| 00:37 | <Hixie> | what's the rule that hsivonen refers to regularly regarding how a technology fragments into the same number of independent components as there are working groups designing it? |
| 00:38 | <Hixie> | Conway's Law, nevermind |
| 00:38 | <Dashiva> | Either that or n^2 interfaces :) |
| 01:02 | <Dashiva> | How does the "new XML specs should not define named entities" jive with MathML? |
| 12:34 | <erlehmann> | haha, randall munroe made i GIANT chart |
| 12:36 | <jgraham> | i? |
| 12:37 | <gsnedders|work> | UnexpectedPronounException |
| 12:39 | <jgraham> | erlehmann: I assume you mean http://xkcd.com/657/large/ which is rather cool |
| 12:39 | <erlehmann> | jgraham, i and a are right next to each other |
| 12:40 | <erlehmann> | jgraham, now that he has proven he'll do huge pointless charts, you can ask him ;) |
| 12:40 | <erlehmann> | although i think we might just relabel the "primer" thingy :D |
| 12:44 | <jgraham> | We should wait until CR though |
| 12:44 | <jgraham> | Or at least until there are a couple of shipping implementations |
| 12:44 | <jgraham> | Otherwise the chart will be quickly wrong |
| 12:45 | gsnedders|work | hopes the parser spec doesn't change too much from hereon to avoid the problems discussed last night |
| 12:46 | hsivonen | wishes Opera didn't implement stuff like XML 1.1, xml:id, entities for RDFa doctype and entities for innerHTML setter in XML |
| 12:47 | <jgraham> | hsivonen: It would be nice if we could just make all XML applications recognise all HTML5 entities |
| 12:47 | <hsivonen> | instability in interop in this area consumes resources while IE just ignores it all and doesn't need to put in similar effort |
| 12:48 | <hsivonen> | jgraham: if we rock the boat like that, I think we should go full XML5 |
| 12:48 | <hsivonen> | instead of rocking the boat several times |
| 12:51 | gsnedders|work | wonders how come we support entities in XHTML+RDFa |
| 12:51 | <gsnedders|work> | Do we just support entities in all application/xhtml+xml? |
| 12:53 | <hsivonen> | gsnedders|work: you don't |
| 12:54 | <jgraham> | hsivonen: A one-step XML5 would have the lowest cost / best benefits if people would go with it. However pragmatically it requires even more buy-in and so has a bigger risk of total failure |
| 12:54 | <gsnedders|work> | hsivonen: We currently just use our HTML parser for innerHTML regardless of whether we're in XML, FYI |
| 12:54 | <gsnedders|work> | hsivonen: That's why we support entities there |
| 12:54 | <hsivonen> | gsnedders|work: oh. that's a sensible reason |
| 12:55 | <gsnedders|work> | hsivonen: I guess we won't once we move to using XML parser for it |
| 12:55 | <hsivonen> | gsnedders|work: would be useful to have these data points in the thread on the list |
| 12:57 | gsnedders|work | emails about innerHTML |
| 12:57 | <Philip`> | data:application/xhtml+xml,<!DOCTYPE html SYSTEM "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd"><html><b>test</b></html>; |
| 12:57 | <Philip`> | Opera puts HTML styles on those elements |
| 12:58 | <Philip`> | and puts them in the XHTML namespace |
| 12:58 | <Philip`> | but doesn't if you remove the doctype |
| 12:58 | <gsnedders|work> | Philip`: You filed a bug on that? |
| 12:58 | <gsnedders|work> | Oh, no, not you. But we have a bug on that. |
| 12:59 | <gsnedders|work> | It's because we act too much like a validating parser, obviously. Yay. |
| 13:01 | <gsnedders|work> | We imply namespaces for XHTML 1.0 (all three DTDs), XHTML 1.1, XHTML-MP 1.0 (as application/xhtml+xml) and WML 2 (application/xml+xml), but not for XHTML Basic 1.0 and 1.1 and XHTML-MP 1.2 |
| 13:01 | <zcorpan_> | do we support entities in basic, 1.1 and mp? |
| 13:02 | <gsnedders|work> | Dunno. I'm just reading the bug report. :) |
| 13:03 | <Philip`> | I hypothesise that you support entities/etc if the public id is one of: |
| 13:03 | <gsnedders|work> | zcorpan_: Comment by, uh, zcorpan_imples we do :) |
| 13:03 | <Philip`> | -//W3C//DTD XHTML 1.0 Strict//EN -//W3C//DTD XHTML 1.0 Transitional//EN -//W3C//DTD XHTML 1.0 Frameset//EN -//W3C//DTD XHTML 1.1//EN -//W3C//DTD SVG 1.1//EN -//W3C//DTD SVG 1.1 Basic//EN -//W3C//DTD SVG 1.1 Tiny//EN -//W3C//DTD SVG 1.0//EN |
| 13:03 | <Philip`> | or the system id is one of: |
| 13:03 | <Philip`> | http://www.w3.org/TR/xhtml11/DTD/xhtml11-flat.dtd http://www.wapforum.org/DTD/xhtml-mobile10-flat.dtd http://www.wapforum.org/DTD/wml_1_1.dtd http://www.wapforum.org/DTD/wml12.dtd http://www.wapforum.org/DTD/wml13.dtd http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-flat.dtd http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-basic.dtd http://www.w3.org/Graphics/SVG/1.1/DTD/svg11-tiny.dtd http://www.w3.org/TR/SVG10/DTD/s |
| 13:04 | Philip` | curses truncation |
| 13:04 | <zcorpan_> | gsnedders|work: ok |
| 13:04 | <gsnedders|work> | Philip`: Where are you looking? |
| 13:04 | <Philip`> | http://www.w3.org/TR/SVG10/DTD/svg10.dtd http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd http://www.wapforum.org/DTD/wml20.dtd http://www.wapforum.org/DTD/xhtml-mobile10.dtd |
| 13:04 | <Philip`> | though I could be wrong, but some quick testing seems to agree with that list |
| 13:04 | <Philip`> | gsnedders|work: "strings opera" |
| 13:05 | <gsnedders|work> | "strings opera"? |
