| 00:42 | <AryehGregor> | roc, random question: about what percentage of your time do you spend reviewing other people's code, versus writing your own? |
| 00:47 | <AryehGregor> | Also volkmar ^^ |
| 00:47 | AryehGregor | is having another argument with Wikimedia managers about the value of reviewing volunteer contributions promptly |
| 00:58 | <volkmar> | AryehGregor: i'm not a reviewer |
| 00:58 | <AryehGregor> | Oh, well, that answers that. :) |
| 00:58 | <volkmar> | AryehGregor: you should ask bz and smaug maybe (and roc) |
| 00:58 | <AryehGregor> | I was going to ask smaug, but he left the channel. |
| 00:58 | <AryehGregor> | No big deal. |
| 01:00 | <AryehGregor> | We have a Wikimedia manager who's suggesting that if we had paid developers make sure to review all submitted patches promptly, they would be there (quote) "purely in a subservient role of approving/rejecting code as it comes in". http://lists.wikimedia.org/pipermail/wikitech-l/2011-March/052524.html |
| 01:04 | <volkmar> | AryehGregor: i've never heard this kind of complaints at Mozilla |
| 01:04 | <AryehGregor> | Nor have I. |
| 01:04 | <AryehGregor> | Nor at any other open-source project in the world. |
| 01:04 | <volkmar> | even if senior devs spend an important part of their time doing reviews |
| 01:05 | <AryehGregor> | It was influenced by the fact that there was a huge code review backlog and a lot of paid developers had to work full-time to clear it. But they only had to do it full-time because they let it accumulate for so long . . . like a year or so. |
| 01:07 | <volkmar> | eh, i'm managing my emails the same way :) |
| 01:08 | <AryehGregor> | I use filters to keep everything manageable. |
| 01:08 | <AryehGregor> | Out of the last 50 e-mails in "All Mail" in Gmail, eight are read. |
| 01:09 | <AryehGregor> | Five I haven't read yet because they're in my spec inbox and I didn't go through that today, and all the rest I'm never going to see. |
| 01:28 | gsnedders | pushes tree-walker API change to html5lib, finally changing it to something he is happening with |
| 02:25 | <roc> | AryehGregor: I spend hours every day reviewing code |
| 02:25 | <roc> | some days as little as one hour |
| 02:25 | <roc> | sometimes it's a drag |
| 02:25 | <roc> | but it's totally worth it |
| 02:26 | <roc> | and yes, it's totally essential to attracting and retaining contributors to review their patches promptly |
| 02:26 | <roc> | so for me it's almost always higher priority than writing my own code |
| 02:32 | <roc> | although we're not perfect; patches slip through the cracks, some developers are not as responsive as they should be, etc |
| 03:25 | <othermaciej> | AryehGregor: that person sounds clueless about open source and perhaps even clueless about software |
| 03:25 | <othermaciej> | code review is power |
| 03:27 | <othermaciej> | AryehGregor: though, looking at the email, perhaps things ar different in a CTR model |
| 03:30 | <othermaciej> | AryehGregor: I also disagree on this: "Reviewing code normally takes *much* less time than writing it." |
| 03:31 | <othermaciej> | sometimes reviewing code properly takes more time than writing it |
| 03:39 | <roc> | depends on whether it was written properly :-) |
| 08:21 | <zcorpan> | hello everyone |
| 08:26 | <jgraham> | zcorpan: Hej! |
| 08:29 | <zcorpan> | apart from being publication time soon, what has happened? |
| 08:31 | <jgraham> | I want to tell you that we usurped Hixie and instated gsnedders as editor. But I would be lying. |
| 08:31 | <jgraham> | Think how awesome that would be though |
| 08:31 | <zcorpan> | heh |
| 08:31 | <jgraham> | We could get an emo version of the spec done out in black |
| 08:32 | <zcorpan> | i take that as nothing in particular has happened |
| 08:32 | <jgraham> | I can't remember anything, but there were some Decisions and stuff |
| 08:33 | <jgraham> | I asusme you were here when <device> died? |
| 08:33 | <jgraham> | Oh and Hixie had some proposal for multi-track media stuff |
| 08:33 | <jgraham> | myabe you missed that? |
| 08:37 | <zcorpan> | yes and yes |
| 08:37 | <zcorpan> | thanks |
| 09:23 | <zcorpan> | where's the execCommand spec? |
| 09:25 | <zcorpan> | http://aryeh.name/spec/editcommands/editcommands.html |
| 09:25 | <zcorpan> | AryehGregor: please spec queryCommandSupported |
| 11:15 | <zcorpan> | does html5lib testsuite have a test for <device> not being special? |
| 11:19 | <jgraham> | Not sure |
| 11:45 | <zcorpan> | shelley++ for weeklies |
| 12:41 | <zcorpan> | sow how would you use custom controls with the MediaController thing? |
| 12:46 | <zcorpan> | oh you just use the API for the MediaController instead of for one of the slaves |
| 12:59 | <MikeSmith> | hsivonen: when you're around, I wanted to ask if it's OK with you if I go ahead and check in the patch submitted for http://bugzilla.validator.nu/show_bug.cgi?id=818 |
| 13:03 | <hsivonen> | MikeSmith: OK |
