| 00:00 | annevk | wonders why this person keeps insisting on changing it to <image> while it has been pointed out repeatedly in the past that it won't work |
| 00:40 | <mcarter> | in section 7.2.3, about parsing SSE, when is it determined what the newline character is? can it change line by line? |
| 00:42 | <annevk> | never, yes |
| 00:43 | <Hixie> | any line can be terminated by 0x0D, 0x0A, 0x0D0A, or the end of the file |
| 00:43 | <annevk> | just like it works for text/html resources |
| 00:46 | <mcarter> | so how do you tell if a particular \r counts as the end of a line. do you see if data: comes immediately after? |
| 00:48 | <mcarter> | i mean, i guess not all lines are data fields (i was just thinking about this in terms of the data) |
| 00:49 | <annevk> | for \r you check if a \n follows it |
| 00:49 | <annevk> | iirc you could first split on lines and then do the rest of the parsing |
| 00:50 | <mcarter> | the way the spec reads, the stream "data: hello\r\ndata: world\r\n\r\n" could be interpreted as either a single event "hello\nworld" or two events "hello" and "world" |
| 00:52 | <Hixie> | when you get a 0x0D, you look to see if the next byte is 0x0A |
| 00:52 | <Hixie> | if it is, you treat them as a single newline |
| 00:54 | <annevk> | section 7.2.4 is actually pretty clear about processing, no? |
| 00:54 | <mcarter> | so then is this payload allowed in SSE: "hello\rdata: world" |
| 00:56 | <annevk> | I don't think so |
| 00:56 | <annevk> | newlines are \n |
| 00:57 | <mcarter> | but this payload is allowed: "hello\rworld" ? |
| 00:58 | <mcarter> | or would that cause an event with data = "hello" and ignore the "world field" ? |
| 00:59 | <annevk> | the latter |
| 01:00 | <annevk> | "data: hello\rworld" becomes "data: hello", "world" per 7.2.4 |
| 01:00 | <mcarter> | so there's no way to send the "\r" character at all over sse? |
| 01:00 | <annevk> | right |
| 01:00 | <annevk> | afaict |
| 01:02 | <mcarter> | do you know what the rationale is for not allowing the '\r' character? |
| 01:03 | <annevk> | it seems like a side effect of allowing it to be a newline character |
| 01:05 | <mcarter> | well, we had to switch away from SSE for orbited (we implemented it in gecko/webkit with xhr streaming) because we weren't able to send protocols like IRC over it (which use '\r') |
| 01:05 | <mcarter> | and its a shame that we wouldn't be able to use native SSE either, without some kind of extra encoding |
| 01:06 | <annevk> | IRC uses \r? lame |
| 01:09 | <mcarter> | i sort of feel like its the perogative of 20 year old protocols to use whatever framing they use |
| 01:12 | <annevk> | I suppose |
| 01:18 | <mcarter> | Hixie, is it your intention to disallow "\r" in the payload, or was it an oversight? (I'm getting ready to send some feedback in about it) |
| 01:18 | <Hixie> | yes |
| 01:18 | <Hixie> | \r in general isn't supported on the web |
| 01:18 | <Hixie> | use \n |
| 01:20 | <mcarter> | Hixie, but then SSE becomes useless for tunneling other text-based protocols |
| 01:21 | <mcarter> | other text-based protocols that require \r, anyway |
| 01:21 | <Hixie> | yes |
| 01:21 | <Hixie> | are there any such protocols where you couldn't translate \r to \n on the server? |
| 01:25 | <mcarter> | No. I can encode any protocol in base64 or hexidecmal or a string of "0" and "1" -- its just not ideal |
| 01:27 | <Hixie> | converting \r to \n seems like a non-issue |
| 01:28 | <Hixie> | it's not like you can use this without some level of conversion anyway |
| 01:28 | <mcarter> | it means though that \n needs to be escaped somehow as well |
| 01:29 | <mcarter> | I think supporting \r as a line ending as catering to a much smaller issue that allowing \r in the protocol would be |
| 01:31 | <mcarter> | I also think that silent stripping out any data after a \n and throwing it away is a pretty big problem |
| 01:31 | <mcarter> | * i mean, any data after a \r |
| 01:32 | <Hixie> | what i mean is, why can't you just convert all \rs to \ns, and all \ns to \ns, losing the distinction |
| 01:35 | <mcarter> | in the case where the data your sending has the characteristic that \r\n is used as framing and \n is allowed in the payload |
| 01:42 | <Hixie> | are there any such cases? |
| 01:42 | <Hixie> | seems like that kind of protocol would be awful |
| 01:45 | <Hixie> | i mean we also can't handle cases where the data contains malformed UTF-8 |
| 01:45 | <Hixie> | it's not really intended for tunneling existing protocols |
| 01:49 | <franksalim> | #actionscript |
| 01:49 | <franksalim> | ha |
| 01:49 | <franksalim> | whoops |
| 01:49 | franksalim | slits wrists |
| 01:57 | <franksalim> | mcarter, i don't even think that is the main problem with the line breaks in SSE |