| 13:05 | <Philip`> | http://unixhelp.ed.ac.uk/CGI/man-cgi?strings |
| 13:05 | <gsnedders|work> | Ah |
| 13:06 | zcorpan_ | tries subscribing to commit-watchers but doesn't get a confirmation email |
| 13:07 | <Philip`> | Hmm, seemingly not the SVG ones |
| 13:07 | <zcorpan_> | oh wait |
| 13:07 | <Philip`> | nor the WML 1.* ones |
| 13:07 | <zcorpan_> | i did get confirmation email |
| 13:07 | <Philip`> | (I guess those trigger different special behaviours) |
| 13:08 | <erlehmann> | i'm trying to get german parliamentary video channel techs to consider moving from embedded flash movies to <video> with flash fallback. anyone here has tips for doing this? |
| 13:09 | <gsnedders|work> | Philip`: We imply for SVG, and that's unlikely to change (due to content deployed when Adobe SVG Viewer didn't require it) |
| 13:09 | <zcorpan_> | erlehmann: tips for convincing them or tips for implementing it? |
| 13:11 | <erlehmann> | zcorpan_, its about convincing. i already met the woman in charge at a "how to make government data more open" session, she referred me to the techs. |
| 13:12 | <Philip`> | erlehmann: Why would it be worthwhile for them to switch? |
| 13:17 | <gsnedders|work> | "RAW DATA NOW"? |
| 13:17 | <erlehmann> | gsnedders|work, something like that. |
| 13:19 | <gsnedders|work> | Then remember the RDF graphs too. |
| 13:20 | <erlehmann> | Philip`, the government has its own data in some encumbered formats. we get laws as untagged PDF, videos as flash. in the end, it's all about accessability. |
| 13:20 | <erlehmann> | but i don't really know how to deal with government types |
| 13:21 | gsnedders|work | wouldn't call PDF encumbered, as there are plenty of readers and there is an open standard |
| 13:21 | <gsnedders|work> | erlehmann: I guess the first thing you need to work out is whether it is the government or whether it is civil servants making the decisions |
| 13:22 | <erlehmann> | gsnedders|work, the decision is one of bureaucracy |
| 13:22 | <gsnedders|work> | erlehmann: But who made the decision to use the formats currently used? |
| 13:22 | <gsnedders|work> | (This does have a real affect about how to go about getting things done) |
| 13:23 | <erlehmann> | interesting question. |
| 13:24 | <gsnedders|work> | You can _try_ and get civil servants (or rather, more generally, government officials) to change things, but they will tend to take the solution that is the least work for them (and may have their own agenda). |
| 13:24 | <Philip`> | erlehmann: If the idea is just about unencumbered formats, it sounds like they could just provide a download link to an .ogv or .mp4 file, and there's no need to mess around with using <video> |
| 13:24 | <jgraham> | gsnedders|work: AIUI the problem with pdf is not that it is encumbered per-se it is that extracting information is typically very difficult |
| 13:25 | <gsnedders|work> | If that fails, you have to push those who the non-elected officials are meant to obey |
| 13:25 | <gsnedders|work> | It's all one big ball-game basically |
| 13:25 | <gsnedders|work> | jgraham: Indeed, but it's certainly got quite a lot better |
| 13:26 | <erlehmann> | Philip`, valid response. but if you have the encoded files, <video> is simple |
| 13:26 | <erlehmann> | and you can always have flash fallback |
| 13:27 | <Philip`> | I guess the simpleness depends on how well the fallback works in practice, and how well <video> works in practice, and how easy it is to copy-and-paste the right markup for it |
| 13:27 | <jgraham> | gsnedders|work: "better" in the sense that it is now theoretically possible to produce PDF files in which getting the text out isn't a major undertaking? |
| 13:27 | <jgraham> | It doesn't really count as better unless people actually do it |
| 13:27 | <gsnedders|work> | jgraham: I think it is easier once you have a parsed copy of the file now |
| 13:28 | Philip` | wishes he could copy-and-paste text from PDFs without getting "fi" ligatures as single characters |
| 13:28 | <jgraham> | and having a format where it is easier to do the wrong thing than the right thing means that people will typicalloy do the wrong thing even if there are laws and stuff saying you should do the right thing |
| 13:29 | <erlehmann> | gsnedders|work, as far as i know the head of that thingy came to the open government session out of personal interest. she readily admitted that some systems are barely usable and was looking for input from local hackers and parliamentary informaticians. |
| 13:29 | <jgraham> | gsnedders|work: eaiser than what? Also what do you mean "parsed copy of the file"? |
| 13:29 | <erlehmann> | Philip`, technical things can be solved easily. its about getting the process into motion |
| 13:29 | <gsnedders|work> | jgraham: You convert the PDF file to an internal representation before drawing. |
| 13:30 | <gsnedders|work> | jgraham: Easier than it used to be |
| 13:30 | <gsnedders|work> | (I could be wrong, I've not looked that much at text extraction) |
| 13:30 | <Philip`> | gsnedders|work: Easier than it used to be for exactly the same input that was used before, or easier for new inputs that use new features? |
| 13:30 | <gsnedders|work> | Philip`: The latter, I think. |
