00:10
<Philip`>
http://fonts.philip.html5.org/
00:11
<Philip`>
Now with more exciting front-end
00:11
<Philip`>
and occasional ligature support, at least for cases like "ff" in the DejaVu fonts
00:12
<Philip`>
Also it violates the licences of some of the fonts, but please don't notice that
00:12
<Philip`>
Oh, and it adds IE support too
00:13
<Hixie>
for the fonts that are available freely, it would be useful to be able to download the fonts so that the author can try them on his site
00:13
<Hixie>
and it would be useful to be able to point the script at a site and have it fetch all the files and figure out what characters are needed
00:13
<Hixie>
by parsing the html and css and applying the css even :-)
00:14
<takkaria>
and executing the JS? :)
00:14
<Hixie>
sure!
00:16
<Philip`>
It wouldn't seem infeasible to let people specify an HTML file and a CSS class name, and have it extract the characters from that
00:16
<Philip`>
but I guess it's more important to have the basics working first, before adding fancy features like that :-)
00:16
Philip`
fixes some bugs so now it should handle Unicode characters too
00:23
<Philip`>
It's already much more useful than http://www.fontembedding.com/eot/ which just lets you take a perfectly good font and then limit it so it can only work on your own domain, and then gives you a "Unable to load DLL 'makeeot.dll': The specified module could not be found. (Exception from HRESULT: 0x8007007E)"
00:35
<Hixie>
christ, as if xslt wasn't enough: http://www.xsharp.org/samples/
00:45
<roc>
it might be better then XSLT if it doesn't pretend to not be a programming language
00:47
<roc>
uh oh
02:54
<MikeSmith>
Hixie: about attributes that are new to HTML5, e.g., ping and sizes
02:56
<MikeSmith>
Hixie: the HTML5 spec doesn't say anything about whether conformant UAs must expose those new attributes as properties, right?
02:58
<MikeSmith>
I mean about supporting them in addition to generic methods; e.g, link.sizes in addition to just link.getAttribute('sizes')
03:01
<Lachy>
MikeSmith, the spec lists sizes in the HTMLLinkElement interface
03:01
<MikeSmith>
ah, yeah, OK
03:01
<Lachy>
and ping in the HTMLAnchorElement interface
03:01
<MikeSmith>
I see
03:02
<MikeSmith>
what about the data-* attributes?
03:03
<Lachy>
there's a dataset api http://www.whatwg.org/specs/web-apps/current-work/#dom-dataset
03:28
<heycam>
annevk, (re event handler attribute definitions) yeah it would be good to have a common place that defines those
03:28
<heycam>
what does html5 have atm, EventListener or so? what makes that not enough?
03:29
<Hixie>
http://www.whatwg.org/specs/web-apps/current-work/#event-handler-attributes is what html5 has
03:30
<Hixie>
i don't think webidl is the right place for it
03:30
<Hixie>
dom events, maybe
03:30
<heycam>
yeah sounds like it
03:31
<Hixie>
i think it's fine to have it in html5 for now
03:31
<heycam>
Hixie, in that section: "The second way is as as an ..."
03:37
<Hixie>
thanks
06:04
<zcorpan__>
hmm, http://fonts.philip.html5.org/ doesn't work for me in opera now
06:08
<zcorpan__>
and ie8
07:57
<hsivonen>
Philip`: shouldn't CSS 'format' for .ttf be format("truetype") and format("opentype") for .otf?
08:34
<Philip`>
hsivonen: Not really - the fonts are OpenType, and the OpenType spec says "OpenType fonts may have the extension .OTF or .TTF"
08:34
<hsivonen>
Philip`: do all of Firefox, Safari and Opera support CFF outlines on all their platforms?
08:34
<Philip`>
(or at least it says that for fonts with TrueType outlines)
08:35
<Philip`>
hsivonen: No idea, I've only been looking at TrueType-outline ones
08:36
<hsivonen>
http://www.w3.org/TR/css3-webfonts/#src
08:36
<Philip`>
zcorpan: Oops, looks like Opera only likes format("truetype") and not format("opentype")
08:36
<hsivonen>
how do I, as an author, tell if a font is "TrueType" or "TrueType Open"?
08:38
<annevk>
DOM Events would work, but that's not moving...
08:38
<annevk>
And since we could invent some special type of interface... On the other hand, that might make Web IDL depend on HTML5 depending on how it's defined...
08:40
<hsivonen>
yay for SSL. fedorahosted.org's cert expired half an hour ago--blocking access to Liberation fonts
08:40
<annevk>
Hixie, where is it defined that things bubble from Document to Window?
08:40
<jwalden>
yay for fedorahosted.org and the provider of its cert, more correctly
08:40
<annevk>
Hixie, and if load is just dispatched on Window, how does document.onload work?
08:42
<Philip`>
zcorpan: I've changed it to lie about being format("truetype") now, so it seems to work in Opera
08:43
<Philip`>
zcorpan: It worked for me in IE8 before these changes, and still works, so I'm not sure why it wouldn't work for you
08:44
<Philip`>
hsivonen: As an author, you probably do what I do and assume you might as well say they're OpenType because that's a superset of TrueType, and then find it doesn't work in Opera and so say they're TrueType instead :-)
08:44
<annevk>
why do you use format() at all?
08:45
<annevk>
and please file bugs :)
08:45
<Philip`>
To stop browsers wasting time downloading formats they can't understand
08:45
<annevk>
and how do you make it work in IE?