| 13:03 | <MikeSmith> | thanks |
| 14:18 | <zcorpan> | hmm. i have opera mobile 11 on android 2.3.3, HTC Gracia which is ARM6. webm.html5.org says it doesn't support WebM |
| 14:29 | <hsivonen> | zcorpan: what does http://webm.html5.org/debug/ say? |
| 14:29 | <hsivonen> | zcorpan: I may have introduced a bug when I tried to blacklist Opera Mobile on Symbian |
| 14:38 | <zcorpan> | hsivonen: it says Android 2.2, (empty), (empty) |
| 14:39 | <zcorpan> | shows a video element but doesn't play anything |
| 14:39 | <zcorpan> | i thought i had upgraded to 2.3.3 |
| 14:39 | <hsivonen> | zcorpan: oh so Opera Mobile now reveals the Android version. interesting |
| 14:40 | <hsivonen> | zcorpan: could you paste the entire UA string, please? |
| 14:42 | <zcorpan> | sent by email |
| 14:43 | <hsivonen> | zcorpan: thanks |
| 14:44 | <zcorpan> | it seems i still have android 2.2, but no new updates are available |
| 14:45 | <AryehGregor> | othermaciej, I agree that sometimes reviewing code properly takes more time than writing it, but normally it takes much less time. In particular, well-written patches should include enough comments, documentation, and explanation that they're reasonably easy to review, and if they don't, the reviewer can ask for them. |
| 14:45 | <AryehGregor> | In most cases. You'll still always have some hard-to-review patches, in the end. |
| 14:45 | <AryehGregor> | (you probably have a lot more experience reviewing patches than I do, though, and maybe you review them more thoroughly) |
| 14:46 | <AryehGregor> | I do think that in commit-then-review, reviewers have a lot less power than in review-then-commit. |
| 14:46 | <hsivonen> | zcorpan: what does the ADR number mean in the UA string? |
| 14:46 | <jgraham> | Is there any software project that uses CtR? |
| 14:46 | <AryehGregor> | zcorpan, I'm going to spec everything execCommand()-related eventually, but there's loads to do. The only thing I'm really working on speccing right now is the algorithm for applying inline markup commands. |
| 14:46 | <zcorpan> | hsivonen: no idea |
| 14:47 | <AryehGregor> | jgraham, MediaWiki, for one. |
| 14:47 | <AryehGregor> | It's probably a bad idea, and most of us seem to be in favor of switching to RtC. |
| 14:47 | <zcorpan> | AryehGregor: cool |
| 14:47 | <jgraham> | AryehGregor: Well that seems like a bad idea |
| 14:47 | <AryehGregor> | zcorpan, I should say, I'm going to do it if time allows. I still have a few months left, though, so it's likely that I'll be able to finish it. |
| 14:47 | <jgraham> | (using CtR) |
| 14:48 | <AryehGregor> | jgraham, yes, it is, which is why a lot of us want to change it. |
| 14:52 | <AryehGregor> | jgraham, if you count projects that have a status of "commit access" but no formal review process as CtR, then I suspect it's more popular overall than RtC. |
| 14:52 | <AryehGregor> | Although big, prominent, well-run projects tend to use RtC. |
| 14:54 | <hsivonen> | zcorpan: I refined the Android advice that webm.html5.org gives |
| 14:54 | <jgraham> | AryehGregor: I think "we don't have review" is a bit different to CtR |
| 14:54 | <AryehGregor> | Maybe, yeah. |
| 14:54 | <hsivonen> | it still sucks that Firefox is tied to CPU and Opera to the Android version |
| 14:54 | <AryehGregor> | You'll still usually have some sort of informal review in such cases, though. |
| 14:54 | <hsivonen> | I wonder how many Android users know what CPU their device has :-( |
| 14:55 | <jgraham> | e.g. html5lib doesn't really have review. Except that I have never committed a patch from a non-core-contributer without reviewing it first |
| 14:55 | <jgraham> | For single person projects the concept of review doesn't even make sense |
| 14:56 | <hsivonen> | I'd love to know if the default browser on Android 2.3.3 supports WebM and if Opera Mobile 11 on Android >= 2.3 but < 2.3.3 supports WebM |
| 15:02 | <hsivonen> | for additional Android fragmentation FAIL, HTC doesn't list the Android version or the CPU instruction set in its product specs |
| 15:03 | <Rik`> | hsivonen: I tried to update my android device but it requires a SIM card… |
| 15:04 | <hsivonen> | Rik`: so the device works without a SIM in general but doesn't update without a SIM? |
| 15:04 | <Rik`> | yes |
| 15:04 | <Rik`> | I'm using it on wifi all the time in the office for testing but never as a phone |
| 15:04 | <hsivonen> | is this some user-hostile mechanism that allows updates to be blocked by operator? |
| 15:04 | <Rik`> | might be |
| 15:05 | <hsivonen> | because otherwise, people with hostile operators could just take out the SIM and upgrade over wifi |
| 15:05 | <hsivonen> | the mobile business is so sad |
| 15:06 | <Rik`> | as someone reminded recently, home ISP was bad 10 years ago |
| 15:06 | <hsivonen> | even Apple, the supposed savior of users from operator hostility, reportedly requires partner operators to impose data transfer caps in order sell iPads |
