09:13
jgraham
only just noticed that all the technical debate about EME is going to public-html-admin. Which is pretty funny. And shows that I'm a bit slow to notice things.
09:29
<sedovsek>
Hey. Is there a chance to achieve the same effect (hw accel.) in Firefox that we do with -webkit-transform: translateZ(0);?
10:34
<hsivonen>
aargh. I forgot the DOM interface for <main>. good catch by smaug
10:43
<MikeSmith>
hsivonen: your tests didn't catch that?
10:43
<hsivonen>
MikeSmith: I'm bad at writing them, apparently. this is why there should be a different QA person writing the tests
12:31
<jgraham>
hsivonen: I sort of imagine that the way that EPUB will handle compat-breaking changes is to offer multiple variations for download during transitional periods. Presumably the half-life of readers isn't going to be super-long so that seems sort of viable. I might be wrong though.
12:37
<MikeSmith>
I think what's fundamentally broken with EPUB to begin with is that I'm required to download and install something called a "Reading System" just to be able to read a book that's essentially just HTML and CSS
12:38
<Ms2ger>
But at least it's polyglot
12:38
<MikeSmith>
or actually I need to download and install multiple such Reading System applications, one for each vendor I buy books from
12:39
Ms2ger
reads dead tree books instead
12:40
<MikeSmith>
the size of my apartment I have no room for any more dead-tree books than the ones I already have
12:44
<MikeSmith>
the things I've learned from Henri have just left me thinking that there's little hope of fixing the problems with ebooks
12:45
<MikeSmith>
the current user experience around ebook reading is appallingly bad and I seems really unlikely to change much any time soon
13:42
<hsivonen>
jgraham: So you expect users to know what version of EPUB the embedded software on their device supports?
13:42
<hsivonen>
jgraham: it would be easier to rebrand the incompatible new EPUB to something other than EPUB then
13:43
<hsivonen>
jgraham: you know, unlike you and me, normal people don't load a bunch of test cases to figure out what the thing they have bought *actually* is capable of doing
13:44
<hsivonen>
(as you might infer from my emails, I have loaded some test cases)
13:46
<hsivonen>
regarding http://lists.w3.org/Archives/Public/www-style/2013Jan/0517.html : Have I misunderstood how Vary and ETag work in HTTP caching?
13:47
<hsivonen>
MikeSmith: note that there are devices with Reading Systems on them and you don’t get to download and install new ones
13:47
<hsivonen>
MikeSmith: except as part of an updated system image your vendor might be kind enough to give you
13:48
<MikeSmith>
hsivonen: yeah I was just going to say that I guess I hadn't thought about those devices enough
13:48
<MikeSmith>
and people hang on to those for a long time
13:48
<MikeSmith>
I still see a lot of older Kindles around
13:49
<hsivonen>
MikeSmith: I got one for Christmas. It has 2 EPUB engines. The Adobe one does not support EPUB3.
13:49
<hsivonen>
MikeSmith: I haven't yet proceeded far enough in my testing to be able to tell if I can load my own content into the other engine to diagnose if it supports EPUB3
13:50
<MikeSmith>
why2 engines as atll?
13:50
<MikeSmith>
*at all?
13:50
<hsivonen>
MikeSmith: two different DRM systems
13:50
<MikeSmith>
ah yeah
13:51
<hsivonen>
MikeSmith: also, I think Kobo wants to be able to do stuff in their engine without being limited by what Adobe is willing to sell them
13:51
<hsivonen>
but AFAICT, there's no way for the user to push a standard EPUB file onto the Kobo engine
13:51
<MikeSmith>
wonderful
13:51
<hsivonen>
I suspect there is a way to push an non-standard EPUB-like file onto it, but I haven't tested that hypothesis
13:52
<hsivonen>
I'd prefer Kobo to catch up with the features of the Adobe engine before implementing some more
13:53
<hsivonen>
it really irks me that the Kobo engine doesn't do ligatures or English hyphenation
13:53
<hsivonen>
my main complaints about the Adobe engine are that it doesn't support OpenType kerning or Finnish hyphenation
13:53
<hsivonen>
also, it's too shy to hyphenate English
13:53
<hsivonen>
that is, the hyphenation threshold does not match my preferences
13:54
<MikeSmith>
better than not doing hyphenation at all I guess
13:54
<MikeSmith>
I wonder why the Kobo engine doesn't do ligatures or English hyphenation
13:54
<MikeSmith>
are both of the engines on your device based on browser engines?
13:55
<hsivonen>
I believe the Kobo thing is based on WebKit. Probably QtWebKit.
