/irc-logs / w3c / #html-wg / 2007-06-21 / end

Options:

# Session Start: Thu Jun 21 00:00:00 2007
# Session Ident: #html-wg
# [00:30] * Quits: Sander (svl@71.57.109.108) (Ping timeout)
# [00:34] * Quits: mjs (mjs@17.255.100.206) (Quit: mjs)
# [00:35] * Joins: mjs (mjs@17.255.100.206)
# [00:37] * Quits: mjs (mjs@17.255.100.206) (Quit: mjs)
# [00:37] * Joins: mjs (mjs@17.255.100.206)
# [00:37] * Joins: Sander (svl@71.57.109.108)
# [00:44] * Joins: heycam (cam@130.194.72.84)
# [00:58] * Quits: mjs (mjs@17.255.100.206) (Ping timeout)
# [01:06] * Joins: mjs (mjs@17.255.100.206)
# [01:11] * Quits: Lachy (Lachlan@203.158.59.119) (Connection reset by peer)
# [01:19] * Quits: billmason (billmason@69.30.57.156) (Quit: .)
# [01:30] * Joins: mjs_ (mjs@17.203.15.179)
# [01:33] * Quits: mjs (mjs@17.255.100.206) (Ping timeout)
# [01:34] <Philip`> Radial gradient rendering tests: Firefox 2: 61%. Firefox 3: 50%. Opera 9.2: 39%. Safari 3: 94%.
# [01:34] * Philip` gives the WebKit developers a cookie
# [01:34] * mjs_ likes cookies
# [01:34] * mjs_ is now known as mjs
# [01:35] <Hixie> Philip`: is that with the recent spec fixes?
# [01:35] <Philip`> Yes
# [01:35] <Hixie> cool
# [01:36] <Hixie> though you could also give a sanity score -- how self-consistent they are :-)
# [01:36] <Hixie> would probably be about the same scores
# [01:37] <Philip`> FF2 and Opera are somewhat lacking in sanity, but FF3 and Safari are both quite sensible (though only Safari is close to the spec)
# [01:37] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [01:38] <Hixie> i think that might be because i picked safari's rendering when i was fixing the spec
# [01:38] <Hixie> (the spec was insane for a bit too)
# [01:39] <Philip`> If I remember correctly, the updated spec was incompatible with Safari in a couple of ways (when the two circles were equal, and when their edges were touching) but their implementation seems to have been updated correctly
# [01:39] <Hixie> cool
# [01:40] <Philip`> There might still be bugs but not ones that I can work out how to find
# [01:44] * Joins: gavin (gavin@74.103.208.221)
# [01:54] * Quits: mw22 (chatzilla@84.41.169.151) (Ping timeout)
# [01:59] * Joins: karl (karlcow@128.30.52.30)
# [02:02] <Philip`> IE7 scores 7% on the radial gradient tests
# [02:03] * Philip` thinks this is not its strong point
# [02:03] * Joins: mjs_ (mjs@17.255.100.206)
# [02:06] * Quits: mjs (mjs@17.203.15.179) (Ping timeout)
# [02:09] * Quits: zcorpan (zcorpan@84.216.42.213) (Ping timeout)
# [02:13] <Hixie> Philip`: you mean excanvas?
# [02:15] <Philip`> Yep
# [02:16] <Philip`> Some people had suggested trying to implement it in Flash or Java, but excanvas seems to be the only choice at the moment :-(
