00:00
<tantek>
general problem with arguments of the form "not something … in my experience"
00:00
<zewt>
(sorry, i just don't think this is worth replying to)
00:02
<Hixie>
ok, critics, start your engines: http://www.whatwg.org/specs/web-apps/current-work/#table-sorting-model
00:02
<zewt>
i need to hack something up in firefox so if it ever is about to load the html spec, it punts the url to chrome
00:03
<annevk>
Hixie: what might be kind of unexpected is that setting a sorted attribute and then removing it will not return the table to its original state
00:03
<Hixie>
that is true, but i have no idea how to change that
00:04
<tantek>
per http://utminers.utep.edu/omwilliamson/ENGL1311/fallacies.htm - Ad Hominem Argument "[…] Also applies to cases where potential opposing arguments are brushed aside without comment or consideration, as *simply not worth arguing* about"
00:05
<annevk>
Hixie: crazy idea: sort the render boxes
00:05
<annevk>
nn
00:05
<Hixie>
yeah, i considered that
00:05
<Hixie>
nn
00:06
<annevk>
oh, and before I forget, happy b-day!
00:06
<Hixie>
thanks :-)
00:06
<tantek>
Hixie, today's your b-day? Happy b-day!
00:07
<Hixie>
today in europe
00:07
<Hixie>
tomorrow here
00:07
<tantek>
UTC?
00:07
<Hixie>
27th
00:11
<Hixie>
if anyone here has IE, could you let me know what the Log says on this page? http://software.hixie.ch/utilities/js/live-dom-viewer/saved/2047 (any version of IE will do)
00:14
<zewt>
log: false
00:14
<zewt>
log: 1000
00:14
<zewt>
log: false
00:14
<zewt>
rendering mode: CSS1Compat
00:14
<zewt>
document has no title
00:14
<zewt>
log: true
00:14
<zewt>
ie9
00:14
<Hixie>
thanks
00:25
<gsnedders>
Hixie: Happy birthday, in the unlikely circumstance you are in this part of the world!
00:29
<Hixie>
thanks :-)
01:18
<Hixie>
https://www.w3.org/Bugs/Public/show_bug.cgi?id=20524 <-- people with concrete ideas for how to sort, please comment here
01:25
<dydx>
annevk: you there?
09:13
<Ms2ger>
Did rniwa have tests for the OverrideBuiltins stuff?
09:17
<Ms2ger>
And happy birthday, Hixie
11:24
annevk
is here now....
11:30
<annevk>
Hixie: concept-node-remove is lacking a <dfn>
11:30
<annevk>
Hixie: what's also lacking is "tree order" where you invoke it
11:30
<annevk>
Hixie: that's kinda important for the mutation records that will be created
11:31
<annevk>
Hixie: did you dismiss sorting layout boxes because Selectors would be fucked? that was a pretty big negative I could think of
11:39
<Velmont>
So IIUC <table sortable><tr sorted=1>Age <tr sorted=2>Name to sort by Age, then Name? :-)
11:39
<annevk>
Hixie: <td><time datetime="2011">MMXIV</time> metadata mismatch
11:40
<annevk>
Velmont: if you add a <tr> and change your tr to th, yes
11:41
<Velmont>
Uh, yeah. I didn't mean to write that... <table sortable><tr><th sorted=1>Age <th sorted=2>Name
11:41
<Velmont>
Let's blame early morning etc :S :D
11:43
<annevk>
Hixie: #type-and-value-of-the-cell should also ignore ProcessingInstruction I think
11:43
<annevk>
Velmont: early? yeah...
11:46
<annevk>
Hixie: disregard the earlier comment on concept-node-remove...
11:46
annevk
forgot to refresh
11:47
<Velmont>
annevk: Well, I just woke up at least :-) I wonder how we'll render the "sorted by first, second" hints. Some authors inevitably want to override that as well.
11:48
<annevk>
Velmont: idea yesterday was to only show the primary key in the UI, put the arrow in ::after, and have some kind of pseudo-class that applies to the <th> element that is the primary key
11:51
<Velmont>
annevk: There's bound to be lots of confusion by that. E.g. using the Django admin, which has that functionality already (but serverside), it shows 1> 2> when you have sorted on several, and you can remove the sort by clicking on it again.
