01:01
<MikeSmith>
hsivonen: ^
10:30
<annevk>
Domenic: added a bunch of README.md files
10:49
<MikeSmith>
annevk: hey so for URL should we just drop the canonicalization tests from the wpt test suite or is what's in there now OK (I mean after the cleanup you did)
10:59
<annevk>
MikeSmith: hmm I missed those
11:00
<annevk>
MikeSmith: what is canonicalization testing? just serializing?
11:03
<MikeSmith>
annevk: Yeah it's only relevant to serialization afaict
11:03
<annevk>
MikeSmith: I guess we should at least rename it then
11:04
<annevk>
MikeSmith: we might at some point define a normalization/canonicalization algorithm I suppose that does certain things
11:07
<MikeSmith>
annevk: are those tests checking anything that's actually specified in the spec?
11:09
<annevk>
MikeSmith: looking
11:10
<annevk>
MikeSmith: what tests are you referring to exactly?
11:12
<MikeSmith>
annevk: the ones you cleaned up in https://github.com/w3c/web-platform-tests/commit/3c5a8de5b0cd49033fd6ef0954cc802b7682c769
11:12
<annevk>
MikeSmith: oh yes, they are
11:12
<MikeSmith>
annevk: OK good
11:12
<MikeSmith>
annevk: I was confused
11:13
<annevk>
Compared to RFC 3986 URL parsers in browsers do a fair bit of "normalization" / "canonicalization" right in the parser
11:13
<annevk>
E.g. something like lowercasing the scheme or the domain
11:18
<annevk>
MikeSmith: if I update the README file of the URL tests, can I do that without PR?
11:19
<MikeSmith>
annevk: yeah sure
11:21
<MikeSmith>
annevk: btw I guess it could be argued that RFC 3986 does actually implicitly deal with normalization/canonicalization by e.g., saying that the host and scheme should be compared case-insensitively
11:22
<MikeSmith>
annevk: though of course it doesn't specify parsing per se
11:25
<annevk>
MikeSmith: it also says %AA and %aa are identical, but it does not say what you have to transmit to the server based on what input
11:25
<MikeSmith>
ah ok
11:26
<annevk>
MikeSmith: so you can say HTTP is equal to hTTp when it's a scheme, but that does not answer what e.g. .protocol will return
11:26
<annevk>
MikeSmith: we could of course assume the parser does none of fixing up and instead define what needs to happen at all the end points
11:26
<MikeSmith>
right, I see now
11:27
<annevk>
but that seems less straightforward (it's kind of what CSS sometimes does, and it's a bit annoying when defining the CSSOM)
14:21
<Manishearth>
annevk: around?
14:22
<Manishearth>
Possible problem with the XHR spec
14:28
<Manishearth>
Compare it with the w3c spec
14:28
<Manishearth>
http://www.w3.org/TR/XMLHttpRequest2/#the-abort-method
14:28
<Manishearth>
if you see step 3, it also removes tasks in the DOM task queue related to the XHR object
14:29
<Manishearth>
the WHATWG spec only says that the fetch algorithm should be terminated
14:29
<Manishearth>
which doesn't terminate any queued load events
14:30
<Manishearth>
eg lets say the XHR object changes state to LOADING
14:30
<Manishearth>
and in the readystatechange handler, we invoke abort()
14:30
<Manishearth>
however, before doing so, we perform a heavy computation
14:31
<Manishearth>
so that the XHR finishes loading and queues a https://xhr.spec.whatwg.org/#handle-response-end-of-file event to the queue
14:31
<Manishearth>
now we do abort()
14:31
<Manishearth>
changing the state to unsent
14:32
<Manishearth>
after abort(), within the same handler, we call open() and send() again
14:33
<Manishearth>
that response-end-of-file event for the previous file will be processed after this
14:33
<Manishearth>
so we'll get the end of the previous file instead of the new file
14:36
<gsnedders>
is it possible to have an arbitrary domain forwarding all email to Gmail in the normal way through Postfix nowadays? My domain seems to constantly get IP blocked by Gmail beause of an "unusual rate of spam"
15:03
<GPHemsley>
gsnedders: Well quit spamming, then!
15:22
<gsnedders>
GPHemsley: but but but I don't send anything from that SMTP server!
15:22
<GPHemsley>
gsnedders: Well quit anti-spamming, then!
15:22
<gsnedders>
bah!
16:10
<annevk>
Manishearth: hmm, file a bug?
16:12
<Manishearth>
annevk: I'll do that, but do you think it's an issue or did I misinterpret the spec?
16:13
<Manishearth>
(probably tomorrow)
16:22
<annevk>
not sure
21:14
<terinjokes>
is there any other spec that has a bunch of methods that do effectively the same thing?
21:19
<annevk>
terinjokes: these are quite similar: https://fetch.spec.whatwg.org/#body
21:20
<terinjokes>
annevk: indeed, and what I was thinking "must be the result of running <<logger>> with <<blah>>"