# [02:17] * Quits: edas (edaspet@88.191.34.123) (Ping timeout)
# [02:44] * Quits: jgraham (jgraham@81.86.214.247) (Connection timed out)
# [03:01] * Quits: dbaron (dbaron@63.245.220.242) (Quit: 8403864 bytes have been tenured, next gc will be global.)
# [03:03] * Quits: inimino (inimino@75.71.88.233) (Client exited)
# [03:42] * Quits: mjs_ (mjs@17.255.100.206) (Quit: mjs_)
# [03:43] * Joins: mjs (mjs@17.255.100.206)
# [03:56] * Quits: Sander (svl@71.57.109.108) (Ping timeout)
# [04:26] * Joins: Lionheart (robin@66.57.69.65)
# [04:33] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [04:38] * Joins: gavin (gavin@74.103.208.221)
# [04:40] * Joins: Lachy (chatzilla@203.158.59.119)
# [04:44] * Joins: sbuluf (uuyxkre@200.49.140.208)
# [04:47] * Quits: heycam (cam@130.194.72.84) (Quit: bye)
# [04:48] * Joins: jgraham (jgraham@81.86.214.247)
# [05:19] * Quits: marcos (chatzilla@131.181.148.226) (Ping timeout)
# [05:47] * Quits: mjs (mjs@17.255.100.206) (Quit: mjs)
# [05:49] * Joins: Sander (svl@71.57.109.108)
# [06:18] * Quits: Zeros (Zeros-Elip@67.154.87.254) (Quit: Leaving)
# [06:41] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [06:46] * Joins: gavin (gavin@74.103.208.221)
# [06:47] * Joins: mjs (mjs@64.81.48.145)
# [06:49] * Joins: heycam (cam@203.214.72.248)
# [07:27] * Quits: Sander (svl@71.57.109.108) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
# [07:56] * Quits: Lachy (chatzilla@203.158.59.119) (Quit: ChatZilla 0.9.78.1 [Firefox 2.0.0.4/2007051502])
# [08:12] * Joins: zcorpan (zcorpan@84.216.43.161)
# [08:20] <zcorpan> html60 is in ur list, spoilin ur atmosphere
# [08:20] <karl> zcorpan: ?
# [08:21] <zcorpan> nm :)
# [08:48] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [08:50] * Joins: Lachy (chatzilla@203.158.59.119)
# [08:53] * Quits: Lachy (chatzilla@203.158.59.119) (Quit: ChatZilla 0.9.78.1 [Firefox 2.0.0.4/2007051502])
# [08:53] * Joins: gavin (gavin@74.103.208.221)
# [08:56] * karl wonders if henri has thought about putting background-color:#fff; on body {
# [08:56] <karl> font: caption;
# [08:56] <karl> padding: 3em;
# [08:56] <karl> margin: 0;
# [08:56] <karl> }
# [08:56] <karl> in http://hsivonen.iki.fi/style/validator.css
# [08:56] <karl> henri = hsivonen
# [08:57] <zcorpan> karl: why?
# [08:58] <karl> zcorpan: the new game in town - http://www.zeldman.com/2007/06/18/bgcolor-follies/
# [08:59] <zcorpan> why is it a problem?
# [09:00] <karl> read the previous sentence
# [09:00] <karl> I said the new *game* in town
# [09:00] <zcorpan> ah
# [09:00] <zcorpan> ok
# [09:20] <hsivonen> karl: it is by design. if a user doesn't like white for accessibility or taste reasons, (s)he should be free to override it
# [09:30] * Quits: karl (karlcow@128.30.52.30) (Quit: Where dwelt Ymir, or wherein did he find sustenance?)
# [09:38] * Joins: edas (edaspet@88.191.34.123)
# [10:11] <anne> plaintext does cause a parse error
# [10:11] <anne> unexpected end-of-file
# [10:12] * anne tried to get that to change
# [10:24] * Quits: mjs (mjs@64.81.48.145) (Ping timeout)
# [10:55] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [10:56] * Joins: Dashimon (noone@129.241.151.35)
# [10:56] * Quits: Dashiva (noone@129.241.151.35) (Connection reset by peer)
# [10:56] * Dashimon is now known as Dashiva
# [11:00] * Joins: gavin (gavin@74.103.208.221)
# [11:05] <anne> Though I have not really explore cross document messaging, the description seems to head in the direction of making life of mash-ups easier.
# [11:05] <anne> oops
# [11:05] <anne> http://dy-verse.blogspot.com/2007/06/html-5-rocks.html
# [11:06] * Joins: ROBOd (robod@86.34.246.154)
# [11:11] <anne> "Rich: All existing XHTMLs have been modular, and HTML5 is not. It's a mess."
# [11:11] <anne> http://www.w3.org/2007/06/20-xhtml-minutes#item05
# [11:12] * Quits: gorme (gorm@213.236.208.22) (Ping timeout)
# [11:12] <anne> The namespace bit is funny too
# [11:12] <anne> HTML5 is less compatible than XHTML2 is
# [11:12] <anne> yes browser vendors are willing to implement HTML5 and are not willing to implement XHTML2
# [11:12] <anne> hmm
# [11:13] * anne ponders
# [11:29] * Joins: mjs (mjs@64.81.48.145)
# [12:15] * Quits: zcorpan (zcorpan@84.216.43.161) (Ping timeout)
# [12:36] <hsivonen> making modularization a precondition for naming makes no sense at all
# [12:37] <anne> not with XHTML 1.0 around
# [12:38] <hsivonen> not with modularization having nothing to do with how desktop browsers implement the language
# [12:39] <anne> as I understand it modularization affects language design, not implementors
# [12:39] <anne> but it's all really unclear
# [12:40] <mjs> modularization is only relevant for internal spec organization, or for allowing easy creation of language profiles or mix & match languages
# [12:40] <mjs> I don't think either of those is an essential hallmark of XHTML
# [12:43] <hsivonen> in practice it will matter relatively little how Hixie (re)defines the the semantics of old elements but politically it might help to tone it down
# [12:44] <hsivonen> s/relatively/remarkably/
# [12:44] <mjs> tone down the redefinition or the discussion thereof?
# [12:45] <anne> tone down the redefinitions
# [12:45] <hsivonen> mjs: the redefinitions
# [12:45] <anne> I'm not sure we should bow soon under pressure though, but I don't really care
# [12:45] <hsivonen> mjs: since they will have near-zero practical effect, losing political capital over them is bad strategy
# [12:46] <mjs> with whom are we losing political capital?
# [12:46] <mjs> personally I wouldn't care if things like <b> and <i> were defined to be purely presentational
# [12:47] <mjs> <p> being conforming only for a standard prose paragraph would be a little annoying though
# [12:47] <hsivonen> mjs: we are losing the hyping mindshare benefit of those who go "OMG, they changed semantics!11!1111!!!"
