| 01:57 | <MikeSmith> | XhmikosR: in vnu.jar 17.1.0 command-line checking of URLs is broken |
| 01:58 | <MikeSmith> | XhmikosR: so just released 17.2.0 with the fix |
| 06:10 | <KiChjang> | can someone explain what the relationship is between WHATWG and IETF? |
| 06:11 | <boogyman> | complementary |
| 07:14 | <MikeSmith> | hsivonen: if you can make a little time to look at https://github.com/validator/validator/issues/456 I would much appreciate it |
| 07:32 | <hsivonen> | MikeSmith: looking |
| 07:37 | <hsivonen> | time flies. It's been over a decade since I wrote that code. |
| 07:47 | <MikeSmith> | hsivonen: wow yeah |
| 07:53 | <gsnedders> | wow. |
| 08:09 | <zcorpan> | we're all getting *old* |
| 08:09 | <tobie> | heycam|away: you around today? Would love to help get you through https://github.com/heycam/webidl/pull/221 |
| 08:09 | <zcorpan> | including gsnedders |
| 08:10 | <tobie> | heycam|away: just installed it for some other repo, can assist to make the process faster |
| 08:37 | <XhmikosR> | MikeSmith: thankd, published a new grunt-html version |
| 08:38 | <MikeSmith> | XhmikosR: cheers |
| 08:57 | <hsivonen> | MikeSmith: my current main suspect is findInsertionPoint() in RowGroup. |
| 09:04 | <MikeSmith> | hsivonen: ok looking at that myself too (and hoping something jumps out…) |
| 09:07 | <MikeSmith> | so yeah seems like a good place to step through |
| 09:07 | <hsivonen> | MikeSmith: yeah. trying to set up debugging for the validator again |
| 09:09 | <MikeSmith> | could just drop in some System.out.println too |
| 09:09 | <MikeSmith> | anyway will try it myself too |
| 09:10 | <hsivonen> | so. many. src. folders. in. jing-trang |
| 09:11 | <MikeSmith> | ah yeah, James has some idiosyncratic ways of doing things |
| 09:12 | <MikeSmith> | as far at the table-checker code I can’t claim to have ever understood itーbut I never have had to, because I’ve never had to fix any bugs |
| 09:12 | <MikeSmith> | in it |
| 09:13 | <MikeSmith> | because other than this exception that gets thrown extremely rarely I do not recall there ever have been any bugs in there |
| 09:13 | <hsivonen> | afaict, the code that throws the exception is OK. |
| 09:13 | <MikeSmith> | sure yeah |
| 09:14 | <MikeSmith> | that code itself is pretty simple and intuitive |
| 09:15 | <hsivonen> | so I accidentally added the jing-trang src folders. should have added src/main folders |
| 09:15 | <MikeSmith> | ah yeah |
| 09:16 | <MikeSmith> | (and for the record that exception has been thrown in the W3C service only maybe 5 times since I started persistently logging the exceptionsーfive times out of tens of millions of requests) |
| 09:24 | <hsivonen> | MikeSmith: which XSLT engine if any does the W3C instance of V.nu use these days? |
| 09:25 | <hsivonen> | MikeSmith: is schematron support even supposed to build anymore? |
| 09:27 | <MikeSmith> | hsivonen: the Web UI for the W3C service does not expose anything that would exercise the schematron/XSLT stuff |
| 09:28 | <MikeSmith> | I guess the V.nu network API does though, so somebody could trigger it using that |
| 09:29 | <MikeSmith> | but anyway whatever is there still builds afaik |
| 09:29 | <MikeSmith> | the XSLT engine that gets used is the Saxon9 I think |
| 09:29 | <MikeSmith> | from the jar of that in the jing-trang repo |
| 09:30 | <hsivonen> | oh. I guess Eclipse's wizard failed to scan for it |
| 09:30 | <MikeSmith> | the stock V.nu build still includes that Saxon9 jar in the runtime classpath I think |
| 09:31 | <MikeSmith> | ah yeah, I don’t set up my Eclipse to handle the jing-trang sources as project |
| 09:32 | <hsivonen> | oh. there's a checked-in .classpath and I wasn't supposed make Eclipse jing-trang-aware |
| 09:35 | <MikeSmith> | oh |
| 09:35 | <MikeSmith> | that is probably my fault then |
| 09:37 | <hsivonen> | Exception in thread "main" java.lang.OutOfMemoryError: GC overhead limit exceeded |
| 09:37 | <hsivonen> | haven't seen that one before |
| 09:40 | <MikeSmith> | ah |
| 09:40 | <MikeSmith> | yeah I have |
| 09:41 | <hsivonen> | looks like my old settings didn't have enough heap |
| 09:42 | <MikeSmith> | yep, upped it to 512 in the build script a while back |
| 09:42 | <MikeSmith> | but Eclipse doesn’t use that |
