| 08:04 | <krijn> | Meh |
| 08:05 | <MikeSmith> | krijn: general comment? |
| 08:07 | <krijn> | Server down |
| 08:07 | <krijn> | That's what happens when you guys don't donate! :] |
| 08:07 | <krijn> | (Can only fix it tonight, sorry!) |
| 08:32 | <Ms2ger> | r? https://critic.hoppipolla.co.uk/r/611 |
| 09:06 | <zcorpan> | jgraham: does trickle delay only the body or the headers also? |
| 09:10 | <jgraham> | zcorpan: Only the body. If you want to do creative stuff with the headers it is possible but there isn;t a script at the moment |
| 09:12 | <zcorpan> | jgraham: so i want a slowly-loading video that will fire error. right now i have an .ogv file with trickle but if the browser doesn't support video/ogg and is strict with content-type it will fail fast instead of slow |
| 09:13 | <zcorpan> | maybe i can use a .py file instead |
| 09:13 | <foolip> | zcorpan: bad luck :( |
| 09:17 | <jgraham> | zcorpan: Yeah, if nothing else you can make a .py file that writes the output directly and take full control |
| 09:24 | <foolip> | jgraham: is it also possible to emulate network errors, as in the connection disappearing or being closed mid-stream? |
| 09:32 | <jgraham> | foolip: You have the socket object if you go down through the layers enough, so anything you can do with that is supported |
| 10:16 | <hsivonen> | annevk-cloud: The command |locale -m| is scary on Linux |
| 10:16 | <hsivonen> | it lists many encoding names that I've never heard of before |
| 10:16 | <Ms2ger> | Mm, EBCDIC-AT-DE |
| 10:17 | <Ms2ger> | And VIDEOTEX-SUPPL |
| 10:19 | <foolip> | jgraham: ok, sounds hopeful |
| 10:23 | <foolip> | jgraham: I tried http://web-platform.test:8000/tools/runner/index.html and it has an error with MANIFEST.json, is that known? |
| 10:23 | <foolip> | error as in doesn't work at all |
| 10:24 | <Ms2ger> | foolip, did you generate the manifest first? |
| 10:27 | <foolip> | Ms2ger: no :) |
| 10:28 | <foolip> | I guess I should read the documentation, and/or catch that error to tell the user to RTFM |
| 10:33 | <foolip> | jgraham: rektide would you have any idea why I'm not allowed to review the top 2 commits in https://critic.hoppipolla.co.uk/r/604 ? |
| 10:34 | <foolip> | rektide: sorry, tab completion failure |
| 10:34 | <Ms2ger> | Did you make them? |
| 10:35 | <foolip> | Ms2ger: nope |
| 10:35 | <Ms2ger> | Did zcorpan click "I'll review this"? |
| 10:37 | <foolip> | Ms2ger: not that I can tell, but add ?user=zcorpan to see that he certainly *can* review |
| 10:38 | <foolip> | looks like maybe I can fix it with "Manage Assignments"" |
| 10:38 | <Ms2ger> | Oh, that's an interesting trick |
| 10:38 | <Ms2ger> | Can you also review as him that way? :) |
| 10:39 | <foolip> | Ms2ger: yes, but *shhh* |
| 10:39 | <foolip> | or at least it used to be possible, who knows if this feature was removed or not :) |
| 10:40 | <Ms2ger> | Whoa |
| 10:41 | Ms2ger | reviews everything as foolip from now on |
| 10:41 | <hsivonen> | annevk-cloud: FWIW, telemetry so far indicates that we instatiate the IBM866 (DOS Cyrillic) way more often than we instantiate the KOI8-U decoder |
| 10:42 | <hsivonen> | though it could be a detector misfiring and detecting non-Cyrillic content as DOS Cyrillic |
| 10:43 | <hsivonen> | I guess eventually I have to write the telemetry code for figuring out what's actually happening with the Russian and Ukrainian decoders |
| 10:43 | <hsivonen> | marking the bug mentored hasn't had the results I hoped |
| 11:07 | <jgraham> | I didn't think you could actually perform actions as another user? |
