| 09:51 | <krijnh> | o/ |
| 09:52 | <zcorpan> | \o/ |
| 09:53 | <krijn> | Sorry about that |
| 09:53 | <nimbu> | wrong window krijn |
| 09:53 | <krijn> | ? |
| 09:53 | <nimbu> | "sorry about that" |
| 09:54 | <annevk> | no |
| 09:54 | <annevk> | he fixed the logs |
| 09:54 | <nimbu> | o |
| 09:56 | <krijn> | Yeah, sorry for not fixing it earlier :) |
| 09:57 | krijn | blames dConstruct and cremations |
| 10:04 | <annevk> | :( |
| 12:26 | <hsivonen> | wow. itemId reflects an URL like href, etc.? That's novel in the URLs-as-identifiers land. |
| 12:27 | <annevk> | it's solely an identifier? |
| 12:27 | <annevk> | seems like a bad idea then |
| 12:38 | <MikeSmith> | so hey |
| 12:39 | <annevk> | hey MikeSmith! |
| 12:39 | <MikeSmith> | hey man |
| 12:39 | <MikeSmith> | I'm trying to list out some likely candidates for new attributes and elements that may make their way into the HTML spec some time in the near future |
| 12:39 | <MikeSmith> | the main one I can think of is the inputmode-like thing |
| 12:39 | <MikeSmith> | http://wiki.whatwg.org/wiki/Text_input_keyboard_mode_control |
| 12:39 | <annevk> | <x- or <data- for the component model? |
| 12:40 | <MikeSmith> | the other I can think of is the <data> |
| 12:40 | <MikeSmith> | element |
| 12:40 | <MikeSmith> | annevk: ah yeah |
| 12:40 | <MikeSmith> | any others come to mind? |
| 12:40 | <annevk> | there was the <input type=range> bug about having two sliders |
| 12:40 | <MikeSmith> | OK |
| 12:40 | <annevk> | so <input type=double-range> maybe? |
| 12:40 | <annevk> | guess that's not really a new attribute |
| 12:41 | <annevk> | oh |
| 12:41 | <annevk> | <dialog> |
| 12:41 | <annevk> | <dialog modal> |
| 12:41 | <annevk> | maybe XBL |
| 12:41 | <MikeSmith> | ah yeah |
| 12:42 | <annevk> | XBL - 30% chance |
| 12:42 | <annevk> | <x-> 50% |
| 12:42 | annevk | tries to think of other arbitrary percentages |
| 12:42 | <zcorpan> | namespaces, 0% |
| 12:42 | <annevk> | we once had that tree thing |
| 12:42 | <annevk> | :) |
| 12:42 | <annevk> | <datatable>? |
| 12:43 | <hsivonen> | annevk: is XBL that much alive still? |
| 12:43 | <annevk> | <datagrid>! |
| 12:43 | <annevk> | hsivonen, if the component model ever gets a declarative side it would be XBL |
| 12:44 | <annevk> | hsivonen, afaict |
| 12:44 | <annevk> | hsivonen, and it seems some Gecko developers and Hixie still think that's a good idea |
| 12:45 | <hsivonen> | annevk: is any of those Gecko developers actively implementing? |
| 12:45 | <annevk> | no, but is Gecko implementing the component model at this point? |
| 12:45 | <MikeSmith> | hsivonen: I'm wondering what your assessment is on <menuitem>, and what you think the odds are on the ever getting into the spec |
| 12:46 | <hsivonen> | annevk: I wouldn't know |
| 12:46 | <hsivonen> | MikeSmith: oh yeah, that |
| 12:46 | <annevk> | MikeSmith, you're trying to do this HTML.next thing? |
| 12:46 | <hsivonen> | MikeSmith: I thought that one was between the Mozilla guys who wanted it and Hixie |
| 12:47 | <hsivonen> | MikeSmith: I guess I'll need to find out what the situation is |
| 12:47 | <MikeSmith> | hsivonen: yeah it is, essentially as you said |
| 12:47 | <MikeSmith> | hsivonen: it's just kind of been hanging for a few weeks now |
| 12:47 | <MikeSmith> | last comment on it from Jonas |
| 12:48 | <MikeSmith> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=13608 |
| 12:48 | <MikeSmith> | annevk: yeah, just trying to make a summary assessment of where things are at overall at this point, as far as possible new markup additions |
| 12:49 | <hsivonen> | I want Gecko and the spec to match. and I want us to be careful about minting new void elements. (IIRC there's now a polyfill for <menuitem> floating around, and that sort of thing is actually easier to make when it's non-void) |
| 12:49 | <hsivonen> | I not that interested into bikeshedding the name |
| 12:49 | <hsivonen> | s/into/in/ |
| 12:50 | <MikeSmith> | yep |
| 12:50 | <MikeSmith> | agreed on all that, fwiw |
| 12:51 | <MikeSmith> | but I think the implementation work that Jan did has at the very least exposed some possible problems with the context-menu/command stuff that should be addressed in some way |
| 12:51 | <MikeSmith> | (if not necessarily with a new element) |
| 12:51 | <hsivonen> | I haven't paid careful enough attention to comment |
| 12:54 | <smaug____> | hsivonen: FYI, menuitem won't be void element |
| 12:54 | <smaug____> | hsivonen: since its handling was made more consistent with <option> |
| 12:55 | <hsivonen> | smaug____: on the DOM level that is? hopefully not in parsing |
| 12:56 | <hsivonen> | smaug____: so we can now back out the #ifdefs for making menuitem void? |
| 12:57 | <smaug____> | hsivonen: <menuitem>label</menuitem> should work |
| 12:57 | <hsivonen> | smaug____: ok |
| 12:57 | <hsivonen> | let's get the #ifdefs backed out, then |
| 13:05 | <foolip> | hsivonen, annevk: itemid isn't quite like itemtype, something like itemid="#foo" isn't completely unreasonable |
| 13:14 | <hsivonen> | foolip: I didn't say unreasonable. just novel in the URIs-are-identifiers land. |
| 13:28 | <MikeSmith> | so to be clear, the <x-foo> custom-element registration piece would require changes to the parsing algorithm, right? |
| 13:29 | <annevk> | maybe |
| 13:29 | <annevk> | depends on the details |
| 13:29 | <MikeSmith> | OK |
| 13:34 | <hsivonen> | http://www.w3.org/Bugs/Public/show_bug.cgi?id=14029#c16 is sad |
| 13:37 | <annevk> | "Where I came from, we have rather colorful names for folk that hide behind walls, including walls of pseudonyms." |
| 13:37 | <annevk> | holy shit |
| 13:37 | <MikeSmith> | hsivonen: yep |
| 13:38 | <MikeSmith> | annevk: yeah, that is even worse |
| 13:38 | <MikeSmith> | dude is out of line |
| 13:38 | <MikeSmith> | too much time on their hands |
| 13:39 | <annevk> | and again I am amazed by bz remaining calm in the face of idiots |
| 13:39 | <MikeSmith> | the stuff that people choose to burn up time agitating about is boggling |
| 13:39 | <hsivonen> | I wonder if Glenn believes all the spam puppets with real-looking Firstname Lastname names on G+ are real people with passports written for those names |
| 13:39 | <MikeSmith> | annevk: yeah, bz is model of how to deal with cases like this |
| 13:40 | <smaug____> | hsivonen: actually, how is <option> defined? |
| 13:40 | <smaug____> | it can be void but also container, right? |
| 13:40 | <hsivonen> | smaug____: it can't be void. it's end tag can be omitted |
| 13:40 | <smaug____> | ah, ok |
| 13:41 | <erlehmann> | hsivonen, also there are people with “fake-looking” names, that get kicked out regularly everywhere. |
| 13:41 | <MikeSmith> | hsivonen: I think he does not, but he doesn't care. The goal is just for the sake of appearances. Or perhaps it's not and they next thing they are going to agitate for is that the W3C start actually verifying the identity of everybody who has a W3C account or who volunteers to write a spec (and actually does it). |
| 13:42 | <erlehmann> | scaring contributors away, yay! |
| 13:48 | <hsivonen> | (speaking of passports, I occasionally find it curious that there are countries that have the audacity to issue passports--i.e. certify a person's existence--without have a database of all people in the country) |
| 13:49 | <hsivonen> | (at least *my* existence is backed by a database line) |
| 13:49 | <hsivonen> | s/have/having/ |
| 13:51 | <hsivonen> | (though countries that don't have a database of all citizens manage to get by much better than one would expect) |
| 13:51 | takkaria | is pretty happy about that |
| 13:51 | <erlehmann> | in before anyone mentions IBM |