| 01:59 | <franksalim> | \r\n is ambiguous...it could be a single break or two breaks (in the unlikely case that the server changed break character mid stream) |
| 01:59 | <franksalim> | in my understanding |
| 02:00 | <annevk> | that's just the server guys not reading the spec, doesn't make the spec ambigious |
| 02:02 | <franksalim> | annevk, i thought the issue was that any line could be terminated by a CR or a LF |
| 02:02 | <franksalim> | the 'look to see if the next character' thing doesn't really work if you get the CR and the LF in two read events |
| 02:03 | <franksalim> | do you just not dispatch a CR terminated event until there is another byte |
| 02:04 | <annevk> | right |
| 02:04 | <franksalim> | annevk, what if that was the last event...maybe another event doesn't arrive for an hour |
| 02:04 | <annevk> | well, you can simply dispatch it actually |
| 02:05 | <annevk> | you just need to ignore the next byte if it's LF |
| 02:06 | <franksalim> | scratch that |
| 02:06 | <franksalim> | consider the case where you receive data:something\r |
| 02:07 | <franksalim> | and then \n a short time later. do you dispatch it? or do you consider the \r\n a single newline? |
| 02:07 | <annevk> | see the spec |
| 02:08 | <franksalim> | annevk, i have seen the spec. i think there is an ambiguity |
| 02:08 | <Hixie> | you dispatch it as "something" |
| 02:08 | <Hixie> | and then when you get the \n later, you ignore the \n |
| 02:08 | <Hixie> | you don't ever need to look ahead |
| 02:08 | <Hixie> | you only ever need to look back |
| 02:08 | <franksalim> | you dispatch it, even though it is not followed by a blank line? |
| 02:09 | <franksalim> | "If the line is empty (a blank line) Dispatch the event, as defined below. " |
| 02:09 | <annevk> | \r is also a valid newline character |
| 02:09 | <franksalim> | so data:something\r should not be dispatched as something |
| 02:09 | <Hixie> | i don't really understand the question |
| 02:09 | <Hixie> | could you give some more background? |
| 02:10 | <annevk> | anyways, bedtime here |
| 02:10 | <franksalim> | annevk, goodnight |
| 02:10 | <franksalim> | Hixie, i will try |
| 02:11 | <franksalim> | data:something\r is not dispatched until a blank line is seen |
| 02:11 | <franksalim> | so data:something\r\r would be dispatched |
| 02:11 | <franksalim> | (as "something") |
| 02:12 | <franksalim> | are we on the same page so far? |
| 02:13 | <franksalim> | (you can tell me i'm crazy, i won't be offended) |
| 02:17 | <franksalim> | annevk, perhaps another time |
| 02:19 | <Hixie> | i don't know what you mean by "dispatched" |
| 02:19 | <Hixie> | are we talking server or client? |
| 02:19 | <Hixie> | where is the data coming from? |
| 02:19 | <franksalim> | client. by dispatched i mean dispatched as a messageevent |
| 02:20 | <Hixie> | so the client receives "data:something\r". |
| 02:20 | <Hixie> | ok |
| 02:20 | <franksalim> | yes. that is the case i am worried about |
| 02:20 | <franksalim> | the client receives "data:something\r". the next byte is "\n" |
| 02:21 | <Hixie> | so what's the problem? |
| 02:21 | <franksalim> | it is not clear if that "\n" was part of a "\r\n" or was a blank line |
| 02:21 | <Hixie> | if a \n follows a \r it is part of the \r\n sequence. |
| 02:21 | <Hixie> | always. |
| 02:21 | <franksalim> | since "\r\n", "\n", and "\r" are all considered valid line breaks |
| 02:23 | <franksalim> | is that implied somehow in the pseudo-BNF? |
| 02:23 | <Hixie> | the bnf is non-normative |
| 02:23 | <Hixie> | the next section says: The stream must then be parsed by reading everything line by line, with a U+000D CARRIAGE RETURN U+000A LINE FEED (CRLF) character pair, a single U+000A LINE FEED (LF) character, a single U+000D CARRIAGE RETURN (CR) character, and the end of the file being the four ways in which a line can end |
| 02:24 | <Hixie> | i can make this more explicit if you have reason to believe that browsers are going to implement thing incorrectly if you like |
| 02:29 | <franksalim> | i think the odds of a server using \r and \n interchangeably (which would break things) are low enough as not not need clarification, especially if you do not think it is worth the time |
| 02:29 | <franksalim> | *as to not |
| 02:36 | <franksalim> | actually, i just read the paragraph on appropriate buffering, and i think that takes care of it. i think that _definitely_ takes care of it |
| 02:36 | <franksalim> | Hixie, sorry |
| 02:37 | <Hixie> | there's a paragraph on buffering? :-) |
| 02:37 | <franksalim> | Hixie, yes. it might make me feel a little less foolish if that went in recently |
| 02:38 | <franksalim> | haha. this is my finest hour |
| 02:39 | <Hixie> | i haven't touched that section in a while :-) |
| 02:39 | <Hixie> | which paragraph do you mean? |