08:46
<hsivonen>
Philip`: also, with Free fonts that have both .ttf and .otf available, I have no idea which one I should want except that CFF outlines don't work in Prince yet
08:46
<Philip`>
though it's not a very successful plan, because it results in IE downloading both 'foo.eot' and 'foo.ttf) format("truetype"), url(foo.eot'
08:47
<Philip`>
Oops, I mean 'foo.ttf) format("truetype"), url(foo.eot) format("embedded-opentype"'
08:48
<hsivonen>
so far, I've noticed that Firefox has superior ligature support compared to Opera and Safari on Mac
08:48
<zcorpan>
Philip`: oh, i had disabled "Font download" in ie
08:48
<Philip`>
annevk: By providing EOT files to IE
08:48
<annevk>
well yeah, that's clear by now :p
08:48
<Philip`>
zcorpan: I can imagine that would hamper its ability to download fonts :-)
08:49
<zcorpan>
Philip`: you can reduce the number of 404s by using conditional comments, but dunno about a css-only solution
08:50
<Philip`>
zcorpan: Could that be done in a forward-compatible way?
08:51
<Philip`>
zcorpan: I want something that'll work in IE9 if they don't change their font support at all (so it needs a @font-face with the .eot, and with no format("...") because IE can't parse that), but will also work if they implement the new @font-face syntax and either do or don't add support for .ttf
08:51
<hsivonen>
still, my main concern is that TrueType rendering on Windows is so different from Mac and Linux that it'll be easy to create pages on Mac and Linux and have the page go unreadable on Windows
08:53
<Philip`>
hsivonen: Look at the FFF Tusj Bold example on my page - it's awful in Safari on Windows, differently awful in IE8 on Windows, even more broken in Opera on Linux, and works fine in Opera on Linux
08:53
<zcorpan>
Philip`: then i guess only by serving IE9 what you have now
08:53
<zcorpan>
which makes it all very crufty
08:53
<Philip`>
zcorpan: Indeed :-(
08:55
<zcorpan>
Philip`: "even more broken in Opera on Linux, and works fine in Opera on Linux" - i guess you meant windows for one of them? :)
08:55
<Philip`>
zcorpan: Oh
08:55
<Philip`>
zcorpan: No, I meant "Firefox" for the second one
08:55
<zcorpan>
ah
08:56
<Philip`>
Opera seems to only use the font for the "us" and "Bol" parts when rendering the font's name
08:57
<Philip`>
though it works fine with the original non-subsetted .ttf, so it could be the result of a bug in my code
08:58
<Philip`>
hsivonen: Ligatures seem to work for me in Firefox on Linux and IE on Windows, and not in Opera on Linux or Safari on Windows
08:59
<hsivonen>
Philip`: my understanding is that Safari intentionally prefers speed over prettiness here
09:00
<hsivonen>
I think it's discriminatory to turn off ligatures for Latin while supporting them for some other scripts
09:01
<Philip`>
Preferring anything over prettiness seems contrary to the whole Mac culture :-)
09:01
<jgraham>
hsivonen: I thought ligatures made a big difference in othr scripts to the shape so the word looked totally wrong whereas in latin it is only a minor aesthetic difference that only matters at small font size
09:02
<Philip`>
hsivonen: It doesn't seem that discriminatory, since for Latin it just makes some character sequences look a bit better, while other scripts are entirely unreadable unless you have ligatures
09:02
<jgraham>
(totally wrong without ligatures)
09:02
<Philip`>
Oh, what jgraham said
09:02
<Philip`>
jgraham: Ligatures matter in Latin at large font sizes too
09:02
<jgraham>
Philip`: I thought they generally looked weird at large font sizes
09:03
<jgraham>
But maybe I am wrong
09:03
<jgraham>
(do you have nice examples?)
09:04
<hsivonen>
jgraham: the word "Quirks" looks great in Linux Libertine in Firefox on Mac
09:04
<hsivonen>
not so great in Safari
09:04
<zcorpan>
Philip`: for some reason opera seems to use the font on the index page instead of the generated font, or at least i get the glyphs for "ABCFTabcdjlosu123;" instead of those for "Testing abc"
09:05
<Philip`>
jgraham: Try http://fonts.philip.html5.org/ and DejaVu Serif and "fluffily affable fish" in recent Firefox, and try changing the text size
09:06
<Philip`>
zcorpan: Oh, I guess it's possible that it's caching fonts by their internal name, or something
09:07
<Philip`>
zcorpan: (and all the subsetted fonts have the same name regardless of which characters are included in them)
09:08
Philip`
sees that his font service is pretty slow for DejaVu fonts :-(
09:14
<zcorpan>
Philip`: i guess we should change that then if font subsetting is supposed to work (or everyone doing font subsetting needs to change the internal name)
09:17
<hsivonen>
zcorpan: caching fonts by font name seems like a security hole
09:17
<hsivonen>
zcorpan: an attacker can serve a font with misleading glyphs, have it stick into the cache and then wait for the user to navigate to another site using the font
09:18
<hsivonen>
the browser should instead generated Origin-scoped unpredictably random names for the fonts
09:18
<hsivonen>
*generate
09:18
<zcorpan>
hsivonen: the cache doesn't work cross origin
09:18
<hsivonen>
ah
09:19
<zcorpan>
actually i think it shouldn't work cross document
09:19
<Philip`>