| 15:06 | <hsivonen> | that's so crazily bizarre. |
| 15:07 | <Rik`> | like you could only use the ISP modem, you couldn't use more than one machine behind the modem, etc |
| 15:07 | <hsivonen> | maybe they want their product offering to suck everywhere as much as it sucks in the U.S. or something |
| 15:07 | <hsivonen> | Rik`: in France? it wasn't like that in Finland in 1995 |
| 15:08 | <Rik`> | hsivonen: yes |
| 15:08 | <Rik`> | they weren't enforcing the "one machine" rule a lot, but it was in the contract |
| 15:09 | <Rik`> | and the plans were limited in time, only 10/20/50 hours a month |
| 15:11 | <bfrohs> | Rik`: It was the same way in the US back then :( |
| 15:12 | <Rik`> | hopefully one ISP was aggressive enough to change a lot of those rules and we now have one of the cheapest internet access |
| 15:12 | <Rik`> | and this ISP (Iliad/Free) should launch mobile plans next year, hopefully changing the market in the same way |
| 15:51 | <volkmar> | is there a reason to use NodeList instead of HTMLCollection except when we might have non-elements in the list? |
| 17:03 | <aho> | background-image:url(foo.svg#bar) <- ff, chrome, and opera discard the fragment. is it supposed to be this way? |
| 17:04 | <aho> | would allow awesome svg spriting otherwise :I |
| 17:05 | <jgraham> | aho: Pure guess: does the # need to be escaped? |
| 17:05 | <wilhelm> | Indeed. But who needs sprites when we have data URIs? :P |
| 17:07 | <aho> | oh lol |
| 17:07 | <aho> | it actually works in firefox |
| 17:07 | <aho> | :) |
| 17:08 | <aho> | unescaped, that is |
| 17:08 | <aho> | replacing it with %35 (right one?) doesn't seem to help with chrome and opera |
| 17:09 | <zewt> | wouldn't escaping it be explicitly making it *not* a fragment, but part of the filename? |
| 17:10 | <jgraham> | Right, I didn't mean % escaping |
| 17:11 | <aho> | lemme upload my test thingy |
| 17:12 | <aho> | http://kaioa.com/k/test/svgsprites/index.html |
| 17:12 | <aho> | http://kaioa.com/k/test/svgsprites/svgsprites.svg <- view source |
| 17:13 | <jgraham> | Escaping is wrong; # should work fine |
| 17:14 | <aho> | <style> svg>g{display:none}...</style> <- having it like that in the svg is fine, right? |
| 17:14 | <aho> | or do i need cdata stuff there? |
| 17:16 | <aho> | mmh nop... adding cdata doesnt change anything |
| 17:16 | <aho> | still works in ff, still doesnt work elsewhere |
| 17:16 | <aho> | anyone got ie9? :> |
| 17:18 | <tomasf> | aho: no go in IE 9 |
| 17:18 | <aho> | kay. ty. :) |
| 17:18 | <aho> | wilhelm, ye, data uris work. however, i think this looks a bit nicer and it's easier to automate |
| 17:19 | <aho> | background-image:url(svgsprites.svg#square); <- e.g. i can tell that this is some square thing |
| 17:21 | <aho> | soooo... is it a bug that it works in firefox or are the other browsers faulty? :) |
| 17:29 | <CvP> | [aho] what do you need? |
| 17:30 | <aho> | fame and money, i guess |
| 17:30 | CvP | flames all aho's money. |
| 17:30 | <aho> | http://kaioa.com/k/test/svgsprites/index.html |
| 17:30 | <aho> | works with firefox |
| 17:30 | <aho> | but i'm not really sure if this should work or not |
| 17:30 | <aho> | it doesnt work with any other browser |
| 17:31 | <CvP> | i can see square, large dot and cross |
| 17:31 | <CvP> | in ie9 |
| 17:32 | <CvP> | it doesn't owrk in chrome dev |
| 17:32 | <CvP> | works just fine in ie9 and latest minefield |
| 17:33 | <aho> | tomasf, can you try it again in ie9? |
| 17:33 | <CvP> | does not work on opera 11.01 |
| 17:33 | <aho> | ie9 probably needs that cdata stuff |
| 17:34 | <nimbupani> | yeah i am pretty surprised it doesnt work on opera aho |
| 17:34 | <aho> | me too |
| 17:34 | <aho> | :) |
| 17:34 | <tomasf> | aho: sorry! I had IE 7 mode turned on. it does work :) |
| 17:34 | <aho> | ah :) |
| 17:35 | <aho> | zooming is weird with ff4 |
| 17:35 | <aho> | the repeated ones behave like bitmaps (pixelated/blurry) while the no-repeat ones (spans at the bottom) scale nicely |
| 17:36 | <CvP> | lol |
| 17:36 | <CvP> | now that i have zoomed it in minefield |
| 17:36 | <CvP> | i remember this picture |
| 17:36 | <CvP> | seeing somewhere before |
| 17:37 | <aho> | triangle is missing |
| 17:37 | <aho> | :> |
| 17:37 | <nimbupani> | umm aho |
| 17:38 | <nimbupani> | why do you have display: none in ur svg? |
| 17:38 | <aho> | why not? |
| 17:38 | <aho> | well, the big idea is to hide all those groups |
| 17:38 | <aho> | and then show them via the :target pseudo class thingy |
| 17:38 | <nimbupani> | but everything is hidden :| |
| 17:38 | <nimbupani> | then how will it work as bg? |
| 17:38 | <aho> | unless targeted, yes |
| 17:39 | <aho> | http://kaioa.com/k/test/svgsprites/svgsprites.svg#dot |
| 17:39 | <aho> | http://kaioa.com/k/test/svgsprites/svgsprites.svg#square |
| 17:39 | <aho> | see? (works fine with opera) |
| 17:39 | <nimbupani> | oh i see |
| 17:39 | <aho> | :) |
| 17:39 | <nimbupani> | thats what the sprites thing is about :) |