| 13:51 | <hsivonen> | takkaria: about countries managing to do stuff without a proper database? |
| 13:53 | <erlehmann> | i think germany has no central database. |
| 13:54 | <erlehmann> | but thousands of local ones. so in theory, the information is there. |
| 13:55 | <takkaria> | hsivonen: about states not having centralised databases of everyone |
| 13:55 | <erlehmann> | this is funny http://en.wikipedia.org/wiki/Resident_registration#France |
| 13:56 | <erlehmann> | >current residency is often verified by showing bills relating to the current home |
| 13:56 | <erlehmann> | haha |
| 13:59 | <hsivonen> | takkaria: it's kinda weird not to have one in any country that has elections, taxes or social security benefits (or a conscription-based army) |
| 14:03 | <hsivonen> | erlehmann: heh. the wikipedia entries for Finland and Sweden have obviously been copied and pasted from each other (as probably the relevant legislation, too). |
| 14:04 | <hsivonen> | subentries on the Resident_registration page, that is |
| 14:08 | <kennyluck> | Does anyone here know how reality-matching the ISO/IEC C standard is? as compared to WHATWG HTML LS. |
| 14:13 | <Philip`> | kennyluck: I think it's reality-matching enough that programmers and compiler developers refer to it to authoritatively determine the correctness of programs and of compilers, mostly successfully (though there's certainly a few cases where it's vague and/or incorrect so you have to rely on common interpretation rather than just the text) |
| 14:15 | <Philip`> | kennyluck: It leaves a lot of stuff as implementation-defined (e.g. the number of bits in a byte) even when pretty much every implementation in the world makes the same decision, so (unlike the general direction of HTML) it lets you write conforming compilers that probably won't work on much real-world code |
| 14:16 | <Philip`> | kennyluck: and compilers implement a load of non-standard features too |
| 14:16 | <kennyluck> | Philip`, it sounds like CSS2.1 :p |
| 14:17 | <kennyluck> | Philip`, is there a supplementary document that talks about these *decisions*? |
| 14:17 | <Philip`> | kennyluck: so I suppose it's less of an attempted comprehensive implementation description than HTML is, and more of a almost-but-not-quite-large-enough-to-be-practical subset that programmers and compilers can agree on |
| 14:19 | <Philip`> | kennyluck: The decisions like the number of bits per byte? It differs by compiler and by platform, and I don't think anyone's collected all the data together, though in theory the compilers ought to document all their choices for implementation-defined behaviour |
| 14:19 | <Philip`> | (Undefined behaviour is different and isn't expected to be documented anywhere) |
| 14:19 | <kennyluck> | Philip`, I mean the hidden rules that all implementation agree upon. |
| 14:20 | <Philip`> | http://gcc.gnu.org/onlinedocs/gcc/C-Implementation.html |
| 14:20 | <kennyluck> | Yes, I was talking about undefined behaviors. |
| 14:21 | <Philip`> | The standard has an appendix listing "unspecified" and "undefined" and "implementation-defined" cases |
| 14:21 | kennyluck | just read an article that mentions the fact that the standard makes NULL == 0 implementation-defined, but he wonders if there's real world exception. |
| 14:22 | <Philip`> | Nobody agrees on handling of undefined behaviour - that's things like "(x++) + (x++)" and even a single compiler with different optimisation settings gives different results |
| 14:24 | <izhak> | Hi. Could you please guys clarify how does most popular engines deal with encoding when parsing html, do they transfer any encoding to a single one to work with it when parsing, or parsing process deals with different encodings? |
| 14:25 | <hsivonen> | Philip`: it seems terrible that undefined stuff compiles instead of causing errors |
| 14:29 | <Philip`> | kennyluck: The standard says 0 is a null pointer constant, and NULL is a null pointer constant, but doesn't say there's only a single null pointer constant or that they're all equal |
| 14:30 | <Philip`> | They'll be equal if converted to pointers, but == will compare them as integers |
| 14:30 | <Philip`> | so I think I agree that expression is implementation-defined |
| 14:31 | <erlehmann> | Philip`, the number of bits in a byte? o.0 |
| 14:31 | <Philip`> | but I can't conceive of any non-intentionally-weird implementation where NULL would not equal 0 |
| 14:31 | <erlehmann> | does x64 have different bytes? |
| 14:31 | <erlehmann> | or what. |
| 14:31 | <kennyluck> | erlehmann, that one has a known exception. |
| 14:32 | <erlehmann> | kennyluck, i wish for a compiler that translates „undefined behaviour“ to „delete everything“ |
| 14:33 | <kennyluck> | erlehmann, here you go http://www.velocityreviews.com/forums/showpost.php?p=2435265&postcount=10 |
| 14:33 | <Philip`> | hsivonen: In general it's impossible to prove that it's undefined, like in "int f(int* a, int* b) { return (*a)++ + (*b)++; }; int x, y; ... f(&x, &y) /* okay */; f(&x, &x) /* undefined */;" |
| 14:41 | <annevk> | oops, almost forgot to define createRange! |
| 14:45 | <Philip`> | kennyluck: http://stackoverflow.com/questions/2597142/when-was-the-null-macro-not-0/2597232#2597232 has some examples where a null pointer is not 0x0000..., though that doesn't mean NULL will be a non-zero integer |
| 14:46 | <Philip`> | (at least per C99) |
| 14:47 | <kennyluck> | interesting |
| 14:50 | <Philip`> | I suppose a particularly security-conscious implementation could do pointer-to-int conversion by generating an incrementing ID number for each new pointer it encounters |
| 14:51 | <Philip`> | and then do int-to-pointer conversion by looking up the pointer with that ID number, or returning the null pointer if the ID wasn't valid |
| 14:52 | <Philip`> | and then it could #define NULL 0xDEADBEEF so it's easily distinguishable in memory dumps etc, and never allocate that as an ID number so it always returns a null pointer |
| 14:53 | <Philip`> | and I think that would be a conforming C implementation |
| 14:54 | <zewt> | in practice any system where null isn't false would be fatally broken |
| 14:55 | <Philip`> | zewt: The NULL macro, or a null pointer? |
| 15:02 | <annevk> | note to self: check if HTML has garbage collection rules for all objects that can do requests |
| 15:02 | <annevk> | e.g. <script>, <img>, <link> |
| 15:03 | <annevk> | and if not, file bugs, I guess |
| 15:03 | <zewt> | Philip`: don't see how they could possibly be different |
| 15:04 | <zewt> | annevk: where is that ever non-transparent? |
| 15:05 | <Philip`> | zewt: NULL can be an integer type, not a pointer type, so different rules apply for conversions and comparisons etc |
| 15:06 | <zewt> | but it still needs to be a false value, whatever its type |
| 15:06 | <zewt> | 0xdeadbeef isn't false |
| 15:07 | <zewt> | so every program that says if(p) instead of if(p != NULL) would break |
| 15:07 | <annevk> | zewt, what? |
| 15:08 | <zewt> | annevk: when are effects of gc rules visible to scripts/users in a way that should be specified? |
| 15:08 | <zewt> | specifying it for workers makes sense, of course, since it's very visible |
| 15:09 | <annevk> | it's specified for XMLHttpRequest, EventSource, WebSocket |
| 15:09 | <Philip`> | zewt: If p is a pointer, "if(p)" will convert the pointer to an integer, which you want to be 0 (which I don't think the standard requires); that doesn't mean NULL has to be the integer 0, it could be a different integer that also happens to become a null pointer when converted to a pointer type, without breaking that code |
| 15:10 | <annevk> | apparently there have been compatibility issues in this area as well |