| 02:41 | <franksalim> | Since connections established to remote servers for such resources are expected to be long-lived, UAs should ensure that appropriate buffering is used. In particular, while line buffering may be safe if lines are defined to end with a single U+000A LINE FEED character, block buffering or line buffering with different expected line endings can cause delays in event dispatch. |
| 02:42 | <franksalim> | it's a pretty good paragraph |
| 02:43 | <Hixie> | ah ok |
| 02:43 | <Hixie> | cool |
| 07:39 | <aboodman3> | hm, sorry about all the aboodmans. I don't know how to operate irc. |
| 07:40 | <Hixie> | no worries |
| 07:40 | <Hixie> | aboodmans are always welcome |
| 07:40 | <aboodman3> | Hixie: has anyone implemented MessageChannels? |
| 07:41 | <Hixie> | not to my knowledge, it was added only recently |
| 07:52 | <aboodman3> | when a port becomes unentangled, do messages start getting queued for it? |
| 07:53 | <Hixie> | no it just returns false |
| 07:53 | <Hixie> | when a port is unentangled that it's, it can never be entangled again |
| 07:54 | <aboodman3> | ok |
| 10:15 | <annevk> | Hixie, you wrote port(); rather than port; in one of the Workers IDL blocks |
| 10:43 | <Philip`> | http://wiki.whatwg.org/wiki/User_talk:Flowerwish - spam |
| 10:46 | <Lachy_> | Philip`, deleted |
| 11:08 | <hsivonen> | looks like marcos is going to have a rough ride practicing extensibility of the URI system: http://lists.w3.org/Archives/Public/www-tag/2008Aug/0068.html |
| 11:10 | <hsivonen> | if URI schemes supported Unilateral Extensibility, he could just proceed without consulting with any authority |
| 11:10 | <annevk> | seems he sort of misses the point talking about media types though |
| 11:11 | <hsivonen> | annevk: he meing stuart or marcos? |
| 11:11 | <annevk> | widget: is about what address a file has *within* a package |
| 11:11 | <annevk> | Stuart |
| 11:11 | <hsivonen> | /meing/being/ |
| 11:11 | <annevk> | with the constrain that the address needs to be consistent regardless of where the package is |
| 11:19 | <annevk> | more URI scheme fun: http://omniplex.blogspot.com/2008/08/leader-of-pack.html |
| 11:22 | <hsivonen> | boo. del.icio.us is now delicious.com. no fun |
| 11:24 | <Philip`> | "We’ve seen a zillion different confusions and misspellings of “del.icio.us” over the years (for example, “de.licio.us”, “del.icio.us.com”, and “del.licio.us”), so moving to delicious.com will make it easier for people to find the site and share it with their friends." - that sounds fairly reasonable |
| 11:24 | <Philip`> | although they could have solved part of the problem by registering names like licio.us themselves |
| 11:25 | <annevk> | haha, markp about his blog.whatwg.org post "i admire my ability to get paid for this" |
| 11:27 | <hsivonen> | I admire markp's ability to restrain himself from posting to the WHATWG list. I mean, no doubt he has opinions about this stuff. |
| 11:28 | <Lachy_> | I upgraded the blog to wordpress 2.6, and wrote a shell script to simply the whole process in the future |
| 11:28 | <Lachy_> | let me know if you see any problems with the upgrade |
| 11:28 | <othermaciej> | I predict at least one problem |
| 11:28 | <othermaciej> | which is that you will still be using wordpress |
| 11:29 | <Lachy_> | othermaciej, do you have an alternative proposal? |
| 11:29 | <Philip`> | All the cool people write their own blogging software |
| 11:29 | <othermaciej> | no |
| 11:29 | <othermaciej> | but it is fun to bitch about wordpress |
| 11:29 | <Lachy> | one day, when I find the time, I will write my own blog |
| 11:30 | <Lachy> | othermaciej, agreed :-) |
| 11:30 | <othermaciej> | the sloppy markup, the hacky implementation, the shoddy security |
| 11:30 | <othermaciej> | no end of fun |
| 11:30 | <hsivonen> | I have wanted to write a blogging system and an online photo management system |
| 11:30 | <hsivonen> | then I figured that Flickr has the photo thing solved |
| 11:31 | <Philip`> | Blogging really ought to be a solved problem by now |
| 11:31 | <hsivonen> | but for the blogging system, I'm still using flat files under Apache and a Jython script in cron |
| 11:32 | <hsivonen> | because I wouldn't be practicing what I preach if I used a PHP system like WP |
| 11:32 | <hsivonen> | also, I become extremely unhappy when I have to fix someone else's PHP |
| 11:32 | <krijnh> | Rolling your own blogging thing definitely is fun :) |
| 11:32 | <hsivonen> | and I'm sure using WP would put me into that situation |
| 11:32 | <annevk> | I'm pretty happy with flickr though, although I want it to be easier to add tags to photos |
| 11:33 | <annevk> | for instance, modifying tags of photos on overview pages would be most welcome |