| 17:42 | <nimbupani> | aha aho something about #dot and #square is not being detected by Opera. |
| 17:43 | <aho> | ye, the fragments don't show up in dragonfly |
| 17:43 | <nimbupani> | yeah thats what. i am filing a bug as we speak :) |
| 17:44 | <aho> | ty :) |
| 17:48 | <aho> | i really should have tried this sooner... had this idea about 2-3 years ago :> |
| 17:49 | <nimbupani> | its a pretty neat idea! |
| 17:49 | <aho> | oct 2007... actually :) |
| 17:49 | <nimbupani> | we could combine this with fonts |
| 17:50 | <aho> | heh |
| 17:50 | <nimbupani> | and all we need is a single SVG file ;) |
| 17:50 | <jgraham> | TabAtkins: I think the cynical answer to that is "yes", and the very cynical answer is "no" |
| 17:54 | <jgraham> | If you don't add the word "formal" the chairs can ignore what you say because it doesn't fall under Process. If you do add the word "formal" than the chairs have to pass it on to TimBL and I imagine you share the same reckoning that I do of the chance of him upholding an objection to remove URI-based indirection |
| 17:58 | <hasather> | nimbupani: did you file a bug? Can you CC me? |
| 17:58 | <hasather> | nimbupani: CC davidh that is |
| 18:01 | <nimbupani> | hasather: oh haha hai david :) |
| 18:01 | <nimbupani> | yes I will ASAP |
| 18:01 | <hasather> | nimbupani: Hey! And great :) |
| 18:11 | <aho> | heh... odd |
| 18:11 | <aho> | firefox4 requests the svg 3 times |
| 18:12 | <nimbupani> | :| |
| 18:13 | <aho> | it really shouldn't do that :] |
| 18:14 | <aho> | in the net panel these requests actually do show up as svgsprites.svg#dot, svgsprites.svg#square, and svgsprites.svg#x |
| 18:14 | <aho> | i also cross-checked with fiddler |
| 18:14 | <aho> | there are indeed 3 requests to svgsprites.svg |
| 18:20 | <aho> | kay... those pointless requests can be suppressed with far future expires headers |
| 18:20 | <aho> | http://kaioa.com/k/test/svgsprites2/index.html <- only one request |
| 18:21 | <AryehGregor> | Hixie, so, can I write an objection on Google's time to your CP in conforming-u? 0:) |
| 18:26 | hsivonen | wonders if gruber is going to write about Firefox 4 now that there's a "no Flash" angle |
| 18:26 | <AryehGregor> | What "no Flash" angle? |
| 18:27 | <Ms2ger> | hsivonen, "Apple did it first"? |
| 18:27 | <hsivonen> | AryehGregor: Firefox 4 on Android doesn't support Flash |
| 18:27 | <AryehGregor> | Oh, interesting. |
| 18:27 | <hsivonen> | maybe that applies to Maemo, too |
| 18:28 | <AryehGregor> | Is that for ideological reasons or practical ones? |
| 18:28 | <hsivonen> | AryehGregor: practical |
| 18:28 | <AryehGregor> | Ah, okay. |
| 18:29 | <hsivonen> | though I expect Apple's reasons to be practical, too |
| 18:35 | <Smylers> | jgraham: So "formal" is to "objection" as "sudo" is to "make me a sandwich"? |
| 18:35 | <aho> | nimbupani, fyi suppressing the pointless requests with far future expires headers did work :) |
| 18:36 | <aho> | http://kaioa.com/k/test/svgsprites2/index.html |
| 18:36 | <nimbupani> | ah sweet aho |
| 18:36 | <nimbupani> | you should like blog about it :) |
| 18:36 | <nimbupani> | its a technique worth implementing once browser support is all fixed :) |
| 18:36 | <aho> | yeaaaa... there are about a dozen things i should blog about :I |
| 18:37 | <aho> | maybe next weekend .) |
| 18:38 | <nimbupani> | :) |
| 18:41 | <Hixie> | AryehGregor: if your objection is something substantial that hasn't already been debunked, sure |
| 18:42 | <AryehGregor> | Hixie, you probably think all my objections are groundless and have been debunked. Doesn't really matter, I'll do it on my own time. |
| 18:42 | <Hixie> | k :-) |
| 18:44 | <hober> | Smylers: yes |
| 19:46 | <AryehGregor> | Hixie, could you please set "X-XSS-Protection: 0" for the Live DOM Viewer? IE's XSS filter is really interfering with my development work. http://blogs.msdn.com/b/ieinternals/archive/2011/01/31/controlling-the-internet-explorer-xss-filter-with-the-x-xss-protection-http-header.aspx |
| 19:48 | <Hixie> | file a bug with microsoft |
| 19:48 | <AryehGregor> | Hixie, . . . |
| 19:49 | <Hixie> | i'm not adding a flag to make a browser follow the specs when they should just follow the specs in the first place |
| 19:49 | <Hixie> | (especially for something as dumb as this) |
| 19:49 | <AryehGregor> | How is this not following the specs? Other browsers have heuristic XSS filters too, like Chrome. |
| 19:49 | <AryehGregor> | The Live DOM Viewer page views that trigger the filter *are* XSS, you're injecting HTML from a query parameter into the document. |
| 19:49 | <Hixie> | if the browser is following the specs, what is interfering with your work? |
| 19:50 | <AryehGregor> | Because I'm trying to make something in Live DOM Viewer in a non-IE browser, save it, and load it in IE. |
| 19:50 | <Hixie> | if IE is following the specs, that'll work fine |