| 15:10 | <annevk> | it also matters for navigation |
| 15:12 | <zewt> | don't think there's any way NULL could be something other than 0 while having boolean expressions behave normally |
| 16:10 | <annevk> | why is it important that "concept-range-bp-node" is a separate concept Ms2ger, AryehGregor? |
| 16:10 | <annevk> | seems to be the same as concept-node |
| 16:11 | <annevk> | we should maybe introduce "start node", "start offset", "end node", and "end offset" as concepts |
| 16:29 | <zewt> | surprising how many people just don't understand the DOM Events model, even people working on specs |
| 16:30 | <zewt> | the webgl guys had similar confusion, by having side-effects of having an event handler registered at all (which fortuantely they fixed) |
| 16:33 | <Ms2ger> | "spam a note with <span>" |
| 16:35 | <zewt> | it's also discouraging that people who should know better go "come on let's just add a few little unexpected side-effects to a core API" :| |
| 16:36 | <dglazkov> | good morning, Whatwg! |
| 16:41 | <kennyluck> | good morning, Whatwg! |
| 16:46 | <hsivonen> | aargh. is polyglot on its way to pubrules if it's up to TimBL? http://lists.w3.org/Archives/Public/www-tag/2011Sep/0008.html |
| 16:49 | <timeless> | hsivonen: g+ has spam puppets? |
| 16:49 | <timeless> | no fair, i can't see them, i got kicked off g+ :) |
| 16:50 | <timeless> | hsivonen: fwiw, the US doesn't really have a database of all US citizens |
| 16:51 | <timeless> | at best it has a database of all people who have filed taxes |
| 16:51 | <timeless> | but chlidren don't file taxes generally, and you can certainly get passports for them |
| 16:52 | <timeless> | on the topic of current residence, i have at least one account w/ 200cad currently suspended because i changed too many fields at the same time |
| 16:52 | <zewt> | birth certificates |
| 16:53 | <timeless> | so i get to scan a utility bill at some point as step 1/5 to try to unlock the account |
| 16:53 | <timeless> | (this isn't actually a bad thing, i don't blame them for thinking my account was stolen, the actions i took sure make it look like i stole the account) |
| 16:53 | <timeless> | zewt: birth certificates aren't actually well backed by a database either fwiw |
| 16:54 | <timeless> | at least not for people going back to say 1901 |
| 16:54 | <timeless> | it might be the case that they're well backed to say 1960 |
| 16:54 | <zewt> | but for people who are children now :) |
| 16:54 | <hsivonen> | timeless: it sucks that they kicked you out, though I had been wondering what took them so long |
| 16:54 | <timeless> | hsivonen: yeah, i was wondering that too :) |
| 16:54 | <timeless> | but i don't think they have a proper database that pairs birth to death |
| 16:55 | <timeless> | which means that if your child is born in one place, and dies elsewhere, i don't think the database record is required to "close" :) |
| 16:55 | <hsivonen> | timeless: did this cause collateral damage to your use of other Google services? |
| 16:55 | <timeless> | hsivonen: my picasa account is suspended |
| 16:55 | <hsivonen> | timeless: and are they now blocking you from seeing public posts if you are logged in? |
| 16:56 | <timeless> | i had to liberate the data |
| 16:56 | <timeless> | otherwise the consequences are minimal |
| 16:56 | <hsivonen> | timeless: did Picasa have a real names policy when you signed up? |
| 16:56 | <timeless> | no |
| 16:56 | <timeless> | the account was tied to timeless |
| 16:56 | <timeless> | and i have an account in my real name too |
| 16:56 | <timeless> | i chose not to use that one for the data.. |
| 16:57 | <hsivonen> | Google is burning goodwill pointlessly here |
| 16:57 | <timeless> | # [15:11] <hsivonen> erlehmann: heh. the wikipedia entries for Finland and Sweden have obviously been copied and pasted from each other (as probably the relevant legislation, too). |
| 16:57 | timeless | chuckls |
| 16:57 | <timeless> | s/ls/les/ |
| 16:59 | <timeless> | kennyluck: my favorite wrt C was the definition of `main` |
| 16:59 | <timeless> | it basically said `this set of main() functions are vaild, plus any documented by the compiler to be valid` |
| 16:59 | <hsivonen> | timeless: it also sucks that the rules are different for Famous People. Madonna gets to use "." as her last name (which is dumb--shouldn't they support one-token names for real Indian names anyway) and MG Siegler doesn't need to expand MG |
| 16:59 | <hsivonen> | (both Madonna and MG have verified accounts that seem to break the rules) |
| 16:59 | <timeless> | hsivonen: oh, she used "."? |
| 17:00 | timeless | should try that |
| 17:00 | <zewt> | why would google support people using names that are literally impossible to search for, heh |
| 17:02 | <timeless> | zewt: that soudns like my opinion on anything outside the us-ascii range :) |
| 17:02 | <hsivonen> | anyway, the way Google is treating other people makes me less enthusiastic to use G+ even though I have a conforming name |
| 17:02 | timeless | wonders if this browser has bookmarked the page talking about the Poker player with the nullish byte in his name |
| 17:02 | <hsivonen> | (conforming name in the ASCII range even) |
| 17:03 | <timeless> | hsivonen: btw, i'm quite glad you do have such a name |
| 17:03 | <zewt> | hsivonen: the fact that they go "we care about privacy!" and then immediately proceed with a privacy-demolishing names policy is borderline hilarious |
| 17:03 | <timeless> | just as i was glad to move from Döbelninkatu to Hietaniemenkatu |
| 17:03 | <timeless> | that first address is still stuck in some places |
| 17:04 | <timeless> | iirc my .eu domain might have it |
| 17:04 | <timeless> | because changing that field was basically impossible :) |
| 17:04 | <hsivonen> | timeless: I went downhill on that point. I'm now on Strömbergintie |
| 17:04 | <timeless> | eww |
| 17:04 | <timeless> | has it caused any pain? |
| 17:05 | <zewt> | at least for locations you can leave off combining marks and stuff will *usually* still get to where it goes |
| 17:05 | <zewt> | of course, sites that try to be "smart" and "verify" addresses may give a headache, but those give everyone headaches |
| 17:06 | <hsivonen> | timeless: no pain so far, but an American company has sent me postal mail to Strömbergintie because they were too incompetent to pull UTF-8 out of Paypal's system correctly |
| 17:06 | <timeless> | sounds familiar |
| 17:06 | <timeless> | i presume that posti.fi is used to handling such things? :) |
| 17:06 | <hsivonen> | (I'm told that Paypal's API does the right thing) |
| 17:06 | <timeless> | wow |
| 17:06 | <zewt> | heh i remember somewhere seeing mail forwarding by russian post offices or something, where they use tables to de-mojibake printed addresses |
| 17:07 | <hsivonen> | timeless: the mailing reached me, alas |
| 17:07 | <timeless> | so far paypal has caused me minor pain because W3C is silly |
| 17:07 | <hsivonen> | timeless: I wish it had been returned to the sender |
| 17:07 | <timeless> | W3C's TPAC billing agency is in Paris |
| 17:07 | <timeless> | the event is in California |
| 17:07 | <timeless> | and the system here automatically fills in (and PINS) the biller for reimbursement |
| 17:08 | <timeless> | and then it sanity checks that the currency (USD) matches the location (EUR/Paris) |
| 17:08 | <timeless> | and complains when it doesn't |
| 17:08 | <timeless> | grr |
| 17:08 | <hsivonen> | (not because of the botched address but because I didn't want to order the thing. it was added to my order in order to have a pretext to charge extra) |
| 17:08 | <timeless> | heh |
| 17:09 | <hsivonen> | conclusion: don't do business with Kagi. If you do do business with Kagi, be sure to delete extra unsolicited items from your order before checking out |