| 11:33 | <hsivonen> | the problem with the Jython script is that it such a legacy setup that I can't update it of move my blog to a different machine without breaking it |
| 11:34 | <hsivonen> | also, my CPython cron jobs depend on PyGenx, which doesn't have a Ubuntu package, so I can't move those scripts, either |
| 11:34 | <hsivonen> | yay for legacy |
| 11:38 | <annevk> | can't you export your content to a new system somehow? |
| 11:38 | <hsivonen> | annevk: content, yes. trivially. cron jobs, not so trivially |
| 11:39 | <hsivonen> | I'd love to have a pure-Python implementation of PyGenx |
| 11:40 | <hsivonen> | although now I haveextra disk space on the Ubuntu server, so I guess I could go ahead and install a compiler along with alla the dev packages |
| 11:41 | <hsivonen> | or devel rather |
| 11:43 | <hsivonen> | not being a bozo with feeds comes at a high price |
| 11:47 | <hsivonen> | someone should write something like Movable Type (generating flat files for Apache) with a TurboGears-based admin front end |
| 12:03 | <Lachy> | I originally planned to write my own blogging system in PHP, but now I want to write my own blogging system in python using html5lib |
| 12:04 | <gDashiva> | And soon you'll want to write it in xhtml2 using xforms |
| 12:04 | <Lachy> | gDashiva, no |
| 12:04 | <gDashiva> | The revolution is coming |
| 12:05 | <Lachy> | it will have XHTML on the backend for templates and authoring, and then serialise it as HTML |
| 12:06 | <zcorpan> | why not HTML for authoring? or what do you mean by authoring? |
| 12:06 | <Lachy> | has anyone written an HTML5 serialiser in python yet? |
| 12:06 | <zcorpan> | i think kid or genshi has one |
| 12:07 | <Lachy> | zcorpan, because I like being able to write markup in XHTML cause if I make a well-formedness errror, then I notice and fix it straight away. |
| 12:07 | <Lachy> | it makes me less reliant on on a validator for syntax checking |
| 12:08 | <virtuelv> | re the alt={foo} syntax: How does it impact existing server-side tools/templating languages? |
| 12:09 | <annevk> | it makes you reliant on XML though |
| 12:09 | <Lachy> | virtuelv, are there any such languages that use braces on their own? I know JSP expression language using ${...}, so that won't be affected. |
| 12:11 | <Lachy> | annevk, is that a problem? |
| 12:12 | <virtuelv> | Lachy: movable type has/had {} in templates |
| 12:13 | <annevk> | dunno |
| 12:13 | <Lachy> | virtuelv, ok. Then that might be a problem. |
| 12:14 | <virtuelv> | Django uses it |
| 12:15 | <virtuelv> | {% block title %} |
| 12:15 | <virtuelv> | http://www.djangoproject.com/documentation/templates/ |
| 12:15 | <Lachy> | virtuelv, movable type uses {{...}} |
| 12:15 | <virtuelv> | IIRC it didn't initially |
| 12:15 | <Lachy> | know of any documentation that shows that? |
| 12:17 | <virtuelv> | Lachy: http://bradchoate.com/weblog/2003/07/19/simple-template |
| 12:21 | <virtuelv> | my memory seemed a bit off, though |
| 12:21 | <Lachy> | virtuelv, that's using them with $ signs, so there doesn't appear to be a problem |
| 12:21 | <virtuelv> | my main concern would be {% something %} though |
| 12:22 | <Lachy> | I don't see how that would be a problem |
| 12:22 | <virtuelv> | also survey, http://www.whenpenguinsattack.com/2006/07/19/php-template-engine-roundup/ |
| 12:22 | <virtuelv> | Lachy: because you introduce ambiguity |
| 12:23 | <Lachy> | how is it ambiguous? {...} is different from {%...%} |
| 12:23 | <virtuelv> | yes, but {%foo%} is not in any way prohibited as alt-text now, is it? |
| 12:25 | <Lachy> | it's highly unlikely that anyone would legitimately use the % signs like that |
| 12:25 | <Lachy> | for alt text |
| 12:25 | <virtuelv> | Lachy: or {{foo}}? |
| 12:26 | <Lachy> | that's also unlikely to be a legitimate alt text |
| 12:27 | <zcorpan> | {{:-{)}} |
| 12:27 | <Lachy> | besides, if someone is using such a templating language, wants alt="{{foo}}" as alt text, then they would have to use some escaping mechanism in their templating langugae |
| 12:27 | <Lachy> | zcorpan, how does that clash with templating languages? |
| 12:28 | <zcorpan> | Lachy: it doesn't but it could be legitemate alt text |
| 12:28 | <Lachy> | well, if they want that as alt text, then it's a problem regardless of what HTML5 says. |
| 12:28 | <zcorpan> | why? |
| 12:28 | <Lachy> | because {{ ... }} clashes with some templating languages |
| 12:28 | <virtuelv> | here's a fun one |
| 12:29 | <virtuelv> | {a, b, c} |
| 12:29 | <virtuelv> | denoting the set a, b, c |
| 12:29 | <Lachy> | virtuelv, is that something from a templating language, or intended to be alt text? |
| 12:30 | <Lachy> | because I'm only interested in something that a) could be used as legitimate alt text, and b) directly clashes with a templating language. |