| 13:32 | <jgraham> | gsnedders|work: That is exactly the problem. If these documents are being produced by some person using word or whatever and "export to pdf" does the wrong thing regardless of how they wrote the document, you have already lost because all the regulation in the world won't stop untagged documents getting produced |
| 13:32 | <jgraham> | erm |
| 13:33 | <jgraham> | by "regardless" I think I mean something quite unlike regardless |
| 13:33 | <erlehmann> | jgraham, our administration has secretaries with word using the print function. absolutely hatable, if there are two columns |
| 13:34 | <jgraham> | Why I'm trying to say is that it has to work for any authoring style, not just for people who use e.g. the styles and formatting feature |
| 13:34 | <jgraham> | (which is approximately no-one) |
| 13:34 | <Philip`> | jgraham: "irregardless"? |
| 13:35 | <jgraham> | iirespective |
| 13:35 | <jgraham> | but with the right number of each letter |
| 13:35 | <jgraham> | in the right order |
| 13:37 | <jgraham> | No that's still not right |
| 13:37 | <jgraham> | Oh dear |
| 13:38 | <gsnedders|work> | erlehmann: As I understand it, they're meant to be moving away from Word, at least :P |
| 13:38 | <Philip`> | How can it not be right when you're allowing any combination of any letters? |
| 13:39 | <Philip`> | unless it's something that can only be expressed through gestures |
| 13:39 | <erlehmann> | gsnedders|work, i can't see that happen. |
| 13:40 | <gsnedders|work> | erlehmann: Several ministries have moved away from Word already, to my knowledge |
| 13:40 | <Philip`> | What have they moved to? |
| 13:40 | <gsnedders|work> | OpenOffice :P |
| 13:40 | <Philip`> | Oh, not Emacs? :-( |
| 13:40 | <erlehmann> | its nanowrimo |
| 13:40 | <erlehmann> | use nano(1) |
| 13:40 | <erlehmann> | the ministry of the interior is big with free software, but they are reviled almost universally in tech circles |
| 13:41 | <erlehmann> | also some cities have switched |
| 13:41 | Philip` | likes nano, it's the only text editor he can understand how to use |
| 13:41 | <erlehmann> | but i, as a consumer of data, am more interested in free formats |
| 13:41 | <workmad3> | Philip`: I feel like that too :) |
| 13:55 | <jgraham> | Understanding your text editor is overrated. The only real criterion is that pressing tab should produce a reasonable indentation much of the time. And not, I repeat not, try to insert a tab character. Ever. |
| 13:56 | <zcorpan_> | hsivonen: afaict opera doesn't support the xhtml+rdfa doctype |
| 13:58 | <hsivonen> | zcorpan_: ok. I guess I should test every allegation myself without trusting anything that has been said in the thread. sorry. |
| 15:17 | <Philip`> | "once we start to add predefined entities, where do we stop?" - we stop once we've reached XML5 |
| 15:18 | <hsivonen> | how do we stop the MathML folks from minting more and more entities? |
| 15:19 | <jgraham> | More to the point who thought it was a good idea to let validators accept a superset of wht UAs accepted in the first place? |
| 15:19 | <hsivonen> | jgraham: you'll find that V.nu accepts a subset by default |
| 15:21 | <hsivonen> | doing a hg clone where the target is an AFP share is insanely slow |
| 15:22 | <Dashiva> | I don't understand this entity stuff at all. What is the rationale for allowing documents to change just depending on whether the parser validates or not? |
| 15:22 | <Philip`> | I don't like how Sourceforge's current design consists almost entirely of colours with >50% brightness, which look indistinguishable from white when I view my monitor from a slight angle |
| 15:22 | <jgraham> | hsivonen: I'm far from surprised that you are doing something sane |
| 15:22 | <Dashiva> | It seems like it's just asking for interop failures |
| 15:23 | <jgraham> | Dashiva: Yes |
| 15:23 | <Dashiva> | At least with draconiainness there was a _perceived_ benefit |
| 15:24 | <Philip`> | Dashiva: Not at all - it ensures your content has maximum interoperability with the specification |
| 15:24 | <hsivonen> | Dashiva: optional features in specs are the collection of unresolved disputes that the WG threw over the fence for the entire world to deal with |
| 15:25 | <hsivonen> | next time I want to back up a hg repo to AFP I'll use tar or something |
| 15:25 | <Dashiva> | Related, what is the term for a document using invalid entities? It's still well-formed, I assume |
| 15:26 | <hsivonen> | Dashiva: well-formedness depends on whether the parser processed external entities |
| 15:27 | <Dashiva> | So a validating parser can make a non-well-formed document into a well-formed one |
| 15:27 | <Dashiva> | *? |
| 15:29 | <gsnedders|work> | What WFC depends upon whether it is validating or not? |
| 15:29 | <gsnedders|work> | For entities, undeclared entities are a well-formedness error iff standalone='yes' |
| 15:29 | <gsnedders|work> | (when dealing with a non-validating processor) |
| 15:31 | <Dashiva> | I think I was asking if <root>&bogus;</root> is well-formed or not |
| 15:32 | <gsnedders|work> | It is well-formed for a non-validating parser, not for a validating one |
| 15:33 | <Philip`> | The definition of well-formedness appears to be just based on the textual object, not the parser |
| 15:33 | <gsnedders|work> | Indeed |
