02:17
<MikeSmith>
https://whatwebcando.today/ is nice but should track a lot more features
02:35
<MikeSmith>
yoav: is there an open webkit bug for adding support for <link rel=preload>
04:35
<yoav>
MikeSmith: I see you opened one
07:01
<annevk>
So yeah, unsubscribed from public-webapps
07:02
<annevk>
Guess that era is over
07:03
<MikeSmith>
annevk: yup
07:03
<MikeSmith>
for better or worse
07:05
<MikeSmith>
yoav: yeah, did not find that you or igrigorik_ had opened one yet
07:06
<yoav>
MikeSmith: Cool. I opened smaller issues as I'm implementing, but always nice to have a meta-issue
07:07
<tantek>
what happened to public-webapps?
07:08
<tantek>
just transition wpwg?
07:08
<annevk>
The bureaucrats took over
07:08
<tantek>
sigh
07:09
<annevk>
"being concerned with procedural correctness at the expense of people's needs" is apt in particular as it was called out a few times
07:12
<tantek>
did any browser implementers respond to the HTML 5.1 CfC for CR? I was busy with conferences at the time and it already closed. https://lists.w3.org/Archives/Public/public-webapps/2016AprJun/thread.html#msg99
07:15
<tantek>
If I don't find any I may reply to that thread pointing out that an HTML CfC is meaningless without any browser implementer participation
07:16
<annevk>
tantek: Microsoft and Yandex
07:16
<tantek>
I didn't know Yandex made a browser
07:16
<MikeSmith>
yoav: I actually wasn’t aware you were already implementing it in WebKit
07:17
<tantek>
annevk: who at Microsoft?
07:17
<annevk>
tantek: Microsoft edits the fork, Adrian, Travis, Aaron, ...
07:17
<annevk>
tantek: I don't think they're doing a particularly good job, but they're making changes and copying our work
07:18
<tantek>
thank annevk. I'll look into it further.
07:18
<MikeSmith>
yoav: I saw one bug about preload and events but wasn’t sure how it was related (and didn’t look into it further)
07:19
<MikeSmith>
tantek: the thing is, they already made a bunch of decisions unilaterally without even attempting to ask the WG and record WG decisions
07:19
<MikeSmith>
as they are supposed to
07:19
<MikeSmith>
and as they would if they were actuallly sincere about operating by consensus
07:19
<tantek>
MikeSmith: so it really is off the rails completely then
07:19
<MikeSmith>
yes
07:19
<MikeSmith>
bingo
07:20
<MikeSmith>
they are operating way out of consensus at this point
07:20
<MikeSmith>
it is just the many people do not even care enough any longer to even bother to speak up
07:21
<tantek>
this is exactly what I need to be hearing. AB f2f meeting is next week.
07:22
<MikeSmith>
well at this point I do not know what difference it would make, because they have already decided to operate so far outside of the W3C’s own process
07:23
<MikeSmith>
they have basically fundamentally de-legitimized everything they are doing
07:23
<tantek>
in particular this is highly problematic since two WPWG co-chairs are now on the AB, so it is even worse to not be following W3C's own process
07:24
<MikeSmith>
I am sure they have some ends-justify-the-means rationalization about it
07:24
<tantek>
that's not going to fly at the AB
07:25
<MikeSmith>
meanwhile chaals in particular is quietly and unilaterally pulling stunts like this: https://github.com/w3c/html/pull/484/commits/19c383bbce5f032f8e66470d44a95f52f4f0f940
07:26
<MikeSmith>
“resetting” the acknowledgements for a document they are hostile-copying against the wishes of the authors who actually created
07:26
<MikeSmith>
“resetting” the acknowledgements by essentially removing 99% of them
07:27
<tantek>
that seems quite rude
07:27
<MikeSmith>
it is gall on top of rudeness motivated by genuine pathology
07:29
<MikeSmith>
he replaced the real acknowledgements with thanks to a list of people who have had zero to do with this actual document but instead worked on ancient previous versions of HTML specs at the W3C
07:29
<MikeSmith>
and on XHTML “family” specs
07:31
<tantek>
that is very odd indeed
07:33
<MikeSmith>
tantek: in the little bizarro universe he has constructed for himself around this, I am sure it all makes perfect sense to him
07:33
<MikeSmith>
it is certainly consistent with the rest of the crazy shit he has been doing there
07:34
<MikeSmith>
loose cannon on personal vendetta as chair of a WG
07:34
<tantek>
I really do not understand the motivations for these actions. Like why is it worth the personal time to bother with.