zcorpan: I'm already intending to change my code to give a random meaningless name to each font, since the Open Font License doesn't let you use any of the words of the original font's name
09:19
<Philip`>
zcorpan: so I don't think it's a practical issue for me
09:19
<zcorpan>
Philip`: ok
09:19
<hsivonen>
Philip`: you could also just do an MD5 on the resulting font data
09:19
<Philip`>
zcorpan: but it does seem confusing and wrong if fonts are shared based on some internal details that the user might not ever see
09:19
<hsivonen>
that would give the same name to the same subset each time
09:20
<Philip`>
hsivonen: That would be hard, because the font's name is part of the font data :-)
09:20
<jgraham>
Philip`: That more or less meets my definition of ugly :) (I like the fl ff and fi ligatures at small font size but not at large font size)
09:20
<Philip`>
zcorpan: (as opposed to URL or something like that)
09:20
<Hixie>
annevk: only opera has document.onload
09:20
<Hixie>
annevk: http://www.whatwg.org/specs/web-apps/current-work/#events-and-the-window-object
09:21
<Philip`>
hsivonen: but md5(original-font-file + list-of-characters-in-subset) seems like it should work fine
09:22
<Philip`>
jgraham: Doesn't the non-ligatured version look uglier with its narrow separated 'f's? :-)
09:23
<zcorpan>
Philip`: yep, i'm filing a bug about it
09:25
<Philip`>
zcorpan: Okay, thanks
09:25
<annevk>
Hixie, I see
09:26
<annevk>
I updated http://wiki.whatwg.org/wiki/HTML5_Presentations by the way, but I think it's far from complete
09:26
<annevk>
maybe I should ask on the blog if people want to contribute to it
09:27
<jgraham>
Philip`: No, I think it looks better
09:27
<jgraham>
:p
09:36
annevk
finds http://www.cs.tut.fi/~jkorpela/HTML4.0/comments.html
09:41
<Philip`>
"In printed form, the HTML 4.0 draft is about 265 printed pages long, and it still contains empty parts to be added later. Compared with the about 77 pages of HTML 2.0, this exhibits an intolerable trend." - seems like the trend is still continuing
09:47
<hsivonen>
looks like Yucca has noticed the bogosity of the term 'internationalization' back then
09:48
<hsivonen>
'Multilingual support should be labeled as such, or as language-specific support, not as "internationalization". (Typically, multilingualism is important for national documents, whereas most international documents use just English.) '
09:51
<hsivonen>
is the part about "Further replacements for EM might be something like the following: " serious and not demonstrating the absurdity of semantics?
10:04
<annevk>
if you ignore the RDF bit http://www.w3.org/2005/Talks/05-steven-xtech/ could basically be a talk about HTML5 :)
10:06
<hsivonen>
annevk: not really. "As generic XML as possible: if a facility exists in XML, try to use that rather than duplicating it."
10:07
<annevk>
oh right, I should've mentioned excluding that line too
10:07
<annevk>
though actually I think we have
10:07
<annevk>
we are using SVG and MathML rather than inventing new vector and math markup
10:07
<annevk>
we encourage people to use xml:lang and xml:base in the XML serialization
10:08
<annevk>
though maybe I'm twisting things too much now :)
10:09
<hsivonen>
xml:base might be a bit mistake, but I guess it's too late to get rid of it now
10:09
<Hixie>
xml:base makes sense in machine-generated xml environments
10:09
<Hixie>
it's a pain on the web
10:10
<hsivonen>
actually, xml:base is just one of the many things that flow out of the problem of URIs being too long
10:12
<Hixie>
not really
10:12
<Hixie>
doesn't matter how long urls are, sometimes relative urls are a good thing
10:14
<Hixie>
uncs are pretty terse, and they'd still need xml:base if you generated compound documents from unrelated bits
10:54
<annevk>
my weblog no fails to render properly in all stable releases of browsers
10:54
<annevk>
zen
10:54
<annevk>
s/no/now/
10:55
<virtuelv>
also "SVG sucks"
10:55
<annevk>
thanks to zcorpan, rubys, and Robbert Broersma
10:56
<zcorpan>
oooh
10:56
<virtuelv>
annevk: you should've used a web font
10:56
<annevk>
virtuelv, I use two
10:56
<virtuelv>
not in the logo?
10:56
<annevk>
how's that?
10:57
<annevk>
name me a font that can render the logo :)
10:57
<annevk>
(Weblog 4.2 is rendered using a normal font though)
10:57
<virtuelv>
annevk: make your own :D
10:58
<annevk>
I think using SVG is cheaper ;)
10:58
<hsivonen>
annevk: SVG fonts!
10:59
<virtuelv>
there's also something odd about an image of Mandrake next to a screenshot from Ubuntu
10:59
<virtuelv>
you should replace it with a trace of Mark Shuttleworth
10:59
<virtuelv>
svg, of course
10:59
<annevk>
you can do that on your own blog ;)
11:00
<virtuelv>
meh, neh
11:00
<virtuelv>
I'll keep the orange to annoy a co-worker
11:01
<Hixie>
svg fonts from the html would be a pretty awesome way to do it
11:01
<Hixie>
if you want things to not work
11:01
<Hixie>
:-)
11:01
<zcorpan>
annevk: what did Robbert Broersma do, btw?
11:01
<annevk>
that actually works in Safari and Opera I think, might even be in Safari 3.1
11:01
<annevk>
zcorpan, making the menu in CSS ages ago
11:02
<zcorpan>
annevk: ah ok
11:02
<annevk>
though he hadn't done the font embedding hting
11:03
<annevk>
it seems Opera requires a reload for SVG images loaded through the content property when the content property is applied to an element?