13:55
<hsivonen>
I'm pretty sure the Adobe thing is an independent impl of CSS
13:55
<MikeSmith>
ok
13:55
<hsivonen>
the Adobe folks at TPAC couldn't confirm :-(
13:56
<MikeSmith>
btw my perspective on this is biased I guess by the fact that I mostly just read books on my Android smartphone
13:57
<MikeSmith>
I used to read at lot on my iPad2 but I quit because it was a chore to carry around and actually was causing my pain in my hands because of the way you have to hold it to read
13:59
<MikeSmith>
whereas I have at least one smartphone with me all the time and holding it and using it are a lot more natural
14:03
<hsivonen>
10" tablet for reading books: bad
14:04
<hsivonen>
smaller, lighter, less reflective screen, optional front lighting FTW
14:04
<hsivonen>
I should get around to writing a review of this thing, but I probably won't
14:05
<hsivonen>
blog posts get long. too long to be finished
14:05
<MikeSmith>
heh
14:05
<MikeSmith>
yep
14:05
<hsivonen>
I have it written in my head, but the bandwidth of braindumping is bad
14:07
<MikeSmith>
kinda like coding
14:08
<hsivonen>
totally. I also have an EPUB creation tool designed in my head
14:10
<MikeSmith>
enjoy that ability at least
14:11
<MikeSmith>
I think as you get older it gets harder to design stuff in your head
14:23
<Ms2ger>
hsivonen, does it do polyglot?
14:23
Ms2ger
runs
14:24
<hsivonen>
Ms2ger: HTML to XHTML conversion without bothering with any polyglot guarantees
14:30
<hsivonen>
so. regarding the www-style thread. I reread the relevant parts of RFC 2616 and I still think I've understood HTTP caching correctly. What I am missing or am I misunderstanding what Ilya says?
14:31
<hsivonen>
Hixie: are you paying attention to the www-style thread? It's relevant to srcset in the sense that it proposes a competing mechanism.
14:31
<Ms2ger>
I have happily ignored the thread, but maybe the IETF is missing the point
14:39
<zewt>
*insert ietf joke here*
14:39
<zewt>
(also, isn't HTTP caching an odd topic for www-style? heh)
15:04
<kamathln>
Hi, I was going through the IndexedDB specification, I tried to find what the intended behaviour is if "onversionchange" of the database object is set to null ? From firefox's behaviour, it seems it will just act like it returned success, but the specification itself does not say anything file:///home/kamathln/Documents/Indexed%20Database%20API.html#widl-IDBDatabase-onversionchange
15:04
<kamathln>
is there anything I missed?
15:06
<kamathln>
Its my first time reading specifications like this, so I wish someone could handhold me for the same
15:07
<hsivonen>
zewt: well, the HTTP-based proposal that is claimed to be cache-friendly (which I don't buy) is trying to work around the way Media Queries behave
15:08
<hsivonen>
zewt: so if MQs behaved a bit differently, perhaps we could avoid this HTTP-based workaround
15:09
<hsivonen>
zewt: not only am I skeptical about Client-Hints' cache-friendliness in general, but I'm particularly skeptical relative to MQ-based solutions if MQs are adjusted a little
15:10
<hsivonen>
cool. now there can be a showdown between the old TAG and the 4 horsemen about how much the TAG should care about closed systems
15:49
<Velmont>
Does anyone with a popular site (but not as popular as Google)
15:49
<Velmont>
actually want to rescale images on the fly (as opposed to letting
15:49
<Velmont>
Opera do it for them)?
15:49
<Velmont>
hsivonen: lol, - didn't have your own browser to use as an example? :P
15:57
<hsivonen>
Velmont: Firefox doesn’t have Turbo/Mini server-side image optimization
16:01
<Velmont>
hsivonen: Ah... It was meant like that. :-)
17:01
<dglazkov>
good morning, Whatwg!
17:09
<yroc>
dglazkov: or good afternoon or evening depending on where you are!
17:18
<zewt>
hsivonen: not familiar with the problem, but is it of the "fetching resources over and over" variety?