| 15:33 | <gsnedders|work> | So that is not well-formed, but only a validating parser will notice it is not |
| 15:34 | <Dashiva> | So that's not even an XML document |
| 15:34 | <gsnedders|work> | Maybe a non-validating one would because it hasn't got a DTD |
| 15:35 | <gsnedders|work> | That is not well-formed, so all UAs will throw a fatal error |
| 15:35 | <Philip`> | gsnedders|work: Why is a well-formedness error? |
| 15:35 | <gsnedders|work> | Philip`: Well-formedness constraint: Entity Declared |
| 15:35 | <Philip`> | I only see relevant things mentioned as validity constraints |
| 15:36 | <Philip`> | Oh, okay |
| 15:37 | <Dashiva> | But if there was a DTD with an external entity declaration... then it would become valid _only_ in validating parsers? |
| 15:37 | <gsnedders|work> | No, the document would be valid. |
| 15:37 | <Dashiva> | Well-formed, I mean |
| 15:38 | <gsnedders|work> | That document is always not well-formed |
| 15:38 | <gsnedders|work> | That's wrong |
| 15:38 | <gsnedders|work> | Yes, it would then always be valid if the external DTD defined &bogus; IIRC |
| 15:39 | <Dashiva> | But it would look non-well-formed to a non-validating parser |
| 15:39 | <Dashiva> | Should it not then refuse to process it? |
| 15:39 | <gsnedders|work> | No, it would look well-formed to a non-validating parser |
| 15:39 | <gsnedders|work> | It just has an unknown entity |
| 15:39 | <Dashiva> | But you said that was a well-formedness requirement earlier? |
| 15:39 | <gsnedders|work> | It isn't. |
| 15:40 | <gsnedders|work> | I was wrong |
| 15:40 | <Dashiva> | Oh, okay |
| 15:41 | <gsnedders|work> | "Certain well-formedness errors, specifically those that require reading external entities, may fail to be detected by a non-validating processor. Examples include the constraints entitled Entity Declared, Parsed Entity, and No Recursion, as well as some of the cases described as forbidden in 4.4 XML Processor Treatment of Entities and References." |
| 15:41 | <rollandburn> | i wonder if someone could give some advice.... I am tasked with creating an overview of our company's accessibility policies and I have done a bit of research on WAI and WCAG2.0 which is apparently the standard to which one should hold their site against... the problem is that I am coming across a lot of conflicting info about the real-world usefulness or attainability of WCAG2.0 and am not... |
| 15:41 | <rollandburn> | ...sure if there is another standard or body of guidelines that would be better suited to modern developers |
| 15:42 | <jgraham> | gsnedders|work: So what are you expected to do in those cases? |
| 15:42 | <gsnedders|work> | jgraham: Your choice. |
| 15:42 | <jgraham> | Seriously? |
| 15:42 | <gsnedders|work> | jgraham: Yeah. |
| 15:42 | <jgraham> | sigh |
| 15:42 | <Dashiva> | XML is great |
| 15:43 | <Dashiva> | Are there any well-known apps or similar that use validating parsers? |
| 15:43 | <gsnedders|work> | Dashiva: http://validator.w3.org is well-known, I guess. |
| 15:43 | <Philip`> | Dashiva: All the ones that hammer w3.org for DTDs |
| 15:43 | <rollandburn> | i guess if I nutshell that huge question - is there anything more recognized or applicable than WCAG2.0 ? |
| 15:43 | <jgraham> | rollandburn: This is probably the wrong place to ask. However WCAG has a techniques document with some practical advice. It is not allup to date wrt HTML5 and other new technologies though |
| 15:44 | <Philip`> | Dashiva: and the ones that broke when Netscape took down the RSS DTD |
| 15:44 | <Philip`> | I guess |
| 15:44 | <Dashiva> | Philip`: So basically validators and bad ideas? :) |
| 15:45 | <rollandburn> | yes ok, not really looking for techniques so much as confirmation that as far as accessibility is concerned WCAG is God |
| 15:45 | <gsnedders|work> | jgraham: No, it's the informative documents that relate to HTML 4.01 and XHTML 1.0 that aren't up to date |
| 15:45 | <Dashiva> | rollandburn: Some people consider it god. Not everyone. |
| 15:45 | <jgraham> | gsnedders|work: Yeah Techniques for WCAG 2.0 is one such document |
| 15:45 | <rollandburn> | ok, are there other guidelines that are similar |
| 15:45 | <jgraham> | Which is the one I meant |
| 15:46 | <Philip`> | Dashiva: I'm not personally aware of any applications that use validating parsers for good reasons |
| 15:46 | <Dashiva> | Sounds like a way forward |
| 15:46 | <gsnedders|work> | But for XHTML 1.0 and RSS (0.91-Netscape) you need to cope with entities. |
| 15:46 | <Philip`> | Dashiva: though partly that's because I don't care about XML except when I can break it, so there might be good uses somewhere |
| 15:46 | <jgraham> | rollandburn: I don't know if there are other documents. It's more that some of the advice given by WCAG may be misguided or the documents may be too hard to read |
| 15:46 | <Dashiva> | DTDs are too weak for real validation anyhow, so just discard the validating parser concept in XML5? |
| 15:47 | <gsnedders|work> | Dashiva: Yeah, just totally get rid of DTDs. |
| 15:47 | <rollandburn> | oh ok, i'm not finding it hard to read... just never got into following any 'guideline' in order to 'conform' before, wanted to make sure I was following the right lead |
| 15:47 | <rollandburn> | thanks! |
| 15:47 | <rollandburn> | (btw... what is the main focus of this channel?) |