| 12:30 | <virtuelv> | Lachy: intended to be alt text |
| 12:30 | <virtuelv> | have a look at http://en.wikipedia.org/wiki/Set |
| 12:31 | <Lachy> | ok, that's a separate problem from the templating language problems we were talking about. |
| 12:31 | <virtuelv> | yes |
| 12:32 | <virtuelv> | templating languages is but one of the problem with microsyntaxes |
| 12:32 | <virtuelv> | WP's alt text seems to be latex-centric |
| 12:32 | <Lachy> | I'm not yet convinced it is a real problem though, since you've yet to demonstrate a real clash |
| 12:33 | <virtuelv> | but I'd have written {1,3} is a proper subset of {1,2,3,4} |
| 12:33 | <Lachy> | the latex problem may be a real problem. I'm not arguing either way on that particular issue |
| 12:34 | <virtuelv> | latex in itself is not the problem |
| 12:34 | <virtuelv> | the problem is that curly brackets (or any other brackets) have history and prior use outside of HTML5 |
| 12:34 | <Lachy> | I didn't say it was. I was just referring to it as the latex problem because latex examples are the most common that have been raised |
| 12:35 | <virtuelv> | http://en.wikipedia.org/wiki/Curly_bracket#Uses_of_.E2.80.9C.7B.E2.80.9D_and_.E2.80.9C.7D.E2.80.9D |
| 12:35 | <Lachy> | I'm just not convinced the templating language issue is a real problem |
| 12:37 | <virtuelv> | Lachy: you will not know until you have actually surveyed them |
| 12:38 | <Lachy> | virtuelv, I know. That's why I've been looking at every template language you mentioned to see if it really does clash |
| 12:38 | <Lachy> | so far, i have seen one that does |
| 12:39 | <Philip`> | All characters have history and prior use outside of HTML5 |
| 12:39 | <Lachy> | oh, here's one. smarty uses: {include file="footer.tpl"} |
| 12:40 | <Lachy> | http://www.smarty.net/crashcourse.php |
| 12:40 | <Lachy> | here's one that could really clash: <tr bgcolor="{cycle values="#aaaaaa,#bbbbbb"}"> (if that was <img alt> instead) |
| 12:41 | <Philip`> | E4X does stuff like 'var x = <img src={imgsrc} alt={imgalt}>;', if I remember correctly |
| 12:41 | <Philip`> | but that's not ambiguous since it's proper XML |
| 12:46 | <Philip`> | It would be easy if UAs weren't expected to implement special processing for not-good-alt images, because then alt="{Photo}" and alt="{...latex etc...}" could both be conforming and it wouldn't matter than a UA couldn't tell them apart |
| 12:46 | <Philip`> | particularly since in practice a UA won't be able to tell them apart anyway, because everyone is going to ignore whatever the spec says |
| 12:47 | <virtuelv> | the alt situation is, I believe, fracked up beyond repair |
| 12:47 | <virtuelv> | s/ck/kk/ |
| 12:48 | <virtuelv> | <img noalt> |
| 12:53 | <Lachy> | virtuelv, noalt by itself doesn't address the problem, and when used with alt, then it's a bad name |
| 13:41 | <annevk> | jgraham, you forgot that no-text-equivalent has a bad fallback scenario |
| 13:42 | <Philip`> | Why is it bad? |
| 13:44 | <Philip`> | I think most current UAs announce that images are images, rather than treating them like <span>...alt text...</alt>, so it won't have the problem that they'll be indistinguishable from plain text (which is the problem that the new feature is trying to solve) |
| 13:44 | <annevk> | that's probably true, indeed |
| 13:55 | <virtuelv> | besides, I don't think the huge amount of DSC_[0-9]{4,}.jpg's on the web are going to get sensible alt's anytime soon |
| 13:56 | <virtuelv> | {DSC_0201.jpg} is no more meaningful than [IMAGE] |
| 13:56 | <jcranmer> | [bright blue swath of pixels in the upper half of the image, dull green swath of pixels in the lower half. Various hues of pixels scattered throughout] ? |
| 13:56 | <annevk> | a lot of my flickr images have meaningless titles |
| 13:57 | <annevk> | I don't really expect this to change |
| 13:57 | <virtuelv> | trying to fix the alt situation is like trying to explain every joke you ever tell |
| 13:58 | <Philip`> | That would be difficult - usually I don't even know whether I'm joking or not |
| 13:58 | <virtuelv> | yet, alt is trying to describe everything |
| 13:59 | <virtuelv> | If I show you a 7-11 cup with the "Oslo" text visible and three coins, you'll get the joke if it was meant for you to get |
| 13:59 | <virtuelv> | I couldn't possibly explain it to you without ruining it |
| 14:00 | <virtuelv> | s/joke/in-joke/ |
| 14:00 | <jcranmer> | [7-11 cup with "Oslo" text visible and three coins], that should be good ALT? :-) |
| 14:00 | <virtuelv> | jcranmer: no |
| 14:02 | <virtuelv> | The proper alt-text would be "Oslo needs to do something about it's begging problem" |
| 14:03 | <virtuelv> | 7-11 are standard-issue beggar cups |
| 14:03 | jcranmer | was expecting one of this rigged game explanations |