| 11:08 | <jgraham> | foolip: Anyway, I think zcorpan had grabbed all the assignments. But I gave you back some (just using "Manage Assignments") |
| 11:45 | <jgraham> | foolip: So your UTF8 BOM removal PR basically traces all the files submitted by Microsoft. I seem to recall that Notepad generates BOMs but surely they aren't using that to write tests? |
| 11:48 | <jgraham> | MikeSmith: https://critic.hoppipolla.co.uk/showcommit?review=591&filter=files&file=35687,35973,35974,35977 is for you |
| 12:08 | <zcorpan> | jgraham: what happens if i use .asis and trickle? |
| 12:10 | <jgraham> | zcorpan: At the moment you get .asis behaviour but not trickle behaviour |
| 12:11 | <zcorpan> | jgraham: ok |
| 12:12 | <jgraham> | The pipe functions only work for serving normal files because they are at a higher level than the asis handler or (possibly, depending on what they do) the python handlers |
| 12:12 | <jgraham> | (the asis handler is basically implemented as socket.write(file.read()) |
| 12:12 | <jgraham> | ) |
| 12:13 | <jgraham> | (whereas at least some of the pipe functions depend on understanding the semantics of the response e.g. to set headers or the status code) |
| 12:14 | <zcorpan> | yeah i guess that makes sense |
| 12:41 | <zcorpan> | foolip: bringing in document.write makes panda sad |
| 12:44 | <zcorpan> | gsnedders: you didn't come up with a better proposal so your glaring has little effect |
| 12:46 | <zcorpan> | foolip: but that might well be a spec bug, not sure it was intentional to provide a stable state (or Perform a microtask checkpoint) for document.written scripts |
| 12:48 | <zcorpan> | filed https://www.w3.org/Bugs/Public/show_bug.cgi?id=24356 |
| 13:08 | <gsnedders> | zcorpan: :) |
| 13:40 | <zcorpan> | gsnedders: also you need to put things into perspective, compare with about:blank or document.write :-) |
| 13:44 | <Ms2ger> | jgraham, didn't you have a PR at some point to rename all the manual tests to have -manual? |
| 14:03 | <jgraham> | Ms2ger: Didn't that alrady happen? I do remember something like that |
| 14:03 | <Ms2ger> | Well, I found http://w3c-test.org/web-platform-tests/master/2dcontext/shadows/canvas_shadows_001.htm |
| 14:04 | <gsnedders> | zcorpan: I just apply an SEP field to them. :) |
| 14:07 | <jgraham> | Ms2ger: Hmmmmm |
| 14:14 | <gsnedders> | "If an attribute is so removed from a token, it, along with the value that gets associated with it, if any, are never subsequently used by the parser, and are therefore effectively discarded. Removing the attribute in this way does not change its status as the "current attribute" for the purposes of the tokenizer, however." --- can anyone in here actually understand that without reading it about five times? |
| 14:15 | <gsnedders> | If I'm not mistaken, the first "are" is attempting to agree with the "it", which makes it grammatically bogus. |
| 14:16 | <Ms2ger> | Yeah, the first two ares should be ises |
| 14:16 | <jgraham> | No, it's fine |
| 14:17 | <gsnedders> | How so? |
| 14:17 | <jgraham> | It's talking about the pair (attribute, value) |
| 14:18 | <jgraham> | Which therefore needs "are" for plural agreement |
| 14:18 | <gsnedders> | You still can't use "it" with "are", though! |
| 14:18 | <Ms2ger> | The spec is en-US |
| 14:18 | <jgraham> | Of course you can |
| 14:18 | <jgraham> | This is an example of doing just that |
| 14:18 | <Ms2ger> | It, the rock band, are? |
| 14:20 | <jgraham> | "it" is just a backward reference to the subject. You can replace the word "it" with "the attribute" |
