| 00:44 | <takkaria> | http://www.shirky.com/writings/semantic_syllogism.html is a really good critique of the Semantic Web project |
| 01:03 | <erlehmann> | http://en.wikipedia.org/wiki/In_popular_culture#Popular_Culture_in_Popular_Culture |
| 01:24 | <Dashiva> | the "this will work because it would be good if it did" fallacy |
| 01:24 | <Dashiva> | ^ Does that one have a more formal name? |
| 01:27 | <takkaria> | not as far as I know |
| 01:30 | <Philip`> | http://en.wikipedia.org/wiki/Wishful_thinking#As_a_logical_fallacy |
| 01:32 | <Dashiva> | "My girlfriend has an XML serialization of her shoe closet (I am not making this up)" |
| 01:33 | <takkaria> | where's that? |
| 01:34 | <Dashiva> | http://diveintomark.org/archives/2002/12/30/the_tag_soup_of_a_new_generation |
| 01:37 | <blooberry> | I think that's called "wishful thinking" |
| 01:37 | <Hixie> | takkaria: yeah, good article. from 2003, interesting. not much has changed since then as far as i can tell. |
| 01:41 | <Dashiva> | Hixie: You could say everything except SW has changed, SW is still stuck where it was ;) |
| 01:41 | <takkaria> | in my philosophy course, I'm currently doing a lot of reading of 20th century philosophers who basically wrote the rules of modern logic |
| 01:42 | <takkaria> | part of the shift towards modern logic was realising that syllogistic logic isn't actually all that helpful |
| 01:42 | <Philip`> | Boolean logic is all we really need |
| 01:43 | <Dashiva> | . |
| 01:43 | <Philip`> | You can prove everything by enumeration, because infinite sets don't exist in reality so you don't need to care about them |
| 04:01 | <Hixie> | anyone remember what various ways people suggested splittig the spec last week? |
| 04:41 | <sicking> | Hixie, there was a lot about splitting out the parts that define what 'valid' markup looks like |
| 04:42 | <sicking> | Hixie, and splitting out some APIs, don't remember which ones though |
| 04:42 | <sicking> | Hixie, oh, 'window' was one of them |
| 04:44 | <Hixie> | ok, that's what i remembered too |
| 04:44 | <Hixie> | thanks |
| 10:07 | <Lachy> | Hixie, there were people asking for some kind of clean language spec that defined only what is considered conforming and a "clean parsing algorithm" that only works with conforming content. |
| 10:09 | <Lachy> | one specific case that was pointed out to me was where it defines microsyntaxes, it mixes the conforming syntax definition with the rules for parsing it a little too closely http://www.whatwg.org/specs/web-apps/current-work/#numbers |
| 12:45 | <MikeSmith> | hsivonen_: http://jing-trang.googlecode.com |
| 12:46 | <MikeSmith> | http://tech.groups.yahoo.com/group/rng-users/message/871 |
| 12:48 | <MikeSmith> | "Another potential area for improvement is the quality of error messages that Jing gives when doing RELAX NG validation. They're pretty bad at the moment, and it's definitely possible to do significantly better with a bit of effort." |
| 16:34 | <annevk2> | Hixie, new Option is a named constructor |
| 19:37 | <Hixie> | anyone got an idea for how to solve the "don't validate for this submit button" use case? (e.g. saving state to the server, or asking the server for completion information, separate from the final submission step) |
| 19:39 | <annevk2> | use submit() |
| 19:39 | <annevk2> | if it's really common, add a dontvalidate attribute or some such |
| 19:40 | <annevk2> | bbiab |
| 19:42 | <Hixie> | "dontvalidate" is kinda ugly |
| 19:42 | <Hixie> | even "novalidate" is ugly |
| 19:46 | <Lachy> | is the attribute intended to go on the form element or the <input type="submit">/<button> elements? |
| 19:49 | <Lachy> | Hixie, validate="no" is less ugly, but it doesn't fit the normal boolean attribute design pattern |
| 19:50 | <gsnedders> | ihatevalidation? |
| 19:59 | <Hixie> | Lachy: if it's an attribute, the button, i guess, but i have no idea how to really solve this |
| 19:59 | <Hixie> | maybe nonvalidating ? |
| 19:59 | <Hixie> | it's better than dontvalidate and novalidate |
| 20:00 | <Lachy> | hmm, that's a rather subjective comparison |
| 20:00 | <Lachy> | I thought novalidate looked better |
| 20:00 | <Hixie> | novalidate is not grammatically correct english |
| 20:00 | <Hixie> | nonvalidating is at least one word |
| 20:01 | <Lachy> | ok |
| 20:01 | <Lachy> | I still don't particularly like nonvalidating |
