| 07:58 | <tantek> | Quick question - the archives for the channel should be here right? http://krijnhoetmer.nl/irc-logs/whatwg/20090515 |
| 08:00 | <Hixie> | yup |
| 08:00 | <Hixie> | but krijnh wasn't in the channel these past few hours so some stuff will have been missed |
| 08:56 | <Hixie> | jesus, the win7 installer makes the linux installers look positively consistent |
| 09:02 | <Hixie> | ok seriously |
| 09:02 | <Hixie> | even the VERY FIRST DIALOG of the installer mixes the win2k skin and the vista skin |
| 09:02 | <Hixie> | wtf |
| 09:04 | <Hixie> | the first window is win2k-style (with non-native buttons) and can be dragged around, the second window in the installer is aero-style with buggy button focus outlines |
| 09:04 | <Hixie> | -_- |
| 09:07 | <othermaciej> | Hixie: just wait until you find the stuff with Windows 98 skins |
| 09:07 | <othermaciej> | or Win 3.1 |
| 09:07 | <Hixie> | really? |
| 09:07 | <Hixie> | in win7 proper? |
| 09:07 | <othermaciej> | the Windows UI is an archeological treasure trove |
| 09:07 | <othermaciej> | well I dunno about Win7 |
| 09:07 | <othermaciej> | in Vista you can still find retro dialogs |
| 09:08 | <othermaciej> | it is possible they fixed it all for 7, though I suspect some of the dialogs were in programs where MS may have lost the source code or something |
| 09:08 | <Hixie> | if the win7 installer is any indication, they didn't. |
| 09:10 | <Philip`> | Hixie: In an all-array API, when you're dealing with data of a form you already know, all you have to do is write "item.properties.name[0]" instead of "item.properties.name", which doesn't seem too hard, and then your code won't randomly inexplicably fail when you make small changes to your data |
| 09:10 | <Hixie> | yeah |
| 09:10 | <Hixie> | it's really ugly |
| 09:10 | <Hixie> | having [0]s everywhere |
| 09:10 | <Hixie> | and i always forget to include them |
| 09:10 | <Hixie> | (e.g. with gEBTN i always forget the [0]) |
| 09:12 | jgraham | always forgets the [0] too but still prefers APIs which are consistent over ones where you always have to put in explicit type checks |
| 09:12 | <Hixie> | you still have to type check even with the [0] |
| 09:12 | <Hixie> | (you have to check for null) |
| 09:12 | <Hixie> | so it doesn't actually gain you much |
| 09:12 | <Philip`> | "[0]" is less ugly than "if (!x instanceof Array) x = [x];"? |
| 09:13 | <Hixie> | yes, but "" is less ugly than "[0]" |
| 09:13 | <Hixie> | and that's the common case |
| 09:13 | <Hixie> | the generic case isn't the common case |
| 09:13 | <Philip`> | It seems a very common pattern to have some property which can occur one or more times |
| 09:13 | <Philip`> | and that's the case where (in the current API) you can't tell whether you're going to get back an array or a string |
| 09:13 | <Philip`> | and you have to do the instanceof thing |
| 09:14 | <Hixie> | so what api changes would you want? |
| 09:14 | <jgraham> | (FWIWW the interal API in my implementation effectively uses property:[values] where [values] is a list of one or more values, so it is more like the API Philip` is proposing |
| 09:15 | <jgraham> | ) |
| 09:15 | <Hixie> | i could make HTMLPropertyCollection always return a PropertyNodeList |
| 09:15 | <Philip`> | I haven't thought about this in any detail so I don't know if there's a better solution, but changing HTMLPropertyCollection.namedItem to always return an array (and maybe an empty array instead of null) would make me happier |
| 09:16 | <Philip`> | The JSON serialisation already makes everything an array, so it's no different to that |
| 09:16 | <Philip`> | (s/array/PropertyNodeList/ or whatever) |
| 09:16 | <jgraham> | Yeah that sounds like a good solution at first glance |