11:03
<annevk>
o_O
11:04
<annevk>
also, the daddy font doesn't work in Opera 10 but does in non-branched stuff
11:04
<annevk>
hmm
11:04
<zcorpan>
annevk: seems to work for me
11:05
<zcorpan>
oh wait
11:05
<zcorpan>
yeah i get the first bug
11:06
<annevk>
the font thing might be Linux only
11:08
<zcorpan>
annevk: should i file a bug for the reload thing?
11:09
<annevk>
please
11:13
<annevk>
zcorpan, it seems to be a bug for 'content' on ::after as well
11:13
<annevk>
zcorpan, so it's slightly less weird I suppose
11:20
<zcorpan>
it just doesn't happen for daddy.svg?
11:25
<annevk>
zcorpan, it does
11:25
<annevk>
at least, when I tested again in core-2.3...
11:27
<zcorpan>
weird, doesn't happen for me
11:34
<Lachy>
annevk, why doesn't Opera 10 render the web font for the left navigation menu?
11:34
<Lachy>
it looks like only webkit is getting that right
11:34
<annevk>
might be a Mac issue
11:34
<annevk>
it does on Linux
11:34
<Lachy>
ok. I'll try windows
11:37
<Philip`>
Someone ought to write a web browser that isn't rubbish
11:37
<Philip`>
If you could write a web page and it would actually work, that'd be a great step forward
11:37
<Lachy>
Philip`, we're hiring if you would like to help with that :-)
11:39
<Philip`>
Lachy: Sadly I've got a PhD I need to finish first, which will keep me occupied for the next couple of years :-)
11:40
<Lachy>
what topic are you doing your PhD on?
11:40
<Philip`>
It's some kind of network routing stuff
11:47
<annevk>
zcorpan, weird indeed, because now it does load again...
11:49
<zcorpan>
annevk: now i could reproduce the :after bug in opera 10
11:50
zcorpan
confused
11:53
<annevk>
cc rune and let him figure it out? :)
11:54
<Lachy>
annevk, the menu font works on windows, so it must be a Mac bug
11:54
jgraham
likes the way that Philip` is so vauge about his PhD
11:55
<Philip`>
jgraham: That's because I don't really know what I'm doing :-p
11:56
<zcorpan>
Philip`: or because you can figure it out when you start working on it?
11:56
<Philip`>
zcorpan: I've been working on it for a year already :-)
11:56
<zcorpan>
Philip`: ah. how far have you come?
11:57
jgraham
found that a lot of his PhD was thinking of cool things that would be useful then finding that they were too hard to actually do so doing easier but less cool things instead
11:58
<jgraham>
Also PhD-related protip: never ask PhD students how far thay have got.
11:59
<Philip`>
I'm working as part of a smallish group (about six people), so mostly I've been learning what other people have been doing and writing some code to make things work and planning what I ought to spend the next couple of years working on :-)
11:59
<jgraham>
(If you don't understand why go waste an afternnon reading Piled Higher and Deeper sometime)
11:59
<jgraham>
http://www.phdcomics.com/
12:02
<annevk>
Lachy, can you file it, since you have a Mac and all? :)
12:03
<annevk>
my Macbook is elsewhere atm
12:03
<Lachy>
ok, I'll add it to my list of other bugs I really need to get filed today
12:13
<annevk>
Opera's fallback for 'content' is also screwed I notice, but I think it might be per the current draft, but not per the agreed upon changes to the current draft
12:13
<annevk>
might be correct*
12:21
<Hixie>
nn
12:21
<annevk>
g'night
12:32
<Lachy>
what's the best way to make a test case for a web font that relies on the correct rendering of the downloaded font, and still make it clear what the result should look like?
12:33
<Lachy>
should I just include an image of the font in the TC for testers to compare it with?
12:33
<hsivonen>
Lachy: depends on how complex the determination of "correct" is
12:34
<hsivonen>
Lachy: for mere @font-face presence checking, you could use the Ahem trick Hixie used for Acid 3
12:37
<annevk>
Lachy, does it rely on correct rendering? you could e.g. say "The following two lines should have different fonts" "line 1" "line 2" first styled with font-family:custom,serif and the second with font-family:serif
12:37
<zcorpan>
Philip`: you could make the bogus 404 url redirect to the eot (or make a copy of the eot file with the bogus file name)
12:37
<annevk>
for correct rendering a screenshot would be best I suppose
12:37
<zcorpan>
Philip`: basically the same idea as http://annevankesteren.nl/2005/10/ie-import-hack
12:40
annevk
reads that post again and almost gets upset again by the comments
12:41
<annevk>
aah, a fresh Firefox nightly renders the fonts for the menu and heading correct, but it doesn't do any of the content property thingies
12:42
<annevk>
and thanks to the new style rules for the search form it no longer overflows, but I think that's a bug with the new style rules :)
12:42
<annevk>
(i.e. not something that was fixed)
12:43
<zcorpan>
annevk: that would be min-width on body
12:45
<hsivonen>
annevk: is there any UA that renders your new blog heading correctly?
12:45
<annevk>
Opera 10 alpha after a refresh
12:46
<annevk>
though maybe only on non-Mac platforms
12:46
<hsivonen>
I see. Refrest fixed your name but "Weblog 4.2" is still ugly
12:46
<hsivonen>
*Refresh
12:47
<annevk>
yeah, apparently that font doesn't work well on Mac yet
12:47
<hsivonen>
annevk: some weird TTF tables?