| 14:03 | <jcranmer> | "find under which coin the cup is under" |
| 14:05 | <virtuelv> | http://xkcd.com/3/ |
| 14:06 | <virtuelv> | {Island (sketch)} or {Cartoon} or "I'm lonely"? |
| 14:09 | <virtuelv> | imo, {Cartoon} would best be replaced by class=Cartoon |
| 14:09 | <virtuelv> | s/class/whatever/ |
| 14:09 | <virtuelv> | it's not alternative |
| 14:09 | <virtuelv> | it represents a classification |
| 14:11 | <Philip`> | http://www.mathtalk.com/demos/Demos/MathTalkforVisuallyImpaired.avi looks kind of painful to use |
| 14:14 | <jcranmer> | I hope your happy |
| 14:14 | <jcranmer> | s/r/'re/ |
| 14:14 | <jcranmer> | I've now spent the past 10 minutes browsing XKCD archives |
| 14:14 | <virtuelv> | gawd, that looks painful |
| 14:18 | <jcranmer> | painful would be an understatement |
| 14:22 | <Philip`> | I would guess it's more bearable when you're just using voice output, not voice input |
| 14:22 | <Philip`> | (and using a normal keyboard) |
| 14:40 | <Philip`> | It took me way too long to find a joke in LaTeX |
| 14:46 | <Philip`> | I'm sure I saw one ages ago in someone's signature on Slashdot, but I can't remember it at all :-/ |
| 14:48 | <hsivonen> | hendry: is this on your radar: http://www.w3.org/TR/2008/WD-ct-guidelines-20080801/ ? |
| 14:48 | <Philip`> | Oh, I think it was some variant of http://www.xs4all.nl/~jcdverha/scijokes/1_11.html#1 |
| 14:53 | hsivonen | predicts: people who think Content Transformation sucks will start putting Cache-Control: no-transform in their headers |
| 14:53 | hsivonen | predicts: next proxy vendors start ignoring the header if they aren't already |
| 14:54 | <hsivonen> | solution: people use browsers that suck less with carriers who offer pure IP routing |
| 14:58 | <annevk> | oh yes, near free tickets (~400 EUR) to the Kilimanjaro thanks to miles |
| 14:58 | <annevk> | apparently they're good for something |
| 14:58 | <annevk> | can even get business class on the way back |
| 15:00 | <hsivonen> | is the Kilimanjaro area tourist-friendly these days? |
| 15:00 | <annevk> | afaik, yes |
| 15:09 | <annevk> | collegues climbed it last summer, I read on some blog of one of the former Mozilla XForms guys that he climbed it too recently; that's enough for me |
| 15:10 | <hsivonen> | I see that Kilimanjaro is entirely on the Tanzanian side. Somehow I had thought the border with Kenya ran through it. (I need to update my geography knowledge.) |
| 15:10 | hsivonen | is assuming the Kenya is not good for tourism these days, but what do I know |
| 15:12 | hsivonen | notices http://www.panoramio.com/ |
| 15:12 | <hsivonen> | it seems it would be better for users if Google Maps pulled geotagged photos from the whole Web--including Flickr |
| 15:13 | <annevk> | oh, dunno about Kenya either; last thing I read was not good |
| 15:14 | <annevk> | Kenya has Mount Kenya which is the second highest mountain of Africa iirc |
| 15:16 | <hsivonen> | wow. Google Maps now has Wikipedia intergration awesomeness |
| 15:16 | <hsivonen> | how new is that? |
| 15:16 | <mpt> | hsivonen, both Wikipedia and (I assume) have spam control. The Web of geotagged photos in general does not. :-) |
| 15:17 | <mpt> | ...(I assume) Panoramio have... |
| 15:17 | <hsivonen> | mpt: well, Google does have pretty good spam control on its own for Image Search |
| 15:18 | <hsivonen> | well, maybe they focus on filtering out porn rather than spam per se |
| 15:19 | <mpt> | I think spam's probably easier to filter out of an image search than out of a map, because you've given the image search engine some indication of what you're searching for |
| 15:20 | <hsivonen> | so far I haven't noticed spammy coordinate data on Flickr, but I haven't exactly looked, either |
| 15:20 | <mpt> | whereas if you're looking at a map, who knows why |
| 15:20 | <annevk> | where do you see this Wikipedia integration in Google maps? |
| 15:21 | <hsivonen> | annevk: there's a rectangle labeled More left to Map/Satellite/Terrain |
| 15:22 | <hsivonen> | it seems to me that Flickr as a data source is just owned by a competitor |
| 15:22 | <hsivonen> | but as far as searching *the Web* goes, getting data from there as well would be cool |
| 15:23 | <annevk> | neat |
| 15:23 | <annevk> | yeah, flickr integration would be nice |
| 15:24 | <annevk> | or random photo data source, even |
| 15:27 | hsivonen | is amused that the Content Transformation Landscape document enumerates Flash and SVGT under "Non Web Applications" |
| 15:54 | <hendry> | hsivonen: certainly is, thanks. have not read it carefully. i sent word to luca wurfl, and he sent in some good comments |
| 16:00 | <hsivonen> | hendry: is your employer in the business of creating/selling tranforming proxies? |
| 16:02 | <hendry> | hsivonen: no |