| 09:18 | jgraham | seems to have got his senses confused there |
| 09:19 | <Hixie> | ok, done |
| 09:21 | <jgraham> | (BTW I made some changes to my implementaion so that it has a slightly different set of bugs. I expect it is still very broken though) |
| 09:30 | <Hixie> | ok, updated the examples too |
| 09:44 | <aja_> | fixed the vcard example, too? |
| 09:45 | <Hixie> | what was wrong with it? |
| 09:45 | aja_ | had system hange after mentioning it to you earlier |
| 09:45 | <aja_> | class=adr instead of itemprop |
| 09:46 | <Hixie> | yeah fixed that when you said it |
| 09:46 | <Hixie> | none of this is checked in yet |
| 09:46 | <aja_> | the jack bauer vcard.....afred's okay |
| 09:46 | <aja_> | alfred |
| 09:47 | <aja_> | cools |
| 09:47 | <aja_> | geez...can't friggin type today |
| 09:51 | <Hixie> | gotta love how win7 has a magical 16x16 hit box in the top left of windows that has no visible affordance but that shows a menu if you click it |
| 09:51 | <gsnedders> | Doesn't that exist going back to at least 95? |
| 09:52 | <gsnedders> | Hixie: Also: stop giving your cats drugs. They're crazy enough as is. |
| 09:53 | <Hixie> | in win95, there's an icon there |
| 09:53 | gsnedders | nowadays only uses the school's crippled version of Windows, and his memory isn't that great |
| 09:54 | gsnedders | only needs a text editor anyway :P |
| 09:59 | <Philip`> | http://en.wikipedia.org/wiki/File:Windows_2.0.png - they've had that top-left-corner icon for a long time |
| 10:03 | <Hixie> | i like(d) the icon, my complaint is just that now they have a ui element with no visual affordance |
| 10:06 | <Philip`> | Maybe it's for compatibility with programs that automatically open the menu by simulating a mouse click in the top-left 16x16 pixels of a window |
| 10:06 | <Hixie> | man, win7 has all kinds of weird bugs |
| 10:06 | <Hixie> | e.g. the UAC window has the wrong cursor for the path label |
| 10:06 | <Hixie> | Philip`: i'm sure it is |
| 10:06 | <Hixie> | Philip`: or compatibility with legacy users that do that, ilke me |
| 10:06 | <Hixie> | like, even |
| 10:07 | <Hixie> | and there's no way to enable UAC on high but have it not dim the desktop...? weird |
| 10:12 | <Philip`> | Does the screen still flash black when it brings up the UAC box? |
| 10:13 | <Hixie> | it does something similar |
| 10:13 | <Hixie> | fades it on display, and then fades back poorly on dismissal |
| 10:21 | <Hixie> | looks like actually most windows do have an icon, it's just explorer windows that don't |
| 10:23 | <virtuelv_> | Hixie: aren't path labels in Windows >= Vista just buttons? |
| 10:23 | <jgraham> | The default windows 7 UI is so /so/ ugly |
| 10:24 | <jgraham> | (that was irrelevant but still) |
| 10:24 | <Philip`> | virtuelv_: They turn into text fields if you click them in the right way, if I remember correctly |
| 10:25 | <virtuelv_> | Gnome/Nautilus is also doing the buttons thing, but never pretends to be both |
| 10:26 | <virtuelv_> | if you want to make it editable, you either press ctrl-l, or a button |
| 10:26 | <virtuelv_> | at which point, it becomes an <input type="text"> |
| 10:28 | <Hixie> | virtuelv_: didn't do anything when i clicked it |
| 10:28 | <Hixie> | virtuelv_: but it had the hand cursor |
| 10:28 | <Hixie> | btw win7 seems to come with more cronjobs preconfigured than even linux installs |
| 10:28 | <Hixie> | which is impressive |
| 10:28 | <Hixie> | (33) |
| 10:30 | <Hixie> | [This content is preliminary and subject to change.] |
| 10:30 | <Hixie> | surely that shouldn't appear in a release candidate :-) |
| 10:30 | <Hixie> | (help for the memory checker tool) |
| 10:48 | <Hixie> | is there any end-user software that uses RDF? something that isn't a semantic web tool per se but actually does tasks a typical end user actually wants to do |
