| 00:29 | <Hixie> | i'm having connection issues, so i'm gonna go home and will be back online later to continue replying to all this mail |
| 08:22 | <Hixie> | hmm |
| 08:22 | <Hixie> | how about a rel value that disables sending of referer headers for a link? |
| 08:23 | <Hixie> | people seem to go to extreme lengths right now with document.open and document.written meta refreshes and all kinds of nonsense |
| 08:27 | <jruderman> | for a long time, i argued that things like "meta refresh doesn't send referrer" and "links from https to http send no referrer at all" are bugs, maybe even security problems |
| 08:28 | <jruderman> | how about making the rel strip out the path&query from the referrer but leave the hostname |
| 08:32 | <Hixie> | seems simpler just to allow the whole thing to be dropped |
| 08:32 | <Hixie> | why would it be a security problem? |
| 08:32 | <othermaciej> | how is not sending referer conceivably a security bug? |
| 08:34 | <Hixie> | rel=noreferer now in the spec. |
| 08:35 | <othermaciej> | is that really a RELationship? |
| 08:36 | <Hixie> | not really, but hey, rel is pretty damn poluted already anyway |
| 08:36 | <Hixie> | i considered rel=untrusted, but i figured why pretend |
| 08:37 | <Hixie> | we could add an attribute, but that's far heavier |
| 08:37 | <Hixie> | (adds an attribute to two elements, DOM attributes to match, etc) |
| 08:45 | hsivonen | hopes that TPAC will have better wifi connectivity than XTech |
| 08:46 | hsivonen | expects GPRS/EDGE/HSDPA roaming to be insanely expensive and perhaps even broken |
| 08:46 | <Hixie> | tps usually have wifi. |
| 08:46 | <Hixie> | better wifi, i wouldn't count on. |
| 08:46 | <Hixie> | at least not at the start of hte week |
| 08:46 | <Hixie> | it usually settles down by about noon on wednesday |
| 08:47 | <hsivonen> | Hixie: XTech had none for people who weren't presenting the same day, so almost any connectivity would be better |
| 08:48 | <Hixie> | heh |
| 08:51 | <Hixie> | huh |
| 08:51 | <Hixie> | HTMLHeadElement.profile in DOM2 HTML is defined to return one URI |
| 08:51 | <Hixie> | but <head profile=""> in HTML4 is defined to take a space-separated list of URIs |
| 08:51 | <othermaciej> | yum |
| 08:52 | <Hixie> | (HTML5 drops both) |
| 08:52 | <Hixie> | are there other attributes that take lists of URIs? |
| 08:52 | <Hixie> | or is it just profile (and now ping)? |
| 08:52 | <hsivonen> | I guess that happens when the operation/DOM nonsense is pushed to another spec |
| 08:53 | <Hixie> | hsivonen: yeah, that's why i think it's totally stupid to suggest they should be split again. but whatever. :-) |
| 08:53 | <Hixie> | i wonder how i should expose ping in the DOM |
| 08:53 | <othermaciej> | did someone suggest that? |
| 08:53 | <Hixie> | i think ff3 just has a DOMString of absolute URIs |
| 08:53 | <hsivonen> | Hixie: the input method attribute on WF2 whatever it was called takes a list on tokens and URIs |
| 08:53 | <Hixie> | othermaciej: oh it's a common request |
| 08:53 | <Hixie> | the xforms thingy? |
| 08:53 | <Hixie> | hmm |
| 08:53 | <Hixie> | i wonder how i handled that |
| 08:54 | <Hixie> | i bet i just ignored it |
| 08:54 | <hsivonen> | othermaciej: "operational/DOM nonsense" is from Roy Fielding's survey "disagree" rationale |
| 08:54 | Hixie | notes searching for "input" and searching for "method" in the wf2 spec... not useful |
| 08:54 | <hsivonen> | Hixie: yeah, the XForms thingy |
| 08:55 | <Hixie> | inputmode |
| 08:55 | <othermaciej> | hsivonen: I wonder if anyone told him about SVG, MathML, XForms... |
| 08:55 | <Hixie> | crap. inputmode is just a DOMString. |
| 08:55 | <Hixie> | screwed myself again. |
| 08:55 | <hsivonen> | Hixie: it's xsd:string in the XForms schema :-) |
| 08:55 | <Hixie> | good times. |
