/irc-logs / w3c / #css / 2014-12-30 / end

Options:

# Session Start: Tue Dec 30 00:00:00 2014
# Session Ident: #css
# [00:35] * Quits: thinkxl (~thinkxl@public.cloak) ("Lingo: www.lingoirc.com")
# [01:19] <sgalineau> leaverou: this sounds like a bug IE itself ran into using the exact same font family
# [01:19] <sgalineau> leaverou: in particular, the Ultra Bold was installed by Office and had this side effect
# [01:19] * leaverou_away is now known as leaverou
# [01:21] <sgalineau> leaverou: it's been long enough the details are fuzzy but this is a Windows issue browsers need to code around
# [01:23] <leaverou> sgalineau: The Windows issue is that users end up with that combination of fonts. It still stands that displaying the whole thing in Ultra bold is correct per spec and it shouldn't be (see my message to the list)
# [01:24] <sgalineau> leaverou: the issue we ran into was that once Ultrabold was installed, reference to the Gill Sans family used the Ultra Bold weight which was wrong for pretty much all sites
# [01:25] <leaverou> sgalineau: exactly, but this is correct behavior according to the spec. Font matching never fails for weight and in those cases it should
# [01:25] <sgalineau> leaverou: well, this is definitely not correct per any sane authors I know so it was fixed.
# [01:26] <leaverou> sgalineau: it was not fixed in the spec. Even in the ED, the algorithm only specifies in which *order* weights will be matched (which is also suboptimal, but that's another issue)
# [01:26] <sgalineau> leaverou: specifically, the installation of a new weight should not screw up existing pages.
# [01:26] <leaverou> and it was not "fixed", except in Firefox, which is a hack (and technically a spec violation, according to the current algorithm)
# [01:27] <sgalineau> leaverou: but I'm really not sure the spec went anywhere near the level of detail where this issue happened at at the time. Again, it was a Windows-specific issue so I doubt the spec had anything to do with it
# [01:28] <leaverou> sgalineau: no, it happens on OSX too if you disable all weights of Gill Sans except Ultra bold.
# [01:28] <leaverou> in fact, it even happens on Firefox OSX, since the "fix" was only for Windows
# [01:28] <sgalineau> leaverou: well yeah, of course.
# [01:28] <leaverou> the issue with Windows is that due to MS Office, users ended up with that specific font configuration a lot
# [01:29] <leaverou> sgalineau: yes, but users might always have incomplete font families for a number of different reasons. The font matching algorithm should fall back to the next family in cases of such huge difference.
# [01:30] <sgalineau> leaverou: that's a different issue. In this case, you get Ultrabold even if MT is there and that's definitely a Windows thing i.e. even if you change the spec algorithm Windows would still only give you Ultrabold i.e. you have to work around it
# [01:30] <sgalineau> leaverou: it was a royal PITA. no wonder I repressed the memory of it :)
# [01:30] <leaverou> well, there is a second issue at play here that that font misreports itself as a 400 weight
# [01:31] <leaverou> but I'm saying that even if it correctly reported itself as 900, this would be correct behavior according to the spec, which needs to be changed.
# [01:31] <sgalineau> mmmm...not sure. Anyway, looking through the archive as I thought this was discussed at the time...
# [01:32] <sgalineau> nope, this was only discussed privately. rats.
# [01:33] <sgalineau> I'm sure jdaggett will remember though
# [01:33] <leaverou> sgalineau: The reply to my post on www-style clarified why this was happening, so I'm not troubled by that anymore
# [01:33] <leaverou> sgalineau: currently I'm more concerned about the font matching algorithm, as I described on said post
# [01:33] <sgalineau> yeah but we discussed the spec angle too, so I'm surprised this is still an issue
# [01:33] <sgalineau> anyway. good times.
# [01:34] <leaverou> sgalineau: the spec angle was glossed over in the public bug report, so I guess it must have been in the private discussion
# [01:35] <sgalineau> leaverou: maybe. years pass, things fall off the table...
# [01:35] <leaverou> sgalineau: it's not even just the fact that it doesn't fall back for big weight differences. It also often doesn't make the best choice of which weight to pick, if the desired one is missing. See my post
# [01:36] <sgalineau> leaverou: yeah, that's been an issue for some time
# [01:36] <leaverou> sgalineau: why? just time resources or is there a reason the algorithm is like that?
# [01:37] <sgalineau> leaverou: I don't really know. font matching tends to be far more complicated than it looks. I learned just enough to know how much I don't know :)
# [01:38] * Joins: estellevw (~estellevw@public.cloak)
# [01:39] <sgalineau> leaverou: rebooted my memory from the Bugzilla issue. Yes, low-level font name substitution is what it took
# [01:40] <sgalineau> leaverou: specifically comment 24. IE had to work around the same bug. Ugly, ugly stuff. Sigh.
# [01:40] <leaverou> sgalineau: thanks, but you're talking about that specific issue (which was caused by a number of things), whereas I've moved on from that. :)
# [01:41] <sgalineau> leaverou: you think you've moved from it. the way this stuff actually works informs many bits of the font matching algorithm
# [01:41] <leaverou> what do you mean, in the context of the specific change I'm proposing?