| 10:48 | <Hixie> | like booking plane flights or whatever |
| 10:52 | <zcorpan> | <form><span></form><input> does not seem to match browsers - browsers associate the input with the form, afaict |
| 10:52 | <jgraham> | Hixie: I don't know but it might help if you defined "uses" better. Do you mean "uses for input", "uses as its internal data model", "has the option to serialize to" or something else? |
| 10:53 | <Hixie> | uses for input |
| 10:56 | <zcorpan> | http://software.hixie.ch/utilities/js/live-dom-viewer/saved/114 |
| 11:00 | <jgraham> | Hixie: Maybe http://www.mkbergman.com/?page_id=325 is helpful (I selected category:harvester which I have to tell you since apparently they don't give queries unique uris) |
| 11:03 | <zcorpan> | though it says "null" in ie8 |
| 11:03 | <zcorpan> | so html5 matches ie8 |
| 11:05 | <zcorpan> | wait now it's null/undefined in webkit and firefox too... must have done something wrong before |
| 11:05 | <zcorpan> | aha |
| 11:05 | <zcorpan> | <form><span>foo</form><input> |
| 11:06 | <zcorpan> | makes it non-null in firefox |
| 11:06 | <zcorpan> | but still null in webkit and ie |
| 11:07 | <zcorpan> | <form><div></form><input> is non-null in webkit |
| 11:13 | <Hixie> | jgraham: i don't understand what that is |
| 11:13 | <Hixie> | jgraham: but it doesn't look like something my mum would use |
| 11:13 | <jgraham> | Hixie: It's a list of semantic web tools |
| 11:13 | <Hixie> | oh |
| 11:13 | <jgraham> | Well RDF tools I guess |
| 11:13 | <gsnedders> | Hixie: Well, in my mum's case, the whole web doesn't look like something she would use :P |
| 11:14 | <jgraham> | (it doesn't seem to work in Opera) |
| 11:14 | <Hixie> | jgraham: does it not work in safari? |
| 11:14 | <gsnedders> | Actually, s/web/computer/ |
| 11:14 | Hixie | opens firefox |
| 11:14 | <jgraham> | Hixie: I don't know and finding out would mean starting my wwindows virtual machine which is... slow |
| 11:15 | <Hixie> | hm, works in ff |
| 11:15 | <Hixie> | doesn't work in safari. |
| 11:17 | <jgraham> | Hixie: To be fair nothing that I have seen listed thee looks like something my Mum would use |
| 11:17 | <jgraham> | (and my Mum is rather more computer literate than gsnedders' Mum) |
| 11:18 | <Hixie> | i can't even find a _category_ that my mum might use |
| 11:18 | <Hixie> | there has to be some average-end-user software SOMEWHERE that uses RDF as its input |
| 11:19 | <gsnedders> | Hixie: Google, nowadays? :P |
| 11:19 | <gsnedders> | Time for me to head off to school to do some last minute revision before a three-and-a-half hour long English exam. Yay. :\ |
| 11:19 | <jgraham> | gsnedders: Good luck |
| 11:19 | <Hixie> | i've already looked at how google handles rdf internally, i'm looking for more data points |
| 11:20 | <gsnedders> | Hixie: Well, it's some average-end-user software, no? :P |
| 11:20 | <Hixie> | yeah but i'm looking for something else |
| 11:20 | <jgraham> | Hixie: There just aren't people publishing RDF for the kind of things that ordinary people do. So there is no need to build tools |
| 11:20 | <Hixie> | there are tools for end users that use xml |
| 11:20 | <Hixie> | why not tools for end users that use rdf? |
| 11:22 | <jgraham> | (I can vaugely imagine my Mum using something that allowed her to annotate web pages and there seem to be several RDF-based tools for doing that but I'm not really sure why the RDF is needed or whether, in fact my Mum would use one of those even in the unlikely event that it had perfect UI) |
| 11:23 | <Hixie> | would or would not? |
| 11:24 | <jgraham> | ? |
| 11:24 | <Hixie> | "in fact my Mum would use one of those even..." |