07:35
<MikeSmith>
neurosis
07:35
<MikeSmith>
is the motivation
07:35
<MikeSmith>
if not something worse
07:35
<MikeSmith>
going back years now
07:35
<tantek>
it just looks petty at this point
07:35
<MikeSmith>
petty payback time
07:35
<MikeSmith>
yeah
07:35
<tantek>
lol
07:36
<MikeSmith>
sad
07:36
Ms2ger
waves
07:36
<tantek>
yeah, I'm off to get breakfast here in London
07:36
<MikeSmith>
and extra sad that he has managed to take of all of public-webapps down with him
07:36
Ms2ger
should do likewise
07:36
<MikeSmith>
hai
07:37
<Ms2ger>
Chaals on an empty stomach is not something I recommend
07:57
<ondras>
annevk: why is there only a very small amount of response http headers available to JS for crossorigin requests, even if the request is explicitely allowed via CORS?
08:00
<Mek>
ondras: you mean why there is a Access-Control-Expose-Headers header?
08:02
<ondras>
Mek: well I mean why is there https://fetch.spec.whatwg.org/#cors-safelisted-response-header-name
08:03
<Mek>
ondras: well, that's kind of the same question. That's the list of headers that is available even without explicitly allowing any via the CORS Access-Control-Expose-Headers header
08:04
<ondras>
Mek: right. so I would like to know why there is a separate way to whitelist headers which somehow duplicates the fact that the response itself -- the body -- was already made available via Access-Control-Allow-Origin
08:04
<ondras>
Mek: or, in other words, why does one want to hide headers but make the body public
08:11
<annevk>
ondras: we want explicit consent for both
08:11
<annevk>
ondras: to avoid exposing debug headers and such
08:17
<ondras>
annevk: interesting, although I probably do not exactly see the motivation/usecase... the server owner already made the guarantee that the client may process his response data
08:31
<annevk>
ondras: it just makes it a little trickier to copy-and-paste yourself into a bad situation
08:32
<annevk>
ondras: a lot of CORS is motivated by that
08:32
<annevk>
ondras: not really my preference, but so be it
08:33
<ondras>
okay, thanks
11:15
<zcorpan>
MikeSmith: i suppose case COMMENT_LESS_THAN_BANG_DASH: should be moved down to
11:15
<zcorpan>
case COMMENT_END_DASH:
11:15
<zcorpan>
case COMMENT_START_DASH:
11:15
<zcorpan>
errEofInComment();
11:15
<zcorpan>
/* Emit the comment token. */
11:15
<zcorpan>
emitComment(1, 0);
11:49
<gsnedders>
zcorpan: how convinced are we that this change introduces no behaviour changes?
11:49
<gsnedders>
zcorpan: modulo parse errors, obviously
11:51
<zcorpan>
gsnedders: i found one bug in MikeSmith's impl from the tests in the PR, see https://github.com/whatwg/html/pull/1356#issuecomment-225839546
11:51
<zcorpan>
gsnedders: but we need to put those tests into html5lib tokenizer tests or something, too
11:51
gsnedders
is quite worried about unintentional changes here
11:52
<zcorpan>
gsnedders: yep, that's a risk. please review :-)
11:52
<gsnedders>
Hard to review every possible transition
11:52
<zcorpan>
i tried to be as careful as i could be, but i could have made a mistake
11:53
<gsnedders>
And I don't really want to add parse errors to formal model of the tokenizer to verify this
11:53
<zcorpan>
doesn't html5lib already test for parse errors in the tokenizer?
