Options:
- # Session Start: Tue Nov 30 00:00:00 2010
- # Session Ident: #css
- # [00:28] * Joins: homata (homata@58.158.182.50)
- # [02:26] * Joins: jdaggett (jdaggett@202.221.217.73)
- # [03:18] * Quits: homata (homata@58.158.182.50) (Ping timeout)
- # [03:19] * Joins: homata (homata@58.158.182.50)
- # [04:53] * Joins: homata_ (homata@58.158.182.50)
- # [04:55] * Quits: homata (homata@58.158.182.50) (Ping timeout)
- # [05:13] * Joins: homata (homata@58.158.182.50)
- # [05:14] * Quits: homata_ (homata@58.158.182.50) (Ping timeout)
- # [05:36] * Joins: kojiishi (kojiishi@222.158.227.129)
- # [05:56] * Joins: homata_ (homata@58.158.182.50)
- # [05:58] * Quits: homata (homata@58.158.182.50) (Ping timeout)
- # [06:19] * Quits: nimbupani (Adium@24.22.131.46) (Quit: Leaving.)
- # [06:28] * Quits: homata_ (homata@58.158.182.50) (Ping timeout)
- # [06:29] * Joins: homata (homata@58.158.182.50)
- # [06:39] * Quits: homata (homata@58.158.182.50) (Ping timeout)
- # [06:39] * Joins: homata (homata@58.158.182.50)
- # [07:08] * Quits: homata (homata@58.158.182.50) (Ping timeout)
- # [07:09] * Joins: homata (homata@58.158.182.50)
- # [07:28] * Joins: homata_ (homata@58.158.182.50)
- # [07:29] * Quits: homata (homata@58.158.182.50) (Ping timeout)
- # [09:45] * Quits: jdaggett (jdaggett@202.221.217.73) (Quit: jdaggett)
- # [12:13] * Joins: rapitivo (rapitivo@201.22.227.7)
- # [12:25] * Joins: anne (annevk@213.236.208.22)
- # [12:46] * Quits: arronei (arronei@131.107.0.87) (Ping timeout)
- # [12:46] * Quits: kojiishi (kojiishi@222.158.227.129) (Quit: Leaving...)
- # [12:51] * Joins: arronei (arronei@131.107.0.102)
- # [13:17] * Quits: homata_ (homata@58.158.182.50) (Quit: Leaving...)
- # [13:25] * Quits: rapitivo (rapitivo@201.22.227.7) (Connection reset by peer)
- # [14:24] * Quits: anne (annevk@213.236.208.22) (Client exited)
- # [14:24] * Joins: anne (annevk@213.236.208.22)
- # [14:38] * Quits: kennyluck (kennyluck@128.30.52.169) (Quit: kennyluck)
- # [14:39] * Joins: miketaylr (miketaylr@24.42.95.108)
- # [14:39] * Joins: kennyluck (kennyluck@128.30.52.169)
- # [15:05] * Quits: kennyluck (kennyluck@128.30.52.169) (Quit: kennyluck)
- # [15:21] * Joins: nimbupani (Adium@24.22.131.46)
- # [16:12] * Joins: kennyluck (kennyluck@128.30.52.169)
- # [16:14] * Joins: bradk (bradk@99.7.175.117)
- # [16:53] * Quits: lhnz (lhnz@188.223.83.48) (Ping timeout)
- # [16:53] * Joins: lhnz (lhnz@188.223.83.48)
- # [17:08] * Quits: bradk (bradk@99.7.175.117) (Quit: Get MacIrssi - http://www.sysctl.co.uk/projects/macirssi/ )
- # [17:13] * Quits: nimbupani (Adium@24.22.131.46) (Quit: Leaving.)
- # [17:17] * Joins: nimbupani (Adium@24.22.131.46)
- # [17:24] * Quits: miketaylr (miketaylr@24.42.95.108) (Quit: miketaylr)
- # [17:27] * Joins: miketaylr (miketaylr@24.42.95.108)
- # [19:12] * Quits: anne (annevk@213.236.208.22) (Quit: anne)
- # [19:47] * Joins: dbaron (dbaron@63.245.220.240)
- # [19:57] * Joins: hyatt (hyatt@98.200.224.166)
- # [20:02] <hyatt> fantasai: ping
- # [20:02] <fantasai> plinss_: Bert's reporting a server problem on csswg.org http://lists.w3.org/Archives/Public/public-css-testsuite/2010Nov/0142.html
- # [20:02] <fantasai> hyatt: pong
- # [20:02] <hyatt> fantasai: should this work...