| 17:09 | <zewt> | and beware of godaddy, heh |
| 17:10 | <zewt> | they set ssl certificates to auto-renew without asking, and the only way to find out is through obscure menus |
| 17:10 | <timeless> | zewt: hrm |
| 17:10 | <timeless> | do they at least send you your new certs somehow? |
| 17:11 | <zewt> | they renew your cert and charge you full price ($50) without permission |
| 17:11 | <zewt> | http://www.jwz.org/blog/2011/09/godaddy/ another case of godaddy keepin' it classy |
| 17:11 | <zewt> | need to find another usable budget CA |
| 17:12 | <timeless> | zewt: nice |
| 17:12 | <timeless> | <jwz> Next time I'll go with DigiNotar, I've been hearing good things. |
| 17:12 | <zewt> | heh |
| 17:13 | <zewt> | i hear you can just get a nice universal wildcard certificate and not have to worry about certs ever again |
| 17:13 | <timeless> | i was surprised they allowed issuing "*.*.com" and "Foo Space Bar" |
| 17:13 | <zewt> | do any browsers actually accept *.com, though? |
| 17:13 | <timeless> | given that afaik no browser will allow more than one * in a Cert |
| 17:13 | <timeless> | dunno, *.com might be legal |
| 17:13 | <timeless> | *.*.anything is not |
| 17:14 | <zewt> | i know at least some ssl code explicitly rejects wildcards at that level |
| 17:14 | <zewt> | sad, of course, that that actually *mattered* |
| 17:14 | <timeless> | it's possible that *.com is only legal if <com> is appropriately listed in the ccTLD cookie file |
| 17:15 | timeless | doesn't really feel like generating the necessary certs to test *.com |
| 17:15 | <timeless> | otoh, if you don't mind a delay in payment, i'll gladly send you 5usd via paypal if you test that condition and publish your findings |
| 17:16 | <timeless> | (by delay, i mean somewhat significant, as i have to get my paypal accounts in order...) |
| 17:16 | timeless | hates passwords |
| 17:17 | <timeless> | http://www.jwz.org/blog/2011/09/godaddy/#comment-93938 |
| 17:17 | <AryehGregor> | annevk5, I guess concept-range-bp-node could be the same as concept-node, yeah. Not sure why they're different, offhand. |
| 18:15 | <Ms2ger> | Looks like css3-values just missed the 5-year mark |
| 18:19 | <timeless> | yeah well |
| 18:19 | <timeless> | that's hardly a big accomplishment :) |
| 18:46 | <gsnedders> | XMLDocument is used by Dojo to sniff whether to be case-sensitive for comparisons on not in Presto… in a buggy manner. |
| 18:46 | <Ms2ger> | \o> |
| 18:49 | <gsnedders> | Removing XMLDocument will only break stuff that relies upon case-senitivity in XHTML documents, I think, so the breakage there is minimal. |
| 18:49 | <gsnedders> | (Though mainly because the behaviour on the Presto code-path is already so broken.) |
| 20:02 | <timeless> | has anyone here played w/ QoS in the last 8 years? |
| 20:02 | <zewt> | only simple token bucket stuff |
| 20:04 | <timeless> | i was wondering if anyone knows of anything related which is actually usefully deployed |
| 20:04 | <timeless> | especially if any have been deployed by telco's or isp's |
| 20:05 | <timeless> | (I2 deployments aren't helpful) |
| 20:05 | <zewt> | well, comcast qos's everyone |
| 20:05 | <zewt> | that's what "speedboost" is |
| 20:05 | <timeless> | well, does it do it in a useful way? |
| 20:05 | <timeless> | rather |
| 20:05 | <timeless> | if one customer sets up 3 connections for three different purposes |
| 20:05 | <zewt> | my connection is about 2mb/sec for a minute or two, then tapers down to 750k/sec |
| 20:05 | <timeless> | can that customer influence behavior by using any QoS tag |
| 20:06 | <zewt> | dunno what you're asking |
| 20:06 | <timeless> | suppose you have 3 connections, one you'd like to download "eventually" |
| 20:06 | <timeless> | one you'd like to watch "in real time" |
| 20:06 | <timeless> | one you'd like to get "as soon as possible, but not necessarily in real time" |
| 20:06 | <timeless> | assume that each connection is backed by 10mb of data |
| 20:07 | <zewt> | a bit too specific for an ISP to do it... |
| 20:07 | <timeless> | is there a way for you to adjust your requests for the 3 resources such that instead of each being given "2mbps for 120s ; 750kbps" |
| 20:07 | <timeless> | you get something else... |
| 20:07 | <zewt> | well, linux qos can do that, but i don't know about it being used in any broad way |
| 20:07 | <timeless> | possibly "100kbps initially" on that first request |
| 20:08 | <timeless> | right, i believe that various QoS things are capable of doing this |
| 20:08 | <timeless> | what interests me is whether anyone has any useful ISPs who have any deployed QoS one can *talk* to |
| 20:08 | <zewt> | (bucket data based on ip/port/tcp flags/whatever) |
| 20:09 | <zewt> | don't know |
| 20:09 | <timeless> | (yes, "useful ISPs" is probably a nonsequitor) |
| 20:11 | <hsivonen> | was SIP designed by people whose top priority was lawful intercept or how else did they manage to make it unencrypted in practice? |
| 20:11 | <timeless> | hsivonen: you'd prefer that they bake in default certs? :) |
| 20:12 | <timeless> | i think the general idea with most things is "develop the basic protocol in a debuggable manner" "apply crypto as an independent layer" |
| 20:12 | <timeless> | it makes things easier to design/debug |
| 20:12 | <timeless> | and you just pray that someone else solves crypto for you |
| 20:13 | <hsivonen> | timeless: except it looks like the crypto layer didn't get applied to SIP |
| 20:13 | <timeless> | um |
| 20:13 | <timeless> | there's definitely a sips |
| 20:13 | <timeless> | nokia was using it |
| 20:13 | <timeless> | http://wiki.freeswitch.org/wiki/SIP_TLS |
| 20:13 | <zewt> | "lawful intercept" sure sounds like a bad euphemism for "spying" |
| 20:13 | <hsivonen> | if you want to turn a legacy Helsinki-area landline number to a VoiP number with Elisa, their solution uses unencrypted SIP |
| 20:13 | <timeless> | by that, i mean that my n900 let me make VoIP calls using sips w/ nokia as my sip provider |
| 20:14 | <timeless> | you can ask sp3000 to play with it since he's still there |
| 20:14 | <timeless> | hsivonen: that sounds more like an implementation detail that the vendor didn't pay for |
| 20:15 | <timeless> | paying here could just be "[x] provide SSL for SIP" which translates into "1 person week" "get Cert, configure Cert, |
| 20:15 | <timeless> | write instructions for using SIPS" |
| 20:16 | <timeless> | hrm, possibly 2 other people who translate those instructions into Fingrish or Finnish or Swedish |
| 20:16 | <timeless> | depending on ... |
| 20:16 | <timeless> | http://translate.google.com/translate?hl=en&sl=et&u=http://mipc.telegrupp.ee/paketid/elisa-sip-trunk&ei=H3JmTq3IAYS80AHHmJn-CQ&sa=X&oi=translate&ct=result&resnum=2&ved=0CCIQ7gEwAQ&prev=/search%3Fq%3Delisa%2Bsip%26hl%3Den%26biw%3D702%26bih%3D780%26prmd%3Divns |
| 20:16 | <timeless> | is all i can find |
| 20:17 | <Philip`> | timeless: Would you count things like http://www.paetec.com/products-services/data/mpls-vpn/overview.html where someone provides a VPN service that you can connect your networks to, with QoS prioritisation support? |
| 20:17 | <hsivonen> | timeless: Well, it can't be a coincidence that some admins want to limit SIP to a VPN |
| 20:18 | <hsivonen> | timeless: presumably SIP along sucks enough that it can't be allowed to cross the public Internet without a VPN |
| 20:18 | <timeless> | Philip`: hrm |
| 20:18 | <timeless> | hsivonen: well, nokia certainly didn't deploy SIPs-less for its emoployees |
| 20:18 | <timeless> | s/emo/em/ |
| 20:18 | <timeless> | we got the SIPS version only :) |