12:47
<hsivonen>
or lack thereof?
12:47
annevk
is a font noob
12:47
<annevk>
Lachy will find a bug so supposedly I'll find out when they debug it :)
12:48
<annevk>
s/find a bug/file a bug/
12:48
<annevk>
duh
12:52
<Philip`>
Lachy: You could make the downloaded font a serif one, and the page's default font a sans-serif one, and say "This line should have a serif font", and hope the testers understand what that means
12:52
<hsivonen>
annevk: I disagree about the gracefulness of the degradation of the blog title
12:52
<Philip`>
http://www.webfonts.info/wiki/forum/viewtopic.php?f=5&t=7 seems a nice way of testing specific font-rendering features
12:53
<hsivonen>
annevk: "Kesteren" is virtually unreadable in Firefox 3.0.x
12:53
<Philip`>
(The font is set up so that e.g. the string "ab" is treated as a ligature and rendered as "liga [tick]" if it's working correctly)
12:54
<hsivonen>
why is "TEST" blinking in Firefox?
12:54
<hsivonen>
annevk: whoa! you actually used text-decoration: blink!!!
12:55
<annevk>
hsivonen, with graceful I meant that you can read the content, not that the heading is necessarily usable
12:55
<hsivonen>
annevk: ok
12:56
<Philip`>
zcorpan: I currently like the idea of having people download the font files and copy them onto their own web server, which should be as easy as possible and shouldn't involve setting up redirects or crazy filenames
12:56
<annevk>
ah yeah, the text-decoration effect only works in Firefox, guess no browser gets it right then :D
12:56
<hsivonen>
in my pending redesign, I'm using CSS generated content with a PUA character for <hr>s. Is that theoretically pure enough or are PUA chars always theoretically wrong?
12:57
<annevk>
seems to me that stuff inside CSS shouldn't matter
12:57
<hsivonen>
the result is OK without CSS or with CSS when @font-face is supported
12:57
<hsivonen>
but it doesn't with CSS without @font-face
12:58
<annevk>
I gave up on making that combination work
12:58
<annevk>
quite fast
12:58
<annevk>
my criteria was that it had to be usable, not pretty
13:02
<annevk>
it's a CSS problem, just as with different fonts, you want to apply a different set of CSS rules when 'content' fails to apply, but you can't
13:07
<zcorpan>
seems webkit doesn't apply the font for daddy.svg when used in 'content'
13:08
<zcorpan>
the 'find' button is also misplaced for some reason
13:09
<annevk>
simonw once had this CSS series called "CSS ain't Rocket Science"
13:10
Philip`
sees CORE-17018 already exists about format("opentype") in Opera
13:10
<hsivonen>
annevk: you should also use some rounded corners :-)
13:11
<Philip`>
annevk: You should write your entire page upside-down, and then use CSS Transformations to turn it round again
13:14
<annevk>
in half a year or so I'll add some transforms and transitions
13:14
<annevk>
to increase the menu buttons for instance
13:14
<annevk>
hsivonen, they are passe :p
13:15
<hsivonen>
annevk: passé or not, you could use some dotted rounded corners. they don't work right in any of the top 4 browsers
13:15
<hsivonen>
double works in Gecko but sucks in WebKit
13:16
<annevk>
under the assumption layout developers want my blog to render correctly? :)
13:17
<hsivonen>
of course
13:18
<zcorpan>
hsivonen: if you go too far ahead then they'll ignore the whole thing :)
13:26
<Lachy>
I don't understand this font bug. The fonts work on http://fonts.philip.html5.org/ but not when I download them and load them from localhost
13:27
<annevk>
do fonts work from localhost at all?
13:29
<Lachy>
They should. I'm running Apache, not just using file:///
13:30
<annevk>
k
13:41
<Lachy>
it seems to be either falling back to some sans-serif font that looks a lot like Helvetica, or it is using Helvetica with a narrower letter-spacing
13:55
<Philip`>
Lachy: Try checking Apache's access_log to see if the fonts are actually being downloaded?
13:56
<Lachy>
ok, I'll have to find out where they're stored
13:56
<Philip`>
Lachy: Also, are these the fonts which are used on the index page there, or the fonts that are generated when you click the button?
13:57
<Philip`>
Lachy: If it's Linuxish then it's probably /var/log/apache/access_log
13:57
<Philip`>
s/apache/apache2/
13:57
<Lachy>
I downloaded one of the fonts you're using on the page
13:58
<Lachy>
but my TC is using Anne's font though
15:44
<Philip`>
Lachy: Okay - I was just thinking you might need to be careful when downloading the fonts used on the index page, since they're gzipped regardless of whether your UA sends Accept-Encoding
15:45
<Philip`>
(I should probably set up mod_gzip properly instead)
15:45
<Lachy>
Philip`, I used wget to download it. I assume it would have unzipped it for me
15:48
<Philip`>
$ wget http://fonts.philip.html5.org/indexfonts/3a3846bc220dd7f0aadb8b12d839e6f4.ttf
15:48
<Philip`>
$ file 3a3846bc220dd7f0aadb8b12d839e6f4.ttf
15:48
<Philip`>
3a3846bc220dd7f0aadb8b12d839e6f4.ttf: gzip compressed data
15:48
<Philip`>
$ gzip -cd 3a3846bc220dd7f0aadb8b12d839e6f4.ttf|file -
15:48
<Philip`>
/dev/stdin: TrueType font data
15:48
<Philip`>
So, wget doesn't unzip it for you
15:49
<Philip`>
Those filenames are a bit ugly, really
15:56
<Lachy>
why do you use such ugly file names?
