Options:
- # Session Start: Thu Jul 04 00:00:01 2013
- # Session Ident: #testing
- # [00:06] * Quits: krisk (~krisk@public.cloak) (Ping timeout: 180 seconds)
- # [00:07] * heycam|away is now known as heycam
- # [00:49] * Quits: tobie (tobie@public.cloak)
- # [04:02] * Joins: gitbot (~gitbot@public.cloak)
- # [04:02] -gitbot:#testing- [web-platform-tests] jungkees pushed 1 new commit to master: https://github.com/w3c/web-platform-tests/commit/9dde2fbacd8fad352b8b1837617f9c8e38274aa4
- # [04:02] -gitbot:#testing- web-platform-tests/master 9dde2fb Jungkee Song: Merge pull request #128 from w3c/hallvors/XHRtestBugFixes2...
- # [04:02] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [04:20] * heycam is now known as heycam|away
- # [05:12] * heycam|away is now known as heycam
- # [06:59] * Quits: jhammel (~jhammel@public.cloak) ("leaving")
- # [09:18] * Joins: darobin (rberjon@public.cloak)
- # [09:33] * Joins: Ms2ger (~Ms2ger@public.cloak)
- # [09:57] * Joins: gitbot (~gitbot@public.cloak)
- # [09:57] -gitbot:#testing- [web-platform-tests] yutak opened pull request #237: shadow-dom: Update a test for shadow root's accessors. (master...shadow-dom/shadow-root-accessors) https://github.com/w3c/web-platform-tests/pull/237
- # [09:57] * Parts: gitbot (~gitbot@public.cloak) (gitbot)
- # [10:06] * heycam is now known as heycam|away
- # [10:21] * Joins: tobie (tobie@public.cloak)
- # [10:33] * Joins: dom (dom@public.cloak)
- # [10:38] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [10:38] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [10:38] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [14:17] * Joins: r12a (rishida@public.cloak)
- # [14:17] <r12a> which list do i write to if i need some help with a test?
- # [14:18] <Ms2ger> What kind of test?
- # [14:18] <r12a> scripted
- # [14:18] <Ms2ger> What about?
- # [14:18] <r12a> ruby in html5 extension
- # [14:18] <Ms2ger> public-html-testsuite, probably
- # [14:18] <darobin> mmmm, not sure
- # [14:19] <r12a> it's how to detect success using javascript that i'm struggling with
- # [14:19] <darobin> didn't we say we wanted everything on public-test-infra?
- # [14:19] <darobin> yeah, that's definitely public-test-infra, it's not at all HTML WG specific
- # [14:19] <Ms2ger> I don't know
- # [14:19] <Ms2ger> I thought not yet
- # [14:19] <Ms2ger> But yeah, test-infra is good too
- # [14:19] <darobin> my understanding is that public-html-testsuite is only for the stuff that's specific to the HTML WG
- # [14:19] <darobin> r12a: I'd go with public-test-infra
- # [14:19] <r12a> ok, thanks
- # [14:19] <Ms2ger> Maybe we should shut them all down for public-test
- # [14:20] <darobin> at some point yeah
- # [14:20] <Ms2ger> Any objections?
- # [14:20] <Ms2ger> Resolved.
- # [14:20] <Ms2ger> darobin, go ahead :)
- # [14:21] <darobin> rm -Rf w3c/lists/public-test-*
- # [14:21] <darobin> there, done
- # [14:47] <r12a> mail sent
- # [15:07] <jgraham> r12a: My advice-you-don't-want is "let Microsoft fix the bug"
- # [15:08] <Ms2ger> Unhelpful yet correct advice?
