IRC logs for #respimg, 2014-06-24 (GMT)

[00:31:41]* gregwhitworth has joined #respimg
[06:32:07]* SteveF has joined #respimg
[08:53:34]* darobin has joined #respimg
[09:16:05]* javery has joined #respimg
[09:17:19]* kasperg has joined #respimg
[09:17:47]* SteveF has joined #respimg
[11:57:36]* yoav has joined #respimg
[12:05:23]* TimWright has joined #respimg
[12:40:18]* yoav has joined #respimg
[12:43:51]* kasperg_ has joined #respimg
[13:22:33]* kasperg has joined #respimg
[13:23:56]* kasperg__ has joined #respimg
[13:24:56]* kasperg has joined #respimg
[13:25:07]<yoav>marcosc_: Around?
[13:36:02]* kasperg_ has joined #respimg
[13:47:02]* kasperg_ has joined #respimg
[13:47:02]* MikeSmith has joined #respimg
[14:11:47]* uniqname has joined #respimg
[14:14:17]* yoav has joined #respimg
[14:18:28]<marcosc__>yoav: yep
[14:19:09]<yoav>marcosc__: Hey. Wanted to talk to you about the manifest, the fact that it's async and the orientation member
[14:19:22]<yoav>But in the mean time, wrote up a GH issue :)
[14:19:38]<marcosc>will take a looksy
[14:20:13]<yoav>marcosc: Awesome, thanks!
[14:20:23]<yoav>Gotta go now, but be back in 30
[14:24:59]* Wilto has joined #respimg
[14:33:53]<Wilto>Whatup; what’ve I missed?
[14:41:25]* newtron_ has joined #respimg
[14:56:16]* gregwhitworth has joined #respimg
[14:57:05]<gregwhitworth>yoav: I think I saw someone say that you will be updating the srcset to trigger downloads when a user resizes the browser. Is this correct?
[14:58:30]* marcosc_ has joined #respimg
[14:59:01]* eeeps has joined #respimg
[15:08:56]<Wilto>gregwhitworth: Yeah, we’re waiting on another patch to land first.
[15:09:23]<Wilto>gregwhitworth: cbiesinger__ has been working on it.
[15:10:03]<gregwhitworth>Ok, so is it all based on viewport not zooming
[15:10:26]<gregwhitworth>basically it will only trigger when I resize the browser. Not when I control + '+'
[15:10:35]<gregwhitworth>currently zooming doesn't affect it
[15:17:29]<gregwhitworth>wilto: forgot to put your handle so it probably isn't alerting you
[15:20:24]<yoav>gregwhitworth: Don't listen to Wilto, he's bad for you :D
[15:20:56]<yoav>The picture sources are planned to get updated when the viewport changes, and cbiesinger__ is working on that
[15:21:22]<gregwhitworth>yoav: Ok, so viewport only then. How it currently behaves now, it just won't require refresh
[15:21:35]<yoav>There aren't such plans for srcset at the moment, but there may be in the future
[15:21:56]<gregwhitworth>yoav: Hmmmm, so you're specifically referring to picture sources then
[15:21:57]<yoav>Yeah, exactly
[15:22:33]<gregwhitworth>yoav: hmmmm, decisions, decisions
[15:22:38]<yoav>For srcset resources, you could make a case either way
[15:22:50]<yoav>And it's the UA's decision
[15:23:19]<gregwhitworth>yoav: yeah that's our problem at the moment. It is the UAs decision but devs usually want a similiar behaivor
[15:23:38]<gregwhitworth>yoav: I think it would be good to detail this if possible to one another
[15:24:37]<yoav>I don't think it'd be bad if UAs are not compatible here, but I'm all for full documentation
[15:25:04]<gregwhitworth>yoav: I understand why they don't have to be and benefits for it (perf, etc)
[15:25:06]<yoav>gregwhitworth: I also have plans to make resource selection (inside srcset) smarter. See
[15:25:07]* nwtn has joined #respimg
[15:26:32]<gregwhitworth>yoav: I do think you should modify the spec, this is a good example, but that's my opinion
[15:27:27]<yoav>gregwhitworth: The current dev advice we give out is "don't assume anything regarding the actual resource that the browser will request inside srcset". Time will tell if devs would listen
[15:27:51]<gregwhitworth>yoav: man that's rough
[15:28:20]<Wilto>what i do
[15:28:55]<gregwhitworth>yoav: I can understand on mobile possible changing how this works, but on desktop I think we should try to be as consistent and intuitive for the devs
[15:29:20]<gregwhitworth>my grammar and spelling are off today
[15:29:32]<Wilto>I mean, `srcset`/`sizes` whole deal is “when you don’t need super explicit control over sources.”
[15:29:47]<Wilto>“Slap some images in here and the browser will take care of it.”
[15:30:09]<Wilto>`picture`/`source` is for cases that require tight control.
[15:30:30]<gregwhitworth>Wilto: ok
[15:30:47]<yoav>gregwhitworth: What would you say be a more intuitive way to go?