| 16:03 | <hendry> | hsivonen: how is validator.nu traffic? are many people using the gnu error format output? |
| 16:03 | <hsivonen> | hendry: I don't have up-to-date hard data, but my impression is that gnu and json are more popular than xml |
| 16:05 | <hsivonen> | lol. XML is so unpopular I might as well remove it |
| 16:06 | <hsivonen> | HTML is the most popular output format (obviously) |
| 16:07 | <hsivonen> | then from more popular to less popular: gnu, text, json, xml, xhtml |
| 16:07 | <hendry> | hsivonen: cool :) |
| 16:08 | <hsivonen> | those were for the HTML5 facet |
| 16:08 | <hsivonen> | note that Hixie uses out=gnu a lot |
| 16:08 | <hsivonen> | for the generic facet: |
| 16:09 | <hsivonen> | html, json, xml, xhtml, gnu, text |
| 16:09 | <hsivonen> | looks like people who use the html5 facet don't love XML |
| 16:09 | <hsivonen> | but XML seems worth keeping because it's used on the generic side |
| 16:10 | <hsivonen> | my sed skills, suck. otherwise, I'd calculate some per-unique-IP numbers |
| 16:11 | <hsivonen> | s/skills,/skills/ |
| 16:12 | hsivonen | tries perl instead |
| 16:18 | <hsivonen> | unique IP order for generic: html, xml, xhtml, gnu, json, text |
| 16:19 | <hsivonen> | unique IP order for HTML5: html, text, gnu, json, xml, xhtml |
| 16:22 | <Lachy> | watching the opening ceremony, I really don't understand why all the countries came out in a seemingly random order, with Australia finally coming out second last. |
| 16:23 | <hsivonen> | interestingly, the HTML5 facet is used more than the generic facet from browsers, but the generic facet is more popular than the HTML5 facet as a Web service |
| 16:23 | <hsivonen> | in terms of unique client IPs |
| 16:23 | <Lachy> | at first, it was on in chinese before switching to a norwegian stream, so I had no chance of undetstanding anything. |
| 16:28 | <gDashiva> | hsivonen: Maybe people don't like writing long URLs in their code :) |
| 16:28 | <billmason> | Lachy: opening ceremony marching order system -- http://en.wikipedia.org/wiki/2008_Summer_Olympics_Opening_Ceremony#Marching_order |
| 16:32 | <annevk> | anyone tips for best compact camera of the year? |
| 16:32 | <annevk> | ideally it fits in my pocket, has > 10M pixels, has > 5 optical zoom, and bonus points for GPS |
| 16:33 | <takkaria> | what price range are you looking at? |
| 16:34 | <hsivonen> | annevk: if there's any compact camera other than phones with built-in GPS, I want to know, too |
| 16:34 | <annevk> | takkaria, no price range as of yet :) |
| 16:35 | <hsivonen> | hmm. this Wikipedia Ogg thing is going to be huge when Firefox 3.1 ships |
| 16:35 | <annevk> | I guess my maximum for a compact would be EUR 500-1000 but that seems quite excessive for such a thing |
| 16:36 | <takkaria> | annevk: the canon g9 is considered one of the best compacts, though no GPS |
| 16:38 | <Philip`> | hsivonen: I can't remember ever having read a Wikipedia article with video in it, so I'm not sure how many people will ever notice the video support |
| 16:38 | <annevk> | takkaria, interesting, I saw that one in store today |
| 16:38 | <annevk> | hsivonen, I just found http://www.dpreview.com/news/0808/08080702nikonp6000.asp |
| 16:39 | <hsivonen> | annevk: thanks |
| 16:39 | <annevk> | from yesterday apparently :) |
| 16:39 | hsivonen | wonders if camera GPS startup sucks as badly as car navigator GPS startup |
| 16:40 | <annevk> | I believe all GPS sucks badly :( |
| 16:40 | <annevk> | at least, the devices I've seen so far did when trying to find out the initial position |
| 16:40 | <takkaria> | hmm, I'd wait for a review to appear on dpreview before buying one-- I don't believe nikon have been leading the compact market at all in recent years |
| 16:41 | <annevk> | takkaria, so the main problem with the G9 is that it's not all that compact |
| 16:41 | <takkaria> | ah. I'm a DSLR person, so lots of things are compact for me. :) |
| 16:42 | <annevk> | heh |
| 16:43 | <annevk> | the G9 comes with 32MiB by default |
| 16:43 | annevk | sighs |
| 16:43 | Philip` | sees that news.bbc.co.uk has (Flash) video on its front page, which is the first time he's ever seen that |
| 16:43 | <hsivonen> | it sucks environmentally that stuff ships with throw-away memory |
| 16:44 | <takkaria> | Philip`: oo, fun |
| 16:44 | <annevk> | hsivonen, good point |
| 16:44 | <Philip`> | Live video, even |
| 16:44 | <hsivonen> | last month, I my opportunity cost and accounting cost for legally disposing of useless MacBook RAM far exceeded the price I was able to sell it for |
| 16:46 | <Lachy> | wow, 15 strokes just to write the the first letter of Australia in chinese is insanely complicated. |
| 16:46 | <annevk> | the store owner told me http://www.dpreview.com/reviews/specs/Panasonic/panasonic_dmctz5.asp was pretty good |