| 15:48 | <Philip`> | Dashiva: What about all the people who use DTDs for default attribute values, or for localisation of text in XML files? |
| 15:49 | <Philip`> | DTDs aren't necessarily a good solution for those things, but people rely on them in certain situations and wouldn't be happy if XML5 didn't support that |
| 15:49 | <rollandburn> | got it...the development of HTML and APIs needed for Web applications. |
| 15:49 | <Dashiva> | Philip`: So they can make their content break in non-validating parsers? |
| 15:49 | <rollandburn> | thanks,,,, cya |
| 15:50 | <Philip`> | Dashiva: So they can use off-the-shelf XML parsers (and validators and editing tools etc) for application-specific data files where interoperability isn't so important and they know they're using a validating parser themselves |
| 15:50 | <gsnedders|work> | Dashiva: Indeed. The Universal Feed Parser docs are an example of such a thing (they are written in DocBook, and rely upon external DTDs, but are distributed as HTML/PDF/etc.) |
| 15:50 | <Dashiva> | Philip`: Then they can just keep using a validating XML1 parser? |
| 15:51 | <Philip`> | Dashiva: And all XML parser developers would have to maintain two parallel versions or modes of operation? |
| 15:52 | <Philip`> | and everyone wanting to use an XML parser would have to learn about and understand the difference and make a reasoned choice? |
| 15:52 | <Dashiva> | No, they'd just use XML5 |
| 15:52 | <Dashiva> | Just like we don't expect people to make a reasoned choice between HTML5 and SGML |
| 15:52 | <gsnedders|work> | But SGML is awesome! |
| 15:53 | <Dashiva> | And I'm sure validating XML parsers are awesome too |
| 15:53 | <Philip`> | Dashiva: There aren't a billion tempting SGML tools they could pick up and use without realising it was obsolete |
| 15:54 | <Dashiva> | There are plenty of tagsoup libs for HTML4, though |
| 15:55 | <Philip`> | Indeed |
| 15:55 | <Philip`> | and lots of people use them |
| 15:55 | <Philip`> | and it's likely to be hard to stop people from using them |
| 15:55 | <Philip`> | and it's not likely that Python will deprecate and remove sgmllib any time soon |
| 15:55 | <Dashiva> | We don't have to stop them, they'll stop themselves as the web corpus gradually becomes influenced by HTML5 |
| 15:56 | <Philip`> | In the meantime it causes confusion and uninteroperability when people unintentionally choose the older version instead of the newer version, and means there's twice as much code to maintain |
| 15:57 | <jgraham> | Philip`: Presumably there were once lots of SGML tools and people who had invested in the SGML toolchain. But it was still worthwhile to simplfy out the good bits and make XML |
| 15:58 | <jgraham> | Now seems like a good time to do the same with XML with a particular focus on the web |
| 15:58 | <Philip`> | so it seems it'd be a better approach in many ways if backward compatibility was not lost and the old tools were simply upgraded to support the new spec, without having to maintain the old ones too |
| 15:58 | <Dashiva> | You can't maintain backwards compatability when one of the changes is removing a misfeature |
| 15:58 | <Philip`> | You can keep the misfeature |
| 15:59 | <Dashiva> | That's perpetuating the problem, rather than suffering a transitory period |
| 15:59 | <jgraham> | Of course. But then we would still have SGML |
| 15:59 | <Philip`> | Dashiva: You sound like you should be in the XHTML2 WG |
| 15:59 | <jgraham> | The question is "is there a path from here to there"? |
| 16:00 | <Dashiva> | Philip`: No, they wanted to change interoperable content |
| 16:00 | <Dashiva> | This is about content that's non-interoperable to begin with |
| 16:00 | <hsivonen> | gsnedders|work: you are right. the text in "Include if Validating" isn't labeled as a WFC even though implementations treat it as such |
| 16:00 | <jgraham> | We haven't discovered a path away from HTML yet even though people have tried |
| 16:00 | <Dashiva> | Your own example specified that, "application-specific data files where interoperability isn't so important" |
| 16:00 | <Philip`> | Dashiva: They wanted a transitory period of suffering rather than perpetuating the problem |
| 16:01 | <Philip`> | so you'd fit in just fine :-) |
| 16:01 | <Dashiva> | Yeah, and so did the people who introduced two-button mice |
| 16:01 | <Dashiva> | Methinks you're overreaching :) |
| 16:01 | <jgraham> | But there might be a path away from XHTML to XML5 if you could specify XML5 so that existing implementations wouldn't break too many XHTML pages if they moved to having just XML5 parsers |
| 16:03 | <hsivonen> | http://www.w3.org/TR/REC-xml/#wf-entdeclared isn't exactly easy to read |
| 16:04 | <gsnedders|work> | Most of the XML spec when it comes to entities isn't exactly easy to read. |
| 16:07 | Philip` | inherited some code which used Xerces and DTDs specifying default attribute values, but got so fed up with Xerces after a few years that he just rewrote it to use libxml2 instead and dropped all the DTDs, which was a good move |
| 16:07 | <hsivonen> | according to Tim Bray's annotated XML spec, browsers weren't supposed to read DTDs |
| 16:08 | <Philip`> | Why did browsers start doing this magical entity-mapping thing? |