[15:30:53]<gregwhitworth>Wilto: that seems somewhat odd to me, but maybe I'm just a type A person and want my x == x
[15:31:12]<Wilto>The 1x/2x syntax blurs the line a little, but you can generally say “`srcset` when you don’t wanna think about responsive images, `picture` when you do.”
[15:31:28]<Wilto>gregwhitworth: Oh, don’t get me wrong—same here.
[15:31:41]<Wilto>Big hero images, art direction, etc. I lean `picture` still.
[15:31:43]<gregwhitworth>yoav: if I say srcset="img1.jpg 500w, img2.jpg 750w, 850w" and my viewport is 860, you better grab the 850
[15:32:19]<gregwhitworth>yoav: That's speaking as a web dev and what I would expect
[15:32:47]<yoav>gregwhitworth: As someone writing this from a desktop, tethered on a sometimes lousy 3G, I disagree :)
[15:33:37]<Wilto>Yeah, my cafe wifi is with yoav.
[15:33:38]<yoav>As a user, I want to be able to tell my browser "just download crappy images now, I don't care"
[15:33:48]<Wilto>But, there are cases for that explicit control too.
[15:34:01]<gregwhitworth>I agree with both of you
[15:34:28]<Wilto>Y’know, gregwhitworth, if I didn’t know any better, I’d say we were discussing some preliminary implementation details here WINK WINK
[15:34:36]<gregwhitworth>A good example of this though, I'm a dev and say that I want to download 4 versions of jquery, two mootools, etc - do you guys do it?
[15:34:46]<gregwhitworth>Just curious
[15:35:12]<gregwhitworth>Even if network constraints are known
[15:35:34]<Wilto>I mean, we’re on weird turf here, comparitively.
[15:35:44]<Wilto>Because we *have* to assume the experience will suffer if we omit a script.
[15:35:47]<Wilto>Something will likely break.
[15:36:15]<Wilto>Big vs. small images on big vs. small screens, most options give us no perceptable difference, as a user. Just more damage to a data plan.
[15:36:48]<Wilto>We’re onto weird new turf with these stretchy images of ours.
[15:37:15]<Wilto>Which is why both cases need to be covered: the explicit heuristics case, and the “look, just figure it out for me” case.
[15:37:28]<Wilto>I mean, I’m pumped to see where things go with the new netinfo API for the sake of these.
[15:37:58]<Wilto>I want that “opt me out of high res images” toggle in the worst way. Again, see: this wifi connection.
[15:38:06]<gregwhitworth>Ok, so Blink is treating srcset as "grab me the best picture you think" and then using <picture> with sources as the art direction piece that you WILL adhere to.
[15:38:19]<Wilto>Yep. `picture` is as strict as things get on the web.
[15:38:29]<Wilto>I mean, there’s some MUSTs in there, I think.
[15:39:37]<yoav>gregwhitworth: Yeah, picture is the "download what the web dev is saying" case, and srcset is the "download what's best for the user" case
[15:39:46]<gregwhitworth>good to know
[15:39:54]<gregwhitworth>Thanks for all of the help guys
[15:40:13]<Wilto>Anytime, dude.
[15:40:27]<Wilto>We’ve made a weird little thing, here. Happy to help make sense of it.
[15:40:27]<yoav>Currently srcset doesn't have that much smarts in it, but I plan on getting some smarts in there soon
[15:42:51]* newtron_ has joined #respimg
[15:45:52]<Wilto>`srcset` if you’re lazy, `picture` if you’re crazy™
[15:54:39]<JonathanNeal>I love srcset.
[16:01:57]* Wilto has joined #respimg
[16:05:55]* uniqname has joined #respimg
[16:15:23]* marcosc has joined #respimg
[16:18:49]* kasperg has joined #respimg
[16:38:17]* jlembeck has joined #respimg
[16:42:19]* gregwhitworth has joined #respimg
[16:42:27]<gregwhitworth>is anyone from apple in here?
[16:47:18]<cbiesinger__>yoav: did you see my conversation with elliott last night?
[16:47:25]<TabAtkins>gregwhitworth: It's not "Blink is doing this", it's "that's what the spec actually says to do".
[16:47:54]<TabAtkins>You *must* pick the first <source> with a matching MQ, and reevaluate that choice when the MQ truthiness changes.
[16:49:02]<TabAtkins>Once you've chosen a <source> (or the <img>), it's entirely browser-choice which source to pick, based on whatever criteria they find relevant. "Screen resolution" is definitely intended to be one, but "bandwidth situation" is also one we anticipate browsers using.
[16:53:36]<gregwhitworth>TabAtkins: Blink currently doesn't update when the truthiness changes
[16:53:52]<TabAtkins>And that's a bug.
[16:53:55]<gregwhitworth>TabAtkins: Specifically in regard to srcset
[16:54:02]<TabAtkins>Wait wait wait.
[16:54:27]<TabAtkins>Be clearer, please? MQ truthiness and srcset are quite unrelated - those are the two distinct directions I just talked about.
[16:55:32]<gregwhitworth>TabAtkins: I figured we were talking about two different things
[16:56:07]<gregwhitworth>TabAtkins: I am merely referring to if I have the following: srcset="http://placehold.i/500 1800w, 1200w"
[16:56:19]<TabAtkins>Okay, yeah, that has nothing to do with MQs
[16:57:18]<TabAtkins>You might be one of those unfortunates who has mixed up "old HTML srcset" with "new <picture> srcset", and have gotten confused about what the XXXw syntax means.