# [01:42] <sgalineau> I mean, wait for jdaggett to explain why things are the way they are :)
# [01:43] <leaverou> sgalineau: but the suspense is killing me :P
# [01:43] <sgalineau> lol. yes, me too.
# [01:43] <sgalineau> fonts/text sure are complex things
# [01:43] <leaverou> everything is complex if you look at it deep enough
# [01:44] <sgalineau> no, text is quite a particular beast
# [01:47] <sgalineau> from font matching and rendering to encoding to line layout, line breaking algorithms, writing modes, all the i18n intricacies, the layers of legacy formats and APIs involved across platforms...it's not just deep it's very, very broad
# [01:48] <leaverou> ok, yeah, I see what you mean
# [01:49] <sgalineau> I think what I'm saying is that the amount of stuff we take for granted using it is huge vs. say flebox, grid, canvas or blending which pretty much work the way you think from reading the spec
# [01:50] <leaverou> not sure layout modules are a good example for this, layout is pretty complex too
# [01:50] <leaverou> but I see what you mean :)
# [01:53] <sgalineau> ok maybe table layout is harder :)
# [02:05] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [02:15] * leaverou is now known as leaverou_away
# [03:06] * leaverou_away is now known as leaverou
# [03:11] * Joins: estellevw (~estellevw@public.cloak)
# [03:15] * leaverou is now known as leaverou_away
# [04:06] * leaverou_away is now known as leaverou
# [04:16] * leaverou is now known as leaverou_away
# [04:43] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [04:47] * Joins: estellevw (~estellevw@public.cloak)
# [05:09] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [05:11] <liam> oops, too late, but agree the font matching algo has some bad edge cases that are not as infrequent as one might hope
# [05:13] * Joins: estellevw (~estellevw@public.cloak)
# [05:21] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [06:06] * leaverou_away is now known as leaverou
# [06:14] * Joins: estellevw (~estellevw@public.cloak)
# [06:15] * leaverou is now known as leaverou_away
# [06:45] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [06:46] * Joins: estellevw (~estellevw@public.cloak)
# [07:13] * Quits: dbaron (~dbaron@public.cloak) (Ping timeout: 180 seconds)
# [09:07] <fantasai> sgalineau: No, I think font matching is a worse problem. There are finite number of quirks to consider for table layout. For fonts, I'm less sure.
# [09:08] <fantasai> sgalineau: Implementations have blacklists for font stuff.
# [09:08] <fantasai> sgalineau: It's that bad.
# [09:09] * fantasai doesn't remember what exactly the blacklists are for, but they're for misbehaving fonts in some capacity or other
# [09:10] <fantasai> I think it's like, imagine a random sampling of the Web, including Geocities. Now use that code as modules in your shiny new page. That's fonts.
# [09:11] * fantasai at least, that's the impression I get from talking to font people
# [09:11] <fantasai> s/Geocities/Geocities and the output of Frontpage/
# [09:13] <liam> for years a lot of frontpage sites used to come out in bold italic on Linux systems
# [09:14] <liam> turned out they shipped with Ariel UltraBold Italic
# [09:14] <liam> and no other fonts in that family
# [09:35] * Joins: Ms2ger (~Ms2ger@public.cloak)
# [10:52] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [10:56] * Joins: lajava (~javi@public.cloak)
# [12:22] * leaverou_away is now known as leaverou
# [13:24] * RRSAgent excuses himself; his presence no longer seems to be needed
# [13:24] * Parts: RRSAgent (rrsagent@public.cloak) (RRSAgent)
# [15:08] * Joins: dbaron (~dbaron@public.cloak)
# [16:06] * Joins: antonp (~Thunderbird@public.cloak)
# [16:23] * Joins: glazou (~glazou@public.cloak)
# [16:34] * Joins: thinkxl (~thinkxl@public.cloak)
# [16:37] * Quits: glazou (~glazou@public.cloak) (glazou)
# [16:57] * Quits: antonp (~Thunderbird@public.cloak) (antonp)
# [17:00] * Quits: dholbert (~dholbert@public.cloak) (Ping timeout: 180 seconds)
# [17:02] * Joins: dholbert (~dholbert@public.cloak)
# [17:45] * Quits: dholbert (~dholbert@public.cloak) (Ping timeout: 180 seconds)
# [17:48] * Joins: dholbert (~dholbert@public.cloak)
# [17:50] * leaverou is now known as leaverou_away
# [18:11] * Joins: estellevw (~estellevw@public.cloak)
# [18:33] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [19:34] * Quits: lajava (~javi@public.cloak) (Ping timeout: 180 seconds)
# [20:21] * Quits: Ms2ger (~Ms2ger@public.cloak) ("nn")
# [20:21] * Joins: estellevw (~estellevw@public.cloak)
# [20:24] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [21:06] * Joins: estellevw (~estellevw@public.cloak)
# [21:37] <sgalineau> fantasai: yeah, it's definitely worse in the gnarly legacy dimension
# [21:38] <sgalineau> fantasai: every time I've talked with John or Sergey about it I came out thinking it was amazing it seemed to work as well as it does. Like, major respect.
# [21:39] <sgalineau> fantasai: it's like this crazy Rube Goldberg machine that does what you expect most of the time
# [21:39] <sgalineau> fantasai: hmm, maybe I just described all browsers....
# [22:02] * Quits: estellevw (~estellevw@public.cloak) ("Snuggling with the puppies")
# [22:46] * leaverou_away is now known as leaverou
# [22:47] * Joins: estellevw (~estellevw@public.cloak)
# [23:56] * Quits: dbaron (~dbaron@public.cloak) (Ping timeout: 180 seconds)
# Session Close: Wed Dec 31 00:00:00 2014

Think these logs are useful? Then please donate to show your gratitude (and keep them up, of course). Thanks! — Krijn