| 16:08 | <Philip`> | Were there original browsers that did load DTDs, and hence content that relied on it? |
| 16:08 | <gsnedders|work> | Because there was a bug report. |
| 16:09 | <hsivonen> | Philip`: I'd expect the blame to go to a MathML-motivated bug filed against Gecko |
| 16:09 | <Philip`> | So it was just somebody trying to help users write pages? |
| 16:09 | <Dashiva> | "The validator says is valid" |
| 16:09 | <Dashiva> | Something like that? |
| 16:10 | <jgraham> | Not having entities in MathML is a big deal |
| 16:11 | <jgraham> | It changes it from "difficult to hand author" |
| 16:11 | <jgraham> | to "impossible to hand author" |
| 16:11 | <hsivonen> | whoa. looks like I have been part of the problem: https://bugzilla.mozilla.org/show_bug.cgi?id=113400 |
| 16:11 | <gsnedders|work> | Blatantly you just need a keyboard will all the various mathematical symbols on it |
| 16:12 | <Dashiva> | Skeletons in the closet |
| 16:12 | <hsivonen> | Sorry. |
| 16:12 | <erlehmann> | jgraham, is that so? can't i just use UTF8? |
| 16:13 | <erlehmann> | i mean i have αζγ all right here on my mod4 |
| 16:13 | <erlehmann> | or what is it that not having entities forbids |
| 16:13 | <gsnedders|work> | erlehmann: How do you type the symbol for union of two sets? |
| 16:13 | <jgraham> | erlehmann: Of course. But good luck finding a convenient way to enter the utf-8 codepoint for counterclockwise contour integral |
| 16:13 | <gsnedders|work> | Or do you know the Unicode codepoint and can use numeric entities? |
| 16:13 | <hsivonen> | gsnedders|work, jgraham: the world needs a math IME |
| 16:13 | <hsivonen> | something like what Mathematica has on the app layer |
| 16:13 | <gsnedders|work> | It's called LaTeX. |
| 16:14 | <jgraham> | hsivonen: That probably exists. But it is a deficiency in the technology if it screws up something that TeX got right years ago |
| 16:14 | <Philip`> | http://www.columbia.edu/acis/history/dmkeyboard.jpg |
| 16:15 | <hsivonen> | jgraham: didn't TeX get just about everything right except the Rule of Least Power? |
| 16:15 | <jgraham> | (actually I'm pretty sure LaTeX-IME exists so you can write \cccint or whatever it happens to be and get the right symbol) |
| 16:15 | <erlehmann> | gsnedders|work, i use lyx :p |
| 16:15 | <Philip`> | Seems they'd solved the problem back in 1978 |
| 16:15 | <Dashiva> | I suppose if everyone starts using keyboards with programmable LED keys |
| 16:16 | <Philip`> | http://www-lipn.univ-paris13.fr/~saiu/apl-keyboard/apl-keyboard-2.jpg is the lower-tech form of programmable keys |
| 16:16 | <jgraham> | hsivonen: Possibly it sacrified on the rule of least power in order to get everything else right |
| 16:16 | <Dashiva> | I can't speak for TeX, but latex manages to get all kinds of things wrong |
| 16:17 | <jgraham> | Yeah it's true that "everything" is a big overstatement |
| 16:17 | <jgraham> | but simple things tend to be simple and complex things are often possible |
| 16:17 | <erlehmann> | this helps http://detexify.kirelabs.org/classify.html |
| 16:19 | <Dashiva> | I suspect there's more copypaste content in latex than in HTML, even :) |
| 16:19 | <Philip`> | jgraham: Sounds like Perl's design principle |
| 16:19 | <erlehmann> | jgraham, i ener unicode codepoints with holding shift + strg + u and then input the number, works fine ∳ |
| 16:20 | <Philip`> | s//one of/; s//s/ |
| 16:20 | <Philip`> | erlehmann: The hard part is knowing the number |
| 16:21 | <erlehmann> | Philip`, now that is taken of. i looked it up and now i can produce counterclockwise countour integrals |
| 16:21 | <erlehmann> | its 2233 :) |
| 16:21 | <erlehmann> | U+2233 |
| 16:21 | <Philip`> | I suppose you can't use MathML entities without looking them up first, but at least you've got a hope of remembering them for the next few days |
| 16:21 | <erlehmann> | thats what i mean. also, it is called “ANTICLOCKWISE” |
| 16:21 | <erlehmann> | weird |
| 16:21 | <Philip`> | Why is that weird? |
| 16:22 | <Philip`> | Seems like proper English to me |
| 16:22 | <erlehmann> | i would say counterclockwise |
| 16:22 | <Philip`> | Weird |
| 16:22 | <erlehmann> | but i am by no means a native speaker |
| 16:23 | <hsivonen> | is anti-clockwise an en-GB thing? |
| 16:23 | <Philip`> | Sadly most Americans aren't native speakers either and so they all make the same mistake :-( |
| 16:23 | <gsnedders|work> | hsivonen: Yes |
| 16:23 | hsivonen | learned "counter-clockwise" |
| 16:24 | <gsnedders|work> | hsivonen: Then you don't speak proper English. |
| 16:24 | gsnedders|work | hides |
| 16:25 | <erlehmann> | gentlemen |
| 16:26 | <erlehmann> | ┌─┐ |
| 16:26 | <erlehmann> | ┴─┴ |
| 16:26 | <erlehmann> | ಠ_̼ರೃ |
| 16:26 | <erlehmann> | teatime !! |
| 16:26 | <gsnedders|work> | What? |
| 16:26 | gsnedders|work | shakes head |
| 16:27 | erlehmann | is gonna help himself to some tea |
| 16:28 | <Philip`> | gsnedders|work: I believe it is merely japes |
| 16:28 | <zcorpan_> | "I only hope it doesn't cause problems later on." - hsivonen in https://bugzilla.mozilla.org/show_bug.cgi?id=113400 |
| 16:29 | <gsnedders|work> | Truth hurts. |
| 16:32 | <TabAtkins> | erlehmann: I *love* those faces. They're so awesome. |
| 16:36 | Philip` | wonders if they would be more awesome if his font had the right glyphs |