- # [20:02] <hyatt> <span style="text-combine: horizontal 2">
- # [20:02] <hyatt> <span style="color:red">2</span><span style="color:blue">0</span>
- # [20:02] <hyatt> </span>
- # [20:02] <fantasai> no
- # [20:02] <hyatt> so basically only raw text is allowed inside the span or you discard the text-combine?
- # [20:02] <fantasai> yes
- # [20:02] <fantasai> otherwise it gets complicated
- # [20:02] <hyatt> ok
- # [20:02] <fantasai> I'm happy to reconsider if you think otherwise :)
- # [20:02] <hyatt> no i don't think otherwise
- # [20:03] <fantasai> hehe
- # [20:03] <hyatt> just working on implementing it
- # [20:03] <hyatt> and the approach can be simpler if it's one text
- # [20:03] <fantasai> that's what I figured
- # [20:03] <fantasai> if someone wants to do something fancy, he can fake it with display: inline-block; writing-mode: horizontal-tb
- # [20:04] <hyatt> i'm trying to figure out if i should do that
- # [20:04] <hyatt> turn the span into an inline block with horizontal-tb
- # [20:04] <hyatt> or if i should just hack the text painting
- # [20:05] <hyatt> and keep the span with the same writing mode etc
- # [20:05] <fantasai> jdaggett said the behavior is different than with display: inline-block; writing-mode: horizontal-tbl;
- # [20:06] <hyatt> well there's choosing a compressed font and fitting it to 1em etc
- # [20:06] <fantasai> I'm not sure what all the issues are -- probably I should get a list of considerations
- # [20:06] <fantasai> there's that, yes
- # [20:06] <hyatt> i'm mainly thinking of how to get it to be upright
- # [20:06] <hyatt> do i hack the writing mode on the span or just hack the painting of the text
- # [20:06] <fantasai> the painting of the text
- # [20:06] <fantasai> it's going to have to behave as one grapheme cluster
- # [20:07] <fantasai> for various purposes
- # [20:07] <fantasai> letter-spacing
- # [20:07] <fantasai> text-emphasis
- # [20:07] <fantasai> etc
- # [20:07] <fantasai> underlining
- # [20:07] <hyatt> yeah
- # [20:19] <hyatt> fantasai: btw i implemented some really smart baseline adjustment... it is working really really well
- # [20:20] <hyatt> fantasai: the heuristic i used was that if any used font on a line contains vertical forms
- # [20:20] <hyatt> then it taints the whole line
- # [20:20] <hyatt> and all the non-vertical fonts move to using centered baselines
- # [20:20] <hyatt> fantasai: pretty simple to implement and works really well
- # [20:21] <hyatt> fantasai: this way the line just does the right thing based off what's on it without having to resort to language detection
- # [20:21] <hyatt> interesting mixes end up working like you'd expect, e.g., mongolian with some mixed-in ideograph content
- # [20:21] <hyatt> normal english text with no ideographs will baseline-align like it does in horizontal too
- # [20:21] <hyatt> etc.
- # [20:22] <fantasai> Mongolian with ideographic is pretty common :)
- # [20:22] * Quits: CSSWG_LogBot (PircBot@173.230.149.95) (Client exited)
- # [20:23] <fantasai> hyatt: Can you email your ideas to www-style?
- # [20:23] <hyatt> fantasai: sure
- # [20:23] <fantasai> hyatt: It's easier for me to track that way
- # [20:23] <hyatt> fantasai: i basically did that already... i outlined 3 approaches
- # [20:23] <hyatt> fantasai: so basically i implemented one of those 3
- # [20:23] <hyatt> but i can elaborate if it's helpful
- # [20:23] <hyatt> i really don't want to see language-based decisions
- # [20:24] <hyatt> since it's all about ideographs
- # [20:24] * Joins: CSSWG_LogBot (PircBot@173.230.149.95)
- # [20:24] -CSSWG_LogBot:#css- This channel is logged - http://log.csswg.org/irc.w3.org/css/
- # [20:24] <hyatt> which are easily detectable
- # [20:24] <fantasai> hyatt: no, definitely not language-based
- # [20:24] * fantasai pulls up your email
- # [20:25] <hyatt> basically think of a line as having a "baseline type", e.g., alphabetic or ideographic
- # [20:25] <hyatt> if any ideographs occur on the line, the "baseline type" of the line becomes ideographic
- # [20:25] <fantasai> http://lists.w3.org/Archives/Public/www-style/2010Nov/0081.html
- # [20:25] <fantasai> option 3?