| 14:20 | <gsnedders> | It's still using a plural part of the verb with a singular subject! |
| 14:20 | <jgraham> | "The attribute, along with the value associated with it, if any, are [...]" |
| 14:21 | <jgraham> | I don't see how there's a problem there |
| 14:21 | <jgraham> | "The cow and the horse are [...]" |
| 14:21 | <jgraham> | There is only one cow |
| 14:22 | <gsnedders> | The subject there is the phrase "the cow and the horse", which is plural. |
| 14:22 | <Ms2ger> | The cow, along with the horse, is |
| 14:22 | <jgraham> | Here it's "the attribute [and] the value associated with it", |
| 14:23 | <jgraham> | which is isomorphic |
| 14:23 | <gsnedders> | Ms2ger: That's totally different, grammatically. That's an adverbial. You can move the "along with the horse" pretty much anywhere in the sentence. |
| 14:24 | <Ms2ger> | gsnedders, that's exactly the case you quoted, though |
| 14:24 | <gsnedders> | Ms2ger: Yes, indeed. |
| 14:25 | <jgraham> | Anyway, I am going to go all native speaker on you and claim it's fine |
| 14:26 | <gsnedders> | If I rewrite the sentence in the spec to, "If an attribute is so removed from a token, it are never subsequently used by the parser, and are therefore effectively discarded, along with the value that gets associated with it, if any", which is merely reordering the parts of the sentence (which is perfectly fine, because it's an adverbial cluase), it seems fairly obviously wrong. |
| 14:26 | <gsnedders> | Well, I'm going to go all native speaker on you and claim it's not. |
| 14:26 | <Ms2ger> | I'm going to go all generalist on you and claim that it's English, hence it's wrong |
| 14:26 | <jgraham> | gsnedders: You're Scottish and so don't count |
| 14:27 | <jgraham> | Also your reworded sentence is nonsense even with "is" |
| 14:27 | <Ms2ger> | No it isn't |
| 14:29 | <jgraham> | Oh yes it is |
| 14:31 | <Ms2ger> | This isn't an argument, it's just contradiction |
| 14:31 | <gsnedders> | COCA includes "it are" 1360 times; and "it is" 360896 times. "it are" is *really* rare in en-us, and the "it" in pretty much never a subject when followed by "are". |
| 14:32 | <gsnedders> | So I'm going all corpus-linguistics on you and claiming "it are" basically is never used as a subject-predicator pair. |
| 14:32 | <jgraham> | Right, because it's not a pair |
| 14:33 | <jgraham> | it's "(it, and other thing), are |
| 14:33 | <jgraham> | " |
| 14:33 | <Ms2ger> | One thing is clear |
| 14:33 | <Ms2ger> | The topic still applies |
| 14:33 | <gsnedders> | Ms2ger: :) |
| 14:33 | <gsnedders> | jgraham: Also, I will point out that neither of us are native speakers of en-us-x-hixie. |
| 14:34 | <gsnedders> | (or, for that matter, en-us) |
| 14:34 | <Ms2ger> | Nobody is |
| 14:34 | <jgraham> | Well if Hixie is then anything he writes in it and considers correct is correct by definition |
| 14:37 | <darobin_> | gsnedders: the sentence breaks if you move "along with the value that gets associated with it" elsewhere |
| 14:37 | <darobin_> | because there are two "are" clauses that apply to it; positioning it elsewhere would make only one of them apply |
| 14:37 | <gsnedders> | darobin_: Why? I don't see why that actually changes anything? |
| 14:38 | <gsnedders> | Oh, because of the comma before the "and"? |
| 14:38 | <darobin_> | "If an attribute is so removed from a token, it is never subsequently used by the parser, and are therefore effectively discarded, along with the value that gets associated with it, if any." |
| 14:38 | <darobin_> | sorry, "and is effectively" |