| 20:02 | <Hixie> | i don't particularly like any of the |
| 20:02 | <Hixie> | m |
| 20:04 | <gsnedders> | non is grammatically wrong so FAIL. |
| 20:04 | <Hixie> | ? |
| 20:04 | <Hixie> | i don't think a sentence using the "fail" meme really should be complaining about grammar, btw :-) |
| 20:04 | <gsnedders> | I'm not thinking, ignore me |
| 20:04 | <Hixie> | k |
| 20:04 | <gsnedders> | (I am thinking, just about typing on a Dvorak keyboard) |
| 20:14 | <blooberry> | gsnedders: thinking about trying dvorak? |
| 20:15 | <nessy> | how about "novalidation" ? |
| 20:17 | <blooberry> | does it have to use a term with "valid" as a substring? |
| 20:17 | <blooberry> | Would something tangential like maybe check" or "nocheck" work instead? |
| 20:18 | <gsnedders> | blooberry: I am trying, that's the problem |
| 20:18 | <Hixie> | it's not clear to me that an attribute is right in the first place |
| 20:19 | <gsnedders> | an element! |
| 20:19 | <blooberry> | gsnedders: keep it up! *cheering from my dvorak layout* |
| 20:20 | <gsnedders> | blooberry: Everytime before I haven't for very long |
| 20:21 | <gsnedders> | But I seem to be picking it up quicker this time |
| 20:21 | Hixie | is using qwerty... but his dvoark keyboard is less than a foot away! |
| 20:21 | <Lachy> | Hixie, what about a new button type? |
| 20:21 | <Lachy> | instead of a new attribute |
| 20:21 | <Hixie> | not bw compatible |
| 20:21 | <Lachy> | <button type="whatever"> is kind of |
| 20:21 | <Hixie> | not in IE |
| 20:22 | <Lachy> | what about IE8? |
| 20:22 | <Hixie> | (and <input type=non-validating-button> isn't) |
| 20:22 | blooberry | would have huge problems going back to qwerty |
| 20:23 | <gsnedders> | blooberry: I have to cope with that at school |
| 20:23 | <Hixie> | i use both |
| 20:23 | <Hixie> | (on different keyboard types) |
| 20:23 | <gsnedders> | blooberry: So if I persist I have to cope with both |
| 20:24 | <Lachy> | does it need to be a declarative solution, rather than just some kind of javascript API that submits without checking validity? |
| 20:24 | <Hixie> | Lachy: i don't know |
| 20:25 | <Lachy> | if so, then there are 3 options: new attribute, new element, or new button type. If the latter is ruled out because of back compat issues, and the new element doesn't seem realistic, you're left with a new attribute |
| 20:25 | <gsnedders> | Hikie doesn't know every thing!? :o |
| 20:25 | <blooberry> | for me, it's touch typing dvorak, hunt-n-peck qwerty. I have no idea where the dvorak keys are located unless I let my brain put the fingers in the right location when asked to hit the key. |
| 20:25 | <gsnedders> | See, I really can't type now |
| 20:25 | <Lachy> | gsnedders, of course, "Hikie" isn't omniscient, but I'm sure Hixie is :-) |
| 20:26 | blooberry | is always amused how many typos are an obvious sign of which keyboard layout a person is using |
| 20:26 | <gsnedders> | blooberry: How so? |
| 20:26 | <blooberry> | hikie->hixie: qwerty the 'k' and 'x' are on opposite sides of the keyboard. On dvorak, they are adjacent |
| 20:27 | <gsnedders> | ah |
| 20:28 | <gsnedders> | But you just said you don't know where keys are! |
| 20:28 | <Hixie> | Lachy: one could equally say "if a new attribute doesn't seem realistic, and a type is ruled out because of back compat issues, you're left with a new element" |
| 20:28 | <gsnedders> | (Yes, it takes me that long to type.) |
| 20:29 | <blooberry> | I lied? 8-P (not for touch typing, no...but I know where the keys are.) |
| 20:30 | <Lachy> | Hixie, I can't think of any logical way of using a new element for this purpose |
| 20:30 | <gsnedders> | I wish there was a room in this house with good speakers and is warm. |
| 20:32 | <Hixie> | Lachy: there are plenty of ways we could do this, e.g. attributes on the button or the form, an API to submit, an API to disable validation, boolean attributes or enumerated attributes, new types, class values, a container element, not supporting it but requiring script of various kinds, etc |
| 20:33 | <Hixie> | they all suck |
| 20:33 | <BenMillard> | I don't understand the use case...in my experience of using and making and auditing websites, I haven't come across a button which submits a form but must bypass any client-side validation |
