| 01:45 | <csarven> | Is there a way for a browser to grab the contents at a URI fragment identifier? I was hoping to use <object data="http://example.org/#foo">; however, the browser seems to only treat the URI excluding the fragment |
| 15:22 | <jacobolus> | using the canvas API, is there any way to make an arbitrary shape via a bunch of transformations, but then have the stroke transform/scale be relative to pixels? |
| 15:23 | <jacobolus> | e.g. if I want to stroke an oval with a 5px wide outline, it'd be nice if the line could have uniform thickness |
| 15:24 | <jacobolus> | instead of the stroke getting distorted along with the shape |
| 15:25 | <tomasf> | not afaik. you'd have to scale the line width yourself to compensate. 1/transformScale, I guess |
| 15:26 | <Philip`> | You could set up the transform, draw the shape, apply the inverse transform, then stroke it, I think |
| 15:26 | <zewt> | tomasf: doesn't help if you have a non-1:1 scale on the object you're drawing and then that scale applies to the stroke as well, i think |
| 15:26 | <tomasf> | zewt: yep |
| 15:35 | <zewt> | just resetting the transform to identity seems to work: http://zewt.org/~glenn/stroke.html |
| 15:35 | <zewt> | simpler version of what philip said, i think |
| 15:41 | <jacobolus> | Philip`: I tried doing setTransform(1, 0, 0, 0, 1, 0) and that seemed to make the whole thing stop working |
| 15:41 | <jacobolus> | zewt: oh, hmm |
| 15:41 | <jacobolus> | maybe I just put a typo in someplace or something |
| 15:41 | <zewt> | 1,0,0,1,0,0, not 1,0,0,0,1,0 |
| 15:42 | <zewt> | the argument order for that function is odd |
| 15:42 | <zewt> | (column-major, rather unintuitive) |
| 15:43 | <jacobolus> | oh, I see |
| 15:43 | <jacobolus> | that is weird |
| 15:43 | <zewt> | i take it stroke is really just transforming the line width by the transform to figure out the resulting width and then doing the actual operation in screen coordinates |
| 15:43 | <jacobolus> | I'm somewhat annoyed that these don't let people use a 3x3 matrix, but that ship has sailed I guess :) |
| 15:44 | <zewt> | it wouldn't have, if it was row-major :) |
| 15:45 | <jacobolus> | heh |
| 15:45 | <zewt> | two ships that sailed together |
| 15:45 | <jacobolus> | maybe they could just order it M00, M10, M20, M21, M11, M01, M02, M12, M22, for maximal confusion |
| 15:46 | <jacobolus> | "snake order" |
| 15:47 | <jacobolus> | great, that seems to work |
| 15:47 | <jacobolus> | zewt: thanks |
| 15:47 | <jacobolus> | Philip`, tomasf: thanks to you too :) |
| 15:49 | <jacobolus> | so now I have something like http://pastie.textmate.org/private/wgeqzabrzkghnsphwydsa |
| 15:50 | <Philip`> | zewt: It's not just multiplying the line width - you can have non-uniform scales or skews that affect the stroke |
| 15:50 | <zewt> | i don't even know what language this is, heh |
| 15:50 | <jacobolus> | zewt: coffeescript |
| 15:50 | <jacobolus> | :) |
| 15:50 | <zewt> | "obscure fadscript" :) |
| 15:51 | <jacobolus> | zewt: output, http://pastie.textmate.org/private/g09sud4d8ewurqptyialuq |
| 15:51 | <jacobolus> | zewt: less and less obscure all the time, and in my (few weeks) experience, a joy to work with |
| 16:02 | <jacobolus> | that the path shape is set according to the original transformation but the stroke distorted by the stroke-time transformation allows some interesting stuff |
| 16:03 | <jacobolus> | for example, in conjunction with round joins and line ends, can be used to have the stroke shape be an arbitrary ellipse |
| 16:03 | <jacobolus> | "caligraphic" style |
| 16:04 | <jacobolus> | er, *calligraphic |
| 16:07 | <jacobolus> | is all the "focus management' stuff in the canvas 2d context spec implemented widely? |
| 16:07 | <Ms2ger> | Not at all |
| 16:07 | <jacobolus> | one of those "someday" things? |
| 16:08 | <Ms2ger> | I hear there's an HTMLWG decision upcoming about it |