# [12:48] <hsivonen> mjs: when they could be cheering for HTML5
# [12:48] <anne> I haven't seen that happening on digg.com yet
# [12:48] <anne> or any other blog
# [12:48] <anne> most people think it is quite sensible actually
# [12:48] <mjs> the people who say that would just say "OMG, they left in presentational elements!11!!111!" instead
# [12:49] <hsivonen> mjs: I'm all for allowing stuff semantically that was previously improper
# [12:49] <anne> they say that they've been using <b> and <i> like that for years or that it makes sense to redefine them as being non-presentational
# [12:49] <anne> I haven't found much other opinions besides on the HTML WG list itself
# [12:49] <hsivonen> mjs: but it would be better to say that it is ok to mark up copyright notices as <small> than to say that <small> denotes small print notices
# [12:50] <hsivonen> mjs: subtle difference but matters to some
# [12:50] <anne> but that was probably because those people didn't study the draft in detail
# [12:50] <hsivonen> mjs: zero practical relevance, of course
# [12:51] <hsivonen> basically, it would be politically better to say that <i>, <b>, <small>, etc. are perfectly OK for some purposes while refraining from saying that previous usage gets in principle any "new" explicit semantics
# [12:52] <hsivonen> basically, it is better to bless usage than to try to narrow the scope
# [12:53] <hsivonen> afk
# [12:54] <mjs> I think making those elements sort-of-semantic might also be winning mindshare in some quarters
# [12:55] <mjs> personally I don't care much what the defined semantics for those elements are
# [13:02] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [13:07] * Joins: gavin (gavin@74.103.208.221)
# [13:47] * Quits: sbuluf (uuyxkre@200.49.140.208) (Ping timeout)
# [13:51] * Joins: mw22 (chatzilla@84.41.169.151)
# [14:00] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
# [14:03] * Quits: heycam (cam@203.214.72.248) (Ping timeout)
# [14:05] * Joins: karl (karlcow@128.30.52.30)
# [14:19] * Joins: heycam (cam@203.214.72.248)
# [14:42] * Joins: schepers (schepers@69.134.24.226)
# [15:01] * Joins: Lachy (chatzilla@203.158.59.119)
# [15:28] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [15:33] * Joins: gavin (gavin@74.103.208.221)
# [15:40] * Quits: karl (karlcow@128.30.52.30) (Quit: Where dwelt Ymir, or wherein did he find sustenance?)
# [15:45] * Joins: tH (Rob@87.102.2.241)
# [16:02] * Parts: DanC (connolly@128.30.52.30) (Leaving)
# [16:09] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
# [16:33] * Joins: billmason (billmason@69.30.57.156)
# [16:35] * Joins: olivier (ot@128.30.52.30)
# [16:45] <Philip`> http://lists.w3.org/Archives/Public/public-html/2007Jun/0460.html - hmm, someone else is now labelling their own messages as spam
# [16:46] <anne> I think the message from jgraham was mared as spam or something
# [16:47] * Joins: myakura (myakura@58.88.37.26)
# [17:12] * Quits: olivier (ot@128.30.52.30) (Quit: Leaving)
# [17:16] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: Less talk, more pimp walk.)
# [17:32] * Quits: myakura (myakura@58.88.37.26) (Quit: Leaving...)
# [17:35] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [17:40] * Joins: gavin (gavin@74.103.208.221)
# [17:47] * Quits: edas (edaspet@88.191.34.123) (Ping timeout)
# [17:56] * Joins: Sander (svl@71.57.109.108)
# [18:16] * Quits: billmason (billmason@69.30.57.156) (Connection reset by peer)
# [18:17] * Joins: billmason (billmason@69.30.57.156)
# [18:33] <beowulf> argh
# [18:34] <Philip`> Is that an 'argh' of happiness?
# [18:37] <beowulf> no
# [18:37] <beowulf> you're not philip taylor are you?
# [18:38] <Philip`> I am, but I'm not the one on public-html :-)
# [18:38] <Lionheart> I'm not certain that either XHTML2 or HTML5 should be using the XHTML1 namespace. Both aren't backward compatible.
# [18:39] <beowulf> html5 should just be written in machine code or something, just so it's hard enough for the elite
# [18:39] <Philip`> Lionheart: What do you mean by "backward compatible" here?
# [18:41] <Lionheart> Okay, that's not an apt term to use, since they are after a fashion.
# [18:41] <Philip`> Aha, another opportunity to point out http://canvex.lazyilluminati.com/misc/sexp.html and how we could totally change HTML's syntax (to e.g. machine code, though obviously it should be a standard machine like the JVM) while still having new documents be compatible with old UAs, and you only need a single magic line at the top so it's less boilerplate than XHTML
# [18:42] <beowulf> just make you sure you don't copy and paste that line
# [18:42] <beowulf> *think* about it
# [18:42] <beowulf> *ponder* it
# [18:43] <beowulf> i like to dwell for an hour on the shebang line of a script before i write anything, i feel it brings a sense of wholeness