| 20:34 | Hixie | uses anathem as a keyboard rest and switches to dvorak |
| 20:34 | <Lachy> | BenMillard, see the whatwg archives for discussion of non-validating submit buttons |
| 20:34 | <Lachy> | or in the WF2 folder http://www.whatwg.org/issues/#WF2 |
| 20:34 | <Hixie> | BenMillard: e.g. a button that takes a postcade and prefills the address |
| 20:35 | <Hixie> | postcode |
| 20:35 | <Hixie> | without requiring all required fields to be set |
| 20:35 | <Hixie> | or a button that cancels the form |
| 20:35 | <Hixie> | or a button that saves the current state |
| 20:37 | <BenMillard> | "Actually, the "Cancel" button could easily be just a link, not a submit control." |
| 20:41 | <BenMillard> | I share Simon Pieters's experience where he says "I have never seen such a form" for the saving state without validation case. Nearest I've seen to that are multi-page forms (such as YouGov in the UK, IIRC) which save your progress as you go but still require the page you're on to be filled in completely. |
| 20:43 | <BenMillard> | oh, I understand the FAFSA case |
| 20:43 | <billmason> | The company I work for does online college applications...probably 99+% of them are multipage, and all of them have a button to save the page without validation. |
| 20:45 | <Lachy> | for the cancel button case, ideally none of the form's content would be submitted to the server since it's a waste of bandwidth, especially if it's a large form |
| 20:47 | <Lachy> | but sometimes a cancel button does actually need to be a button rather than a link because the server needs to be told to interrupt the process. e.g. cancelling half way through a multi step process |
| 20:49 | <BenMillard> | Lachy, the Cancel button can submit to a different <form>? |
| 20:49 | <Philip`> | The button that takes the postcode and prefills the address should be using AJAX, not form submission, because everyone loves AJAX |
| 20:49 | <Hixie> | the cancel button is not the compelling case |
| 20:50 | <Philip`> | billmason: Do those pages ever do client-side validation? |
| 20:51 | Philip` | vaguely remembers multipage forms like that only doing server-side validation of the entire form at the end of the process (so it can't be client-side because it checks all the pages at once) |
| 20:51 | <Lachy> | BenMillard, yeah, that works |
| 20:51 | <billmason> | Philip`: Eventually to submit the completed form (when you complete all pages), yes. The forms typically validate as you go from page to page, although there is often an option to suspend validation until the final submission. |
| 20:53 | <BenMillard> | billmason, I recognise that use-case as being similar to YouGov, except that you can only save YouGov surveys at the first completed page and not at an arbitrary point |
| 20:55 | <BenMillard> | Philip`, AJAX for address look-up makes sense. Then again, if you've entered your postcode you don't need to see the address because it will be the correct address by definition. |
| 20:57 | <BenMillard> | indeed, if the app can do look-up from a postcode there's no need to ask for other address fields? |
| 21:00 | <Philip`> | BenMillard: It can't determine the precise address from the postcode, because it's not a one-to-one mapping, so it needs to give the user the opportunity to correct their house number or whatever |
| 21:00 | <Philip`> | and the postcode lookup databases are not infallible, so the user has to be able to override it |
| 21:02 | <BenMillard> | Philip`, that's disappointing. :( |
| 21:02 | <Philip`> | BenMillard: That's life :-) |
| 21:03 | <Philip`> | Even if nobody ever makes any mistakes, new houses will be built and they won't be in the existing databases |
| 21:05 | <BenMillard> | Philip`, house take a while to build so getting that far out of date is still disappointing. |
| 21:05 | <BenMillard> | s/house/houses/ |
| 21:08 | <annevk2> | Hixie, has it been shown that this problem needs addressing? |
| 21:08 | <Hixie> | dunno |
| 21:09 | <annevk2> | Hixie, <input type=submit onclick=this.form.submit()> works |
| 21:09 | <annevk2> | (maybe you need return false) |
| 21:43 | gsnedders | notes there have been far fewer xref typos on Anolis |
| 23:02 | <john_fallows> | Hixie: as promised, I sent separate emails to the comments list regarding SSE proxies and SSE load balancing |
| 23:10 | <Hixie> | thanks |
| 23:10 | <Hixie> | saw those |