| 16:39 | <erlehmann> | TabAtkins, i like how different charsets breed different smilie styles, like ヽ(｀Д´)ﾉ |
| 16:44 | <Philip`> | http://webaim.org/projects/screenreadersurvey2/ - "The vast majority of respondents updated their primary screen reader within the previous year" - that sounds more positive than the impression I've often got about adoption rate |
| 16:47 | <jgraham> | Philip`: Maybe the majority of screenreader users don't take geeky surveys |
| 16:48 | <jgraham> | (note the heavy weighting to self-reporting as advanced users) |
| 16:51 | <jgraham> | Seems like these screenreader users generally don't like longdesc |
| 16:53 | <jgraham> | And we should put much less time into markup for tables and much more time into markup for captchas or their replacement |
| 16:53 | <Philip`> | I don't think CAPTCHAs are a markup problem |
| 16:55 | <jgraham> | Well nor do I really but it's not clear that there is nothing that could be done at the paltform level to at least make the experience more bearable |
| 16:55 | jgraham | hates captchas and he can see them |
| 16:56 | <Philip`> | (The problem is the server wants to verify the client has certain capabilities which are closely correlated with being a human, but humans are sufficiently diverse that the correlation can never be really good) |
| 16:56 | <jgraham> | well specifically those silly "here is some hard to read text" ones |
| 16:57 | <Philip`> | (and the server has to err on the side of false negatives, because if it allows any false positives then a million people will start exploiting the false positive they found) |
| 16:57 | <Philip`> | ((or one person a million times)) |
| 19:52 | <bugfux> | I have some lines that are stroked and they are being drawn in front of some arcs (even though the arcs are drawn last) |
| 19:52 | <bugfux> | any way I can force the arcs to the front |
| 19:53 | <Philip`> | bugfux: Hmm, not quite sure what you mean |
| 19:53 | <Philip`> | Whatever's drawn last ought to appear in the front |
| 19:54 | <bugfux> | wait what defines drawn |
| 19:54 | <Philip`> | The stroke()/fill() calls |
| 19:54 | <Philip`> | (which draw everything since the last beginPath()) |
| 19:54 | <bugfux> | yeah okay so I am definately filling the arcs last because they are filled on a hover event |
| 19:54 | <bugfux> | and all the lines are drawn on render |
| 19:55 | <bugfux> | does restore() count as a draw? |
| 19:55 | <Philip`> | restore() does not affect the current path in any way |
| 19:55 | <Philip`> | beginPath is the only way to start a new path |
| 19:56 | <bugfux> | hmm okay |
| 19:56 | <Philip`> | Do you have an example of code that demonstrates the problem? |
| 19:58 | <bugfux> | i mean its not much help because it just shows the library |
| 19:58 | <bugfux> | http://danvk.org/dygraphs/tests/grid_dot.html |
| 19:58 | <bugfux> | but that higlights the problem |
| 19:58 | <bugfux> | those arcs are drawn on mouseover |
| 19:58 | <bugfux> | yet they are drawn behind almost everything |
| 20:01 | <Philip`> | bugfux: There are two <canvas>es there |
| 20:01 | <bugfux> | oh really? |
| 20:01 | <Philip`> | and the graph is drawn on the second one, and the mouseover circles are drawn on the first one |
| 20:01 | <bugfux> | oh crap, your awesome |
| 20:01 | Philip` | just looked at it in Firebug :-) |
| 20:01 | <bugfux> | no i know, but i didnt even think of htat |
| 20:02 | <bugfux> | and I have been knee deep in this library for days |
| 20:04 | <bugfux> | can I give a canvas a z-index or does that make it blow up |
| 20:05 | <Philip`> | That ought to work the same as with any other element |
| 20:06 | <bugfux> | hmm it seems to mess with hover nad stuff |
| 20:06 | <bugfux> | oh well |
| 20:07 | <Philip`> | Ah, yes, that might happen |
| 20:08 | Philip` | doesn't know if there's a way to avoid that being a problem |
| 20:08 | <Philip`> | (other than catching mouse events on the foremost object) |
| 20:13 | <bugfux> | hmmm |
| 20:13 | <bugfux> | so if i draw the canvas with the dots first, it will always be on the bottom, correct? |
| 20:17 | <Philip`> | If you don't change z-index, then yes |
| 20:49 | <TabAtkins> | AryehGregor: You around? |
| 21:45 | gsnedders | does like how HTML5 is off-topic in #html |
| 21:45 | <gsnedders> | There's something badly wrong with that. |
| 22:02 | <jarib> | anyone reading the implementors list? just realized i made the first post in months |
| 22:03 | <MikeSmith> | jarib: yeah, people do read that list |
| 22:03 | <Philip`> | I'm subscribed to it, but it's so low-traffic that I've never even bothered setting up any mail filters for it |
| 22:03 | <jarib> | ah, that's a good thing then :) |
| 22:06 | <gsnedders> | jarib: Most browsers automatically create the actual interface code from IDL (though normally each in their own proprietary format) then have the actual implementation of the interface separate |
| 22:07 | <gsnedders> | On a somewhat related note, is Watir 2.0 going to be API compatible with prior releases? |
| 22:07 | <jarib> | probably not 100% |
| 22:08 | <gsnedders> | (I ask that primarily with Opera hat on with a view of OperaWatir) |
| 22:08 | <jarib> | ah |
| 22:09 | <jarib> | it'll be mostly compatible, but more "compliant" if my IDL idea works out |
| 22:09 | <jarib> | i met Deniz from Opera at GTAC |
