 | <dglazkov> | I mean URL? |
| 22:21 | <Hixie> | http://www.google.com/search?q=xbl2 :-) |
| 22:21 | <Hixie> | http://www.mozilla.org/projects/xbl/xbl2.html |
| 22:21 | <dglazkov> | is that the shadow content part? |
| 22:22 | <Hixie> | right |
| 22:22 | <Hixie> | holy crap |
| 22:22 | <Hixie> | holidays are over |
| 22:22 | Hixie | wades into his inbox |
| 22:22 | <dglazkov> | (gut feeling) I really dislike this whole XMLsy cruft |
| 22:22 | gsnedders | wonders what spec Hixie isn't editing till 2017 |
| 22:23 | <Hixie> | yeah xbl2 is rather heavy handed |
| 22:23 | <Hixie> | i don't really like the way it turned out |
| 22:23 | <Hixie> | not sure how i'd have done it any different though |
| 22:23 | <Hixie> | gsnedders: css3 content (and lists) |
| 22:23 | <dglazkov> | why can't I just go and add crap to CSS, vigilante-style ) |
| 22:23 | <gsnedders> | Hixie: ah |
| 22:23 | <dglazkov> | worked for dhyatt, right? :) |
| 22:24 | <Hixie> | dglazkov: xbl is hyatt's brainchild too :-) |
| 22:24 | <Hixie> | i was just the midwife |
| 22:24 | <Hixie> | and then the doctor and adoptive parent for 6 years... |
| 22:24 | <dglazkov> | he seemed to have abandoned the child :) |
| 22:25 | <Hixie> | he paid child support |
| 22:27 | dglazkov | feels he can add simple and effective render tree manipulation with a handful of new CSS attributes and perhaps a few pseudo-selectors. This should really ease the pain of Web dev who now have to weave these shims and wrappers in markup or with JS. |
| 22:28 | <Hixie> | this is an area where we have to make sure we do things the right way, because there are a huge number of subtlties that make things rather complicated |
| 22:28 | <takkaria> | did hyatt invent the adoption agency algorithm, too? |
| 22:29 | <dglazkov> | Hixie: oh, I already kind of discovered this |
| 22:29 | <dglazkov> | I don't like "complicated" :) |
| 22:29 | <Hixie> | takkaria: a simplified version of it, yeah |
| 22:30 | <Hixie> | dglazkov: agreed |
| 22:31 | <dglazkov> | really, honestly -- I am not pursuing counters, quotations, named string, and such. I just want wrap,insert before, after, and move to. I'll go think on this. |
| 22:32 | <dglazkov> | + clipboard interaction |
| 22:32 | <dglazkov> | + dom/selector effects |
| 22:33 | <Hixie> | well insert before and after you can do now with ::before/::after |
| 22:33 | <Hixie> | wrap would be ::inside, right? |
| 22:33 | <Hixie> | the move-to mechanism in the css3 content draft has more holes in it than a french gruyere |
| 22:34 | <dglazkov> | mmm... cheese |
| 22:49 | <roc> | this stuff is very hard to get right |
| 22:50 | <roc> | If you want DOM access to the generated content, then I think you'll head in the direction of XBL2 pretty fast |
| 22:50 | <roc> | we removed everything from sicking's plate so he can work on XBL2 in Gecko in the next cycle |
| 22:50 | <Hixie> | nice |
| 22:51 | <roc> | and isn't there a project doing XBL2 in Webkit? |
| 22:52 | <Hixie> | do you know which quarter will more or less map to "the next cycle"? I'd like to leave some breathing room for me in my timetable so i can have time to fix xbl for sicking when he finds problems |
| 22:53 | <roc> | Hopefully he'll start working on it seriously in Q1, and then it will be "as long as it takes" |
| 22:53 | <Hixie> | ah ok |
| 22:53 | <Hixie> | well my q1 is pretty locked down already with plans to deal with splitting things from html5 into the ietf |
| 22:54 | <Hixie> | but i'll make sure to leave room in q2 for xbl |
| 22:54 | <roc> | I'm sure he's not promising to have it done for the next Firefox release post 3.1 |
| 22:54 | <Hixie> | annevk: do you know who adam barth should talk to at opera regarding content sniffing stuff? |
| 22:55 | gsnedders | would guess that'd be slumped in with HTTP and therefore be Yngve (sp?) |
| 22:56 | <gsnedders> | It seems I did spell that right. |
| 22:56 | <sicking> | Hixie, it'll probably take a while before i have oodles of feedback anyhow |
| 22:56 | <Hixie> | k |
| 22:56 | <sicking> | Hixie, there'll be a lot to figure out just to get XBL1 and XBL2 to play together nicely |
| 22:56 | <Hixie> | feel free to ask questions btw |
| 22:56 | <Hixie> | i just can't promise to do much editing before q2 |
| 22:56 | <Hixie> | yeah |
| 22:56 | <sicking> | right |
| 22:57 | <Hixie> | they're different namespaces, if that helps :-D |
| 22:57 | <gsnedders> | Hixie: Q2 of which year? :P |
| 22:57 | <Hixie> | 2009 |
| 22:57 | <sicking> | until then i'll just do whatever i think is right ;) |
| 23:04 | <Hixie> | sicking: :-P |
| 23:04 | <Hixie> | ok bbiab |