17:19
<zewt>
firefox is weird with subpixel widths
17:20
<zewt>
http://zewt.org/~glenn/firefox-wrapping.html chrome always sizes the box to the size of the contents; firefox makes it a pixel smaller (when the fractional part of the width is less than .5), causing the spans to be wrapped
17:22
<zewt>
i guess the span wrapping happens when the width would be greater than the truncated width, instead of the rounded width that it'll actually use
17:22
<zewt>
or ... no, but something like that
17:29
<zewt>
looks like it's one of those bugs that someone went "yeah let's just leave it" 5 years ago and left the whole world to get headaches because of their horrible bugs
17:29
<zewt>
thanks firefox
17:30
<Ms2ger>
Yw
17:33
<zewt>
awesome, someone putting "bug" in scare quotes, as if it isn't obviously a bug
17:46
<kamathln>
hi folks, please reply to my query above even if I seem offline, I internet is flaky and I keep checking the logs to know if anyone has replied anyway
17:49
<MikeSmith>
kamathln: you might consider sending that question to the public-webapps⊙wo mailing list which is the main place where the IndexedDB spec is discussed
17:51
<kamathln>
MikeSmith: thnaks
17:53
<yroc>
\quit
17:56
<MikeSmith>
Am I the only one around here who thinks it's a bad idea to hardcode path names into build files? https://bugs.webkit.org/show_bug.cgi?id=108089
18:02
<hsivonen>
trololol: http://lists.w3.org/Archives/Public/public-html/2013Jan/0202.html
18:04
Ms2ger
wonders who got the idea there was a torch that needed carrying
18:05
<MikeSmith>
hsivonen: think about how much different things would be if similar to that Jack guy he was offering a $10 bounty for somebody to implement polyglot support in the validator
18:25
<jgraham>
hsivonen: I was sort of imagining that a backwards-incompatible change in EPUB would be handled by device detection in whatever marketplace people are using to get their books, or by listing popular devices that consume each format on Project Gutenburg
18:26
<jgraham>
Possibly that would still be a terrible UX, but not more terrible than the fact that EPUB doesn't work on kindles at all
18:27
<gsnedders>
From a consumer POV, I wonder whether it'd be better just to call it something else.
18:27
<Ms2ger>
Do consumers actually know "epub"?
18:28
<gsnedders>
(I expect backwards incompatible changes would happen the way they do elsewhere: very, very slowly)
18:28
<TabAtkins>
Depends on what services they buy from.
18:28
<TabAtkins>
If from Amazon or B&N, likely not.
18:28
<TabAtkins>
Smaller sellers often advertise and offer ePubs directly.
18:28
<gsnedders>
Google Play explicitly has "EPUB" and "PDF" downloads for eReaders.
19:00
<Hixie>
hsivonen: i will look out for it
19:22
<Hixie>
hsivonen: what subject?
19:24
<hsivonen>
Hixie: http://lists.w3.org/Archives/Public/www-style/2013Jan/0434.html
19:29
<Hixie>
i don't understand, how does this affect srcset?
19:30
<Hixie>
srcset already doesn't download pictures that aren't relevant
19:34
<Hixie>
hsivonen: on an unrelated note, do you know what they're referring to in https://www.w3.org/Bugs/Public/show_bug.cgi?id=19590 ? i can't work out how you'd get an infinite loop
19:38
<SimonSapin>
TabAtkins: I don’t understand the part about "url() tokenization rules when writing variables"
19:51
<TabAtkins>
It should probably be valid to do things like "var-special-js-thing: url("example.com" more-stuff-here);", if it's meant to be interpreted by JS.
19:51
<TabAtkins>
But that's a bad-url token.
19:52
<TabAtkins>
Hm, though, I don't keep any details of what the token is, though.
19:52
<TabAtkins>
So nevermind, I guess.
19:56
<SimonSapin>
TabAtkins: is a serializer supposed to round-trip bad-url?
19:56
<TabAtkins>
SimonSapin: Given that it'll drop the declaration or block containing it in every other circumstance, obviously not.
19:58
<SimonSapin>
TabAtkins: I’ve had requests to preserve invalid declarations such as `*width: 100px` in tinycss for a minifier
19:59
<TabAtkins>
Ah yeah, that's possible. If you just skip the grammar-checking step, it wont' drop anything.
19:59
<TabAtkins>
But still, bad-url tokens don't have contents (neither do bad-strings), so they won't roundtrip - you can't serialize them as anything.
20:00
<SimonSapin>
btw, when you say a declaration is "grammatically valid", is that for a given property?
20:00
<SimonSapin>
is `color: 4px` grammatically valid?
20:01
<TabAtkins>
Depends on what level we're talking about. Some grammar violations are caught by the parser, like "*width:100px". Most are caught by the grammar-checking step, which is property-specific.
20:01
<SimonSapin>
"If the current declaration is grammatically valid, append it to the value of the current rule."
20:01
<TabAtkins>
(There's a line in the spec when you append a valid declaration about this, and another for when you pop a rule."