| 20:19 | <timeless> | and we could use it freely from anywhere in the world |
| 20:19 | <timeless> | in fact, we were encouraged to do so |
| 20:19 | <hsivonen> | timeless: isn't SIPS encrypted just for signaling? what about the speech payload? |
| 20:19 | <timeless> | (the same more or less applies to RIM's solution, the underlying details i don't know) |
| 20:19 | <Philip`> | (I'd guess QoS is pretty useless unless it's end-to-end, and it's really hard getting multiple providers to cooperate, and if you want a single provider end-to-end network then it seems more likely to be an MPLS VPN than plain old IP, as far as I'm aware) |
| 20:19 | <timeless> | hsivonen: dunno, i don't have a working SIPS provider anymore, talk to sp3000 |
| 20:20 | <timeless> | Philip`: well |
| 20:20 | <timeless> | lemme outline a use case |
| 20:20 | <timeless> | 3 users set up connections at a home in NAm |
| 20:20 | <hsivonen> | timeless: according to the wiki page you linked to, SIPS is for signaling |
| 20:20 | <timeless> | they can choose from Rogers, Bell, AT&T, and a couple of others |
| 20:20 | <hsivonen> | timeless: you also need SRTP if you want encrypted payloads |
| 20:20 | <timeless> | hsivonen: i'd certainly hope that the nokia deployed thing did the right thing |
| 20:21 | <timeless> | if not, well... |
| 20:21 | <hsivonen> | timeless: so if the other end of your call doesn't jump through all these hoops, you end up having to use the least encrypted mode that works |
| 20:21 | <timeless> | but really, i don't have logs of my sips initiated sessions |
| 20:21 | <timeless> | hsivonen: well |
| 20:21 | <timeless> | in our case, typically it was SIPS => POTS |
| 20:21 | <timeless> | which meant we were supposed to be encrypted between our n900 and the POTS gateway |
| 20:22 | <timeless> | which then hopefully was " |
| 20:22 | <timeless> | magically secure" |
| 20:22 | <timeless> | along the POTS path |
| 20:22 | <timeless> | (government wiretapping included at no extra charge) |
| 20:22 | <hsivonen> | timeless: Elisa says that if you also use Elisa as your ISP, the SIP hop is as secure as the POTS part |
| 20:22 | <timeless> | (and national laws against anyone else doing the wiretapping...) |
| 20:22 | <hsivonen> | timeless: but if you don't use Elisa as your ISP, you don't know (a priori) where the SIP hop travels |
| 20:23 | <timeless> | hsivonen: this is Elisa DSL? |
| 20:23 | <timeless> | or Elisa Cellular? |
| 20:23 | <timeless> | elisa doesn't do Cable, right? |
| 20:23 | <hsivonen> | timeless: DSL or fiber |
| 20:23 | <timeless> | and presumably they aren't talking about SIP while on <someone-else's-wifi> |
| 20:24 | <hsivonen> | timeless: it seems they do cable now, too, at least in some locations |
| 20:24 | <timeless> | joy |
| 20:24 | <timeless> | so they're like Rogers here |
| 20:24 | <timeless> | generally speaking, Cable = shared local link |
| 20:24 | <hsivonen> | timeless: (well, their subsidiary does) |
| 20:25 | <hsivonen> | also, SIP is terribly brittle compared to Skype |
| 20:25 | timeless | ponders |
| 20:25 | <timeless> | I've used Nokia SIPS, Skype, Gtalk, and BlackBerry MVS to talk to my parents |
| 20:25 | <timeless> | it turns out they weren't particularly impressed w/ the quality of some of those |
| 20:26 | <hsivonen> | as in, the Elisa-provided SIP box didn't work with network arrangements than Skype was fine with |
| 20:26 | <timeless> | but they didn't give me enough pushback for me to know about it |
| 20:26 | <timeless> | oh |
| 20:26 | timeless | wouldn't call that brittle |
| 20:26 | <timeless> | brittle typically means "likely to break after/in deployment" |
| 20:26 | <hsivonen> | timeless: well, picky |
| 20:26 | <timeless> | yeah, picky is a better word |
| 20:27 | timeless | is still trying to think of the more correct term but has blanked |
| 20:27 | <timeless> | anyway, Skype was designed for all sorts of rude network topographies |
| 20:28 | <timeless> | ("robust") |
| 20:28 | <hsivonen> | well, it sucks that the IETF protocol is neither secure nor robust |
| 20:28 | <timeless> | being robust tends to require overengineering and additional endpoints |
| 20:29 | <hsivonen> | the only redeeming factors is federation and open spec |
| 20:29 | timeless | thinks skype may have resolved some of those points recently |
| 20:29 | <hsivonen> | but federation doesn't really matter when SIP is only used together with a POTS bridge |
| 20:31 | <timeless> | right |
| 20:32 | <timeless> | iirc i was able to use nokia's sips to call zakim⊙wo |
| 20:32 | <hsivonen> | timeless: fwiw, AFAICT, all the three major operators now offer fiber, cable, DSL and 3G |
| 20:32 | <timeless> | which was nice, i.e. i had both POTS bridging and federation |
| 20:33 | <hsivonen> | though curiously, their regulatory statuses aren't symmetrical |
| 20:33 | <timeless> | (technically Skype has both) |
| 20:33 | <timeless> | lol |
| 20:33 | <timeless> | this is Elisa, Sonera, Welho, right? |
| 20:33 | <hsivonen> | timeless: right |
| 20:33 | <timeless> | Welho is treated as a monopoly and the others aren't? |
| 20:34 | <hsivonen> | timeless: so Elisa has to rent their phone copper to everyone but Welho doesn't need to rent their fiber |
| 20:34 | <timeless> | nice |
| 20:34 | <hsivonen> | timeless: dunno if Welho is required to rent their cable |
| 20:34 | <timeless> | yeah, i haven't quite figured out the story of the operators here |
| 20:34 | <hsivonen> | (I called the regulator to find out that Welho isn't required to rent out their fiber) |
| 20:35 | <timeless> | we have Bell and Rogers as the original Phone/Cable providers |
| 20:35 | <timeless> | hsivonen: nice |
| 20:35 | timeless | remembers calling the US Census Bureau from Finland to ask about a small town |
| 20:36 | <timeless> | http://maps.google.com/maps/place?q=Guam,+Missouri,+United+States&hl=en&ftid=0x88786da513501111:0xd3f08770817ac3c8 |
| 20:38 | <timeless> | ah, http://en.wikipedia.org/wiki/Guam,_Missouri is the relevant page |
| 20:38 | <timeless> | hsivonen: wanna guess why we looked that one up? :) |
| 20:39 | <timeless> | http://translate.google.com/translate?hl=en&sl=fi&u=http://www.elisa.fi/saunalahti/asiakaspalvelu/740/saunalahti-nettipuhelin/10100/kaytto-voip-tai-lankapuhelimella/&ei=f3JmTu2vOMG80AGe_OSoCg&sa=X&oi=translate&ct=result&resnum=1&ved=0CBsQ7gEwAA&prev=/search%3Fq%3Dsip%2Bsite:elisa.fi%26hl%3Den%26biw%3D1440%26bih%3D809%26prmd%3Divns |
| 20:39 | <timeless> | seems like it was your SIP thing.. |
| 20:40 | <timeless> | oh joy |
| 20:40 | <hsivonen> | timeless: you thought Guam, Missouri was inconsistent data? |
| 20:40 | <timeless> | it's always nice to see devices that ask you to *confirm* your password |
| 20:40 | <timeless> | hsivonen: the n900 original firmware had two entries for Guam |
| 20:40 | <timeless> | one was for Guam, USA |
| 20:41 | <hsivonen> | timeless: that's a different SIP service. |
| 20:41 | <timeless> | we wanted to confirm that no one in their right mind would want to have that Guam listed in their world map |
| 20:42 | <timeless> | so, one of the steps was determining a population :) |
| 20:48 | <timeless> | hsivonen: i like how some of them have you entering an IP into a field |
| 20:48 | <timeless> | Outbound Proxy: 195.197.95.4 |
| 20:56 | <timeless> | hsivonen: speaking of which |
| 20:57 | <timeless> | do you have your elisa-sip service, or are you just researching? |
| 20:57 | <timeless> | http://www.w3.org/2006/tools/wiki/Zakim-SIP |
| 20:58 | <timeless> | i'm wondering if you can actually call zakim (zakim⊙vwo) using it |
| 20:59 | <hsivonen> | timeless: the would-be user for the SIP service was my grandmother, but she died before my dad and I got the SIP thing working, so we returned it without getting it to work |