# [18:43] * beowulf stops self
# [18:43] <Lionheart> But I think you know what I meant.
# [18:45] <Philip`> Lionheart: I'm not quite sure what you meant, or in what ways things are non-compatible so that there would be any benefit in changing namespaces
# [18:46] <Lachy> Philip`: you'd still need a DOCTYPE to trigger standards mode for that "machine code"
# [18:47] <Philip`> Lachy: Not if you define the new standard to match quirks mode
# [18:47] <Lachy> ah, no!
# [18:48] * Joins: zcorpan (zcorpan@81.233.253.112)
# [18:48] <Lachy> quirks mode is too unpredictable and insane (even if we were to define it)
# [18:48] <Philip`> And standards mode is predictable and sane? ;-)
# [18:48] <Lachy> slightly more than quirks mode
# [18:49] <Lionheart> Phillip: For one thing, elements in that namespace now have different semantics than same-named elements in XHTML2 and HTML5
# [18:49] <Philip`> Lionheart: I guess the main problem with HTML5 using a different namespace is that it would completely break compatibility (of old documents with new UAs, and new documents with old UAs) because things would be referring to elements in the wrong namespace
# [18:50] <gsnedders> Lionheart: but the new semantics are more likely to be correct
# [18:50] <gsnedders> (in HTML 5)
# [18:50] <Philip`> and completely breaking compatibility has worse practical consequences than changing some elements' semantics which most people tend to ignore anyway
# [18:50] <Lachy> Lionheart: the semantic changes made in HTML5 really aren't backwards incompatible. It's the processing requirements that matter more anyway
# [18:53] <Philip`> Maybe it's the difference between 'semantics as described by the specification', and 'semantics as what you can deduce about content or about processing models when you see something claiming to follow the specification' - it would be good if those aligned better, and the second type can't change (since old content will never go away), so the first should change
# [18:54] <Lionheart> Phillip: How would old documents using an old namespace be broken in new UAs?
# [18:55] <Lionheart> Phillip: The other issue, new documents in a new namespace on old UAs, yes, that's evidently an issue
# [18:55] <gsnedders> Lionheart: would new UAs support the old namespace?
# [18:56] <Lionheart> gsnedders: Why wouldn't they? There's a Web full of documents in that old namespace
# [18:57] <gsnedders> Lionheart: of documents that actually use the namespace? there are _very_ few pages that actually use it.
# [18:57] <Philip`> Lionheart: I'm assuming the new UAs would process all documents according to the HTML5 rules (because most browser developers won't accept any other way), and that new HTML5 documents would still look like HTML (e.g. not with XML namespaces), and given a fragment of valid HTML you wouldn't be able to tell what namespace its elements should go in
# [18:57] <gsnedders> Lionheart: how many pages use an XML MIME type?
# [18:57] <Lionheart> Hixie might now, I've no stats on that.
# [18:57] <Philip`> Er, the first two bits were assumptions and the third bit was a probably result of those assumptions
# [18:57] <Lachy> it's possible to support both languages if they use different namespaces, but it's undesirable to introduce a new namespace for XHTML5 because it would break compat with legacy UAs
# [18:58] * Philip` has already been doing document.createElementNS('http://www.w3.org/1999/xhtml', 'canvas') and would prefer that not to break :-)
# [18:59] <Philip`> s/probably/probable/
# [19:00] <Lionheart> Phillip: HTML fragments without any DOCTYPE or namespace are going to be an issue regardless, no?
# [19:01] <Philip`> They will always be parsed as HTML5, and their elements all put in the http://www.w3.org/1999/xhtml namespace, as I understand it
# [19:02] <Philip`> I'm unsure what issues there are regarding that
# [19:03] <Lionheart> Actually, I imagine UAs will parse unversioned HTML as some HTML 3.2/4.0/5.0 melange that accepts <font> and <frame> and <article> as equally acceptable.
# [19:03] <gsnedders> Lionheart: HTML5 allows all those elements from a UA POV?
# [19:03] <gsnedders> *.
# [19:03] <zcorpan> Lionheart: html5 will define how all html elements are to be processed
# [19:03] <zcorpan> Lionheart: including <font> and <frame>
# [19:03] <gsnedders> Lionheart: the UA conformance requirements aren't the same as the document ones.
# [19:04] <zcorpan> Lionheart: and they will process documents the same, regardless of what the doctype says
# [19:05] <zcorpan> that seems to be extremely hard to grasp for many authors
# [19:06] <Lachy> just consider that HTML 2.0 documents are treated exactly the same as HTML 4.01 Transitional documents (when the doctype doesn't use the system ID). They use quirks mode for both.