20:01
<SimonSapin>
that’s a property-specific check, right?
20:01
<TabAtkins>
Yes.
20:02
<SimonSapin>
I skip that in tinycss and leave it to the caller to filter (or not)
20:02
<TabAtkins>
Yes, that's acceptable. Officially, the full "CSS parser" is the combination of tinycss and the caller's filter.
20:03
<SimonSapin>
but `*width: 100px` is not even a declaration so preserving it for minifiers is not obvious
20:04
<SimonSapin>
it might even be a non-goal in rust-cssparser if I want servo to use it eventually
20:05
<TabAtkins>
It's a non-goal in the current Syntax spec, as that declaration would be thrown away early as a syntax error (switching over to next-declaration error state).
20:05
<SimonSapin>
yes
20:06
<SimonSapin>
the spec probably doesn’t need to care about this, but a tokenizer also needs to track source line numbers if good error reporting is desired
20:07
<TabAtkins>
(Which is now present in my cssparser, thanks to a nice patch.)
20:07
<TabAtkins>
(By espadrine, iirc.)
20:07
<SimonSapin>
nice
20:10
<SimonSapin>
doesn’t reconsume() need to check for newline too?
20:11
<Mike_2>
hi everybody
20:17
<Ms2ger>
http://lists.w3.org/Archives/Public/public-webapps/2013JanMar/0165.html
20:17
<espadrine>
SimonSapin: I thought about it, but I believe the first consume() already incremented the line number, didn't it?
20:18
<SimonSapin>
espadrine: yes, but won’t it increment again for the same newline on the next call?
20:18
<SimonSapin>
let’s write a test case :)
20:19
<espadrine>
:) good habit
20:19
<SimonSapin>
I see none has touched https://github.com/tabatkins/css-parser/blob/master/tests.js after me :p
20:20
<SimonSapin>
no I’m exaggerating, Tab has a commit
20:27
<SimonSapin>
my test framework sucks
20:27
<SimonSapin>
it can’t represent token’s line/column numbers
20:29
<SimonSapin>
espadrine: how do you extract/use these location information from a parsed stylesheet?
20:29
<espadrine>
SimonSapin: I just found a case where you do need to reset the line number
20:29
<espadrine>
but it's a parse error
20:29
<espadrine>
foo {bar:"foo\nbar"}
20:30
<espadrine>
(the \n being an actual newline character)
20:30
<SimonSapin>
espadrine: but escaped newline are valid in quoted strings
20:31
<espadrine>
those don't trigger reconsume()
20:31
<espadrine>
it needs to be a parser error
20:31
<espadrine>
that's probably why I didn't think to add that case
20:35
<espadrine>
SimonSapin: I just went through all uses of reconsume() in the tokenizer, that's the only possibility where the line number can go wrong
20:35
<SimonSapin>
ok
20:35
<SimonSapin>
I gave up on the test case. I’d basically need another test framework
20:36
<espadrine>
too bad
23:16
<a-ja>
Hixie: very low-priority ping re: rendering
23:28
<Hixie>
a-ja: pong
23:28
<a-ja>
a nit
23:29
<a-ja>
the implementation status popup thing next to sections-and-headings has wrong links in it
23:30
<Hixie>
woah
23:30
<Hixie>
oh, i see
23:30
<Hixie>
it's just that sections-and-headings doesn't have one
23:30
<Hixie>
if you want one for sections-and-headings, just alt-double-click there
23:30
<Hixie>
and it'll create one
23:31
<a-ja>
? i see links to following section
23:32
<Hixie>
how do you mean?
23:34
<a-ja>
shows link to #quotes and the edit link point to quotes, too
23:35
<Hixie>
yes, #quotes is the previous section
23:35
<Hixie>
there's always a box at the top-left, it is the box for the most recent section to have one
23:36
<a-ja>
yep...preview, not following.
23:36
<a-ja>
k
23:38
<a-ja>
h1 ~ h2 thru h5 looks better now.....though awfully damned verbose when coding for -webkit-any, -moz-any, and :matches :/
23:41
<Hixie>
a-ja: yeah
23:54
<Hixie>
can anyone work out why http://www.hixie.ch/tests/adhoc/dom/workers/shared/001.html fails in webkit?
23:54
<Hixie>
works in opera...
23:57
<othermaciej>
if the Web Inspector is to be trusted 001.js never gets loaded
23:59
<Hixie>
gets loaded for me (safari and chrome) according to web inspector
23:59
<Hixie>
but i can't work out how to debug a worker
23:59
<Hixie>
oh maybe event.source isn't implemented in webkit...