[16:58:09]<TabAtkins>XXXw is just an alternate way of writing NNNx, when the precise x value varies because the image's size is variable.
[16:58:36]<TabAtkins>You just give up, tell the browser how big the source is (the w number) and how big the image will be (the sizes attribute) and let it figure things out.
[16:59:08]<gregwhitworth>hmmmm, it definately still feels dimension based not pixel density based
[16:59:24]<gregwhitworth>And yes I'm looking at the W3C srcset spec
[16:59:29]<TabAtkins>I suggest rereading the spec for it, then, because it's nothing to do with the size of the viewport. ^_^
[16:59:35]<TabAtkins>Why. Why are you looking there.
[16:59:48]<gregwhitworth>You're being sarcastic right
[16:59:56]<TabAtkins>The W3C spec hasn't updated to contain the <picture> information, I think?
[17:00:17]<TabAtkins>Just read
[17:01:02]<TabAtkins>Until we're sure that the W3C copy contains the correct text from the WHATWG copy, the original source is fine.
[17:02:33]<TabAtkins>bbiab - have to do an interview
[17:06:26]<Wilto>Man, that URL is KILLING me.
[17:07:21]<Wilto>gregwhitworth: That doc is left over from the now-ancient WHATWG `srcset` proposal and REFUSES TO DIE.
[17:08:02]<Wilto>I knew we should have made the attribute `srcset-v2` in the `picture` spec.
[17:12:06]* kasperg has joined #respimg
[17:14:39]* SteveF has joined #respimg
[17:19:32]* marcosc has joined #respimg
[17:23:14]* Wilto has joined #respimg
[17:30:40]* marcosc has joined #respimg
[17:32:25]* Wilto has joined #respimg
[17:55:41]* Wilto has joined #respimg
[17:56:25]* kasperg has joined #respimg
[18:07:59]<JonathanNeal>I could just imagine if attribute names had versions and or vendor prefixes.
[18:37:41]<gregwhitworth>Wilto: Ok, so we should be using srcset from the picture spec then as the definitive guide
[18:39:19]<gregwhitworth>Wilto: It says it was updated June of this year
[18:45:08]<Wilto>Yeah, not sure what the deal is there, gregwhitworth.
[18:45:26]<Wilto>Definitely defunct.
[18:48:10]<gregwhitworth>We really need to figure this out
[18:48:21]<Wilto>But it shouldn’t be around much longer with things merged into the WHATWG spec.
[18:48:40]<Wilto>This is a darobin/hober question.
[18:49:06]<gregwhitworth>I hate them being in two places
[18:49:13]<Wilto>You’re tellin’ me.
[18:49:16]<gregwhitworth>I know I'm not the first
[18:49:26]* uniqname has joined #respimg
[18:49:45]<gregwhitworth>Thanks for the info though
[18:49:46]* kasperg_ has joined #respimg
[18:50:27]* SteveF has joined #respimg
[18:53:58]* darobin has joined #respimg
[19:02:26]* uniqname1 has joined #respimg
[19:13:48]* gregwhitworth_ has joined #respimg
[19:23:47]<TabAtkins>gregwhitworth: We did figure this out, but MS won't let you use the solution. ^_^
[19:27:36]* zcorpan has joined #respimg
[19:28:43]<zcorpan>gregwhitworth: w3c HTML 5.1 has new srcset and picture
[19:33:26]<zcorpan>The w3c srcset extension spec is broken until hober gets his act together
[19:52:25]* marcosc has joined #respimg
[20:13:52]<gregwhitworth>I really wasn't wanting to get into a turf war here. There being two locations for "official" specs is problematic
[20:16:31]<gregwhitworth>TabAtkins: I really need to meet you as it is quite hard to discern your attitude in text. :)
[20:26:26]<johns>" Chrome Canary is currently not available on the linux platform. " srs
[20:50:52]* kasperg has joined #respimg
[20:59:07]* Wilto has joined #respimg
[22:49:43]* gregwhitworth has joined #respimg
[22:59:27]<johns>does <media-condition> in the picture text refer to the CSS4 spec?
[22:59:40]<johns>It doesn't seem to exist in the CSS3 MQ text
[22:59:45]<johns>but the spec links to that
[23:09:55]<MikeSmith>johns: yes
[23:10:11]<MikeSmith>johns: the level3 spec is obsolete
[23:11:52]<TabAtkins>johns: Where does <picture> link to MQ3?
[23:13:31]<johns>TabAtkins: The "<media-condition>" leads to which links
[23:16:42]<TabAtkins>Oh, whoops, we do indeed biblio over to MQ3. Darn.
[23:23:13]<MikeSmith>I just now raised a bug
[23:23:22]<MikeSmith>see also
[23:25:03]<MikeSmith>we'll need for Hixie to make a fix but in the mean time also need zcorpan to add some text to work around it for now until Hixie has time to make the general update
[23:39:08]* marcosc has joined #respimg