- # [15:08] <Ms2ger> That's my job
- # [15:08] <r12a> hmm, interesting idea, except that the bug is in the test mechanism, rather than the thing being tested - i'd like to show that they DO support ruby
- # [15:09] <jgraham> It is an unfortunate truth that tests often fail for reasons unrelated to what they nominally test
- # [15:09] <jgraham> I can see why you might consider this unacceptable though
- # [15:09] <jgraham> *this case
- # [15:10] <r12a> between this and trying to do reftests for paragraph handling in textarea for bidi text i'm seriously wondering whether it's realistic to expect all tests to be automated
- # [15:11] <r12a> or at least within a reasonable timeframe
- # [15:11] <jgraham> Well it is clearly unrealistic to expect all tests to be automated
- # [15:11] <jgraham> But it is essential that all tests that *can* be automated are
- # [15:12] <r12a> yes, that's fair enough
- # [15:12] <jgraham> An automated test will be run hundreds of times a day. A non automated one a handful of times a year
- # [15:17] <darobin> r12a: I wonder if there isn't a way to test this that wouldn't fall on the same bugs
- # [15:17] <Ms2ger> s/will/can/ ;)
- # [15:17] <r12a> so do i ;-)
- # [15:18] <darobin> r12a: if you look at the height of the box that contains the ruby without any rt, then measure it again with rt, it ought to increase if the rt is properly on top
- # [15:18] <darobin> at least, I'd hope so (the alternative is that CSS does something pretty screwed up, which isn't out of the question but certainly possible)
- # [15:19] <jgraham> Heh
- # [15:19] <r12a> yes, i thought of that, but i'm going to need to tell whether the box is above or below at some point
- # [15:19] <jgraham> But that's an interesting alternative
- # [15:19] <Ms2ger> darobin, now, do you have a spec to back that? :)
- # [15:19] <r12a> it would be nice if there was one approach that works for all
- # [15:20] <r12a> i'd also much rather find a solution that actually tests the right thing, if possible ;-)
- # [15:20] <Ms2ger> That seems like a nice thing to have, yes :)
- # [15:20] <r12a> i'm curious about what IE is actually doing to produce these odd results though
- # [15:21] <darobin> "which isn't out of the question but certainly possible)" well done me
- # [15:21] * darobin has to step out
- # [15:21] <Ms2ger> (I know why I prefer testing DOM APIs)
- # [15:21] <r12a> i wonder why i got into testing i18n stuff (lots of on screen rendering magic) ;-)
- # [15:22] <r12a> changing or non-existent fonts
- # [15:22] <r12a> etc...
- # [15:22] <jgraham> Well if you need to test that it's on top
- # [15:22] <Ms2ger> Maybe because your name is r12a and you felt a connection?
- # [15:22] <jgraham> and IE is reporting that it's below
- # [15:22] <jgraham> Even though it is actually on top
- # [15:22] <jgraham> I refer you to my earlier comment
- # [15:23] <jgraham> Microsoft should fix their bug
- # [15:24] <jgraham> In either case you should certainly make a test that fails in IE]
- # [15:24] <jgraham> Since you have found a bug
- # [15:24] <r12a> k
- # [15:26] <jgraham> (and report it to Microsoft I guess)
- # [15:34] <jgraham> r12a: have you tried with getBoundingClientRect?
- # [15:34] <r12a> didn't know about that
- # [15:35] <jgraham> I assume it will produce the same wrong information
- # [15:35] <jgraham> But you never know
- # [15:40] <jgraham> I presume that ruby rendering isn't well enough defined that you could write a reftest using ahem?