| 19:51 | <othermaciej> | Safari also has a heuristic XSS filter |
| 19:51 | <othermaciej> | I think Chrome doesn't actually ship with it on by default yet, but Safari does |
| 19:51 | <AryehGregor> | Are you saying that no browser should block page script if it heuristically detects XSS on the page? |
| 19:51 | AryehGregor | tries in Safari |
| 19:52 | <Hixie> | yes |
| 19:52 | <AryehGregor> | It doesn't work in Safari either. |
| 19:52 | <Hixie> | i have considered just base64-encoding the argument |
| 19:52 | <othermaciej> | the specs do not actually permit browsers to protect users from Web sites with reflective XSS vulnerabilities |
| 19:52 | <othermaciej> | but it's still a good thing to do |
| 19:53 | <AryehGregor> | Hixie, um . . . why? XSS is one of the biggest security problems web authors face, and a simple scheme that blocks URL parameters can potentially stop many attacks. |
| 19:53 | <othermaciej> | sufficiently encoding the argument would probably defeat XSS filters |
| 19:53 | <othermaciej> | even rot13 would work, I expect |
| 19:53 | <Hixie> | base64 is easier from js :-) |
| 19:53 | <AryehGregor> | Not in IE. :) |
| 19:53 | <Hixie> | oh right |
| 19:53 | <othermaciej> | AryehGregor: XSS filters are not foolproof |
| 19:53 | <Ms2ger> | Would rot13 encode <s? |
| 19:53 | <othermaciej> | so it's not really wise for authors to rely on them |
| 19:54 | <othermaciej> | it's a protection for users when authors screw up |
| 19:54 | <othermaciej> | not a protection for authors |
| 19:54 | <AryehGregor> | othermaciej, no, but if you can prevent a substantial percentage of author bugs from being escalated to XSS, that's a good thing for everyone. |
| 19:54 | <Hixie> | not for you apparently :-) |
| 19:54 | <othermaciej> | well the downside is that there are false positive matches and/or sites that are deliberately willing to be XSS'd |
| 19:55 | <AryehGregor> | Okay, fine. But pragmatically speaking, IE does have this feature and I do need to do work in IE and it would be nice if the relevant tools were written with the goal of actually working in significant browsers instead of adhering to some idealized world in which everything is completely standardized. |
| 19:56 | <AryehGregor> | (which doesn't seem practical for things as heuristic as XSS filtering, offhand) |
| 19:56 | <Hixie> | i'll fool their protection at some point |
| 19:56 | <Hixie> | in the meantime just use the upload/download feature |
| 19:56 | <AryehGregor> | Also: <script> |
| 19:56 | <AryehGregor> | if (navigator.userAgent.match('Gecko/(\\d+)') && RegExp.$1 == '20060217' && RegExp.$1 != '00000000') { |
| 19:56 | <AryehGregor> | var style = document.getElementsByTagName('style')[1]; |
| 19:56 | <AryehGregor> | style.parentNode.removeChild(style); |
| 19:56 | <AryehGregor> | } |
| 19:56 | <AryehGregor> | </script> |
| 19:57 | <AryehGregor> | That does not look like you're excessively concerned about only supporting standards-compliant browsers. |
| 19:57 | <Hixie> | that wasn't a non-standards-compliant browser |
| 19:57 | <Hixie> | that was a browser that followed CSS2 rather than CSS2.1 |
| 19:57 | <Hixie> | and therefore made ugly trees |
| 19:57 | <Hixie> | we changed the spec on them, not their fault |
| 19:58 | <AryehGregor> | What's upload/download supposed to do? |
| 19:58 | <AryehGregor> | It doesn't seem to do anything but clear the page. |
| 19:58 | <Hixie> | it's like a clipboard |
| 19:58 | <Hixie> | upload copies the file into the clipboard |
| 19:58 | <Hixie> | download copies the clipboard into the file |
| 19:58 | <Hixie> | file = first textarea |
| 19:59 | <zewt> | browser testing tools that only work on bug-free browsers seem no more useful than a debugger that only works on bug-free programs |
| 19:59 | <AryehGregor> | Which doesn't help me much if my primary machine is Linux and I'm using Windows to test IE, does it? |
| 19:59 | <Hixie> | it's a server-side clipboard |
| 19:59 | <AryehGregor> | Wait, then how does it work? |
| 19:59 | <Hixie> | it does an XHR to upload the file |
| 19:59 | <Hixie> | there's a single shared clipboard for all of tus |
| 19:59 | <AryehGregor> | Oh, weird. |
| 19:59 | AryehGregor | tries |
| 20:00 | <AryehGregor> | Hey, that works. Thanks. |
| 20:00 | <Hixie> | i'll fix the save thing at some point |
| 20:00 | AryehGregor | notices if (n.nodeType == 3 /* text node */) n = n.nextSibling; // we should always do this but in IE, text nodes vanish |
| 20:00 | Ms2ger | uploads random junk while AryehGregor tests |
| 20:01 | <AryehGregor> | But upload/download is more convenient than save anyway, so I'll just use that. |
| 20:01 | <Hixie> | yeah save is mostly for when refering to tests in e-mail |
| 20:14 | <zcorpan> | Hixie: if you're making changes to live dom viewer, how about using the fragment identifier instead of query so that the page can be loaded from cache? |
| 20:15 | <zcorpan> | and not need a reload when saving |