16:01
<Philip`>
Lachy: I needed something unique, so I just used an MD5 of the font name and the included characters
16:02
<Philip`>
Lachy: and it would be a bad idea to use the original font's filename, because then the subsets would get confused with each other and with the original
16:19
<hsivonen>
Philip`, Lachy: it seems gzip works as a copying deterrent after all :-)
16:20
<Philip`>
hsivonen: But it only deters competent users who know about wget, not normal users who would paste the URL into their browser, so it's not a hugely effective deterrent :-p
16:21
<Philip`>
...at least when it's just HTTP gzip compressed transmission
16:21
<Philip`>
If the actual files were distributed as compressed data, over an uncompressed HTTP transmission, then it would indeed be more effective :-)
16:32
<Lachy>
Philip`, it works in WebKit when I unzip the file.
16:35
<Lachy>
That test partially works in Opera too, except only some of the letters rendered using that font.
16:37
<Lachy>
oh, Opera has trouble rendering strings comprising all 26 letters in the right font, though it works better if I put in a few spaces
16:39
<Lachy>
ah, I see. once Opera hits a character not included in the web font, it won't revert back to that font for subsequent letters until after a space
16:42
Philip`
hopes the sensible thing to do with Apple Advanced Typographic tables in fonts is simply to delete them, as an attempt to promote cross-platformness
16:42
<Lachy>
Philip`, are the font files you're serving from that page completely valid, or are they likely to contain errors?
16:42
<Lachy>
I mean the ones you're using, not the copies the generator will spit out
16:43
<Philip`>
(and keep the OpenType tables which allow similar functionality, and are supported on Windows and Linux and apparently at least partly on OS X, so hopefully the world will converge on that)
16:43
<Philip`>
Lachy: The ones on the index page are spat out by the generator too
16:43
<Philip`>
Lachy: so they're likely to have similar bugs
16:43
<Lachy>
ok. Is there a font validator somewhere?
16:44
<Philip`>
Use the font and write some text, and if it looks okay then it's valid enough :-)
16:44
<Philip`>
I'm not aware of any tools that do any kind of comprehensive checking
16:45
<Lachy>
ok. Perhaps someone (you!) should develop one, so we can help reduce the occurrence of erroneous fonts occuring on the web once webfonts take off
16:45
<Philip`>
but I'm pretty sure the subsetted fonts I'm generating are buggy
16:46
<Lachy>
which tool are you using to generate them?
16:46
<Philip`>
It doesn't seem so useful to have a validator for fonts since they're all generated by software, so there won't be millions of people introducing millions of syntax errors
16:47
<Philip`>
I'm using the Font::TTF Perl module, plus a load of custom code that strips out lots of glyphs and then tries to update all the internal references to be consistent
16:47
<Lachy>
having validators will still allow people to check that the fonts their tools are generating are valid and will help the developers of those tools find and fix bugs
16:48
<gsnedders>
Lachy: Assuming the validator works correctly
16:48
<Lachy>
and it will also help browser vendors diagnose bugs
16:48
<Lachy>
Validators should exist for image and video formats too.
16:49
<Lachy>
(I did find one for MP4 once, but it cost thousands of dollars and was aimed at the pro market)
16:51
<jgraham>
gsnedders: BTW Black Box Recorder (music)
16:51
<gsnedders>
jgraham: What?
16:51
<jgraham>
Bands you should lsten to. Black Box Recorder
16:51
<gsnedders>
ah
16:52
<jgraham>
gsnedders: http://se.youtube.com/watch?v=FP-2VLQEv4c
16:53
<zcorpan>
http://browsershots.org/http://annevankesteren.nl/
16:56
<Philip`>
http://www.adobe.com/devnet/opentype/afdko/ includes TTX, which converts fonts to an XML format, which can be useful for debugging
16:58
Philip`
is attempting to more comprehensively look through all the interesting OpenType tables, to work out which ones he really needs to process and which can just be ignored or dropped
16:58
<Philip`>
and then I should be more confident that I'm producing correct output
16:58
<Philip`>
(The main problem is forgetting to update all the references to glyphs that have been removed)
17:00
<gsnedders>
Philip`: n00b
17:03
<Philip`>
gsnedders: If you want to read http://www.microsoft.com/typography/otspec/gsub.htm and let me know exactly how to keep everything consistent when some glyphs are removed, please feel free :-)
17:04
<gsnedders>
Nah
17:09
<gsnedders>
Oh shit.
17:10
<gsnedders>
I said I'd do one thing for Hixie by today.
17:11
<Philip`>
One thing in particular, or just any thing?
17:12
jgraham
starts humming 'I'd do anything' or whatever that song from 'Oliver' is called
17:12
<gsnedders>
Philip`: Anolis template
19:25
<Hixie>
gsnedders: i think i have an idea for how to do cross-spec references btw
19:25
<takkaria>
any opera people around?
19:25
<Hixie>
without having to load both specs
19:25
<gsnedders>
Hixie: How?
19:26
<gsnedders>
Hixie: and: email.
19:26
<Hixie>
just have a declaration mechanism that "imports" some names
19:52
<Hixie>
Lachy: yt?
19:52
<Hixie>
annevk: yt?
19:56
<annevk>
am now
20:05
<takkaria>
annevk: do you know who deals with the applications for jobs at opera?