| 09:43 | <MikeSmith> | I think the language-detector code is what pushed it over the limit |
| 10:02 | <hsivonen> | MikeSmith: anyway, now that I'm stepping through this, it's clear that findInsertionPoint() is wrong. Specifically, the second break. |
| 10:04 | <MikeSmith> | hsivonen: not clear to me that it’s wrong :) |
| 10:05 | <MikeSmith> | but I’ve not been able to concentrate on it seriously yet |
| 10:05 | <MikeSmith> | at least from the code standpoint (as opposed to troubleshooting) it does not seem clearly wrong, looking at it |
| 10:06 | <MikeSmith> | nad it ou |
| 10:06 | <MikeSmith> | oofs |
| 10:06 | <MikeSmith> | it does work pretty much all the time, only not in some very rare exceptions |
| 10:09 | <hsivonen> | sadly, it seems that the file was moved in the CVS era, so we don't have (easily findable?) history for the introduction of the second break |
| 10:09 | <hsivonen> | is the second break a premature optimization or does it serve some deeper purpose? |
| 10:10 | <hsivonen> | it looks a whole lot like a mere premature optimization to me |
| 10:12 | <hsivonen> | indeed, this whole nextOldCell thing seems to be an optimization |
| 10:14 | <hsivonen> | I suppose it's somewhat reasonable try not to get posted on http://accidentallyquadratic.tumblr.com/ |
| 10:14 | <MikeSmith> | haha |
| 10:15 | <hsivonen> | I guess the root cause design bug is trying handle both less than left and greater or equal than right in the same freeSlot() method |
| 10:16 | <MikeSmith> | never knew about that site but if there’s ever been a person who deserves to be memorialized there, it’s me |
| 10:16 | <MikeSmith> | well some of those are pretty sophisticated flubs |
| 10:17 | <MikeSmith> | compared to mine which relative to those are more like produced-by-monkey-banging-away-at-keyboard |
| 10:18 | <MikeSmith> | anyway as far as that code, there should be some wise saying that warns us about the dangers of premature optimization |
| 10:19 | <MikeSmith> | so we would know ahead of time there was some various kinds of risk from doing that |
| 10:21 | Ms2ger | wishes anyone cared about FileAPI |
| 10:25 | <MikeSmith> | I wonder if it’s ironic or instead fitting that “premature optimization is the root of all evil” came from same hand that blessed the world with the creation of null pointers |
| 10:26 | <MikeSmith> | hsivonen: about the code trying handle both less than left etc in the same method, yeah, at least that makes it a bit hard to unpack when reading it |
| 10:28 | <hsivonen> | MikeSmith: I have a fix. Do we have unit tests these days? |
| 10:30 | <hsivonen> | seems so |
| 10:32 | <hsivonen> | hmm. it seems that the unit test expectations are not up-to-date |
| 10:45 | <hsivonen> | MikeSmith: Travis fails due to submodule issues. I don't know what to do about that. My changeset was not supposed to change submodule config. |
| 10:56 | <MikeSmith> | hsivonen: OK I will look at the submodule thing in a bit |
| 10:56 | <MikeSmith> | bathtime for my toddler is in the way in the mean time |
| 10:57 | <MikeSmith> | as far as tests, those are in the tests/ submodule, and the test runner is in nu/validator/client/TestRunner.java and the test expectations are in tests/messages.json |
| 11:19 | <smaug____> | hmm, need to sort our fullscreen events and their prefixes in spec level |
| 11:42 | <MikeSmith> | hsivonen: the problem you had with the tests is due to the tablecheck branch not being (re)based against current master |
| 11:43 | <MikeSmith> | there are some relatively recent changes on master that the tests are checking |
| 11:43 | <MikeSmith> | so the tablecheck branch just needs to be rebased against master |
| 11:44 | <hsivonen> | whoa. I thought I was using current master |
| 11:45 | <MikeSmith> | yah not sure what happened but looks like you branch from the state of master circa December 15 |
| 11:50 | <MikeSmith> | hsivonen: anyway I think the only current tests we have that exercise this code are in tests/xhtml/elements/table/ |
| 11:51 | <MikeSmith> | and your fix doesn’t regress any of those |
| 11:53 | <MikeSmith> | and along with fixing the test case in #456 it also fixes the ones in https://github.com/validator/validator/issues/378 for HorizontalCellComparator too |