| 08:56 | <Hixie> | ok so what do i do for ping. Make it a DOMString with a parallel pingList that uses the DOMTokenList thingy, and then never actually put pingList in the spec because, you know, why bother? |
| 08:56 | <Hixie> | or do i actually do URI resolution |
| 08:56 | <Hixie> | do i have a DOMStringList for it? |
| 08:56 | <Hixie> | aaah so many options |
| 08:57 | <hsivonen> | aside: IIRC, XForms wasn't clear on whether IRIs are OK as inputmodes |
| 08:57 | hsivonen | looks up the new ed with so many diffs |
| 08:59 | <Hixie> | i think i'll just make .ping return a DOMString of resolved absolute URIs |
| 08:59 | <Hixie> | which is a dumb API |
| 08:59 | <Hixie> | but i don't see what else to do |
| 09:00 | hsivonen | doesn't find "IRI" in XForms 3rd ed. |
| 09:00 | <hsivonen> | does this mean XForms isn't properly internationalized? :-) |
| 09:01 | <hsivonen> | no diffs in the inputmode section |
| 09:02 | <hsivonen> | (schema-wise inputmode is the most complex attribute in HTML5) |
| 09:02 | <hsivonen> | (and in XForms they dodged it all with xsd:string) |
| 09:03 | <annevk2> | Hixie, a DOMTokenList if resolved URIs? |
| 09:03 | <hsivonen> | hmm. I have written this in comments: "# XForms seems to allow IRIs when it talks about URIs." |
| 09:04 | <hsivonen> | I wonder where I got that from |
| 09:04 | <hsivonen> | I also have "# Let's assume case, digits and symbols are mutually exclusive to cut down on possible permutations." |
| 09:04 | <hsivonen> | # "halfWidth" is deprecated |
| 09:04 | <hsivonen> | # Just assuming at least one token is required. The spec does not say. |
| 09:05 | <hsivonen> | I guess the spec wasn't unambiguous :-) |
| 09:05 | <hsivonen> | and WF2 imports the ambiguity |
| 09:06 | <hsivonen> | Hixie: btw, I've made inputmode enumerated tokens case-sensitive |
| 09:07 | <hsivonen> | Hixie: HTML5 enumerated tokens are otherwise ascii-case-insensitive |
| 09:07 | <hsivonen> | Hixie: the spec should probably say something about case-sensitivity when importing inputmode from XForms |
| 09:08 | hsivonen | wonders if the S60 version of Opera will have actual inputmode support |
| 09:08 | <hsivonen> | Opera 9 that is |
| 09:08 | <Hixie> | hsivonen: i expect to define inputmode explicitly or drop it completely when wf2 becomes part of html5. |
| 09:09 | <Hixie> | annevk2: but then you can't do .ping = 'x y z' |
| 09:09 | <hsivonen> | Hixie: ok |
| 09:09 | <Hixie> | annevk2: you have to use the API to add and remove URIs |
| 09:09 | <annevk2> | you can do setAttribute("ping", "x y z") |
| 09:10 | <Hixie> | annevk2: sure, but that's inconsistent, you can set every other attribute |
| 09:11 | <annevk2> | ok, dunno really then |
| 09:11 | <Hixie> | yeah me either |
| 09:11 | <annevk2> | i'm not sure what the use case is for a DOM attribute here anyway |
| 09:11 | <Hixie> | consistency |
| 09:11 | <hsivonen> | aside: something else profile='' and inputmode='' have in common is that they require absolute IRIs |
| 09:12 | <annevk2> | Hixie, in that case do the resolve URI thingie |
| 09:12 | <Hixie> | oh yeah i guess profile="" didn't allow relative ones |
| 09:12 | <Hixie> | annevk2: yeah that's what the spec now says |
| 09:12 | <Hixie> | annevk2: :-( |
| 09:13 | <annevk2> | how does the spec deal with space separated lists of URIs then in "reflect" |
| 09:14 | <Hixie> | i added a paragraph there |
| 09:14 | <Hixie> | in fact that's all i changed |
| 09:14 | <Hixie> | see diff |
| 09:15 | <annevk2> | ah, refresh helps |
| 09:16 | <Hixie> | i'm confused by issue-2 |
| 09:16 | <Hixie> | not sure what to do about it |
| 09:16 | <hsivonen> | Hixie: about the ping UI req |
| 09:17 | <hsivonen> | whether the spec can/should require ping to be exposed in the UI |