20:05
<annevk>
http://lists.w3.org/Archives/Public/public-xhtml2/2009Jan/0059.html I hope no market leader will ever attempt to natively support XHTML2; that'll become a huge mess to sort out
20:06
<annevk>
takkaria, HR department
20:08
<annevk>
takkaria, they'll pass it along to the appropriate people who will then get back to you (I think it works like that more or less)
20:10
<takkaria>
right, ta
20:20
gsnedders
has one slight issue wrt applying to Opera
20:20
<gsnedders>
I don't know if I'm going to take a gap year or not; if I don't I could get a summer internship, if I do I could try and get a job for nearer a year
20:21
<annevk>
you can apply for a summer internship and if you like it and we like you stay for a year
20:21
<dglazkov>
gsnedders: apply to Google :)
20:21
<annevk>
that sentence requires another you and a comma, but you get the idea
20:22
<gsnedders>
annevk: If I take a gap year, I really need to be around in the UK running off to places all the time in September, which makes it rather not sensible to start before Oct.
20:23
<gsnedders>
dglazkov: And quite how low are my chances at getting a job without a degree? :)
20:23
<annevk>
gsnedders, planes exist for a reason
20:23
<dglazkov>
gsnedders: a degree of what? ;)
20:24
<gsnedders>
annevk: Yeah, but taking, say, four weeks off might not make me popular :)
20:24
<gsnedders>
dglazkov: Oh, I dunno. A university one? :)
20:24
<Lachy>
Hixie, yo
20:24
<annevk>
gsnedders, it doesn't seem like a huge deal to me
20:25
<annevk>
gsnedders, finished your internship, go home, and then come back again later to work on something else...
20:25
<dglazkov>
universities are just deprecated facilities for teaching young people about procrastination and the value of borrowing someone else's work :)
20:25
<gsnedders>
annevk: Also, I'm not really meant to apply for the internship until I'm enrolled in a uni course :P
20:25
<hsivonen>
hmm. the deadline for comments on CSS3 Web Fonts is 2002-08-30
20:26
<gsnedders>
dglazkov: But pretty much everything in this industry requires a degree :(
20:26
<gsnedders>
hsivonen: Think you can make that, then?
20:26
<dglazkov>
gsnedders: :(
20:27
<Philip`>
hsivonen: Perhaps you want http://dev.w3.org/csswg/css3-fonts/ instead? :-)
20:27
<annevk>
hsivonen, it's being moved back to WD
20:27
<annevk>
hsivonen, and you want the URL Philip` points out
20:27
<hsivonen>
thanks
20:28
<gsnedders>
annevk: Also, if I take a gap year, it'd be nice to have a couple of months at the start to try and do all these damned things I keep saying I'll do
20:34
<gsnedders>
(If anyone wants to hire me for Oct–Aug/Sep, let me know :P)
20:34
<annevk>
gsnedders, so start later, or convince someone at Opera to pay for it :)
20:36
<gsnedders>
annevk: A fair amount of the things it is almost certain that Opera wouldn't :)
20:37
<Hixie>
ok time to get ready. bbl.
20:38
<annevk>
gsnedders, it almost seems like you don't want to apply
20:39
<Philip`>
gsnedders: I'll hire you, as long as you don't expect any financial compensation for your time
20:39
<gsnedders>
annevk: I'm indecisive :P
23:13
Philip`
discovers that Opera doesn't like changing styles (e.g. colour) in the middle of a ligature
23:14
<Philip`>
(whereas Firefox handles it pretty nicely)
23:27
<Philip`>
(Hmm... Opera/Safari never do Latin ligatures, but do do Arabic ligatures (probably not what they're called but the same idea) that get broken when you change style in the middle; IE and Firefox do both Latin and Arabic ligatures, but IE breaks Latin ligatures when you change style)
23:27
<Philip`>
(So it's not entirely consistent :-( )
23:35
<weinig>
Hixie: is it defined currently whether an onload handler for a frame needs to wait for all subresources to load before firing?
23:35
<Hixie>
yes
23:35
<Hixie>
search for "delay the load event"
23:35
<weinig>
thanks
23:35
<Hixie>
the spec isn't very thorough yet though, if you spot anything missing let me know
23:36
<weinig>
Hixie: I am thinking about external references to SVGFonts
23:36
<weinig>
Hixie: so that is probably not covered
23:37
<weinig>
Hixie: but is required by Acid 3
23:37
<Hixie>
external from what, css?
23:37
<weinig>
Hixie: no, other svg
23:37
<weinig>
<font-face-uri xlink:href="font.svg#mini"/></font-face-src>
23:37
<weinig>
type things
23:37
<Hixie>
oh then ask the svgwg
23:38
<Hixie>
there's a hook in html5 if they want to define it
23:40
<roc>
Philip`: glad someone noticed, an awful lot of pain went into that. My pain.
23:40
<roc>
the Arabic forms are correctly called ligatures, BTW
23:41
<Hixie>
i've been tempted to use Zapfino's "Zapfino" ligature in Acid4
23:44
<Hixie>
huh, firefox isn't rendering Zapfino's Zapfino ligature for me
23:44
<Hixie>
it does the "pf" ligature inside "Zapfino"
23:44
<Hixie>
safari doesn't even do that
23:44
<heycam>
weinig, so that'd be: <iframe src=something.svg>, and in something.svg you have that reference to an SVG font resource in a different file, and you'd like to know whether that impacts the load event dispatched to the iframe?
23:44
<Hixie>
and opera does both
23:45
<weinig>
heycam: exactly
23:45
<Hixie>
Philip`: how do you mean, it doesn't handle colour changes?