| 20:15 | <zcorpan> | or maybe save doesn't reload now, but anyway |
| 20:16 | <deane> | zcorpan: Hi. would it be ok if I 'copied' some of your html5 elements page, changed it around a bit and made my own one? I just wanted to do something similar but show more details of the elements :) |
| 20:16 | <zcorpan> | deane: go ahead |
| 20:17 | <zcorpan> | deane: there's also an appendix in the spec with all elements and more details |
| 20:17 | <zcorpan> | deane: you could copy that instead, apart from having more details it's less likely to have broken links :) |
| 20:17 | <deane> | zcorpan: Thanks, I'll give a link back to your one with credit to you. cool |
| 20:18 | <zcorpan> | deane: if you find any broken links let me know and i'll fix |
| 20:22 | <zcorpan> | http://dev.w3.org/html5/status/issue-status.html - no decisions are about the html5 differences document, right? |
| 20:22 | <zcorpan> | would be nice with a column about which document the issue is about |
| 20:22 | <jgraham> | zcorpan: I guess in theory they could be |
| 20:23 | <jgraham> | So, is there some reason that having a registry of predefined prefixes wouldn't work for RDFa? |
| 20:23 | <zcorpan> | jgraham: you mean indirectly? |
| 20:24 | <jgraham> | zcorpan: Well, no I mean someone could file a bug with the document that you WONTFIXed and they escallated |
| 20:24 | <jgraham> | Or something |
| 20:24 | <zcorpan> | i mean in the current list |
| 20:25 | <jgraham> | zcorpan: I don't know of anything in the current list that is about that document, but I haven't checked |
| 20:28 | <jgraham> | In particular the Facebook example leads me to believe that one already needs a registry to correctly process "in the wild" RDFa both in order to find all the OpenGraph content that screws up the prefix mapping stuff, nad to avoid the "og" prefix for non-OpenGraph uses |
| 20:28 | <jgraham> | (since that will be misinterpreted by clients that do literal prefix matching) |
| 20:30 | <jgraham> | So if one proposed a setup where one MUST register prefixes, HTML tools MUST only use literal string matching against registered prefixes and other prefix binding mechanisms MAY be supplied for compatibility with non-HTML tools, why would that be bad? |
| 20:31 | <zcorpan> | it could lead to a spec that matched reality |
| 20:31 | zcorpan | now has 4999 mailing list email |
| 20:32 | <zcorpan> | wonder if i should do 'mark all as read' on some mailing lists |
| 20:34 | <AryehGregor> | Opera people, here's a test that makes Opera fail to render the page, with an OOM error: data:text/html,<!doctype html><script>document.createComment("foo").deleteData(0, 3);</script> |
| 20:34 | <AryehGregor> | I get "Out of memory; script terminated." in the error log, and script execution terminates. |
| 20:36 | <jgraham> | AryehGregor: Awesome |
| 20:37 | <jgraham> | AryehGregor: Would you like to file a bug? Otherwise I can |
| 20:37 | <AryehGregor> | jgraham, I don't file bugs in bug trackers where I can't see the status of the bug and have never been contacted for a follow-up. |
| 20:37 | <AryehGregor> | So go ahead. |
| 20:38 | <jgraham> | AryehGregor: We do contact people for follow-up information sometimes fwiw |
| 20:38 | <AryehGregor> | Hasn't happened to me. |
| 20:39 | <jgraham> | Well I guess you haven't filed hard-to-reproduce bugs, or something |
| 20:41 | jgraham | points out to anyone listening that the new set of polls is very unclear since the email is titled "Straw Poll For Objections", then says "In this case we are NOT |
| 20:41 | <jgraham> | asking WG members to give objections to a particular change proposal" |
| 20:41 | <jgraham> | and then goes back to talking about objections everywhere |
| 20:42 | <jgraham> | Including via the biolerplate text about not having to object if someone else already made your objection |
| 20:42 | <jgraham> | and in the text of the poll itself |
| 20:44 | <jgraham> | Oh, so I guess the point is that when it says "we are asking WG members to indicate their position", it actually means "give objections" |
| 20:45 | <jgraham> | Either that or I am more confused than I thought |
| 20:46 | <jamesr> | maybe objections are the only valid position |
| 20:47 | <Ms2ger> | jamesr, Objection! |
| 20:51 | <jgraham> | jamesr: Yes, I think a social scientist would have a field day documenting how we have devolved to the level where we can only communicate in the form of objections |
| 20:52 | <Ms2ger> | s/day/career/ |
| 20:53 | <zcorpan> | what i don't like with the phpbb3 theme is that it's hard to tell apart threads with new posts from threads without |
| 20:53 | <jgraham> | zcorpan: Surely you don't like more than just that? |
| 20:54 | <zcorpan> | yeah i don't like how approving/disapproving new posts is more effort than necessary |
| 20:55 | <zcorpan> | or did you mean just the theme? |
| 20:55 | <jgraham> | AryehGregor: Is there a spec for deleteData anywhere? |
| 20:55 | <jgraham> | Even a broken one? |