11:52
<Velmont>
Either the sort should only do one-column when invoked by the user, -- so that clicking a new header would cancel the old one. Or it needs UI.
12:07
<Ms2ger>
Hey Velmont, want to write some tests? :)
12:07
<Ms2ger>
Anne here had an interesting conversation about OverrideBuiltins yesterday
12:23
<Velmont>
Ms2ger: Working on MediaGoblin right now. :]
12:43
<annevk>
I think I'm finally learning enough about JavaScript to understand where Alex is coming from...
12:44
<annevk>
new document.constructor not working is indeed kinda odd
14:52
<annevk>
I wonder if there's some way we could make a custom sort comparator work by executing the callback in some kind of sandboxed environment...
14:53
<annevk>
But maybe that's too much complexity for little gain...
17:25
<zewt>
annevk: workers are pretty much our sandboxed environment (but yeah, that'd be very complex)
17:28
jwalden
is reminded of a testcase he needs to write/patch, maybe
18:31
<Hixie>
annevk: concept-node-remove has a dfn as far as I can tell, did I make a mistake?
18:33
<Hixie>
Velmont: you can do the rendering of hints for subkeys using [sorted=~"2"] etc
18:34
<Hixie>
annevk: the comparator needs access to the HTMLTableCellElements, unfortunately, which pretty much precludes any sandboxing
19:05
<jwalden>
any spec experts on GSP and maybe WebIDL such around? I'm wondering what's supposed to be the specified printout for http://software.hixie.ch/utilities/js/live-dom-viewer/saved/2049 -- would guess .constructor would be directly on |window|, and the frame-name would only appear if property lookup hit the GSP, which it thus wouldn't, but that's pure guesswork
19:05
<jwalden>
right now it appears engines save for WebKit print [object Window]
19:06
<Ms2ger>
Correct, I think
19:08
<jwalden>
one of my patches removing a use of JSRESOLVE_ASSIGNING I think fixes this for Gecko, noticed while thinking about testcase-writing and such
19:08
Ms2ger
looks
19:08
<jwalden>
interesting to note that window.constructor in WebKit isn't a function, apparently
19:08
<Ms2ger>
I might have a test for this...
19:08
<jwalden>
I haven't posted the patch yet
19:08
<jwalden>
because of not-tested yet :-)
19:09
<jwalden>
well, not-automatic-testcase-tested yet :-)
19:12
<Ms2ger>
Hrm
19:12
<Ms2ger>
My test claims constructor should be an own property on Window.prototype
19:13
<jwalden>
erm, actually, yes, that's correcter behavior :-)
19:14
<jwalden>
as concerns that specific test I don't think it distinguishes, just that it's on an object before the GSP in |window|'s [[Prototype]] chain
21:48
<GPHemsley>
Emory University offers a continuing education class that teaches XHTML, described as "the newest version of HTML"
21:50
<jwalden>
could be an obsolescent description
21:50
<jwalden>
one hopes
21:51
<GPHemsley>
maybe
21:51
<GPHemsley>
I've seen it in multiple catalogs
21:51
<GPHemsley>
and the funny thing is, they also offer a separate class specifically for "HTML5"
22:03
<jwalden>
...
22:13
<annevk>
zewt: workers don't work because they're async
22:14
<annevk>
Hixie: could operate on textContent or some JSON-based tree... but yeah, very much not v1 material and prolly not at all until we get that for free somehow
22:16
<annevk>
Hixie: the concept-node-remove thing was my mistake, I forgot to reload
22:17
<annevk>
Hixie: there was some feedback on twitter too
22:18
<annevk>
Hixie: https://twitter.com/slicknet/status/284355352252805120 https://twitter.com/dascritch/status/284300438969405440 https://twitter.com/dstorey/status/284359061695250432
22:19
<annevk>
Hixie: or 1) are <th> elements in sortable tables focusable? 2) would like explicit typing to catch mistakes 3) can we have <ul sortable>?