| 11:24 | <Hixie> | my dad does all kinds of things with data graphs, like geneology, which i presume is well suited to being handled in RDF if anything is |
| 11:24 | <jgraham> | oh s/infact/in fact,/ |
| 11:24 | <Hixie> | but i'm not aware of any end-user software that does geneology using rdf |
| 11:24 | <jgraham> | er s/in fact/in fact,/ |
| 11:25 | <Hixie> | that still doesn't make much sense :-) |
| 11:26 | <Hixie> | (most geneology software seems to use GEDCOM as an interchange format) |
| 11:26 | <Philip`> | Are you missing the "whether"? :-) |
| 11:26 | <Hixie> | oh, i get it now |
| 11:26 | <Hixie> | thanks :-) |
| 11:26 | <pesla> | Does anyone of you know wether its possible to load HTML from a file on a FTP server using Javascript? |
| 11:26 | <pesla> | (Maybe a bit offtopic) |
| 11:27 | <Hixie> | it's possible if the javascript is itself running in a page from the same ftp server, i believe |
| 11:27 | <Hixie> | just use xhr |
| 11:28 | <pesla> | Is there cross browser support for xhr already? |
| 11:28 | <Hixie> | sure |
| 11:28 | <pesla> | Thx for the suggestion, didn't come to my mind yet ;) |
| 11:29 | <Hixie> | oh hey, that's exciting, i have a meeting in 7 hours and i haven't slept yet |
| 11:29 | <Hixie> | crap |
| 11:29 | Hixie | rushes to bed |
| 11:29 | <Hixie> | nn! |
| 11:30 | jgraham | guesses Hixie no longer wants to know that http://www.jenitennison.com/blog/taxonomy/term/34 has a lot of references to a geneology web 2.0 app using RDF but, afaict no actual link |
| 11:35 | <beowulf> | what's the point of the RDF in HTML5 stuff? |
| 11:36 | <Philip`> | There isn't RDF in HTML5 stuff |
| 11:36 | <beowulf> | oh |
| 11:36 | <Philip`> | There's just data in HTML5 stuff, and you can represent that data with RDF if you really want to |
| 11:38 | <beowulf> | why would you do that, then? |
| 11:39 | <jgraham> | Want to represent data in RDF? Not sure but some people seem to like it |
| 11:39 | <jgraham> | (I guess it might be appropriate for some applications where your model is a graph of interconnected nodes) |
| 11:40 | <beowulf> | i've worked on a couple of sites that used the UK eGMS standard for data in html, but i've never really understood to what end this activity was aimed |
| 11:42 | <jgraham> | beowulf: Isn't this the sort of thing that goverments just like even if it has no actual value? At least that was always the impression I got... |
| 11:44 | <beowulf> | jgraham: that's the impression i have, should i have the same impression of the microsyntaxes section of html5? (sorry if i've confused two seperate techs) |
| 11:46 | <jgraham> | beowulf: I can imagine some cases where it has the potential to be useful. e.g. if academic journals stopped using their own made up classnames for citation data and used the bibtex microdata spec instead it would be trivial to write a tool that extracted tose citations for inclusion in a LaTeX document |
| 11:46 | <Philip`> | <meta name="e-GMS.accessibility" content="NEED ICRA RATING. PLACE IN WEBENGINE.CFG" /> - hooray for metadata |
| 11:46 | <jgraham> | Which I would probably have done if people had been doing that two years ago |
| 11:47 | <jgraham> | s/done/writeen a tool to do/ |
| 11:47 | <jgraham> | argh |
| 11:49 | <Philip`> | In fact, "NEED ICRA RATING. PLACE IN WEBENGINE.CFG" is the second most common value for e-GMS.accessibility, coming just behind "A" and just ahead of "Triple-A" |
| 11:49 | <Philip`> | (with Double-A falling in last place) |
| 11:50 | <Philip`> | jgraham: Why wouldn't you have just clicked the "view as BibTeX" links that online journals seem to all provide? |
| 11:50 | <jgraham> | Philip`: IIRC Nature doesn't |
| 11:50 | <jgraham> | and some others too |
| 11:50 | <Philip`> | Oh |