| 16:46 | <hsivonen> | s/I my/my/ |
| 16:47 | <hsivonen> | my text input really sucks these days |
| 16:47 | <takkaria> | annevk: ah, I have a friend with one of the Lumix line, and it does seem to be pretty good. has fairly decent low-light performance if I remember right |
| 16:51 | <annevk> | ok |
| 16:57 | <annevk> | http://www.dpreview.com/reviews/specs/Fujifilm/fujifilm_F100fd.asp also looks reasonable (successor to my current camera it seems) |
| 17:03 | <hendry> | whoa that opening ceremony was amazing |
| 17:09 | <Philip`> | That flame has got to be environmentally unfriendly |
| 17:17 | <takkaria> | hmm, google's cache pages have changed their look |
| 19:36 | <jgraham> | annevk: You probably don't want >10 MP; the laws of physics say it gives worse picture quality on compacts than smaller numbers of MP |
| 19:37 | <jgraham> | (but is still useful for marketing) |
| 19:42 | <Philip`> | Most cameras seem to let you select the image size, like choosing 640x480 or 1280x960 or 4000x3000 or whatever they do. How does that work? |
| 19:42 | <Philip`> | like, do they only use a small fraction of the CCD or whatever it's called? |
| 19:43 | <Philip`> | or do they do some clever combination of adjacent pixel detectors to get more detail in those fewer output pixels? |
| 19:43 | <Philip`> | or do they save a maximum-resolution image as JPEG and then resize it before storing it, or something? |
| 19:44 | <jgraham> | I guess they resize before storing but I don't actually know (I'm pretty sure they don't use less of the CCD) |
| 19:44 | <Hixie> | is mikesmith back from holiday yet? |
| 20:12 | <Hixie> | the alt text debate no longer seems to include any accessibility "experts" |
| 20:28 | <Philip`> | Now it's just people who don't even pretend to know what they're talking about :-) |
| 20:32 | hsivonen | thinks Raman knows what he is talking about |
| 21:23 | <Philip`> | hsivonen: Well, at least I don't know what I'm talking about :-) |
| 22:04 | <annevk> | jgraham, so my 5MP might be ok? |
| 22:04 | <annevk> | why is buying something so simple so tricky :/ |
| 22:11 | <hsivonen> | annevk: I thought buying one would mean just buying the latest Ixus, but there's a choice even between Ixus models |
| 22:12 | <hsivonen> | clearly, it would all be easier if Steve offered us just one camera |
| 22:13 | <annevk> | true, that would help |
| 22:13 | <jgraham> | annevk: There are other reaons it might not be OK but adding more pixels won't make much difference |
| 22:13 | <annevk> | though I heard the iPhone camera sucks |
| 22:13 | <annevk> | jgraham, ok, grmbl |
| 22:15 | <jgraham> | annevk: If I were shopping for a compact camera I'd look for a) good low light performance b) image stabilisation of some sort or another c) reasonable optical zoom (bearing in mind that wide angles are just as useful as close ups) |
| 22:15 | <jgraham> | And good reviews of image quality |
| 22:17 | <jgraham> | (in I would also look at price, non proprietry storage format, ergonomics and perhaps battery life) |
| 22:17 | <jgraham> | s/in// |
| 22:17 | <hsivonen> | annevk: http://luminous-landscape.com/whatsnew/#264 |
| 22:19 | <annevk> | thanks jgraham, hsivonen |
| 22:19 | <annevk> | jgraham, any camera in mind btw? all sounds good to me :) |
| 22:19 | <jgraham> | I guess RAW is nice to hve but it doen't seem essential in a compact (like if you care enough to spend time postprocessing the photos you probably care enough to lug a DSLR about) |
| 22:20 | <annevk> | I don't really want to do postprocessing other than adding tags |
| 22:22 | <jgraham> | annevk: Not really. We bought a Canon A710IS for my g/f about 18 months ago mainly because it was the cheapest camera with image stabalisation. It has been pretty good but I guess not perfect |
| 22:24 | <annevk> | does it perform well with low light? |
| 22:25 | <jgraham> | OK but not spectacular. IS helps but I think some of the other people have CCDs that work better undr those conditions. I guess if you check dpreview it will give some clues |
| 22:26 | <annevk> | hmm, the Sony models come with 4GiB of internal memory |
| 22:29 | <jgraham> | Do Sony still use their proprietry memory stick things? |
| 22:32 | <annevk> | I believe so, yes |
| 22:57 | <annevk> | hsivonen, reading that site, sounds bad :/ |
| 22:59 | <annevk> | I might try to buy http://www.dpreview.com/reviews/panasonictz5/ as it seems reasonably good, except at high ISO values but apparently all compacts aren't very good in that area |
| 22:59 | <annevk> | excluding extras it's about ~350EUR which seems acceptable |
| 23:55 | <annevk> | hsivonen, http://www.w3.org/QA/2008/08/markup_validator_updated.html |
| 23:58 | <annevk> | hmm, they removed the XML media type warning for mislabeled XHTML |