# [19:07] <Philip`> With all documents parsed as HTML5, I don't know if it would matter that existing pages probably do <!DOCTYPE blah blah html4 whatever><html><script>document.body.appendChild(document.createElementNS('http://www.w3.org/1999/xhtml', 'p')).innerHTML='Hello world'</script> and may get confused if elements get mixed between the old XHTML and new HTML5 namespaces
# [19:07] <Lachy> it's the same for treating HTML5 and HTML 4.01 Strict documents the same
# [19:07] * Philip` wonders if many sites really do depend on what namespace something is in
# [19:08] <Lachy> HTML documents don't use namespaces, so it won't affect them. XHTML docs will break if the namespace no longer worked
# [19:09] <Philip`> There's still the problem that if HTML5 uses a separate namespace and defines entirely new elements, which just happen to have the same name as XHTML elements (except for the namespace), the old XHTML elements would still be as under-defined as they are in HTML4, and that under-definition is a large part of what HTML5 is trying to fix
# [19:13] <Philip`> Lachy: Scripts in HTML documents can use namespaces, and I guess people copy-and-paste XHTML scripts into their HTML pages and see that it still works and are happy
# [19:13] <Philip`> (but I am just guessing)
# [19:13] <Lachy> yeah. I think it's the XHTML2 WG that needs to use a different namespace, though as mjs has pointed out, it makes little practical difference until someone tries to implement both
# [19:15] <Lionheart> zcorpan: Yeah, I remember having a little trouble grasping that XHTML 1.0 Transitional and XHTML 1.0 Strict, which have different schemas, shared the same namespace, because it initially seemed to me like each namespace should be unambiguously described by one schema.
# [19:16] <Lionheart> As if http://www.w3.org/1999/xhtml should have some schema that combines all the elements from XHTML1.0 Transitional, XHTML1.1, and XHTML1.5, to describe all the documents within the namespace
# [19:16] <Philip`> Isn't the whole point of namespaces that you can have two separate technologies and then later you can decide to implement both of them and mix them together, without going "whoops, we can't do that because they conflict and nobody had planned ahead for us combining these two things"?
# [19:16] <gsnedders> Philip`: pretty much
# [19:17] <Philip`> in which case it's totally against the point of namespaces, to put two things in the same namespace just because people don't currently think they'll ever be combined
# [19:17] <Lachy> Philip`: yeah, but it's highly unlikely that anyone is going to implement XHTML2, let alone HTML5 in the same app
# [19:18] <Lachy> the point I think mjs is making is that they can do whatever they like, since it won't matter in the long run anyway
# [19:19] <Philip`> Maybe people will like XHTML2 for writing documents (where direct compatibility is irrelevant) and XSLTing it into browser-compatible HTML on the server side
# [19:19] <Philip`> which sounds less unlikely than browsers supporting XHTML2
# [19:19] <Lachy> that's a possibility
# [19:19] <Lionheart> Phillip: Well, that's why it gets muddled when you don't know if http://www.w3.org/1999/xhtml:abbr is xhtml:abbr or xhtml2:abbr/html5:abbr
# [19:20] <gsnedders> Lionheart: what is different between xhtml:abbr and html5:abbr? both are defined as abbreviations.
# [19:20] <zcorpan> Lachy: indeed, we don't need to worry too much about what the xhtml2 wg thinks or are up to
# [19:21] <gsnedders> Lionheart: the fact that in XHTML there is another element for a subset of that doesn't change what UAs need to implement it as
# [19:22] <Philip`> Lionheart: You would need to know the difference between e.g. xhtml2:img and xhtml:img, so you could move the element's text content into its alt attribute. I'm not aware of any elements where you would want to do the same for html5: compared to xhtml:
# [19:22] <Lionheart> Phillip: Thank you, that is a much better example
# [19:25] <Lionheart> That's a fine feature of XHTML2, by the way, and really how <img> ought to have been specified in the first place.
# [19:27] <Philip`> It's also an example of why XHTML2 is incompatible with XHTML1, while HTML5 is compatible
# [19:30] <Lionheart> Would it cause problems if HTML5 supported both <img>replacement text</img> and <img alt="replacement text" />?
# [19:31] <zcorpan> Lionheart: non-serializability in text/html
# [19:31] <Lionheart> Or <meta>content text</meta> and <meta content="content text" />
# [19:31] <zcorpan> same
# [19:32] <Philip`> Lionheart: It would make HTML5 documents incompatible with HTML4 UAs (which would display both the image and the text), and one of the goals of HTML5 is that HTML5 documents should work as correctly as possible in old UAs
# [19:33] <Philip`> Although you could say: It would give authors the choice of writing HTML5 documents that are incompatible with HTML4 UAs
# [19:33] <Lionheart> Pity, replacement text with inline markup would be rather preferable. Though I suppose there's still <object> for that.
# [19:34] <Philip`> I guess the problem there is that for the next ten years, you'd have to teach people "You can write <img>alt text</img>, except actually you mustn't do that and you have to write <img alt="alt text"> instead else a significant proportion of your site's users will see nasty breakage"
# [19:35] <Philip`> (which is a lot of wasted effort)
# [19:35] <Lionheart> Sort of like teaching XHTML 1.0 writers to put a space before the slash in " />" if they want their documents to be backward compatible with HTML UAs.