| 11:51 | <jgraham> | At one point Monthly Notices of The Royal Astronomical society got a new website that removed all the useful features of the old one, including BibTeX data |
| 11:51 | Philip` | almost never uses anything other than the ACM Portal and sometimes CiteSeer and Google Scholar |
| 11:52 | <beowulf> | Philip`: so most cases or that metadata are wrong and useless, and yet you're probably the only one who noticed |
| 11:52 | <beowulf> | s/or/for |
| 11:52 | <jgraham> | (but fortunatley NASA provided the Atrsophysics Data System which almost always had BibTeX data, but it did mean finding the article again) |
| 11:53 | beowulf | is still scared from having to read the eGMS stuff, never mind use it |
| 11:53 | <jgraham> | beowulf: metadata is sure to be wrong if no one is using it and merely likely to be wrong otherwise |
| 11:53 | <Philip`> | beowulf: (Admittedly this is only 11 pages, so the margin of error in the results are quite high) |
| 11:53 | <Philip`> | *margins |
| 11:54 | <Philip`> | ((by which I mean this is 425K pages, of which 11 use e-GMS.accessibility)) |
| 11:55 | <Philip`> | (and a handful of others use e-GMS for some keywords) |
| 11:56 | <Philip`> | (Actually, all of those handful are on the site http://www.youngscot.org/ which repeats the huge list of keywords five times on every page) |
| 11:57 | <Philip`> | (Metadata is great in theory) |
| 11:59 | <beowulf> | like a laser, then? |
| 12:01 | <Philip`> | HTML5 microdata seems to be more about replacing microformats with something that can parsed with a generic parser |
| 12:01 | <Philip`> | s/can/can be/ |
| 12:01 | <Philip`> | and that doesn't have to squeeze itself into attributes where it doesn't really fit |
| 12:02 | <Philip`> | and tha has a decentralised mechanism for avoiding name conflicts |
| 12:02 | <Philip`> | *that |
| 12:07 | Philip` | can never decide whether something relying on DNS is centralised or decentralised |
| 12:08 | <hober> | I think @class has the same story as @item insofar as disambiguation is concerned. Which is to say, those concerned about disambiguation can choose sufficiently disambiguous names for their own vocabularies. |
| 13:01 | <jgraham> | "I will never support microdata, because no case has been made for its existence." That seems logically flawed |
| 15:15 | <Philip`> | XMLLiteral handling in text/html seems a bit inconsistent at the moment - for the same input in three implementations I get: |
| 15:15 | <Philip`> | "Test<i>Test<br/>Test</i>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>; |
| 15:15 | <Philip`> | "Test<i>Test<br>Test</i>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>; |
| 15:16 | <Philip`> | "Test<I>Test<BR>Test</I>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>; |
| 15:16 | <Philip`> | "TestTestTestTest"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>; |
| 15:16 | <zcorpan> | 3 impl, 4 results? |
| 15:17 | <Philip`> | Yes (since some were JS implementations, and vary between browsers) |
| 15:17 | <zcorpan> | ah |
| 15:21 | <zcorpan> | Philip`: btw, had you sent email about triples changing after following an internal link? should the fragment be omitted from "the document's current address"? |
| 15:21 | <Philip`> | Using the same implementations on XHTML, I get: |
| 15:21 | <Philip`> | "Test<i xmlns=\"http://www.w3.org/1999/xhtml\">Test<BR/>Test</i>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>; |
| 15:21 | <Philip`> | "Test<i xmlns="http://www.w3.org/1999/xhtml">Test<BR xmlns="http://www.w3.org/1999/xhtml"/>Test</i>Test"^^<http://www.w3.org/1999/02/22-rdf-syntax-ns#XMLLiteral>; |
| 15:22 | <Philip`> | zcorpan: Oh, I forgot about that and haven't sent anything |
| 15:25 | Philip` | notes that those XHTML bits are correct, because the input had a "<BR/>" in it |
