freenode :: #whatwg

20 Apr 2017
05:02annevkjyasskin: ta, wonder if that is something we should offer control over at some point
06:54zcorpanTabAtkins: i now get a link error in cssom-view that isn't marked for export, but it looks like it should be?
07:13TabAtkinszcorpan: Looks good to me with a fresh update?
07:15zcorpanSimons-MBP:cssom-view zcorpan$ bikeshed
07:15zcorpanLINK ERROR: No 'dictionary' refs found for 'MouseEventInit' that are marked for export.
07:15zcorpanTabAtkins: i just updated
07:34zcorpanwonder what&#39;s up with this id: <p id=&quot;date-state-%28type=date%29&quot;>When an <code>input</code> element&#39;s <code data-x=&quot;attr-input-type&quot;>type</code> attribute is in
07:34zcorpan(only id with % in source)
07:43annevkzcorpan: would ( and ) trip something up?
07:45zcorpanannevk: a few lines above has <h6 id=&quot;date-state-(type=date)&quot;><dfn data-x=&quot;attr-input-type-date&quot;>Date</dfn> state (<code data-x=&quot;&quot;>type=date</code>)</h6>
07:45* zcorpan runs a blame
07:47annevkoh, I wonder if it&#39;s a workaround for crappy browsers
07:50zcorpanb29fb58ce6139a1e297a915e7291e637c3e1b561 added these ids
07:51zcorpanthe link on caniuse is - not percent-escaped
08:49MikeSmithdo we have branch snapshots for HTML?
08:50annevkMikeSmith: I don&#39;t think so, we should probably start poking at the tooling again a bit and maybe move more of it to Travis
08:51MikeSmithannevk: OK
08:51MikeSmithis there a way I can force a snapshot build for a PR?
08:51MikeSmiththat is, by hacking on the .travis.yml on the branch maybe?
08:51MikeSmithor on the branch?
08:52* MikeSmith looks while hes waiting
08:53annevkMikeSmith: no, just locally
08:53MikeSmithoh I see no deploy stuff gets called from Travis for the HTML spec
08:53annevkMikeSmith: right
08:53annevkMikeSmith: it&#39;s the only spec we build on the server making it a bit of a liability
08:54MikeSmithwell for now I just want to be able to generate multipage output with a change for yall to look at
08:55MikeSmithI guess I can just scp it to my own server
09:07annevkAh yeah, that&#39;s what I do
09:23* MikeSmith adds a bunch of hashes to his CSP policy for all the inline style elements the HTML spec has
10:13* MikeSmith finaally had to resort to &#39;unsafe-inline&quot; and &#39;unsafe-eval&#39;
11:21timwismorning folks. had a thought regarding custom elements.. couldn&#39;t i theoretically make my own custom elements via $(&#39;my-el-name&#39;).html(&#39;....&#39;) ? and maybe set some css on it to display: inline-block etc.
11:22timwisi know custom element spec provides more features but for basic stuff like abstracting a complex widget&#39;s html, it seems like this is a much lighter solution with a lot of browser support, no?
11:26annevktimwis: if you don&#39;t need complexity then you don&#39;t need it...
11:27timwisannevk: it being custom element spec?
11:30annevktimwis: yeah, there&#39;s always multiple ways to do things, what is best depends on the requirements and use cases
11:31timwisannevk: i guess my question is whether there&#39;s anything bad about using a fake tag. pretty sure that&#39;s an HTMLUnknownElement. and just adding functionality to it
11:32annevktimwis: can&#39;t think of much right now
11:59timwisthanks annevk. Just threw together a proof of concept: (demo:
12:48Domenictimwis: it&#39;s confusing for other developers looking at your code. If you&#39;re just using a glorified div, then you should use the div tag.
12:49annevkThat was the one thing I could think of, but I figured it might be fine since the semantics are locally-bound anyway
12:53annevkweb-platform-tests doesn&#39;t appear to have tests for Navigator objects...
12:56DomenicIt does
12:57annevkOh right, I somehow skipped right over that in my grep results
13:16annevkDomenic: so the way we deal with tearoff objects from Window seems really quite bad, they all have their own language and model, some based on document, some on the Window object, but no consistent language
13:36Domenicannevk: agreed
13:36timwisDomenic: ah, you mean like <div data-component=&quot;phl-button&quot;></div>
13:37Domenictimwis: sure, or just use class=&quot;&quot;
13:37Domenicor data-button
13:37timwisis the only downside to using <phl-button> instead that it&#39;s less obvious what&#39;s going on to people reading the code?
13:37DomenicThere are probably a few other minor ones, including it being non-conformant HTML, but that&#39;s the major one.
13:38Domenicannevk: the one thing I&#39;m on sure on is whether there should be a central location saying &quot;Window has an associated X, Y, and Z&quot; or if each object should define that in its definition.
13:38timwisDomenic: sorry if this is an obvious answer, but what&#39;s wrong with it being non-conformant HTML?
13:39annevkDomenic: not sure on?
13:40annevkDomenic: I&#39;m fairly convinced still we want formalized slots and informal slots until that day comes
13:40annevkDomenic: the current haphazard way of describing relations between objects and state makes that very clear to me
13:42Domenicannevk: not sure whether WIndow should describe all its slots, or if e.g. Navigator should say &quot;each window has an associated Navigator object&quot;
13:45annevkDomenic: I think the latter style is confusing
13:46annevkDomenic: it&#39;s acceptable, but not how I&#39;d like things to be
13:47annevkDomenic: what might work is that IDL takes care of most of it once we have formalized slots as I proposed long ago
13:47DomenicI&#39;m still unclear on the benefits of formalized ones over informal ones.
13:47annevkDomenic: so that when a Window object is allocated, IDL allocates a Navigator and Location object as well
13:47annevkDomenic: in the same global
13:48annevkDomenic: and then puts them in their respective slots, which navigator and location would end up returning from
13:48DomenicHmm, I never realized there was an implicit allocation of slot values implied.
13:48annevk(most of that could effectively be codegen at that point)
13:49annevkThat&#39;s an idea for extending formal slots
13:49annevkAnd making things easier to define and more robust by default
14:10jyasskinWhy is it easier to auto-create Navigator and Location into a formal slot than into their readonly attributes?
14:15annevkjyasskin: the latter doesn&#39;t work? That&#39;s just a getter
14:17jyasskinannevk: It seems like the same sort of webidl wording that would auto-create into slots could also specify that readonly attributes, maybe marked by an [ExtendedAttribute], have an instance in an implicit slot by default.
14:17annevkjyasskin: sure, that&#39;s a valid impl of formalized slots
14:17jyasskinAh, SG.
14:18annevkjyasskin: I&#39;m just talking at a higher-level here
16:04DomenicTabAtkins: what would be a good way to get on the CSSWG&#39;s radar? File an issue on maybe? Mailing list? Trying to get some broad discussion.
17:59jsbellannevk: Since it&#39;s part of the Blink ship process I filed a TAG spec review request for Storage. Hopefully any feedback given will be useful/constructive.
18:05annevkjsbell: I thought we already did that once?
18:05annevkjsbell: nothing really changed since then
18:06jsbellannevk; Oh! Hrm....
18:08jsbellannevk: Huh... for some reason I thought I saw a note saying we intentionally hadn&#39;t. Maybe I was dreaming. :P Awesome, I&#39;ll close, never mind and thx
21:58astearnsDomenic: I&#39;ll add 5625 to today&#39;s meeting agenda (TabAtkins)
21 Apr 2017
No messages
Last message: 156 days and 18 hours ago