| 09:18 | hsivonen | is seeing the death of an old business model: my mobile Internet is now fast enough to route Skype calls through for fraction of the cost circuit-switched calls cost |
| 09:19 | <Hixie> | hsivonen: oh i understand the issue. i just don't know what to do about it. |
| 09:19 | <hsivonen> | the telco sells me flatrate packet data but charges 0.25 euros per minute for circuit-switched video calls |
| 09:19 | <hsivonen> | oh sorry |
| 09:19 | <Hixie> | and there are lots of business models dying at the moment |
| 09:19 | <Hixie> | it's great |
| 09:19 | <Hixie> | :-) |
| 09:19 | <Hixie> | breathing new blood into the human race |
| 09:19 | <Hixie> | or at least the capitalist part |
| 09:21 | <hsivonen> | off-topic, but I've never like the attitude of telco folks. |
| 09:22 | <hsivonen> | back at the university, we had a course where the IP part was lectured by a guy who had been extending the Internet to Finland early on |
| 09:22 | <hsivonen> | and had later built the "Saudi Internet" |
| 09:23 | <hsivonen> | and then there were guest lectures from telcos talking about phone stuff |
| 09:23 | <hsivonen> | the different in attitude was striking |
| 09:23 | <hsivonen> | a telco person showed a slide pondering what they can charge for |
| 09:23 | <hsivonen> | and it had a "per bullet" (in a game) bullet point |
| 09:24 | <hsivonen> | (now that I think of it, that would be the game version of Chris Rock's "bullet control" :-) |
| 09:24 | <Hixie> | what's shocking isn't so much their gall or greed but that it works |
| 09:34 | <hsivonen> | Hixie: actually, the failure of WAP and the rise of real browser engines and Internet connectivity on phones is some indication that the telco gall is no longer working like it used to |
| 09:48 | <Hixie> | yeah they're slowly dying |
| 09:48 | <Hixie> | but not very fast |
| 09:50 | <hsivonen> | the main problem is that many countries allow telcos to control the handset features |
| 09:50 | <hsivonen> | that's evil |
| 09:52 | <hsivonen> | it's frustrating that the business side of Nokia is blocking some technical goodness because Nokia needs to appease to telcos abroad where the telcos call the shots |
| 09:52 | <Lachy> | Another problem is that some countries allow locked handsets to be sold without an unlocked version avaliable |
| 09:52 | <Hixie> | am i really not explaining myself in this ping='' discussion, or is julian just not reading what i write? |
| 09:53 | <Hixie> | for the past like 3 e-mails he's asked me the same question, as far as i can tell |
| 09:53 | <hsivonen> | Lachy: that's the mechanism that they use to control handset features |
| 09:54 | Hixie | throws the ping="" mail onto the links pile and goes to bed |
| 09:54 | <Hixie> | nn |
| 09:57 | <hsivonen> | the major *business* innovation in Nokia 770, N800 and N810 is that the device doesn't have a GSM unit and, therefore, can bypass the phone sales channel in countries with bad laws |
| 09:58 | <Lachy> | how do they work without GSM? |
| 09:58 | <hsivonen> | Lachy: through wifi or routing via Bluetooth through a separate GSM device |
| 09:59 | <hsivonen> | Lachy: so I get to run VoIP software on the N800 and route the packets through my phone |
| 09:59 | <Lachy> | ok. So it's only really useful in areas where there's sufficient open wifi coverage |
| 10:00 | <hsivonen> | Lachy: in Helsinki, I use my N800 with the traffic routed through my HSDPA phone |
| 10:00 | <Lachy> | ok |
| 10:01 | <hsivonen> | Lachy: that I can't get proper VoIP software for the S60 handset is just an artificial barries |
| 10:01 | <hsivonen> | barrier |
| 10:01 | <hsivonen> | people can ship all the VoIP software they want for the N800 |
| 10:09 | <Hixie> | gah, i couldn't help replying. |
| 10:09 | <Hixie> | now i'm really going to bed |