- # [20:25] <hyatt> yeah
- # [20:25] <hyatt> with "or alternatively just synthesize a baseline that is halfway through the text while still keeping the vertical-align value the same"
- # [20:26] <hyatt> i did the synthesis
- # [20:26] <hyatt> so that vertical-align:baseline does the right thing
- # [20:26] <fantasai> I don't like making it font-dependent
- # [20:26] <fantasai> means if I use Arial Unicode MS vs. Arial, I get different results
- # [20:26] <hyatt> it can be either font-dependent or ideograph-dependent
- # [20:26] <hyatt> i can't speak to other operating systems
- # [20:26] <hyatt> but every single font on the mac is correct
- # [20:27] <hyatt> there is no such thing as rendering japanese without having the font have vertical tables
- # [20:27] <fantasai> right, but rendering English?
- # [20:28] <fantasai> If you use a font that supports vertical rendering vs. one that doesn't, you'd get different results
- # [20:28] <fantasai> for no apparent reason
- # [20:28] <fantasai> hyatt: btw, wrt text-combine, there have been some requests for a sort of auto-text-combine feature
- # [20:28] <fantasai> hyatt: in which case, you wouldn't be manipulating an actual element
- # [20:29] <hyatt> well, the fonts that support vertical rendering on the mac at least are just the ideographic ones
- # [20:29] <hyatt> note i'm talking about used fonts not specified fonts
- # [20:29] <hyatt> if you say <span style="font-family:Times">... japanese ideographs here... </span>
- # [20:29] <hyatt> those are going to render in Hiragino
- # [20:29] <hyatt> etc
- # [20:30] <fantasai> If you say <span style="font-family: Arial">...English text here...</span>
- # [20:30] <fantasai> those will render in a font without vertical tables
- # [20:30] <hyatt> exactly
- # [20:30] <hyatt> and so will do the right thing and not center baseline
- # [20:30] <fantasai> If you say <span style="font-family: Arial Unicode MS">... English text here...</span> ?
- # [20:30] <hyatt> the idea is you get smart baseline behavior based off what fonts you use
- # [20:31] <hyatt> fantasai: but making it ideograph based would probably be even better than font based
- # [20:32] <hyatt> i did font based since for all practical intents and purposes it works great, but the whole "universal font that does both english and japanese" is not what japanese authors typically want to use
- # [20:32] <hyatt> on os x at least
- # [20:33] <hyatt> ideograph-based would probably be better though
- # [20:34] <hyatt> but yeah i agree if you only use english text and you use a font with vertical tables that also happens to do english
- # [20:34] <fantasai> There are some very nice English+Japanese fonts out there.
- # [20:34] <hyatt> that it should not center baseline align
- # [20:34] <fantasai> but anyway, this is something I need to work on
- # [20:34] <hyatt> fantasai: yeah, there are also degenerate japanese fonts
- # [20:34] <hyatt> fantasai: although none installed by default on os x
- # [20:34] <hyatt> but third party ones like Makiba
- # [20:34] <hyatt> don't have vertical tables
- # [20:35] <hyatt> i synthesized vertical rendering for broken fonts like that
- # [20:35] <hyatt> (the punctuation looks wrong obviously)
- # [20:35] <hyatt> but the ideographs are at least upright
- # [20:35] <hyatt> but with a font-based choice, i would not baseline align
- # [20:35] <hyatt> using the center
- # [20:35] <hyatt> like i should
- # [20:36] <hyatt> so i guess ideograph usage would be a better heuristic
- # [20:36] <hyatt> browsers already have to have an isCJKIdeograph method for a character anyway
- # [20:36] <hyatt> in order to do text-orientation
- # [20:36] <hyatt> this is something you already have to know heh
- # [20:37] <hyatt> fantasai: so i guess i'd suggest ideograph-based rather than font-based.
- # [20:37] <hyatt> but the concept of a "baseline type" per line seems good... just pick it based off whether ideographs got used on the line.
- # [20:43] <hyatt> fantasai: anyway, things are looking really nice now... only major subsystems left to convert are scrolling/overflow and multicol/pagination
- # [21:21] * Joins: dydz (dydz@76.202.118.219)
- # [21:58] * Quits: kennyluck (kennyluck@128.30.52.169) (Ping timeout)
- # [22:33] <plinss> fantasai: fixed. If you see any other issues, please let me know.
- # [22:36] * Quits: miketaylr (miketaylr@24.42.95.108) (Quit: miketaylr)
- # [23:51] * Joins: anne (annevk@95.34.115.33)
- # [23:55] * Joins: homata (homata@58.158.182.50)
- # Session Close: Wed Dec 01 00:00:00 2010
The end :)