| 14:39 | <gsnedders> | I don't see how that changes anything? |
| 14:39 | <darobin> | logically you can still reconstruct the original, but it's clearer in the original formulation |
| 14:39 | <gsnedders> | I find the original /really/ hard to understand. |
| 14:40 | <darobin> | I'm not disputing that it could perhaps be rephrased |
| 14:40 | <darobin> | just that jgraham is right on grammar |
| 14:40 | <gsnedders> | (As an aside, am I the only one who cannot currently login to the spec annotation system?) |
| 14:40 | <gsnedders> | darobin: I still disagree here. :) |
| 14:40 | <GPHemsley> | Never fear, the linguist is here! |
| 14:41 | GPHemsley | reads the scrollback. |
| 14:41 | <Ms2ger> | Oh, I thought smaug____ |
| 14:41 | <Ms2ger> | gsnedders, I'm logged in |
| 14:41 | <gsnedders> | GPHemsley: You aren't the only linguist here. ;) |
| 14:42 | <GPHemsley> | gsnedders: Oh? :( |
| 14:42 | <gsnedders> | GPHemsley: I do joint linguistics and CS. |
| 14:43 | <GPHemsley> | Shucks. |
| 14:44 | <GPHemsley> | Nevertheless, I agree with you and Ms2ger. Those two ares should be ises. |
| 14:44 | <gsnedders> | https://www.w3.org/Bugs/Public/show_bug.cgi?id=24360 is now filed. |
| 14:45 | <gsnedders> | Given I realized I had my username wrong for the spec annotation system :P |
| 14:45 | <gsnedders> | GPHemsley: On the other hand, you've probably a better idea about contemporary American views on en-us grammar. :) |
| 14:46 | <GPHemsley> | gsnedders: Indeed. Such as the fact that your use of "you've" in that context is ungrammatical in en-US. :P |
| 14:48 | <gsnedders> | GPHemsley: Just unacceptable as a contraction? Or you want the "have" after the probably? |
| 14:48 | <Ms2ger> | "you probably have"? |
| 14:48 | <GPHemsley> | In the case of the spec, the value associated with the attribute is incidental. It is the attribute itself that is not used and is therefor discarded. |
| 14:49 | <GPHemsley> | gsnedders: Auxiliary verbs that are the main verb of the clause cannot be contracted in en-US. |
| 14:49 | <gsnedders> | GPHemsley: Because COCA seems to show "you've probably" is used in similar contexts to the above. |
| 14:49 | <gsnedders> | GPHemsley: Though uncommon. |
| 14:49 | <Ms2ger> | "You have probably" still sounds weird |
| 14:50 | <Ms2ger> | (in that context) |
| 14:50 | <GPHemsley> | Ms2ger: Yeah, there are also adverbial placement differences. |
| 14:51 | <GPHemsley> | Well, actually, I think it's just the contraction over the adverbial. |
| 14:51 | <gsnedders> | GPHemsley: Seems not that different to "he's not" v. "he isn't", where the former is massively more common in en-us. |
| 14:51 | <gsnedders> | GPHemsley: (though the latter still occurs ~2% of the time) |
| 14:52 | <GPHemsley> | gsnedders: Could you give some examples of the fuller context of those occurrences? |
| 14:52 | <gsnedders> | GPHemsley: http://corpus.byu.edu/coca/ - go look yourself. |
| 14:52 | <GPHemsley> | meh :P |
| 14:53 | <GPHemsley> | gsnedders: Are you running Firefox ESR? |
| 14:55 | <gsnedders> | GPHemsley: Yes. |
| 14:55 | <gsnedders> | GPHemsley: I'm in uni. :) |
| 14:55 | <GPHemsley> | gsnedders: I should note also that "not" is not an adverbial. |
| 14:55 | <GPHemsley> | negation is a different grammatical category |
| 14:56 | <gsnedders> | GPHemsley: This is true. And gets especially interesting when you look at Scots. |
| 14:56 | <GPHemsley> | Oh yeah, negation gets interesting when you look at a whole lot of things. :P |
| 14:56 | <gsnedders> | GPHemsley: Have "no" as a negation, "-nae" as an negating prefix on verbs... |