| 20:55 | <AryehGregor> | jgraham, http://dvcs.w3.org/hg/domcore/raw-file/tip/Overview.html#dom-characterdata-deletedata |
| 20:55 | <jgraham> | zcorpan: I mostly meant that it was ugly |
| 20:55 | <AryehGregor> | It doesn't look broken to me, looks reasonable. |
| 20:55 | <AryehGregor> | A pretty simple method, no? What's the question? |
| 20:56 | <jgraham> | AryehGregor: "where's the spec" was the question |
| 20:56 | <AryehGregor> | I meant, what did you need the spec for? |
| 20:56 | <zcorpan> | maybe Xdega will make a new theme with a nice design |
| 20:57 | <jgraham> | AryehGregor: To understand what deleteDatais supposed to do |
| 20:57 | <AryehGregor> | Ah, okay. That's a good reason to want a spec. :) |
| 20:58 | <zcorpan> | deleteData must run these steps: 1. Don't crash. 2. ??? 3. Profit! |
| 20:59 | <jgraham> | AryehGregor: OK, filed |
| 20:59 | <jgraham> | Thanks |
| 21:07 | <kennyluck> | jgraham: Did you check http://www.w3.org/profile/rdfa-1.1(re. So, is there some reason that having a registry of predefined prefixes wouldn't work for RDFa?) ? |
| 21:07 | <kennyluck> | http://www.w3.org/profile/rdfa-1.1 |
| 21:08 | <jgraham> | kennyluck: Fascinating |
| 21:13 | <AryehGregor> | Hmm, WebKit doesn't seem to let you made eval()d anonymous functions? data:text/html,<!doctype html><script>var f = eval("function() { alert('hello'); }"); f();</script> |
| 21:13 | <AryehGregor> | This works, though: data:text/html,<!doctype html><script>eval("var f = function() { alert('hello'); }"); f();</script> |
| 21:15 | <zewt> | stick it in parens |
| 21:15 | <zewt> | eg. same as how you have to say (function() { })() instead of function(){} () |
| 21:15 | <AryehGregor> | Weird, no other browser requires that. |
| 21:16 | <AryehGregor> | Thanks for the suggestion, that works. |
| 21:43 | <jgraham> | AryehGregor: That isn't a "WebKit" thing since different WebKit based browsers have different scripting engines |
| 21:43 | <AryehGregor> | Oh, right. |
| 21:43 | <AryehGregor> | I guess I'm just not testing Safari, then. :) |
| 21:57 | <AryehGregor> | I just constructed a test case in IE9 where a Node object has a non-null parent, but is not equal to any child of that parent. |
| 21:57 | <AryehGregor> | . . . |
| 21:58 | <AryehGregor> | Is it just me, or do these insane type of bugs really mostly occur in IE? |
| 21:59 | bfrohs | is going to remain silent because he bashes IE far too often |
| 21:59 | <Ms2ger> | The others fix them? :) |
| 22:08 | <karlcow> | Or people try harder to find bugs in IE |
| 22:09 | <AryehGregor> | I'm running the exact same test suite on all browsers. |
| 22:09 | <AryehGregor> | I generally review the results and try to get some idea of what's causing the failures in all browsers, in case it's a bug in the test or spec. |
| 22:09 | <jamesr> | it's not impossible to get that sort of crazy in WebKit (although i think we've fixed all the issues of that sorts i'm aware of) |
| 22:10 | <jamesr> | if that makes you feel any better |
| 22:10 | <AryehGregor> | jamesr, I found a case the other day where WebKit will produce a range with an offset longer than the length of the corresponding node. |
| 22:10 | <AryehGregor> | But that sort of bug is at least comprehensible. |
| 22:11 | <Ms2ger> | Though potentially scary |
| 22:11 | <AryehGregor> | data:text/html,<!doctype html><body><script>var text = document.createTextNode("Abcdef"); var range = document.createRange(); range.setStart(text, 0); range.setEnd(text, 2); text.splitText(1); document.body.textContent = range.startContainer.data + range.startOffset + range.endContainer.data + range.endOffset;</script> |
| 22:11 | <AryehGregor> | Didn't test in Safari, but in Chrome it outputs "A0A2". |
| 22:16 | <Lachy> | AryehGregor, what's the test case that shows the bug in IE? |
| 22:17 | <AryehGregor> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/895 |
| 22:17 | <AryehGregor> | Subject to the proviso that you'll have to extract the contents using some other browser and then paste them into IE, because otherwise it will trigger the XSS filter. |
| 22:18 | <AryehGregor> | It outputs "false" twice. |
| 22:18 | <AryehGregor> | Hmm, Chrome behaves oddly here too. |
| 22:19 | <Lachy> | ok. I'll run it in a sec, just got to try and load windows 7 in a virtual machine without it slowing everything else down |
| 22:20 | <AryehGregor> | Chrome doesn't output the w()'s unless you add w("foo") or something before the for loop. |
| 22:20 | <Ms2ger> | That's really annoying |
| 22:20 | <Ms2ger> | Aryeh, don't you know anyone at Google who could fix that? ^^ |
| 22:20 | <AryehGregor> | Ms2ger, fix what? |
| 22:21 | <AryehGregor> | I don't know anyone more at Google than you do. |
| 22:21 | <Ms2ger> | The w() thing |
| 22:21 | <AryehGregor> | Also, what's the story with getting DOM Range into the W3C these days? |