| 16:11 | <smaug____> | focus management in canvas? |
| 16:11 | <smaug____> | or a11y? |
| 16:11 | <jacobolus> | smaug____: http://dev.w3.org/html5/2dcontext/#focus-management |
| 16:11 | <jacobolus> | oh |
| 16:11 | <smaug____> | oh, huh. yet more to review |
| 16:12 | <jacobolus> | trying to draw focus rings in canvas seems like somewhat silly idea to me |
| 16:13 | <zewt> | yeah, it's not like native focus styles apply to curves, and many of them aren't simply dotted lines around objects like links |
| 16:14 | <zewt> | giving UA some control over it isn't a bad idea, but i doubt there's much it could actually do other than pick a stroke style |
| 16:15 | <smaug____> | drawFocusRing looks pretty useless to me |
| 16:15 | <zewt> | heard something about that being simplified, but i havn't been following canvas stuff |
| 16:16 | <jacobolus> | Hixie: by the way, CIECAM02 implementation https://github.com/jrus/chromatist |
| 17:25 | <AryehGregor> | There is a link on a page to a PDF I want to read, but it goes to http://10.58.170.5/. |
| 17:25 | <AryehGregor> | I mean, seriously? |
| 17:26 | <AryehGregor> | This is a public website: http://www1.cbs.gov.il/reader/?MIval=cw_usr_view_SHTML&ID=576 |
| 17:27 | AryehGregor | guesses the correct URL from nearby URLs |
| 17:42 | <zewt> | the "worksforme" testing methodology |
| 17:45 | <Philip`> | I like how pressing shift+down several times in Opera on a Google search results page causes it to select the Google logo image, and then to begin scrolling down within the image so you see a load of random icons in its place |
| 17:46 | Philip` | wonders if there will ever be non-hacky ways to do spriting |
| 17:46 | <zewt> | "no hacks to see here, folks" |
| 23:26 | <jacobolus> | is there a good way to clear a <canvas> without altering its context? |
| 23:28 | <jacobolus> | I guess I can do something like save context, zero out transform, clearRect the whole width/height, restore context |
| 23:30 | <bga_> | canvas.width = canvas.width |
| 23:31 | <jacobolus> | bga_: that kills my context |
| 23:31 | <bga_> | dirty hack by @kangax |
| 23:31 | <jacobolus> | bga_: I don't think I can do a save/restore around that can I? |
| 23:31 | <jacobolus> | I guess I can try and see |
| 23:32 | <jacobolus> | also, no longer too much a "dirty hack" when it's widely documented as the right way to do it :) |
| 23:33 | <bga_> | jacobolus gc do not kills your context => you will not lost it |
| 23:33 | <bga_> | i guess |
| 23:33 | <jacobolus> | I'm not sure what you mean |
| 23:34 | <bga_> | gc == garbage collector |
| 23:34 | <jacobolus> | if I do canvas.width = canvas.width, at least in Safari, my transform matrix is reset to the identity |
| 23:34 | <bga_> | heh |
| 23:35 | <jacobolus> | I can't save/restore around it either |
| 23:35 | <jacobolus> | so I'm going to try my way |
| 23:35 | <jacobolus> | :) |
| 23:36 | <bga_> | but you can get second clear context :) |
| 23:36 | <bga_> | may be its faster way |
| 23:36 | <jacobolus> | hm? |
| 23:36 | <jacobolus> | not sure what you mean |
| 23:37 | <jacobolus> | I have a context stack that I don't want to destroy, I just want to clear the whole canvas image |
| 23:37 | <bga_> | one canvas can have multi contexes |
| 23:37 | <bga_> | with different states |
| 23:38 | <bga_> | and types(2в/3d) |
| 23:38 | <jacobolus> | I can have multiple 2d contexts on a canvas? |
| 23:38 | <bga_> | *2d |
| 23:38 | <bga_> | yeah |
| 23:38 | <jacobolus> | how does that work? |
| 23:39 | <jacobolus> | I don't see any mechanism in the spec for switching contexts |
| 23:39 | <bga_> | sec |
| 23:41 | <bga_> | oh |
| 23:41 | <bga_> | fail in chome |
| 23:41 | <bga_> | context is singleton :( |
| 23:42 | <bga_> | soкry |
| 23:42 | <bga_> | *sorry |
| 23:45 | <bga_> | jacobolus another way to avoid save/restore state is do `clearRect` with infinite bounds %) |