| 14:56 | <GPHemsley> | Similar to French, I imagine? |
| 14:57 | <GPHemsley> | (Standard French, not necessarily colloquial French) |
| 14:57 | <gsnedders> | (Like, "he's no goin' to Glasgow" or "he isnae a good lad") |
| 14:57 | <gsnedders> | Not really that similar to French. |
| 14:57 | <GPHemsley> | ah |
| 14:58 | <GPHemsley> | hmm |
| 14:58 | <GPHemsley> | interesting |
| 14:58 | <GPHemsley> | is the -nae suffix only on auxiliaries? |
| 14:58 | <gsnedders> | "he isnae" is fine, "he'snae" sounds wrong to me, but there's a fair bit of dialectal variation. |
| 14:58 | <gsnedders> | Um... yes? Maybe? |
| 14:58 | <GPHemsley> | :) |
| 14:58 | <gsnedders> | [citation needed] |
| 14:58 | <GPHemsley> | have, be |
| 14:59 | <GPHemsley> | ,do |
| 14:59 | <GPHemsley> | oh, here's a nice list: https://en.wikipedia.org/wiki/Auxiliary_verb#A_list_of_auxiliaries_in_English |
| 15:00 | <GPHemsley> | though it kinda conflates modals and auxiliaries; not sure if I'm OK with that |
| 15:00 | gsnedders | may have just caused the Scots corpus to stop responding |
| 15:00 | <GPHemsley> | (though that's not really relevant to this discussion) |
| 15:00 | <GPHemsley> | copula may also have different properties |
| 15:00 | <gsnedders> | GPHemsley: Well, in some grammars of English modals and auxiliaries are the same category |
| 15:01 | <GPHemsley> | yeah, but they have different properties, so I don't think I like that |
| 15:01 | <GPHemsley> | oh, there's a whole separate article on this: https://en.wikipedia.org/wiki/English_auxiliaries_and_contractions |
| 15:03 | <gsnedders> | cannae, couldnae, didnae, dinnae, disnae, doesnae, gaunae, gaunnae, gonae, gonnae, havenae, shouldnae, wasnae, willnae... |
| 15:03 | <gsnedders> | Modals and auxilaries, maybe? |
| 15:05 | <GPHemsley> | what about may and might? |
| 15:06 | <GPHemsley> | or must? |
| 15:07 | <gsnedders> | michtnae and maunnae are the only things in the corpus beginning with "m" (and ending in -nae). |
| 15:08 | <gsnedders> | "michtnae" is relatively obviously "might not" and "maunnae" I'm not sure what that is? |
| 15:08 | <gsnedders> | Anyhow, lab is being taken over by course. Away! |
| 15:08 | <GPHemsley> | :) |
| 15:21 | <TabAtkins> | gsnedders: At least in my dialect of en-US, the correct construction would be "You've probably got a better idea", or "You probably have a better idea". |
| 15:24 | <jgraham> | Pretty sure the correct construction is "this conversation has been going on for an hour now" |
| 15:26 | <TabAtkins> | 45 minutes, come on. |
| 15:26 | <TabAtkins> | And I'm sorry linguistics bothers you. ^_^ |
| 15:27 | <jgraham> | Linguistics doesn't bother me. |
| 15:36 | <gsnedders> | GPHemsley: So my intuition says that "must", "might", and "may" are scarcely used in Scots. |
| 15:36 | <krijnh> | (Back up!) |
| 15:41 | <gsnedders> | GPHemsley: Quick look through corpus shows "must" pretty much only being used in "Standard English" (however you wish to define that :)). |
| 15:41 | <Ms2ger> | Her Majesty's English |
| 15:42 | <gsnedders> | Ms2ger: Not clear that isn't something different again. :) |
| 15:43 | <Ms2ger> | Not if you define it that way :) |
| 15:43 | <gsnedders> | Ms2ger: Consider the royal "we" and similar |
| 15:44 | <Ms2ger> | Well, we disagree with you on that one |
| 16:12 | <Ms2ger> | Hmm, http://web-platform.test:8000//2dcontext/shadows/canvas_shadows_002.htm doesn't work |
| 16:15 | <jgraham> | Ms2ger: In which way? |
| 16:15 | <jgraham> | The //? |
| 16:15 | <Ms2ger> | Yep |