| 20:59 | <timeless> | oh, sorry to hear that |
| 21:01 | <hsivonen> | anyway, I learned that there now exists at least one building in Helsinki that doesn't have phone copper |
| 21:02 | <hsivonen> | which meant that the phone options were GSM or VoIP |
| 21:05 | <hsivonen> | throughout the episode, Elisa's customer service was terrible |
| 21:06 | <timeless> | bad customer service in Finland |
| 21:06 | <timeless> | i don't believe it |
| 21:06 | <timeless> | oh wait, i lived there for 5 years |
| 21:06 | <timeless> | :) |
| 21:06 | <timeless> | good customer service is generally unheard of there :) |
| 21:06 | <hsivonen> | well, Welho's customer service was fine, when I eventually arranged Welho's fiber for my grandfather for the same apartment |
| 21:07 | <timeless> | fine!=good fwiw |
| 21:07 | <timeless> | although `fine` is a typical finnish description |
| 21:07 | <timeless> | fine is somewhere between a D+ and a C |
| 21:07 | <hsivonen> | well, the Welho people answered all my questions and AFAICT answered them correctly and fast |
| 21:08 | <timeless> | did they suggest answers to questions you didn't think to ask but would have needed to ask in later round trips? |
| 21:08 | <hsivonen> | whereas the Elisa people didn't answer or answered incorrectly or were unreasonably slow |
| 21:08 | <timeless> | my experiences w/ elisa have been bad |
| 21:09 | <timeless> | and they sent me another bill recently |
| 21:09 | timeless | grumbles |
| 21:09 | <hsivonen> | timeless: no, but there isn't really much to ask if you are buying Internet connectivity itself from a company that owns the fiber |
| 21:10 | timeless | nods |
| 21:10 | <timeless> | this mostly applies to cases where you end up doing 10 round trips where there were some questions they could have asked or things they could have told you to save you round trips |
| 21:11 | <timeless> | giving you the list of hostnames for all possible fields and suggested alternate ip addresses in case your device sucks |
| 21:11 | <timeless> | are ways of doing that |
| 21:11 | <timeless> | (or offering to do that, instead of just flooding you with them) |
| 21:12 | <hsivonen> | fwiw, our apartment has Elisa's fiber, and purchasing that was uneventful, because, again, there's nothing special about buying just connectivity from the company that own the fiber |
| 21:12 | <timeless> | how long did it take for that? |
| 21:12 | <hsivonen> | I don't remember |
| 21:12 | <timeless> | w/ rogers it sounds like it can take a week for a guy to come to "install" your cable |
| 21:13 | <hsivonen> | with Welho, both fiber and cable are instant-on |
| 21:13 | <hsivonen> | I can't recall if Elisa's fiber was instant or took a day or so |
| 21:13 | <timeless> | "bring your own router/box"? |
| 21:13 | <timeless> | for my Elisa DSL, they had to schedule a guy to come |
| 21:13 | <timeless> | i think he did some line tests (since the building is rather old, that isn't inappropriate) |
| 21:13 | <hsivonen> | timeless: Ethernet in the apartment |
| 21:14 | <timeless> | and basically deployed a filter or two and plugged in a router |
| 21:14 | <hsivonen> | timeless: so own router required if more than one device |
| 21:14 | <timeless> | ethernet in apartment? nice |
| 21:14 | <timeless> | so no stupid extra box |
| 21:15 | <timeless> | i think the strange thing w/ finland is that the classic phone system doesn't use typical RJ- connectors |
| 21:15 | <timeless> | http://en.wikipedia.org/wiki/File:Telefonkontakt.jpg |
| 21:15 | <hsivonen> | our apartment and my grandfather's apartment have the same arrangement: there's a electricity/telecom cabinet in the apartment that has an RJ-45 to the outside world and a bunch of RJ-45 sockets that connect to wall outlets around the apartment |
| 21:16 | <hsivonen> | so you either use a short Ethernet cable to connect the outside world to the wall socket of your choice |
| 21:17 | <hsivonen> | or you put a router in the cabinet (which also has electricity sockets for this precise purpose) and connect the router to several wall sockets |
| 21:19 | <hsivonen> | timeless: our telecom cabinet has also outgoing RJ-11 |
| 21:20 | <timeless> | is there anything that actually talks about the rj-11 3prog thing used in finland? |
| 21:20 | <timeless> | the wikipedia page just says that it's a national standard |
| 21:20 | <timeless> | but no explanation of why |
| 21:20 | <hsivonen> | timeless: I haven't bothered to find out if there's actually a wire behind RJ-11 |
| 21:20 | <timeless> | lol |
| 21:21 | <hsivonen> | timeless: the 3-pronged thing is ancient since at least the 1950s |
| 21:21 | <hsivonen> | timeless: dunno where it came from |
| 21:21 | <hsivonen> | our apartment doesn't have the 3-pronged socket anywhere |
| 21:21 | <hsivonen> | and the wall sockets are RJ-45 |
| 21:22 | <hsivonen> | so I suppose if one bought service to the outgoing RJ-11 (if there's wire!), one would have to use the RJ-45 cabling from there onwards |
| 21:26 | <timeless> | http://en.wikipedia.org/wiki/BS_6312 is distressing |
| 21:26 | <timeless> | (especially: Connector on phone) |
| 21:28 | <hsivonen> | timeless: hooray pointlessly slightly different standard |
| 21:28 | <timeless> | yep |
| 21:30 | <timeless> | oh awesome |
| 21:30 | <timeless> | check out IPlate => BT Broadband Accelerator |
| 21:30 | <zewt> | heh uk wiring is pretty hilarious |
| 21:30 | <zewt> | i can't help but laugh whenever i see uk power plugs |
| 21:31 | <timeless> | i tend to cry |
| 21:31 | <zewt> | i don't have to use them so |
| 21:31 | <timeless> | i have 1 generic => UK adapter, and 1 fairly standard laptop => UK cable |
| 21:32 | <hsivonen> | the design of the UK electricity plug is mind-blowing |
| 21:33 | <hsivonen> | it's not just different shape or different current/voltage. the electrical design is different |
| 21:33 | <hsivonen> | the wiring topology inside apartments is different |
| 21:34 | <hsivonen> | and the fuses are in different places |
| 21:35 | <timeless> | http://en.wikipedia.org/wiki/AC_power_plugs_and_sockets is only 34 printed pages :) |
| 21:35 | <timeless> | hsivonen: grr |
| 21:35 | <zewt> | what we need |
| 21:35 | <zewt> | is for one country to take over the whole world for one year |
| 21:35 | <timeless> | did i mention that when my movers came to packet my apartment in Helsinki, they packed my spare Fuses? |
| 21:35 | <timeless> | ... |
| 21:35 | <timeless> | and then a fuse blew |
| 21:35 | <zewt> | just long enough to force everyone onto the same power standards, paper standards, and so on |
| 21:35 | <zewt> | then topple it and go back |
| 21:36 | <timeless> | zewt: en.wikipedia.org/wiki/The_Mouse_That_Roared ? |
| 21:36 | <zewt> | my bookshelf won't fit A4 books :( |
| 21:37 | <timeless> | heh |
| 21:38 | <zewt> | also it's been over 48 hours, isn't it time for microusb to be declared obsolete and yet another usb connector to come into fashion |
| 21:38 | <gsnedders> | hsivonen: Well, the aim with the wiring was to limit the amount of wires needed within houses, AFAIK |
| 21:39 | <hsivonen> | timeless: hah. fortunately, this apartment doesn't need spare fuses |
| 21:39 | <timeless> | hsivonen: the fuses arrived here in Toronto |
| 21:39 | <zewt> | i'd sooner not live somewhere old enough to not have breakers |
| 21:39 | <timeless> | along with assorted other unhelpful items |
| 21:39 | <hsivonen> | timeless: are they compatible with what you need there? |
| 21:39 | <timeless> | including a classic-finnish-key (actually, two!) |
| 21:40 | <timeless> | and a couple of finnish light bulbs |
| 21:40 | <timeless> | hsivonen: of course not |