| 15:28 | <hsivonen> | where's the parsing spec for XML literals (that don't seem to have a root element)? |
| 15:41 | <hsivonen> | http://www.w3.org/TR/rdf-concepts/#section-XMLLiteral |
| 15:42 | <hsivonen> | wow. is that really saying that XMLLiterals must be exclusively canonicalized to be conforming? |
| 15:44 | <hsivonen> | it seems there's no explicit parsing spec but it's implied that the precessing is similar to setting innerHTML in XHTML |
| 16:07 | <zcorpan> | wouldn't it be <span item><span itemprop=w3c.time propvalue="12-25-2009"> ? |
| 16:09 | <hsivonen> | zcorpan: why the reverse endianness? |
| 16:09 | <hsivonen> | or permuted rather than reversed |
| 16:12 | <Philip`> | zcorpan: It would (according to the current spec) |
| 16:12 | <Philip`> | (since itemprops correspond to the nearest ancestor item, and an element is not its own ancestor) |
| 16:14 | <zcorpan> | hsivonen: copy-paste from what jonas wrote |
| 17:22 | <gsnedders> | w00t! English! Done! |
| 17:24 | <hsivonen> | gsnedders: did it go well? |
| 17:25 | <Philip`> | gsnedders: Did you fail miserably? |
| 17:31 | <gsnedders> | hsivonen: The first part, definitely. The second part, not so well, but I think well enough. |
| 17:32 | <gsnedders> | Philip`: Hence no. |
| 17:32 | <gsnedders> | Philip`: Also, seeming I apparently had the best dissertation it'd be hard to fail badly, seeming I scarcely had to get any marks to pass |
| 17:32 | <hsivonen> | gsnedders: btw, there's no "XHTML WG" |
| 17:32 | <hsivonen> | (Re: twitter) |
| 17:33 | <gsnedders> | Peh. XHTML2 WG. You know what I mean. |
| 17:36 | <Philip`> | gsnedders: Were the people marking your dissertation impressed by the markup? |
| 17:36 | <gsnedders> | Philip`: It wasn't markup :P |
| 17:37 | <Philip`> | gsnedders: I thought that's what you were spending all your time doing? :-) |
| 17:37 | <gsnedders> | Philip`: No. Markup for my notes for English in general |
| 17:37 | <gsnedders> | Also: my fingers hurt from writing for three hours so I don't want to type too much |
| 17:38 | <Philip`> | Type with other appendages |
| 17:41 | Philip` | gets an impression from the "Link rot is not dangerous" topic that namespace URIs are quite a fragile foundation |
| 17:42 | <Philip`> | so they suggest building other structures on top of that, like caching and redirecting and hardcoding override lists and reminding people not to accidentally let their domains expire and making local subclasses |
| 17:42 | <hsivonen> | Philip`: it seems to be that believing in Follow your Nose and believing in Link Rot not being dangerous are contradictory beliefs but you can pick either one and argue coherently |
| 17:43 | <Philip`> | and I suppose it makes me wonder instead whether it'd be a good reason to not use that foundation at all |
| 17:44 | <Philip`> | (though I don't know what other foundations would be better) |
| 17:45 | <hsivonen> | Philip`: Java-style FQNs avoid the temptation to succumb to Follow Your Nose |
| 17:46 | <Philip`> | hsivonen: They still suffer from a mild form of Link Rot, because if I start using com.example.foo and then accidentally let com.example expire and someone else registers it and builds a new application based on that domain then they might conflict with my names and people will think my names belong to them |
| 17:47 | <Philip`> | The problem is that an object with a certain lifetime is depending on an object with a potentially shorter lifetime |
| 17:48 | <Philip`> | and C++ teaches us that such relationships will eventually cause horrible pain and memory corruption |
| 17:48 | <hsivonen> | Philip`: more likely, the owner of com.example is acquired by another company or the product is made open source under org.apache |
| 17:49 | <Philip`> | That too |