# [19:38] <Lionheart> Yeah, that works for me, have a "Compatibility Guidelines" appendix like the XHTML spec.
# [19:39] <Lionheart> Which says things like <br></br> and <p /> are allowed, but will break in existing UAs.
# [19:40] <Philip`> Once people implement HTML5 and CSS, you could do <style>graphic { content: attr(src, url) }</style> <graphic src="image.png">alt text</graphic> and then it could get added in HTML6 later :-)
# [19:40] <gsnedders> Lionheart: they'd never work in the HTML serialisation anyway
# [19:41] <gsnedders> (unless browsers added a different parsing mode for certain versions of HTML, which they won't)
# [19:41] <Philip`> What UAs actually needed the space before the slash, anyway?
# [19:41] <gsnedders> NN4, IIRC
# [19:41] <Lionheart> I think we may already be past needing the space
# [19:42] <gsnedders> who here still tests in NN4?
# [19:42] <Philip`> Hmph, the Live DOM Viewer doesn't work in NN4
# [19:43] <zcorpan> Philip`: would be better idea to use <object data="image.png">alt text</object>
# [19:44] <Philip`> Aha, <br/> does indeed not work in NN4
# [19:44] <Philip`> (I assume it's parsed as an element called "br/")
# [19:45] <zcorpan> Philip`: yeah
# [19:45] <Philip`> IE3 does the same
# [19:46] <Lionheart> zcorp: At the risk of making IE crazy.
# [19:46] <zcorpan> Lionheart: compared to <graphic> with graphic { content: ?
# [19:46] <zcorpan> i think ie7 actually supports images with object
# [19:47] <Philip`> That would just fall back to the alt text, which isn't too bad
# [19:47] <Philip`> (<graphic>, that is)
# [19:47] <zcorpan> object does the same
# [19:48] <Philip`> At least in IE6, it loads the image into a zero-sized frame
# [19:48] <Philip`> which is not as nice
# [19:49] <Lionheart> I rather think that <img> ought to do the same.
# [19:50] <anne> can't change <img>
# [19:50] <anne> or <image>
# [19:50] <Philip`> Hmm, in IE7 it doesn't load the image and shows the fallback text instead
# [19:50] <Lionheart> nor <image>? Is that defined by something?
# [19:50] * Philip` wonders if that's a real difference, or if he's just doing something silly
# [19:51] <Philip`> Oh, I'm doing something silly - it draws the image when I run IE6 in Wine, but not in IE6/IE7 in Windows
# [19:52] <Philip`> Lionheart: <image> is parsed into an "img" element
# [19:53] <Philip`> (and </image> is parsed into a "/image" element in IE, but ignored by everyone else)
# [19:53] <Philip`> Uh
# [19:53] <Philip`> (and </image> is parsed into a "/img" element in IE, but ignored by everyone else)
# [19:54] <Lionheart> I've never heard of this. This is an IE proprietary thing?
# [19:54] <Philip`> (This is why everyone loves HTML so much)
# [19:56] <Lionheart> Hm, putting <img src="image">te</img> into Hixie's Live DOM viewer is informative. Yeah, Firefox & Opera both render image and contents both.