| 16:16 | <jgraham> | Well if that's supposed to work then patches welcome, I guess? |
| 16:17 | <Ms2ger> | Don't think I care enough :) |
| 16:38 | <Ms2ger> | jgraham, okay, this one is fun |
| 16:38 | <Ms2ger> | Clicking a link in http://web-platform.test:8000/foo gives a 404 |
| 16:39 | <Ms2ger> | The same link in http://web-platform.test:8000/foo/ works |
| 16:43 | <jgraham> | Ms2ger: Yeah, that sounds quite plausible. |
| 17:21 | <dglazkov> | good morning, Whatwg! |
| 17:22 | <jwalden> | 'afternoon, dglazkov |
| 17:22 | <Ms2ger> | Hi dglazkov! |
| 17:23 | <Ms2ger> | <insert daily question> |
| 17:23 | <dglazkov> | ah yes! |
| 17:34 | <foolip> | Hixie: it would be nice if you could decide or say what you need to decide on https://www.w3.org/Bugs/Public/show_bug.cgi?id=19619 since I intent to update Blink's implementation of the RSA to match spec soonish |
| 19:23 | <Hixie> | foolip: last paragraph of https://www.w3.org/Bugs/Public/show_bug.cgi?id=19619#c20 is still my position -- if browsers drop it, then i'll remove it from the spec, but at the moment everyone implements it, so keeping it is the logical action for me |
| 20:18 | <Hixie> | ok, so, what have we learnt from the rel wiki and the meta name wiki |
| 20:18 | <Hixie> | we have learnt that: |
| 20:19 | <Hixie> | 1. the long tail of needs is very long |
| 20:19 | <Hixie> | 2. there are many duplicate ideas |
| 20:19 | <Hixie> | 3. many authors have no idea what these really are for, but they are very sure they want to have them and don't want validators to complain about it |
| 20:19 | <Hixie> | 4. specifications rarely exist for these things |
| 20:20 | <Hixie> | 5. if we make certain values non-conforming, we'd better have a damn convincing message to give the validators to help authors pick the right alternative |
| 20:21 | <Hixie> | 6. even major vendors like microsoft do crazy things, as in <meta name="msapplication-square70x70logo" content="images/tinylogo.png"> |
| 20:38 | <tantek> | Hixie you forgot: |
| 20:38 | <tantek> | 7. Nearly anything is better than a registry maintained by IETF. |
| 20:38 | <Hixie> | yeah, the list wasn't suposed to be comprehensive |
| 20:38 | <Hixie> | contributions from everyone are very welcome! |
| 20:39 | <Hixie> | ideally i think we should find some solution that is more nuanced and less burdensome than a registry |
| 20:40 | <tantek> | Hixie, the previous solution was no registry at all - any user values were allowed for rel attrs and meta names |
| 20:40 | <Hixie> | i have a vision of a two-tier system, where anyone can freely add values or add notes to existing values, and after the value has been in the list for a while, someone examines it and decides whether it should go in the spec to be recommended, or go in a list of things validators should reject, or go in a list of things that are proprietary solutions |
| 20:40 | <tantek> | (conforming to the syntax of course) |
| 20:40 | <Hixie> | well, html4's conformance approach is suboptimal in general, i'm not sure what that tells us |
| 20:41 | <tantek> | btw - that's basically what happens with rel values in the microformats wiki |
| 20:41 | <tantek> | people add them, and once in a while someone more experienced goes through and cleans them up and/or does work to promote them (write a better spec etc.) |
| 20:41 | <Hixie> | it's similar, yeah |
| 20:42 | <tantek> | but for that to work you need some semblance of a community that cares about them, which #microformats does about rel values |