| 17:49 | <hsivonen> | but yeah, Java FQNs suck because code outlasts organization structure |
| 17:49 | <Philip`> | I wonder if all the com.sun.* Java classes will be renamed to com.oracle.* |
| 17:53 | <Philip`> | Has anyone (in distributed systems in general, not just the web) solved the problem of having identifiers that are globally unique and eternally unique, and that don't depend on a centralised registry, and that are human-readable? |
| 17:54 | <Philip`> | (I suppose I consider mildly distributed registries, like the IP number assignment system, to be similar enough to a centralised registry that it's not acceptable either) |
| 17:55 | <Philip`> | If not, I guess the question is which of those requirements to sacrifice |
| 17:55 | <hsivonen> | Philip`: depends on probablility of collision |
| 17:55 | <hsivonen> | Philip`: openid.delegate seems to strike the right balance |
| 18:05 | <hober> | Python package names seem to strike a good balance |
| 18:09 | <hsivonen> | hober: yeah, they are basically the same as openid.* and openid2.* rel values |
| 18:10 | <Philip`> | But Python lets you manually override the mapping from identifier to implementation (e.g. by changing PYTHONPATH or moving libraries into the current directory) if you need to resolve conflicts in exceptional cases |
| 18:10 | <Philip`> | just like XML Namespace prefixes do |
| 18:11 | <Philip`> | Without some mechanism for conflict resolution, it's not a robust way to avoid problems caused by naming conflicts |
| 18:12 | <hsivonen> | Philip`: how often do python names actually collide? |
| 18:15 | <Philip`> | It's not terribly unlikely that I would e.g. create a physics simulation program with a file called "atom.py" containing my Atom class, and say "import atom" |
| 18:15 | <Philip`> | and it would collide with the 'atom' package (for RFC 4287 style Atom) |
| 18:16 | <hober> | I think python package names, or java FQNs, etc. provide enough disambiguation |
| 18:16 | <Philip`> | (The conflict there is easily resolved because Python will look for ./atom.py before it looks for /usr/lib/python2.5/site-packages/atom, if I remember correctly) |
| 18:16 | <hober> | People seem to want more disambiguation than they really need, to get some kind of warm-fuzzy feeling of false security |
| 18:17 | <Philip`> | Why is it false security? (as opposed to unnecessary (but true) security) |
| 18:17 | <hober> | But given author error, cut-and-paste, etc. the straightjacket solutions to disambiguation don't really get you any more disambiguation than the "sufficiently-interesting prefix" solution |
| 18:27 | <gsnedders> | Is there any way to have a form that can be submitted to multiple URLs depending on what button is pressed, without script? |
| 18:29 | <gsnedders> | I guess @value of the submit button |
| 18:52 | <Philip`> | "link rot within the RDF world is an extremely rare and unlikely occurrence" |
| 18:52 | <Philip`> | That sounds like a testable observation |
| 18:52 | <Philip`> | and the test seems to refute that observation |
| 18:54 | jgraham | had python package name conflicts when he tried to build V8 so it does happen and can be confusing |
| 18:54 | <Philip`> | I see several 404s and several servers-not-responding, from the few dozen mosts common RDF namespace URIs |
| 18:54 | <Philip`> | s/mosts/most/ |
| 18:56 | <Philip`> | (There are so many servers-not-responding that my download script is taking forever :-/ ) |
| 19:06 | <Philip`> | I really should have reduced curl's timeout |
| 19:51 | gsnedders | dreams of a world in which HTML was sane |
| 20:03 | <gsnedders> | http://direct.tesco.com/search/default.aspx?search=%EF%BF%BF |
| 20:03 | <gsnedders> | Oh how boring. They fixed it. |
| 20:03 | <Dashiva> | "Unexpected error" is a fix? :) |