# [19:56] <Philip`> It's not IE - it was there at least as far back as NN1.22
# [19:57] <Lionheart> One of those undocumented features like <tc> ?
# [19:57] <anne> Lionheart, <image> is HTML parsing legacy
# [19:57] <Philip`> They must have had good intentions when choosing to correct authors' errors
# [19:57] <anne> and appears relatively frequent
# [19:58] <Philip`> <tc>?
# [19:58] * Philip` hasn't heard of that...
# [19:58] <Lionheart> Instead of <tr>, you make a table with <tc>s
# [19:59] <Philip`> I see "TC elements" in http://www.w3.org/MarkUp/html3/tables.html but that looks like a typo
# [19:59] <anne> I don't think <tc> is actually implemented where <image> is
# [20:00] <Philip`> IE6 seems to handle <tc> the same as any other unrecognised element, as far as I can tell
# [20:00] <Lionheart> I seem to remember it working in NN4 ,but I'll have to check
# [20:01] <anne> NN4 is not relevant for HTML5 though
# [20:01] * Joins: olivier (ot@128.30.52.30)
# [20:02] <Lionheart> Oh, I'm not proposing we add <tc>, I'm just likening <image> to it
# [20:02] <Philip`> I just tried it in NN4 and it doesn't seem to make cells
# [20:02] <zcorpan> didn't some version of nn4 not render tables if they contained <tbody> tags? :)
# [20:02] * Quits: tH (Rob@87.102.2.241) (Ping timeout)
# [20:03] <Lionheart> Huh, I wonder what it was that I found it working in then.
# [20:03] * Joins: tH (Rob@87.102.2.241)
# [20:04] <Philip`> Tried it in NN1, 2, 3 and 4, and IE3, and <tc> doesn't seem to do anything in any of them :-(
# [20:04] <Philip`> or at least nothing different to what <zz> does
# [20:05] <Philip`> zcorpan: Looks like NN2, 3, 4 don't render tables if they've got a <tbody>
# [20:06] <Philip`> NN1 and IE3 do
# [20:06] <zcorpan> amusing that you have all those browsers :)
# [20:07] <Philip`> I just got a few from http://browsers.evolt.org/ a while ago :-)
# [20:08] <Philip`> I think that was when I was trying to work out why people do <script></style> to hide content from 'old' browsers
# [20:08] <Lionheart> It was NN3 and IE3 I found <TC> working in
# [20:09] <Lionheart> I remember now, that when I went to work as a webmaster in 1997, I was very surprised to find my predecessor had used <TC> to mark up tables throughout the existing site
# [20:10] <Lionheart> And that it was working
# [20:10] <Philip`> (<style> was supported from NN4, <script> from NN2)
# [20:10] <Lionheart> I recoded them all to <TR>, of course.
# [20:11] <Philip`> Were they tables with multiple columns?
# [20:12] <Lionheart> They were used for layout, so perhaps not.
# [20:12] <Philip`> <table><tr><tc>Site header <tr><tc>Body <tr><tc>Footer <tr><tc><a href="netscape.com"><img src=nn.gif alt="Get Netscape Today!"></a></table> would do what you'd except
# [20:13] <Philip`> (just by the <tc> being ignored and the <td> being implied)
# [20:14] <Lionheart> Hm, now I wish I had a specimen of those original pages
# [20:15] <Philip`> Hmm, NN3 just eats the whole table if you do <table><tc>...
# [20:16] <zcorpan> same as the <tbody> case?
# [20:16] <zcorpan> <table><foo>
# [20:16] <Philip`> Yep, and the same as the <unrecognised> case
# [20:16] <Philip`> It accepts <table><any-recognised-element-name>
# [20:16] <Philip`> but nothing else
# [20:16] <zcorpan> ok
# [20:16] <zcorpan> that's what i thought
# [20:17] <Philip`> It also accepts <table></anything-at-all>
# [20:17] <anne> you'd think people have better stuff to do than reverse engineering NN3
# [20:17] <Philip`> :-P
# [20:18] <Philip`> It's a reminder of kinder days, when you just wrote stuff for Netscape and FTPed it up and that was it - nowadays it's all far too complex :-(
# [20:18] <Lionheart> Hm. I still have a 1997 e-mail I wrote describing my <tc> incredulity, but I suppose I'll have to d/l NN3 and see if I can reproduce it.
# [20:19] <Philip`> http://browsers.evolt.org/?navigator/32bit/3.04/gold - works fine for me in Windows 2000
# [20:19] <anne> hmm, my e-mail English is quite terrible
# [20:19] <anne> oh well, people will get it
# [20:20] <Philip`> (I don't think I managed to get any Netscape working in Wine, unfortunately)
# [20:20] <Lionheart> Well, yeah, I really do have better things to do, but this gives me something to procrastinate with
# [20:21] * Philip` knows the feeling
# [20:24] * Quits: gavin (gavin@74.103.208.221) (Ping timeout)
# [20:27] <Lionheart> Okay, I think perhaps I understand
# [20:28] * Quits: olivier (ot@128.30.52.30) (Quit: Leaving)
# [20:28] <Lionheart> Though <table><tc><td>1</td></tc><tc><td>2</td></tc><tc><td>3</td></tc></table> attempts to define three columns column by column...
# [20:29] <Lionheart> It ends up defining a single row to the same effect
# [20:30] <Philip`> When I look at that in NN3, the whole table's content disappears
# [20:30] * Quits: tH (Rob@87.102.2.241) (Ping timeout)
# [20:30] <Philip`> (NN4/IE3 are fine, though)
# [20:31] <Philip`> That seems kind of a logical thing to write - you start by copying someone's <table><tr>..., but you want columns instead of rows, so <table><tc>... is a perfectly obvious thing to try, and it does in fact work when you test it, so clearly that's correct and you should use it all over your site
# [20:32] <Philip`> Iterative testing is a much better way of learning HTML than looking in a book
# [20:33] <anne> hmm, we're supposed to publish something this month
# [20:33] <anne> somehow I doubt that'll work out
# [20:35] <Lionheart> Okay, I'm reasonably satisfied this mystery is solved, 10 years later. :) Now I suppose I shall have to go back to work.
# [20:35] <beowulf> i can't for the life of me see how shortening the doctype is designing "a markup language for those too lazy to think "
# [20:35] <Lionheart> ...after I eat lunch. :)
# [20:35] <Philip`> Lionheart: If you've got the original designer's email address, you'll have to send a reply and see if he's realised his errors yet ;-)
# [20:36] <anne> HTML is for those too lazy to think
# [20:36] <anne> why people want to require other people to think is beyond me... hasn't history shown that that doesn't work?
# [20:36] <beowulf> :)
# [20:37] <Lionheart> Phillip: No idea, but knowing my boss at the time, it was probably his nephew or some other relation. Nepotism was rampant there.