| 22:09 | <jarib> | Watir 2.0 will be built on WebDriver |
| 22:10 | <gsnedders> | Yeah, he's the guy who does most of the work on OperaWatir, so he probably has more clue than me :P |
| 22:10 | <jarib> | we're meeting in Oslo soon to work on the ruby bindings |
| 22:10 | <jarib> | but you're saying I could find IDLs in the browsers? |
| 22:11 | <jarib> | that's interesting |
| 22:11 | <jarib> | perhaps I should ask Deniz about this |
| 22:12 | <gsnedders> | At least in their codebase, it's gone before it even gets compiled |
| 22:12 | <jarib> | i'll look into that, thanks |
| 22:13 | <gsnedders> | WebIDL even has a defined mapping to Java :P |
| 22:14 | <gsnedders> | (Which I guess if you're going to use WebDriver is relevant) |
| 22:14 | <jarib> | nope, not using Java |
| 22:15 | <jarib> | i have ruby bindings that implement the WebDriver API (talking directly to the browser extensions) |
| 22:15 | <gsnedders> | ah |
| 22:16 | <jarib> | so for opera, we'll just use Deniz' ruby implementaation of STP0 (and perhaps do STP1 down the road) |
| 22:17 | gsnedders | doesn't really know how much about OperaWatir he can say, which is rather annoying |
| 22:18 | <gsnedders> | How much did Deniz tell you, at a fairly high level? |
| 22:19 | <jarib> | he didn't say much, just that the documentation for the protocol is outdated :) |
| 22:19 | <gsnedders> | Yeah. It's awesome documentation. |
| 22:20 | <jarib> | but his java driver for WebDriver is almost ready to ship I think, so will use that as a reference for the ruby bindings |
| 22:21 | <gsnedders> | Ah, so you do indeed know that is Java. OK. That's what I was really wondering. |
| 22:21 | <jarib> | ah, yes |
| 22:22 | <jarib> | i think his main concern was having to maintain both OperaDriver and OperaWatir, so he's pretty happy Watir 2 will be built on webdriver |
| 22:22 | <gsnedders> | My understanding is that his impl relies upon things only added to Scope after 2.2, so it's highly unlikely to ship until we have at least a public release of desktop with 2.4 |
| 22:23 | <jarib> | could be, he said he has WebDriver impls using both STP0 and STP1 |
| 22:25 | <jarib> | we have a wave about it, if you're on Wave and interested to join in |
| 22:26 | <gsnedders> | I do, though I have looked at Wave once. :) |
| 22:26 | <gsnedders> | geoffers, if you want to add me |
| 22:26 | <jarib> | will do. Wave really showed its potential at GTAC |
| 22:28 | gsnedders | really wants totally offline clients |
| 22:29 | jarib | looks at .idl files from WebKit |
| 22:30 | <jarib> | so i guess these are written by hand |
| 22:30 | <gsnedders> | Right, they will be |
| 22:31 | <gsnedders> | Probably primarily down to the fact there are about five different forms of IDL |
| 22:31 | <gsnedders> | (All of which are incompatible with one another) |
| 22:31 | <jarib> | hmm |
| 22:32 | <ment> | gsnedders: almost any modern browser has "offline mode" |
| 22:32 | <gsnedders> | ment: That doesn't mean I have absolutely all my data offline in a form I can readily export |
| 22:33 | <erlehmann> | full stop !! dump the databases !!! everyone uses sqlite, rite? |
| 22:34 | <gsnedders> | That's still fairly evil. |
| 22:34 | <ment> | gsnedders: oh yeah, that sqlite-in-javascript-in-browser concept |
| 22:34 | <gsnedders> | And only works if the web site stores all the data in a local database, which is might not (it is highly likely it'll only download a subset, like the most recent 100) |
| 22:35 | <ment> | client-side sql-injection, here i come! |
| 22:38 | <gsnedders> | jarib: Hmm, I'll probably ping Deniz about one or two things tomorrow at work |
| 22:39 | <jarib> | gsnedders: ok. do you work on anything related to this? |
| 22:39 | <gsnedders> | jarib: Not really, I've used OperaWatir a bit internally, and I've been dealing with one or two STP things recently |
| 22:40 | <gsnedders> | jarib: Probably going to spend a fair amount of time doing stuff with it next month, though |
| 22:40 | <jarib> | i see |
| 22:41 | <jarib> | if you have any suggestion as to what idls would be best fit to my purpose, i'd love to hear |
| 22:42 | <gsnedders> | Hmm, I appear to have created three blank messages in Wave accidently. |
| 22:42 | <gsnedders> | Hmm, empty messages that can't be deleted. Great. |
| 22:42 | <gsnedders> | Oh, they can be |
| 22:44 | <gsnedders> | jarib: If you're starting with something today, I'd use WebIDL. |
| 22:45 | <jarib> | gsnedders: so when you mentioned there being five different formats, are any of the open source browsers using plain WebIDL? |
| 22:46 | <gsnedders> | jarib: WebIDL is only just now becoming stable, but will almost certainly become what is widely used, so no browser uses it internally yet. |
| 22:47 | <jarib> | i see |
| 22:47 | <jarib> | what's the format used in WebKit? looks pretty similar |
| 22:47 | <gsnedders> | I dunno, I haven't looked at that for a while |
| 22:48 | <jarib> | ok, thanks again. this is exactly what i need. |
| 22:48 | <gsnedders> | as far as I know it's their own custom format. Asking in #webkit is probably a better bet. |
| 22:48 | <jarib> | will do |
| 22:48 | <gsnedders> | I'm a mere QA guy, what do I know? ;P |
| 22:49 | <jarib> | hehe, so am i :) |
| 22:49 | <gsnedders> | Anyhow, I really need to sleep |