| 20:42 | <tantek> | however I don't know if there is any community that actually *cares* about meta name values |
| 20:42 | <tantek> | other than "don't reject me validator" type caring |
| 20:45 | <JonathanNeal> | "Nearly anything is better than a registry maintained by IETF" +1 |
| 20:46 | <Hixie> | i envision a world where validators categorise values into five groups: non-conforming unknown (i.e. never registered), conforming interoperable good practice (i.e. in the HTML spec or wherever, with a full spec), provisionally conforming (i.e. registered but not examined), conforming proprietary (i.e. registered, deemed not useful for wide use, not implemented by widely used UAs, but used in private and not harmful), and non-conforming (registered, has bette |
| 20:46 | <Hixie> | alternatives, has a pre-canned message for validators to spit out) |
| 20:47 | <arunranga> | annevk, are you still around? |
| 20:47 | <Hixie> | validators would have an API that lets them take values from non-conforming unknown to provisionally conforming without requiring that people register in a wiki or whatever |
| 20:48 | Ms2ger | wonders how much value there is in making validators complain about any of them |
| 20:49 | <JonathanNeal> | Hixie: what are you hoping to solve by organizing these meta properties? |
| 20:50 | <Hixie> | JonathanNeal: a valid question. my main purpose here is making validators useful for authors who use these features, to enable them to catch typos, bad practices, and the like (same as the usual purpose for validators) |
| 20:50 | <Hixie> | JonathanNeal: e.g. it would be good to help authors avoid wasting time putting in <meta name=keywords> |
| 20:57 | <JonathanNeal> | I have observed meta names change with the wind as they're "created" by things like open graph, Twitter cards, or browser feature detection. It would be difficult to have any proactive specification. |
| 20:58 | <JonathanNeal> | browser feature detection err I mean ... vendor specific configurations? viewport, touch icon, etc. |
| 20:58 | <Hixie> | JonathanNeal: right, hence the reactive specification in the proposal above |
| 21:00 | <JonathanNeal> | I wish I could better describe how this feels out of scope for HTML. It feels a bit like conforming data attributes or class names. |
| 21:01 | <JonathanNeal> | Because I have no data that suggests developers write better or faster when a validator returns the kind of information you're proposing. |
| 21:02 | <Hixie> | JonathanNeal: it seems intuitive, though i admit that it might be misleading intuition, that if an author spends time doing something that will have no effect (like giving meta name=keywords), it is a net negative compred to not doing that. |
| 21:02 | <JonathanNeal> | And in my own experience, I've only ever wanted a "don't reject me" validator. |
| 21:03 | <Hixie> | what's a "don't reject me" validator? |
| 21:03 | <tantek> | Hixie, what we have today. |
| 21:03 | <Hixie> | how is that different than what i'm proposing? |
| 21:03 | <tantek> | people registering things just so the validator doesn't complain ("reject me") |
| 21:04 | <Hixie> | do people want a validator that says <b></i> is valid? |
| 21:04 | <JonathanNeal> | Hixie: If I'm remembering correctly, I had issues with the validator whining after I first implemented twitter cards on my site. |
| 21:05 | <Hixie> | right. what i'm proposing is that validators have a button so you just say "yep, that's valid, don't complain again" |
| 21:05 | <JonathanNeal> | I added it to some wiki, but that didn't immediately resolve the validator - I want to say that someone was manually copying over the values from the wiki. Is that right? |