| 20:03 | <gsnedders> | Dashiva: Well, at least what it returns is well-formed |
| 20:04 | <Dashiva> | Fair enough |
| 20:19 | jgraham | wonders ifmoving the predefined microformats into their own small specs would keep people happy whilst having few downsides |
| 20:22 | <hsivonen> | jgraham: what WG would publish those? |
| 20:22 | <jgraham> | HTML WG? I guess it might be outside our charter... |
| 20:23 | <gsnedders> | webapps. Everything that doesn't have a proper WG goes there. :P |
| 20:23 | <jgraham> | (In practice I can't imagine that Sam would dicourage it at least) |
| 20:28 | <Philip`> | Is Shelley failing to notice that any random person could start using xmlns:foo="http://www.w3.org/foo"; even if the W3C has got nothing to do with it? I don't see why URLs would prevent that whereas reversed domains wouldn't |
| 20:28 | <Philip`> | hsivonen: Publish them on wiki.whatwg.org |
| 20:30 | <gsnedders> | Can you reply to multiple emails at once in Thunderbird? |
| 20:30 | gsnedders | needs a sane email client |
| 20:30 | gsnedders | is on the edge of giving up on GUIs and moving to alpine |
| 20:32 | <Dashiva> | Philip`: I'd guess so. Sacred URLs seem to be a common thing in semweb :) |
| 20:32 | <hsivonen> | is there a corollary for Godwin's law for "I can't wait for the lawsuits on this one."? |
| 20:37 | <jgraham> | I have concluded that there is no point in discussing things with Shelly. She has stated that she is not interested in technical merit and will not change her opinion regardless of facts so I don't see that she has anything productive to contribute |
| 20:38 | <jgraham> | Or at least there is nothing productive to be gained from calling her out when she is making falicious arguments because she is uninterested in facts, only in seeing her chosen solution suceed |
| 20:41 | <jgraham> | (I guess there might be something gained by the side effect of someone who is interested in rational discussion seeing the issues but I think overall it is better to discourage people who are unwilling to be reasonable) |
| 20:44 | <Philip`> | Even if someone will not be convinced by any arguments, they might still bring up relevant points that are worth discussing |
| 20:45 | <Philip`> | so it seems useful to respond to those points, but not to respond if you're just repeating past arguments rather than finding anything new |
| 20:47 | <jgraham> | Philip`: That was what I ws trying to say in the parenthetical above |
| 20:48 | <jgraham> | Although I'm not sure I am right that the value of discouraging pure advocacy outweighs the value of discussing all relevent points when those two goals conflict |
| 20:49 | <Philip`> | jgraham: Indeed, but I started responding during a respawn break in TF2 before your parenthetical, and didn't finish responding until the next respawn |
| 20:50 | <jgraham> | gsnedders: http://mailtweak.mozdev.org/tweaks.html#collect looks like it might help you |
| 21:04 | <Hixie> | oh hey, shelley quit in a huff again |
| 21:05 | <Hixie> | it must be 1pm |
| 21:37 | <Philip`> | http://bnode.org/blog/2009/05/15/could-microdata-work-better-for-me-than-rdfa - "it took me quite some time to admit that (intuitively desirable) URI abbreviations in HTML do have negative practical implications" |
| 23:31 | <gsnedders> | http://twitter.com/brucel/statuses/1810429136 |
| 23:31 | <gsnedders> | Oh lovely. |
| 23:43 | <Philip`> | Just for reference purpopses: the context for jgraham's remarks was http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2009-May/019767.html, which seems to me to be a rant |
| 23:44 | <Philip`> | which isn't necessarily a bad thing, but in this case it makes reasonable debate seem very hard to achieve |
| 23:54 | <Philip`> | (...At least, I assume that was the context) |