- # [15:43] <r12a> jgraham, getBoundingClientRect produces the same result
- # [15:45] <r12a> wrt reftest, the actual distance between ruby text and base and the size of the ruby text are not defined exactly, so i'm leery about going the reftest route
- # [15:45] <r12a> for automatic checking, at least
- # [15:54] <jgraham> OK, both of those are what I expected, sadly
- # [15:55] <jgraham> Then I return to my previous position of leaning on Microsoft until they fix their bug :)
- # [15:59] <darobin> r12a: re the font stuff you should really use a single, predictable font that you impose on the page; anything else will indeed be unreliable
- # [15:59] <darobin> especially on mobile
- # [15:59] <r12a> understood
- # [16:00] <r12a> it's the distances between ruby and the size of scaling of the rt that are the key issues
- # [16:00] <r12a> those are implementation dependent, and not font related
- # [16:00] <r12a> s/between ruby/between ruby text and base/
- # [16:01] <darobin> r12a: also, I know I'm going to get heckled for this, but you might want to give one of these a shot to see if they work better with IE: http://api.jquery.com/category/offset/
- # [16:01] <darobin> fair enough
- # [16:01] <darobin> and if jQuery happens to work, you don't have to make your tests depend on it, we can extract whatever it is they do to paper over IE's bug
- # [16:01] <darobin> I would in fact expect it to work
- # [16:02] <r12a> ok, i'll give it a try later
- # [16:02] <r12a> (i'm deep in css character encoding test rewrites now)
- # [16:10] <jgraham> darobin: jQuery.offset seems to use getBoundingClientRect
- # [16:11] <darobin> jgraham: I wouldn't be violently shocked if IE had a bug on offset and not gBCR
- # [16:12] <jgraham> darobin: 13:40 < r12a> jgraham, getBoundingClientRect produces the same result
- # [16:12] <jgraham> (I wouldn't have been shocked either)
- # [16:13] <darobin> ah
- # [16:14] <darobin> I haven't looked at what jquery does, but I'd be surprised if they exposed something that doesn't work in IE
- # [16:14] <darobin> that said, it may be a bug that only affects some ruby elements, in which case jQuery won't do any better
- # [16:15] <darobin> at that point, it's worth looking at what horrible default CSS they're applying in order to position the element. Maybe some padding set to the proper line height
- # [16:15] <jgraham> Right, it seems likely that someone just got a plus and a minus sign confused when setting the position of ruby elements in the DOM, or something
- # [16:15] <darobin> you'd sort of hope such code would not be element-specific...
- # [16:15] <jgraham> I wouldn't be entirely surprised if there was a ruby-specific part
- # [16:16] <jgraham> Although, yes, it does seem surprising that the position on the screen and the position reported in the DOM don't match
- # [16:17] <jgraham> (alterntaively, they are doing something weird like making the ruby text have the same baseline as the normal text but a large bottom margin)
- # [16:29] * jgraham looks in the IE developer tools
- # [16:29] <jgraham> They aren't, it's just broken
- # [16:29] <jgraham> Or, maybe they are internally
- # [16:30] <jgraham> But the bottom of the <rt> box lines up with the bottom of the <rb> box
- # [16:30] <jgraham> Oh, not quite
- # [16:30] <jgraham> The bottom of the <rb> text
- # [16:41] <jgraham> (so I retract my previous guess. It looks like some part of the layout engine lays out the text at the baseline of the normal text, and that sets the position in the DOM, and later it is shunted upwards, but the DOM isn't updated to reflect that)
- # [17:03] * Quits: dom (dom@public.cloak) ("")
- # [17:30] <darobin> jgraham: that's just absolutely lovely
- # [17:31] <darobin> but then again, ruby being broken isn't shocking either
- # [17:31] <darobin> lots of non-shocking information was uncovered here today
- # [17:33] * Quits: darobin (rberjon@public.cloak) (Client closed connection)
- # [17:35] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [18:23] * Quits: Lachy (~Lachy@public.cloak) (Ping timeout: 180 seconds)
- # [20:24] * Joins: AutomatedTester (~AutomatedTester@public.cloak)
- # [21:56] * Quits: tobie (tobie@public.cloak)
- # [22:19] * Joins: tobie (tobie@public.cloak)
- # [22:28] * Quits: AutomatedTester (~AutomatedTester@public.cloak) (Client closed connection)
- # [23:12] * Quits: tobie (tobie@public.cloak)
- # [23:19] * Quits: Ms2ger (~Ms2ger@public.cloak) ("nn")
- # [23:32] * Joins: tobie (tobie@public.cloak)
- # [23:46] * Quits: tobie (tobie@public.cloak) (Client closed connection)
- # [23:47] * Joins: tobie (tobie@public.cloak)
- # Session Close: Fri Jul 05 00:00:00 2013
The end :)