23:45
<Philip`>
Hixie: Seems kind of difficult to rely on Zapfino when most computers don't have that font and it's presumably not freely redistributable
23:45
<weinig>
heycam: ala Acid3
23:45
<roc>
Hixie: the "Zapfino" ligature is a "discretionary" ligature
23:45
<Hixie>
roc: aah
23:45
<roc>
we disabled discretionary ligatures because in many situations authors don't want them
23:45
<Hixie>
fair enough
23:46
<Hixie>
wow, you do handle the colour change in a ligature well
23:46
<roc>
we need a CSS extension to provide this control to authors ... one of our guys has it on his to-do list
23:46
<heycam>
weinig, ok. we don't say anything on the matter currently. how does it work normally with HTML in an iframe? does the load on the iframe happen as a result of the load dispatched to the html in the frame?
23:46
<heycam>
weinig_, ok. we don't say anything on the matter currently. how does it work normally with HTML in an iframe? does the load on the iframe happen as a result of the load dispatched to the html in the frame?
23:47
<Hixie>
Philip`: i was considering asking apple to license a tiny subset of it for ahem purposes. but i doubt i will go down that road in acid4 anyway.
23:47
<roc>
Philip`: yeah, but we could just do something like a Hixifino font with a Hixifino ligature
23:47
<Hixie>
right
23:47
<Hixie>
if we want to test this at all
23:47
<Philip`>
roc: If I was really picky, I could complain that e.g. "f<font color=red>f</font>i" in Linux Libertine doesn't work perfectly, since the redness is shifted a bit too far left and covers part of the first 'f', but at least it's nicer than deligaturising them entirely :-)
23:47
<Hixie>
which isn't clear given the state of the relevant specs
23:48
<roc>
ligatures don't need a spec IMHO, at least for "required ligatures"
23:48
<Philip`>
Sounds like you're wanting http://www.webfonts.info/wiki/forum/viewtopic.php?f=5&t=7
23:48
<roc>
Philip`: there's just no way to do it perfectly
23:48
<roc>
the fonts don't have enough info
23:49
<Hixie>
roc: well, i mean there's nothing in CSS that says how to colour one or the other part of a ligature
23:49
<Hixie>
roc: i agree that TTF+CSS is enough to require that ligatures be shown at all
23:49
<roc>
yeah
23:49
<roc>
but you could test <span>f</span><span>i</span>
23:49
<weinig_>
heycam: the onload happens when the main resource for the iframe is done loading and all subresources defined to delay the onload have also finished
23:49
<roc>
and see if that creates a ligature
23:49
<Hixie>
roc: right, if there are any browsers that fail thatbut pass fi (are there? i assume there won't be by acid4)
23:50
<Philip`>
roc: It looks like OpenType has the features for that (in GDEF's "ligature caret positioning information"), but I haven't seen any fonts that provide that data (though I've only looked at a handful of fonts)
23:50
<roc>
Hixie: I don't know ... getting from "fi" working to "<span>f</span>i" working is a ton of work
23:50
<Philip`>
Oh, actually, Linux Libertine does appear to provide that data
23:51
<roc>
well, it depends on how the engine is implemented, but for most "obvious" implementations I imagine it's hard
23:51
<roc>
Philip`: it does? what table?
23:52
<heycam>
weinig_, ok so that's possibly later than the load event dispatched to the iframe's window object?
23:52
<Philip`>
roc: http://www.microsoft.com/typography/otspec/gdef.htm - "The LigatureCaretList table contains positioning data for ligature carets, which the text-processing client uses on screen to select and highlight the individual components of a ligature glyph."
23:52
<weinig_>
heycam: I don't think so
23:52
<heycam>
oh
23:52
<Hixie>
roc: both opera and mozilla do it, and safari doesn't do ligatures at all
23:53
<roc>
yay us I guess
23:53
<heycam>
weinig_, load dispatched to the <iframe> and load dispatched to the iframe content's window object are distinct, yes?
23:53
<roc>
but I assure you it was hard :-)
23:54
Philip`
notes that IE dynamically deligaturises ligatures when you attempt to select individual characters
23:54
<weinig_>
heycam: I am not sure, but I didn't think so
23:54
<heycam>
weinig_, ok well i don't have a good understanding of html-ish things like this, so i'm probably wrong
23:54
<Philip`>
...or even when you simply click on the ligature and don't move the mouse
23:54
<roc>
Philip`: Libertine has that eh? Cool ... unfortunately, however, platform APIs don't provide that data AFAIK so I'll have to wait until we have our own shaping code before we can use it
23:55
<roc>
Philip`: oh, so selection can change the layout? Urgh
23:55
<Hixie>
roc: i believe you :-)
23:55
<heycam>
weinig_, when you do <iframe onload=...> does that actually register a listener on the iframe content's window object then?
23:56
<Philip`>
roc: The layout of the rest of the line doesn't seem to change, it just renders the ligature as individual glyphs
23:56
heycam
finds this in the spec: "When content loads in an iframe, after any load events are fired within the content itself, the user agent must fire a load event at the iframe element."
23:56
<roc>
ick
23:56
heycam
wonders then if the load dispatched to the iframe content's window object is considered to be "within the content itself"
23:56
<Philip`>
Hmm, also it seems to have broken spacing after ligatures, like it's using the spacing for the non-ligature glyphs, though that might be because my font subsetter is buggy
23:59
<Hixie>
heycam: that text is horrible, i should fix that to be well-defined