| 22:21 | <AryehGregor> | I don't know, it's a random glitch of some kind, no idea what causes it and no real desire to investigate. |
| 22:21 | <Ms2ger> | I'm wondering if I should wait until I can publish under a reasonable license |
| 22:22 | <AryehGregor> | Why don't you do like with HTML5 and just keep a separate version under a reasonable license? |
| 22:22 | <AryehGregor> | Also, why do you expect that the W3C will ever publish anything under a reasonable license? |
| 22:23 | <Ms2ger> | I'm not yet *that* cynical |
| 22:23 | <Ms2ger> | Also, busy |
| 22:24 | <Hixie> | christ these alt="" polls are confusing |
| 22:24 | <Hixie> | even more confusing that usual |
| 22:26 | <AryehGregor> | Do you view it as cynical to think that the W3C will refuse to publish specs under a reasonable license, when they have repeatedly refused to do so for HTML5? |
| 22:27 | <Ms2ger> | I like to AGF sometimes |
| 22:28 | <AryehGregor> | The W3C is acting in good faith, they just have a different idea of what constitutes a reasonable license. |
| 22:28 | <AryehGregor> | Viz., they don't believe in forking. Or at least some parts of the W3C don't, and so far they've prevailed. |
| 22:30 | <Philip`> | The W3C should publish the spec under an open license and let the people who don't believe in forking fork the spec and release their version under a more restrictive license |
| 22:31 | <Philip`> | That way everyone can be happy |
| 22:32 | <AryehGregor> | I encourage you to propose that solution to the PSIG. |
| 22:33 | <Ms2ger> | Call it option pi |
| 22:33 | <Lachy> | AryehGregor, the problem doesn't seem to be as bad as you thought. The problem is that this returns false: w(range.startContainer == range.startContainer); |
| 22:33 | <AryehGregor> | Oh, that's fine, then. |
| 22:34 | <Lachy> | which is still quite insane having an object that is not equal to itself, but at least is not a malformed DOM |
| 22:34 | <jgraham> | Philip`: Well the current situation is that the people who write the specs publish them under open licenses and then W3C forks the specs and releases them under restrictive licenses |
| 22:35 | <zewt> | jgraham: ... why bother? that's a no-op |
| 22:35 | <zewt> | since you can always fork from the open original, heh |
| 22:36 | <Ms2ger> | Not with a W3C logo |
| 22:37 | <zewt> | that should be a trademark/logo licensing thing, not a license on the spec itself |
| 22:38 | <jamesr> | you also couldn't take the W3C contributions, which they view as significant |
| 22:39 | <Philip`> | Lachy: What about "var s = range.startContainer; w(s == s)"? |
| 22:39 | <Philip`> | Lachy: (Maybe the object is equal to itself, but the 'startContainer' getter returns a new object each time) |
| 22:39 | <Philip`> | Lachy: (which sounds fairly sane to me) |
| 22:39 | <jgraham> | jamesr: Well we seem to, at least with HTML5. Maybe that wouldn't be permitted for less high profile specs |
| 22:40 | <Hixie> | there haven't been any significant contributions |
| 22:40 | <Hixie> | so that rather neatly solves that |
| 22:41 | <Lachy> | Philip`, s==s returns true |
| 22:43 | <AryehGregor> | Philip`, a new object that points to the same thing in the DOM? That doesn't sound at all sane to me. |
| 22:46 | <jgraham> | Hixie: If you have a moment, can you explain to me whether http://software.hixie.ch/utilities/js/live-dom-viewer/saved/897 can be explained by the current spec. Try also uncommenting the commented line. |
| 22:47 | <Hixie> | i have no idea what's going on there |
| 22:48 | <Hixie> | what am i looking for in particular? |
| 22:48 | <Hixie> | i get a different result each time i add a space to the source, in chrome |
| 22:48 | <Hixie> | oh it seems to just be the initial load that's acting differently |
| 22:49 | <Hixie> | jgraham: dunno what i'm looking for, please elaborate :-) |
| 22:51 | <jgraham> | Hixie: It seems that setting document.domain on the parent also affects the about:blank iframe so that one can still access e.g. the <body> in the iframe |
| 22:52 | <jgraham> | But messing with the parent dom seems to prevent this |
| 22:52 | <Hixie> | currently the spec cannot explain that, no |
| 22:53 | <Hixie> | though it probably wouldn't be too much work to fix that, if you can work out exactly what the rules are for the propagation |
| 22:53 | <Hixie> | should about:blank just use the origin of the parent directly rather than copying it when created? |
| 22:54 | <jgraham> | I don't think it is that simple otherwise setting innerHTML on the parent wouldn't affect the abillity to access the child |
| 22:54 | <jgraham> | I will look at this more tomorrow |
| 23:39 | <AryehGregor> | So obviously Microsoft is looking at implementing <iframe sandbox> for IE10. |
| 23:50 | <zewt> | gar i wish there was just a bit more design put into this new floaty-status-bar fad |
| 23:51 | <zewt> | both chrome's and ff4's assume every page has a light background and that it's okay to cover up the bottom corners of the page |