# [20:37] <beowulf> i may use that as a reply
# [20:37] <Philip`> HTML is (in some people's views) for everybody to use; and half of everybody is below average intelligence or above average laziness, so their needs have to be taken into account
# [20:38] <beowulf> i want it taken into account that I can't think about the boilerplate everytime
# [20:38] <anne> it's not clear to me how we improve the world by requiring people to be very precise
# [20:38] <beowulf> i want my laziness acknowledged!
# [20:39] <beowulf> anne: nor me
# [20:39] <beowulf> making it harder won't improve the quality of html
# [20:40] <Philip`> Perl is nice in acknowledging laziness as a virtue
# [20:41] <beowulf> easy things should be easy, hard things should be hard
# [20:41] <beowulf> the boilerplate should be easy
# [20:41] <beowulf> <html>  </html>
# [20:42] <Philip`> The <html> and </html> should be optional too
# [20:42] <beowulf> yes
# [20:42] <anne> the boilerplate is 9 characters which come mostly by default: text/html
# [20:42] <Philip`> (and they are, so that's good)
# [20:42] <Philip`>
# [20:42] <Philip`> Everything else is just redundant
# [20:43] <Philip`> (or, it should be)
# [20:43] <beowulf> i agree
# [20:43] * Joins: olivier (ot@128.30.52.30)
# [20:44] <beowulf> and none of this is in anyway related to standard of markup, accessibility or anything really
# [20:44] <Lionheart> anne: Precision doesn't improve the world. If Dubya said, "The British government has learned that Saddam Hussein recently sought exactly 270 tons of uranium from Niger", we'd still be just as screwed.
# [20:46] <anne> I'd love for some people to just provide clear examples of what they would like to see changed instead of talking in vague terms
# [20:46] <beowulf> vague, mildly insulting, terms
# [20:47] <anne> As long as they talk in vague terms it's hard to acknowledge their comment or prove them wrong
# [20:47] <anne> I think most people would love to fix the specification, but as long as we don't know what's broken it's hard to fix it
# [20:47] <Philip`> Maybe you could make an analogy with archery - precision is what you should aim for, but there's a good reason why the target is three feet wide
# [20:50] <anne> :)
# [20:50] <gsnedders> I'd miss anyway.
# [20:50] <beowulf> I'd try until i hit the target, then I'd make a machine to repeat that for me
# [20:51] <beowulf> i'd call the machine "Boiler Plate"
# [20:51] <gsnedders> I'd just get a sheet of paper and work out what is needed and make the machine from that
# [20:51] <anne> If you see that even Jacques Distler and Sam Ruby have trouble getting it right I wonder how other people can claim it should be more strict...
# [20:51] <gsnedders> It should be more strict.
# [20:51] <gsnedders> There. Done.
# [20:52] <beowulf> gsnedders: nothing beats real experience, my machine would be better :)
# [20:52] <Philip`> (It's really annoying when arrows bury themselves invisibly in the grass twenty metres away - much easier indoors when you've got a whole wall to aim at...)
# [20:53] * Philip` wonders how profitable it would be to sneak into fields after less-competent archers have been practising, with a metal detector, then wipe the mud off everything you find and sell it
# [20:54] <beowulf> arrows are about 5p each i think?
# [20:54] <gsnedders> Philip`: doing anything over the summer? :P
# [20:54] <Philip`> beowulf: I can't remember any details now, but decent arrows are far more than that
# [20:55] <Philip`> http://www.hoddywell.com.au/ARROWSCOMPLETE-Aluminium.html has some between $2.60 and $13.60 each
# [20:55] <Philip`> Uh, I guess those are funny dollars rather than normal ones
# [21:01] <anne> hmm, lots of feedback on my html4-differences doc is not in english
# [21:20] * Quits: zcorpan (zcorpan@81.233.253.112) (Ping timeout)
# [21:28] * Quits: bogi (bogi@153.19.120.250) (Ping timeout)
# [21:29] * Joins: bogi (bogi@153.19.120.250)
# [21:29] * Joins: mjs (mjs@64.81.48.145)
# [21:47] * Quits: olivier (ot@128.30.52.30) (Quit: Leaving)
# [22:19] * Joins: dbaron (dbaron@63.245.220.242)
# [22:24] * Joins: hasather (hasather@80.203.71.22)
# [22:40] * Quits: mjs (mjs@64.81.48.145) (Quit: mjs)
# [22:44] * Quits: ROBOd (robod@86.34.246.154) (Quit: http://www.robodesign.ro )
# [23:21] * Parts: hasather (hasather@80.203.71.22)
# [23:30] * Joins: myakura (myakura@58.88.37.26)
# [23:32] * Joins: gavin (gavin@74.103.208.221)
# [23:37] * Joins: mjs (mjs@17.255.96.159)
# [23:48] <Hixie> hm, danc isn't here
# [23:48] <Hixie> i didn't realise that the naming debate was what was blocking our publication
# [23:49] <mjs> it is?
# [23:49] <Hixie> http://lists.w3.org/Archives/Member/w3c-html-cg/2007AprJun/0203.html
# [23:49] <Hixie> apparently
# [23:50] <Hixie> i suggest that we accept, with one condition, the xhtml2 wg's proposal for us to not use the term "xhtml"
# [23:50] <Hixie> that condition being that they in turn stop using the four-letter acronym "html" anywhere in the names of their specifications
# Session Close: Fri Jun 22 00:00:00 2007

The end :)