22:23
<annevk>
Hixie: you forgot to fix <td><time datetime="2011">MMXIV</time> in the spec (metadata mismatch)
22:29
<annevk>
jwalden: looks like [NamedPropertiesObject] has a similar effect to [OverrideBuiltins] so it appears the spec would suggest it returns the Window object there
22:30
<jwalden>
ugh
22:30
<jwalden>
maybe
22:33
<annevk>
jwalden: sorry, I think http://dev.w3.org/2006/webapi/WebIDL/#getownproperty suggests I'm mistaken
22:34
<jwalden>
/o\^H^H^H\o/
22:34
<jwalden>
:-D
22:35
<annevk>
if you do window.constructor per ES you invoke [[GetOwnProperty]] of window?
22:35
<jwalden>
yeah, you should
22:35
<jwalden>
among other operations, depending
22:35
<annevk>
okay then yes, you would not hit the named property in this case
22:36
annevk
is trying to learn more of ECMAScript
22:38
<jwalden>
the algorithms for property lookups and gets and sets are...in flux, right now
22:38
<jwalden>
the newer algorithms that are going into ES6 are the ones you want to know more of, more or less
22:39
<jwalden>
but in most cases you have [[GetOwnProperty]] called on the object, then its [[Prototype]], and recursively until the property is found, at *some* point in each algorithm
22:39
<jwalden>
the [[GetOwnProperty]] on Window.prototype is what would find the .constructor === Window property
22:39
<jwalden>
well
22:40
<annevk>
do you know when https://people.mozilla.com/~jorendorff/es6-draft.html will be updated?
22:40
<jwalden>
annevk: "would not hit the named property" means would get .constructor === Window, or means it'd get the <iframe> window?
22:40
<jwalden>
annevk: unsure, I could ping him tho
22:40
<annevk>
jwalden: the former
22:40
<jwalden>
okay
22:41
<annevk>
jwalden: if you read that algorithm it says that for Window, which has [NamedPropertiesObject] it would use the "internal" [[GetOwnProperty]] method, which would never get the <iframe> of course...
22:42
<Hixie>
annevk: 1) haven't really figure that out yet, want implementor feedback (should the cell be focusable? the contents? some separate UI? obviously something needs to be focusable). 2) that seems like a general table problem, not sorting-specific. Would be nice though. See if hsivonen or MikeSmith want to support it... 3) How would that work?
22:42
<annevk>
ideally once ES6 is out heycam changes IDL to define thigns in terms of proxies or some such
22:42
<jwalden>
window wouldn't have anything in this case, Window.prototype would have .constructor === Window, the GSP would have .constructor === the iframe (maybe, not sure exactly if you can frob things just right to make that happen or not)
22:42
<Hixie>
don't understand "metadata mismatch"
22:42
<Hixie>
what metadata?
22:43
<Hixie>
oh you mean the data is inconsistent?
22:43
<jwalden>
well, as long as IDL hits all the [[Internal]] properties, I think it shouldn't need much tweaking
22:43
<annevk>
Hixie: 2011 != 2014
22:43
<jwalden>
the ES6 tweaks are to the algorithms that invoke those, I think
22:43
<annevk>
Hixie: I thought we'd call that kind of stuff hidden metadata
22:44
<Hixie>
just seems like data to me
22:44
<Hixie>
metadata is things like what type the data is
22:44
<Hixie>
so e.g. in that example, the <time> element is metadata (says that it's a time), and the contents of datetime and of the element are data.
22:44
<Hixie>
https://twitter.com/dascritch/status/284300438969405440 doesn't make sense to me
22:54
<annevk>
jwalden: so I don't understand the algorithm :/
22:54
<annevk>
jwalden: in particular reading http://dev.w3.org/2006/webapi/WebIDL/#getownproperty I don't understand how with [NamedPropertiesObject] declared you can ever end up returning a value based on a named property
22:56
<annevk>
Hixie: <ul sortable> I guess would work in the same way you sort a table, except you have only one column, not entirely convinced it's very useful though
22:56
<Hixie>
yeah i don't really see how that work work. why not just sort it on the server?
22:57
<Hixie>
if a browser wants to sort the list, they can, i mean, the spec says that's ok
22:57
<Hixie>
the whole point of <ol> is that the order doesn't matter to the page
22:57
<Hixie>
er
22:57
<Hixie>
<ul>
23:46
<jwalden>
annevk: ...yeah, I probably need to do some deep reading of that soon; not today, tho, out of time for the day