| 11:53 | <MikeSmith> | so I will add those cases to the test suite |
| 11:54 | <MikeSmith> | after this fix is merged |
| 12:39 | mathiasbynens | grabs popcorn while reading the responses to https://daniel.haxx.se/blog/2017/01/30/one-url-standard-please/ |
| 12:40 | <zcorpan> | hsivonen: soooooooooo.... <menuitem> parsing. should we change the spec so it parses like an unknown element? |
| 12:48 | <zcorpan> | errata... do we have a stock reply somewhere about why not errata? |
| 12:52 | <MikeSmith> | zcorpan: not that I know of |
| 12:53 | <MikeSmith> | but yeah we should have one |
| 12:59 | <Ms2ger> | mathiasbynens, is there anything else in there than "I want one standard... But not this one"? |
| 13:00 | <mathiasbynens> | Ms2ger: there are some entertainingly confusing false arguments as to why the URL Standard would not be sufficient, but yeah — that’s the TL;DR |
| 13:02 | <nox> | I like the scare quotes around 'spec'. |
| 13:03 | <nox> | I also don't get why we need a new group, when said authors of URL tools don't lurk here. |
| 13:14 | <MikeSmith> | hsivonen: thanks very muchーfix merged (along with the tests) |
| 13:15 | <MikeSmith> | Ms2ger: mathiasbynens nox context? twitter? |
| 13:15 | <nox> | MikeSmith: 13:40 — mathiasbynens grabs popcorn while reading the responses to https://daniel.haxx.se/blog/2017/01/30/one-url-standard-please/ |
| 13:15 | <Ms2ger> | https://daniel.haxx.se/blog/2017/01/30/one-url-standard-please/ |
| 13:15 | <MikeSmith> | ah god |
| 13:15 | <MikeSmith> | sorry he is still banging away at that drum |
| 13:16 | <MikeSmith> | it does not do him credit |
| 13:18 | <MikeSmith> | gotta give the guy a pretty massive amount of credit for curl. it is truly great. but he cuts pretty deep into that every time he trots out this anti-URL/anti-browsers hobbyhorse |
| 13:20 | <annevk> | jgraham: gsnedders: it seems wptserve normalizes header names, is there a way to get to the actual casing used in the request? |
| 13:25 | <MikeSmith> | as far as “I have the feeling that WHATWG people would not like this work to be done within the IETF and vice versa” maybe somebody who knows bagder can fill him on the actual history: The fact we helped get an IETF group chartered to work on updating the IRI spec, and the group went on for years, literallly, without doing any actual work at all, and tried also directly with mnot, etc etc |
| 13:26 | <MikeSmith> | “Any year now I think we will see some movement for real trying to fix this.” |
| 13:27 | <MikeSmith> | ...will be the same year as “The Year of Linux on the Desktop” or whatever |
| 13:28 | <nox> | Are you people going to FOSDEM? |
| 13:28 | <nox> | If yes, tell that to him IRL. :) |
| 13:41 | <Ms2ger> | Possibly, and no :) |
| 13:44 | <nox> | Ms2ger: Ah ah. |
| 14:03 | <nox> | https://twitter.com/nokusu/status/826060665358020608 I don't get it. |
| 14:03 | <nox> | If no one decides to participate to the group, nothing is gonna change. |
| 14:32 | <MikeSmith> | nox: thanks for trying at least |
| 14:32 | <nox> | MikeSmith: You're welcome, I like arguing too much for my own good. |
| 14:32 | <MikeSmith> | haha |
| 14:32 | <nox> | MikeSmith: Quite proud of myself for my reply on the Y THOUSANDS SLASHES thread though. :P |
| 14:33 | <MikeSmith> | “I channel what I've sensed and heard from others.” seems like a kinda odd thing to say in any context |
| 14:33 | <MikeSmith> | nox: Y THOUSANDS SLASHES thread? |
| 14:33 | <nox> | MikeSmith: https://github.com/whatwg/url/issues/118#issuecomment-276072218 |
| 14:36 | <MikeSmith> | nox: nice work but the clear winner in that thread is Domenic https://github.com/whatwg/url/issues/118#issuecomment-217866196 |
| 14:36 | <MikeSmith> | or as maybe we should call him from now on, Domenic “69 thumbs down” Denicola |
| 14:36 | <nox> | MikeSmith: Yeah but that doesn't make the URL standard not seem to focus on browsers. :P |
| 14:36 | <nox> | MikeSmith: AH AH |
| 14:37 | <nox> | Didn't notice the emojis. |
| 14:37 | <nox> | Most passive-aggressive GH feature ever. |
| 14:37 | <MikeSmith> | yeah |
| 14:37 | <MikeSmith> | and he was even really trying |
| 14:37 | <nox> | wasn't*? |
| 14:37 | <MikeSmith> | *wasn’t, yeah |