| 21:05 | <Hixie> | today? probably |
| 21:05 | <Hixie> | (or manual during the build process) |
| 21:06 | <Hixie> | s/manual/automatic/ |
| 21:11 | <tantek> | JonathanNeal - typically a bug must be filed to get new meta or rel values added to the validator. |
| 21:12 | <tantek> | adding to the wiki page may eventually (years?) cause a change in the validator but certainly not immediately |
| 21:12 | <tantek> | note the paragraph before the ToC here: http://wiki.whatwg.org/wiki/MetaExtensions |
| 21:13 | <tantek> | Hixie, I will note that there seems to be a lot more proprietary and/or redundant meta name values than rel values. |
| 21:13 | <Hixie> | yeah |
| 21:14 | <tantek> | I feel like "synonyms" doesn't capture enough for the validator to say something actionable. |
| 21:15 | <tantek> | I.e. we need a way to indicate that a particular value is proprietary, and which (if any) "standard" value to use instead. |
| 21:15 | <tantek> | or even use some other method altogether |
| 21:16 | <tantek> | (rather than stuffing ever more into metacrap) |
| 21:16 | <Hixie> | right, that was the "validator message" in the proposal above |
| 21:16 | <tantek> | makes sense |
| 21:23 | <Hixie> | MikeSmith, hsivonen: you around? |
| 21:24 | <annevk> | arunranga: still around |
| 21:36 | <Hixie> | hsivonen, MikeSmith: if you're around... i'm curious to know your thoughts on the proposal above. |
| 21:36 | <Hixie> | hsivonen, MikeSmith: assuming you like the general direction, i'm also interested in more concretely whether you would prefer an HTTP-based request/response mechanism to register values and poll for updated registrations, or whether you'd prefer a long-lived TCP connection for that kind of thing. |
| 22:11 | <annevk> | I suspect HTTP-based |
| 22:11 | <annevk> | so it can be part of a build process |
| 22:13 | <annevk> | Hixie: if you're going to set up a registry, please allow for multiple registries |
| 22:50 | <Hixie> | annevk: sounds like an interesting idea. can you elaborate? |
| 22:50 | <annevk> | Hixie: e.g. rel / meta / URL schemes |
| 22:50 | <annevk> | might all benefit |
| 22:52 | <Hixie> | oh you mean multiple registries for different things, not multiple registries for the same thing |
| 22:53 | <Hixie> | yeah, if i write any code for this i'll definitely make it reusable, for sure |
| 22:57 | <annevk> | oh heh |
| 22:57 | <annevk> | yeah, didn't really want to propose competing registries maintained centrally :-) |
| 22:58 | <Hixie> | my main questions initially are whether the whole model seems sane; what API the validators would prefer; and what kind of interface we should expose for administration. |
| 22:58 | <Hixie> | that will then decide the implementation approach. |
| 23:03 | <annevk> | arv: cool, happy with you guys maintaining that code |
| 23:04 | <annevk> | arv: I'll mention that somewhere |
| 23:35 | <annevk> | marcosc: lol, way to stir some drama over semicolons |
| 23:47 | <Domenic_> | tangentially related, can we get some syntax highlighting in WHATWG spec examples? |
| 23:47 | <Hixie> | that sounds like a maintenance nightmare |
| 23:47 | <Hixie> | but patches welcome :-) |
| 23:47 | <Domenic_> | well, JS-applied, I'd assume |
| 23:47 | <Hixie> | oh |
| 23:47 | <Hixie> | sure |
| 23:47 | <Hixie> | patches welcome :-) |
| 23:47 | <Hixie> | (but please be ready to fix any mistakes the code makes) |
| 23:47 | <gsnedders> | Nah, make Hixie hand-write the syntax highlighting! |