| 21:40 | <timeless> | the civilized world uses circuit breakers! |
| 21:40 | <hsivonen> | timeless: does Canada have different light bulb sockets? |
| 21:40 | <timeless> | hsivonen: voltage issue |
| 21:40 | <zewt> | canadians see a different spectrum of light |
| 21:40 | <timeless> | Canada is on the NAm standard |
| 21:40 | <timeless> | Finland is on the EU standard |
| 21:41 | <hsivonen> | yay standards |
| 21:41 | <timeless> | and yes, at least using a NAm light bulb in my Finnish apartment blew a fuse |
| 21:41 | <timeless> | actually, iirc i managed to blow two fuses that way |
| 21:41 | <timeless> | (And a pair of bulbs?) |
| 21:41 | <timeless> | hsivonen: exactly! |
| 21:41 | <gsnedders> | (in the UK it is standard to have circuit breakers and fuses in each socket, not just in the socket, which is advantageous insofar as each socket can be fused for a maximum more suited to the specific advice) |
| 21:41 | <Philip`> | zewt: Surely it'd take longer than a year for that country to migrate all the infrastructure to a new standard; a better solution would be to have a war and destroy every country (preferably one at a time, not all at once) so they have to rebuild and then they might as well rebuild with a universal standard that was previously agreed |
| 21:41 | <hsivonen> | timeless: well, new buildings in Finland are fuseless |
| 21:42 | <timeless> | hsivonen: yeah, that helped me a lot... |
| 21:42 | <timeless> | what's your definition of new btw? |
| 21:42 | <hsivonen> | (as in when a "fuse" "blows", you just flip a switch back) |
| 21:42 | <timeless> | hsivonen: that's a "circuit breaker" (see above for civilized world) |
| 21:42 | <hsivonen> | timeless: I don't know what the cut-off is. 2001 is new enough and 1970s is still old |
| 21:42 | <timeless> | http://en.wikipedia.org/wiki/Circuit_breaker |
| 21:43 | <gsnedders> | I don't know how many houses in the UK still have fuses at the entrance point to the house — most have circuit breakers AFAIK. |
| 21:43 | <timeless> | and yeah, finland is weird, you call it "fuseless" instead of having a word to actually describe what's going on |
| 21:43 | <timeless> | yay broken languages |
| 21:43 | <hsivonen> | gsnedders: the mind-blowing thing is the wiring topology and the fuses in the electrical plugs |
| 21:43 | <timeless> | gsnedders: i remember seeing those on a visit or two |
| 21:43 | <zewt> | we have breakers, but they're in the basement, where we have to go outside, through the bulkhead, around the basement and up a spiderweb-infested stairway-to-nothing |
| 21:44 | <zewt> | so please don't trip the breaker |
| 21:44 | <timeless> | ouch |
| 21:44 | <timeless> | the breaker in my place is in the bedroom |
| 21:44 | <gsnedders> | hsivonen: Well, with a ring circuit you can't have one thing in a circuit box per socket, because where the circuit box is you have no concept of sockets yet. |
| 21:44 | <zewt> | also no lights once you're on the stairway |
| 21:44 | <timeless> | zewt: of course |
| 21:44 | <zewt> | which is probably for the best, so you don't know how many spiderwebs you're going through |
| 21:44 | <gsnedders> | hsivonen: But the wiring topology saves wire, which post-WW2 was a big deal, given the amount that had to be rebuilt. |
| 21:45 | <timeless> | gsnedders: but to do wiring work do you have to kill the "mains power"? |
| 21:45 | <timeless> | Philip`: that's more or less what we did around 1944 |
| 21:46 | <gsnedders> | timeless: You just kill whatever ring you're working on, or kill it at the socket-level if you're not touching the ring itself. |
| 21:46 | <hsivonen> | timeless: it turns out that there is a Finnish word for circuit breaker but I had never heard it (I had heard a misnomer that I realized was enough of a misnomer than I didn't try to translate it directly into English) |
| 21:46 | <gsnedders> | Most British houses have multiple rings, AFAIK |
| 21:46 | <timeless> | hsivonen: interesting |
| 21:46 | <timeless> | not surprising, but interesting |
| 21:46 | <timeless> | i mentioned that because you're not the first person to use fuseless in conversation w/ me when describing modern finnish housing |
| 21:47 | <timeless> | (and then power the ring) |
| 21:47 | <timeless> | gsnedders: so... whenever you want to add something to a ring, you kill the ring, pick the closest point along the ring, and splice into the ring? |
| 21:48 | <gsnedders> | timeless: Pretty much. |
| 21:48 | <timeless> | interesting |
| 21:48 | timeless | isn't really sure that it's much different from how things work in NAm |
| 21:49 | <timeless> | practically speaking, your ring sounds fairly close to a NAm circuit |
| 21:49 | <timeless> | and given the basic definition of circuit... |
| 21:49 | <gsnedders> | Got any link for that? Wikipedia doesn't quickly find anything for that. |
| 21:52 | <timeless> | try http://en.wikipedia.org/wiki/Electrical_wiring_in_North_America |
| 21:52 | <timeless> | grr, an intranet site killed my web browser |
| 21:55 | <timeless> | Philip`: iirc you asked about QoS VPN |
| 21:55 | <timeless> | i think the answer is that it doesn't satisfy my requirements |
| 21:55 | <timeless> | i could set up a local QoS system for my own testing if i wanted to |
| 21:56 | <timeless> | but in order to do something useful, i'd want to be able to talk to a vendor |
| 22:00 | <Philip`> | timeless: The idea is that you talk to a vendor who sets you up a VPN (running entirely on their physical network, not over the internet) which does the useful things, though the idea is also that you're an enterprise network with large amounts of money, as far as I'm aware |
| 22:01 | <timeless> | Philip`: oh, i can understand it |
| 22:01 | <timeless> | but it doesn't really help me much |
| 22:01 | <timeless> | lemme pastebin a problem space |
| 22:02 | <Philip`> | I imagine it's hard to do QoS over the internet since the internet can't tell separate customers apart, and everyone would set all of their own packets to 'highest priority' |
| 22:02 | <timeless> | http://pastebin.mozilla.org/1324125 |
| 22:03 | <timeless> | yeah, QoS runs into prisoner problems |
| 22:03 | <timeless> | http://en.wikipedia.org/wiki/Prisoner%27s_dilemma |
| 22:04 | <timeless> | i think in theory it might be possible to do slightly better than that |
| 22:05 | <timeless> | in cases where you have an enforcer and can ask a question and be given a "no" |
| 22:05 | <timeless> | at least when you deal w/ <prisoner=end user> <guard=ISP which wants to be mean anyway> |
| 22:11 | <timeless> | using Comcast/AT&T as examples |
| 22:11 | <timeless> | they generally tend to give you an initial burst and then taper you off |
| 22:11 | <timeless> | but they could easily choose not to do that if you (individual customer) included some QoS |
| 22:11 | <timeless> | instead they could try to honor it up to a quota and penalize anything that isn't QoS'd |
| 22:13 | <timeless> | > As of 1 July 2008, wall outlets for Type E (French 2-pin, female earth) are permitted for installations in Denmark.[22] This was done because no electrical equipment sold to private users is equipped with a Type K plug, and to break the monopoly of Lauritz Knudsen — the only company making Type K sockets and plugs. |
| 22:13 | timeless | chuckles |
| 22:14 | <timeless> | hsivonen: i'm glad to see one of Finland's neighbors caved to basic economics |
| 22:15 | <timeless> | > Sockets for the Schuko Type F will not be permitted. The reason is that a large number of currently used Danish plugs (coincidentally made by the afore mentioned Lauritz Knudsen monopoly) will jam when inserted into a Schuko socket. This may cause damage to the socket. It may also result in a bad connection of the pins, with resultant risk of overheating and fire. Broken Type F sockets are often seen in German hotels visited by Danes |