| 14:38 | <MikeSmith> | he is like a Zen master as evoking the thumbs-down pile-on without even trying |
| 14:38 | <MikeSmith> | very Wu Wei |
| 14:39 | <MikeSmith> | god but people do love them some curl |
| 14:41 | <MikeSmith> | bagder may have missed his calling, should have been a cult leader instead |
| 14:42 | <MikeSmith> | only losers waste their time and talents on software development and creating new Web/Internet technologies |
| 14:53 | <nox> | This whole thread sounds like the ultimate bikeshed. |
| 14:58 | <jgraham> | annevk: The code looks like it tries to preserve case on output? What are you actually doing that doesn't work? |
| 14:58 | <annevk> | jgraham: I did request.headers.iteritems() and got header names lowercased |
| 15:01 | <jgraham> | Oh request? |
| 15:01 | <jgraham> | I was looking at response |
| 15:03 | <jgraham> | annevk: You an look at request._raw_headers |
| 15:14 | <annevk> | jgraham: cool |
| 15:15 | <annevk> | jgraham: not documented at http://wptserve.readthedocs.io/en/latest/request.html though |
| 15:18 | <ato> | Presumably because of the _. |
| 15:28 | <jgraham> | annevk: Yeah it's sort of technically an internal api, but if it's needed we can either make it visible or change the real API in some way |
| 18:46 | <annevk> | I haven't been able to try it yet, will give it a go tomorrow |
| 18:47 | <annevk> | Not really gotten feedback thus far on the various related issues anyway |
| 19:08 | <rektide> | i dunno if this was whatwg or other- i seem to recall there having been talk of a worker that had a (non-visible) DOM it could use? anyone know what i'm talking about? is there a name for this? |
| 20:30 | <annevk> | rektide: it's not clear what benefits it provides over a small library |
| 22:15 | <Mek> | is there any way to find out which reference bikeshed might be giving a link error for? Even the -l option doesn't seem to actually give me any kind of line number info, so no idea how to find out what it might be complaining about... |
| 22:19 | <Mek> | hmm, and the link errors I'm getting started somewhere with "update widlparser" (not sure exactly which commit, since of three commits in a row two cause bikeshed to fail even worse) |
| 22:23 | <Mek> | TabAtkins: is -l supposed to give line number information for link errors? |
| 22:24 | <Mek> | also I'm a bit confused that the link error complains about multiple possible local refs, but then only lists one possibility... |
| 22:29 | <Mek> | ah, seems -l only works for links that don't use any kind of shorthand, since line numbers are attached to elements before the shorthand is processed and turned into html elements... |
| 22:39 | <jyasskin> | Double-checking: sourcemaps don't affect execution at all, right? Even if you load a malicious sourcemap, the worst that can happen is that you're confused while debugging, not that your site data leaks? |
| 22:41 | <jyasskin> | Mek: I've seen bugs with link error reporting in Bikeshed. Might you have both a definition and a link-defaults with the same key? |
| 22:42 | <gsnedders> | jyasskin: yes |
| 22:42 | <jyasskin> | gsnedders: Thanks. |
| 22:43 | <Mek> | jyasskin: it turned out that somehow bikeshed wasn't matching up my <dfn> with the idl generated definition of a method anymore. Which then caused errors when trying to link to that method... |
| 22:43 | <jyasskin> | Yeah. Did you get the right entity type? |
| 22:45 | <jyasskin> | (https://tabatkins.github.io/bikeshed/#dfn-types) |
| 22:45 | <Mek> | yes I did. and this used to work before some widlparser update. Chaning my dfn from <dfn method for=FileReaderSync>readAsText(blob, optional label)</dfn> to <dfn ...>readAsText(blob, label)</dfn> fixed the error... |
| 22:47 | <jyasskin> | Fun. |
| 22:47 | <Mek> | also would be helpful if bikeshed could at least warn for the duplicate definition for FileReaderSync/readAsText() rather than complain only when trying to refernce it... |
| 22:47 | <jyasskin> | +1 |
| 22:47 | <jyasskin> | File a bug? |
| 22:56 | <Mek> | ah, even more interesting, the errors I get are different depending on if the dfn has an explicit id or not.... but yeah, I'll file a bug... |
| 23:03 | <Domenic> | jyasskin: it's possible they might affect error.stack, but I'd guess they wouldn't. |