11:54
<gsnedders>
yeah, sure
11:55
<zcorpan>
https://github.com/whatwg/html/pull/1356#issuecomment-222793570 is intended to cover all relevant cases (except U+0000)
11:55
<gsnedders>
but proving equivalence modulo parse errors based on an implementation instead of a formal model is virtually impossible
11:56
<zcorpan>
ah, ok. yeah i suppose i don't have a mathematical proof
11:57
<gsnedders>
and given we know the current spec is web compat I'm very hestitant about introducing changes to it that we /think/ are right but haven't really tested web compat affect of
12:00
<zcorpan>
it's 3 new states, there are a finite number of new cases to consider. no? if the parsed tree is the same for those tests, there's no effect for web compat
12:02
<gsnedders>
yeah, mostly
12:02
<gsnedders>
still needs a kinda handwavey argument around "current node" though
12:04
<zcorpan>
what about current node?
12:07
<gsnedders>
you need to prove it's dead after that state
12:09
<MikeSmith>
zcorpan: yeah, thanks for catching that (moving the error case for COMMENT_LESS_THAN_BANG_DASH)
12:09
<MikeSmith>
was just careless there
12:10
<MikeSmith>
gsnedders: what zcorpan said (about your concern this introduces no behaviour changes)
12:10
<gsnedders>
MikeSmith: which thing he said?
12:11
<MikeSmith>
gsnedders: despite that modulo the one flub I made in my implementation, this is extremely low risk
12:12
<MikeSmith>
it is really only affecting the parse-error behavior, which most implementations do not implement anyway
12:12
<MikeSmith>
the mistake I made was not because of any lack of clarity in the spec text
12:15
<MikeSmith>
and no risk of new problems with web compat, because as zcorpan said there is not actually any effect on web compat
12:15
<MikeSmith>
if you (re)review the whole thing I think you will see that for yourself
14:16
<MikeSmith>
zcorpan: fixed that tokenizing issue and pushed
14:16
<MikeSmith>
https://checker.html5.org/parsetree/?parser=html5&content=%3C%21--%3C%21-&submit=Print+Tree
14:16
<zcorpan>
MikeSmith: yep saw that. writing a comment in the PR :-)
14:18
<MikeSmith>
k
14:38
<MikeSmith>
botie, inform zcorpan before merging that PR I wonder if we should try to secure agreement from more implementors for it, or else flag it in the spec with a big editorial note for now
15:50
<Domenic>
This PR doesn't impact implementers at all, does it?
15:51
<gsnedders>
Domenic: it *shouldn't*, but we should at least have implementers convinced that changing the state machine as it is in the spec is worth the risk and have them agree that the changes don't have any affect
15:52
<Domenic>
I don't really agree. I think the editors should be able to make editorial changes and we should be able to trust them to do so.
15:57
<gsnedders>
I think plenty of us would much rather some parts of the spec, including the parser, be pretty much frozen
16:01
<MikeSmith>
it does affect those who implement the parse errors and expose them
16:01
<MikeSmith>
which Henri does for View Source in Firefox
16:02
<MikeSmith>
and I thought the parse5 implementor was planning to support all the parse errors
16:02
<MikeSmith>
once we have explicit identifiers for all of them in the spec
16:03
<gsnedders>
I'll try and work on that over the summer, while I update html5lib to match the spec.
16:04
<MikeSmith>
fwiw I believe that because this is user-facing change, it should be motivated by trying to do what is best for users
16:04
<MikeSmith>
and in that regard it is not just editorial
16:05
<MikeSmith>
certainly it makes a significant difference in the HTML checker behavior
16:05
<gsnedders>
Now I don't expect it to be totally frozen, but I think we should have a pretty high barrier before changes
16:05
<MikeSmith>
I behavior change that I assert is a change for the better for users
16:07
<MikeSmith>
gsnedders: I think another way to look at it is that is maybe the last chance we might still have to get in a change like this before we are really stuck forever with the current legacy suboptimal-for-users behavior
16:08
<MikeSmith>
let’s try to change it now while we still have some chance
16:09
<gsnedders>
MikeSmith: why would we not have the chance to make it lateR?
16:26
<MikeSmith>
gsnedders: because I think the longer we wait to make a change like this, the more likely it is that the answer will be, “Too late to make this kind of change”
17:40
<MikeSmith>
botie, inform zcorpan before merging that PR I wonder if we should try to secure agreement from more implementors for it, or else flag it in the spec with a big editorial note for now
17:40
<botie>
will do