| 22:15 | timeless | rotfl |
| 22:20 | <zcorpan> | wait, the orientation information isn't on the event? when it should be a no-brainer to not fire an event on registration... |
| 22:20 | <Philip`> | timeless: Ah - intentional throttling of connections sounds different to what I'd usually assume QoS to be (i.e. unequal prioritisation of packets when congestion occurs) |
| 22:21 | <zewt> | at least in android's webkit, on a quick look: doc->dispatchWindowEvent(Event::create(eventNames().orientationchangeEvent, false, false)); |
| 22:21 | <zewt> | don't know about iOS |
| 22:21 | <zewt> | (seems like a good argument for not putting info inside the event itself unless there's a specific reason to need to know what the state was at the time the event was queued) |
| 22:22 | <timeless> | zewt: so... |
| 22:22 | <zcorpan> | hmm looks like the spec has the information on the events |
| 22:22 | <timeless> | the problem is that some things require resources to retrieve |
| 22:22 | <timeless> | e.g. battery crud |
| 22:22 | <timeless> | or orientation |
| 22:22 | <zewt> | how does that matter here? |
| 22:22 | <timeless> | the assumption i had was that one would calculate those as needed and fire the information when available |
| 22:22 | <zewt> | putting it in the event would be even worse in that case, since you'd be computing it even if it's not needed |
| 22:23 | <zewt> | (i havn't looked at this spec; i'm assuming it preserves window.orientation, since discarding what's already widely-implemented would be insane) |
| 22:23 | <zcorpan> | does it matter if the author asks for it by registering a listener or by getting a property? |
| 22:23 | <timeless> | zcorpan: kinda |
| 22:23 | <timeless> | in the case of GPS |
| 22:23 | <timeless> | the time to get a location is 10minutes |
| 22:24 | <timeless> | so if the author asks for it by getting a property, the app hangs for 10minutes |
| 22:24 | <timeless> | which isn't really "ok" |
| 22:24 | <zewt> | (i'm only referring to orientation; gps is a different beast entirely) |
| 22:24 | <timeless> | i'm not sure about orientation, but it wouldn't shock me if the stall was semi significant |
| 22:24 | <timeless> | would 2ms bother you? |
| 22:25 | <zewt> | yes, but there should be no delay at all--the OS always knows the orientation already |
| 22:25 | <zcorpan> | timeless: ok, so for cases that would take a long time you would want an async api with a method that has a callback or causes an event to fire when it's available |
| 22:25 | <zewt> | that's persistent state that phones (at least) need to keep track of all the time (it's not a simple matter of atomically peeking at an accelerometer) |
| 22:26 | <zewt> | (at least on Android--i happen to have looked over the orientation-state-handling code in android the other day--no clue about iOS) |
| 22:26 | <zewt> | (rather, i'm sure iOS is the same but can't see the code to say for certain) |
| 22:26 | <zewt> | afk |
| 22:29 | <timeless> | the battery case is amusing |
| 22:29 | <timeless> | the nokia devices i worked on didn't know their power level while charging |
| 22:29 | <timeless> | they knew they were plugged in |
| 22:29 | <timeless> | but they couldn't tell you how much was in the battery |
| 22:29 | <timeless> | and to make things more exciting, the cpu was fully capable of consuming more power than the charger was capable of providing |
| 22:30 | <timeless> | i.e. one could run out of power while plugged into a/c :) |
| 22:30 | <timeless> | well, to be fair, that wasn't precisely the 'cpu' so much as 'the complete device package', but.. |
| 22:31 | <timeless> | zewt: was your doc->dispatch... thing dispatching the genuine event or a synthetic one? |
| 22:34 | <timeless> | interesting |
| 22:34 | <timeless> | EV-DO is using QoS |
| 22:35 | <zewt> | WebCore/page/Frame.cpp Frame::sendOrientationChangeEvent |
| 22:35 | <zewt> | looks like the place where the engine fires the real event |
| 22:35 | <zewt> | not a webkit expert, though from what i've seen of webkit it usually looks like fairly straightforward code |
| 22:36 | <zewt> | i don't think any chargers can actually tell battery level while charging |
| 22:36 | <zewt> | only estimate |
| 22:37 | <timeless> | laptops seem to give relatively accurate percentages |
| 22:37 | <timeless> | i'm not sure how that works |
| 22:37 | <zewt> | maybe they charge each cell separately? |
| 22:37 | <zewt> | so it can measure the levels of cells that aren't under charge |
| 22:37 | <zewt> | (no idea, just throwing darts) |
| 22:38 | <timeless> | maybe, certainly ibm (lenovo?) thinkpads used to show icons which might actually represent individual cells |
| 22:38 | <zewt> | of course, measuring the charge level of a li-ion battery is hard no matter what |
| 22:38 | timeless | nods |
| 22:39 | <timeless> | but laptops generally give a percentage reading which is much better than 16% |
| 22:39 | <zewt> | i don't know anything about high-end batteries, i've dealt with firmware for simple AA-cell charging |
| 22:39 | <timeless> | (while charging) |
| 22:39 | <timeless> | now, they might be cheating heavily |
| 22:39 | <zewt> | presumably there's more tech involved in expensive laptop batteries |
| 22:39 | <timeless> | i think that laptops can estimate reasonably well how much power they're consuming |
| 22:40 | <timeless> | and can know how much power they're getting from a/c |
| 22:40 | <zewt> | well, they could just measure it, i'd imagine |
| 22:40 | <timeless> | and if they know how much was in the battery before charging, they can do math to get reasonable numbers... |
| 22:40 | <zewt> | my nexus s likes to use as much power as it can draw, it's pretty great to be in an airport and unable to use my phone because if i do it won't charge for the flight, heh |
| 22:41 | <timeless> | cute |
| 22:41 | <zewt> | i think at the time i was on a 500mA adapter though, should be better now that i'm on a 1A one |
| 22:41 | <timeless> | yeah |
| 22:41 | <zewt> | the high-power-adapter situation is pathetic |
| 22:41 | <timeless> | the nokia charges are 1.2A |
| 22:41 | <timeless> | whereas usb chargers are .1 .. .5A |
| 22:41 | <zewt> | tons of "1A" usb wall adapters that don't actually short the data pins to enable it |
| 22:41 | <timeless> | depending on whether your device is smart or stupid |
| 22:41 | <zewt> | usb can do up to 1.5A |
| 22:42 | <zewt> | if they don't enable high-power charging you're stuck at 500mA |
| 22:42 | <timeless> | i'm not sure why the nokia chargers only did 1.2 instead of 1.5 |
| 22:42 | <zewt> | well, the usb spec says 1.5A (iirc, i think it was 1.8A then dropped to 1.5A for some reason), but lots of adapters only say 1A |
| 22:43 | <zewt> | recently i just grabbed http://www.amazon.com/dp/B003ZBZ64Q so i don't have to care |
| 22:44 | <Philip`> | My laptop gives capacity in mAh with a resolution of seemingly at least 50mAh, whether charging or discharging, and if it's running on battery it gives the discharge rate in mA with a resolution of at least 3mA (else it gives the charge rate instead) |
| 22:44 | <timeless> | ouch |
| 22:44 | <timeless> | Philip`: nice |
| 22:44 | <Philip`> | (via /proc/acpi/battery/BAT1/state) |
| 22:45 | <Philip`> | and it's a pretty cheap laptop so I presume anything that's not very old can do this |
| 22:46 | <Philip`> | (No idea whether it's all based on actual measurement or if it integrates current to get charge or whatever, though) |
| 22:47 | <Philip`> | (but it seems accurate enough) |
