| 00:24:52
| * Fishrock123 | joined |
| 00:25:39
| * AtumT | quit (Remote host closed the connection) |
| 00:42:04
| * Fishrock123 | quit (Remote host closed the connection) |
| 00:47:44
| * Fishrock123 | joined |
| 01:27:35
| * Fishrock123 | quit (Quit: Leaving...) |
| 02:22:57
| * jwalden | quit (Quit: ChatZilla 0.9.92-rdmsoft [XULRunner 35.0.1/20150122214805]) |
| 04:11:09
| * jmdyck | quit (Remote host closed the connection) |
| 07:38:11
| * wycats | quit (Read error: Connection reset by peer) |
| 07:38:34
| * wycats | joined |
| 09:11:02
| * not-an-aardvark | quit (Quit: Connection closed for inactivity) |
| 10:25:08
| * mylesborins | quit (Quit: farewell for now) |
| 10:25:39
| * mylesborins | joined |
| 10:25:50
| * bathos | quit (Quit: bathos) |
| 12:04:49
| * jmdyck | joined |
| 12:20:11
| * AtumT | joined |
| 12:20:43
| * bradleymeck | joined |
| 12:43:50
| * Jasuruzakovgmail | quit (*.net *.split) |
| 12:53:54
| * Jasuruzakovgmail | joined |
| 13:01:18
| * gibson042 | joined |
| 13:39:08
| * gibson042 | quit (Ping timeout: 255 seconds) |
| 13:49:47
| * bathos | joined |
| 13:57:27
| * gibson042 | joined |
| 14:39:36
| * bathos | quit (Quit: bathos) |
| 15:34:47
| * bathos | joined |
| 15:58:26
| * bradleymeck | quit (Quit: bradleymeck) |
| 16:13:50
| * bathos | quit (Quit: bathos) |
| 16:26:30
| * Fishrock123 | joined |
| 16:27:24
| * bathos | joined |
| 16:27:26
| * bathos | quit (Client Quit) |
| 17:15:46
| * bathos | joined |
| 17:40:31
| * jwalden | joined |
| 18:06:00
| * bathos | quit (Quit: bathos) |
| 18:07:40
| * bathos | joined |
| 18:40:36
| <Bakkot> | bterlson: I understand Edge has module support now |
| 18:40:50
| <Bakkot> | you may be interested in, I just updated https://bakkot.github.io/test262-web-runner/ to support modules |
| 18:40:59
| * jwalden | spams test262 with more important tests like https://github.com/tc39/test262/pull/1299 |
| 18:41:01
| <jwalden> | :-) |
| 18:41:03
| <Bakkot> | via service workers |
| 18:45:25
| <Bakkot> | jwalden: your indentation style hurts me :( |
| 18:45:40
| <jwalden> | 2space4lyf |
| 18:45:52
| <Bakkot> | also, wait, this looks like it would pass even if the iterator doesn’t throw |
| 18:46:03
| <Bakkot> | or, for that matter, even if it does |
| 18:46:22
| <jwalden> | hrm, would it? I tested without my bugfix and it failed, with it and it passed |
| 18:46:31
| <Bakkot> | Maybe I’m just confused, then |
| 18:46:38
| <jwalden> | but I am certainly open to the possibility that I don't know what I'm doing here :-) |
| 18:47:33
| <Bakkot> | Oh, wait, right, I forgot how $DONE works. |
| 18:47:40
| <jwalden> | so if I read the async docs correctly, if |async function f| returns a truthy value -- namely either of those two strings -- that'll get passed as argument to $DONE and that...yeah |
| 18:47:58
| <Bakkot> | (i.e. it requires its argument to be falsy) |
| 18:48:41
| <Bakkot> | I don’t _think_ the style is generally to rely on the condition in $DONE, but haven’t read too many of the async tests |
| 18:48:44
| <Bakkot> | ignore my rambling |
| 18:49:38
| <jwalden> | :-) |
| 18:49:45
| <jwalden> | blind leading the blind here |
| 18:49:55
| <jwalden> | where are the adults? |
| 18:50:08
| <Bakkot> | leobalter, rwaldron ^ :) |
| 18:50:17
| <bterlson> | lol I was just about to ping leobalter |
| 18:50:43
| <bterlson> | should we call him "papa leo" from now on? |
| 18:51:32
| * Fishrock123 | quit (Remote host closed the connection) |
| 18:52:09
| * Fishrock123 | joined |
| 18:54:30
| <rwaldron> | jswalden, your test looks correct to me! |
| 18:54:45
| <rwaldron> | FWIW, I compared it to this .case: https://github.com/tc39/test262/blob/master/src/dstr-assignment-for-await/array-elem-trlg-iter-list-nrml-close-err.case#L53-L57 |
| 18:55:24
| <rwaldron> | 👍🏽 |
| 18:55:47
| <rwaldron> | Oh wait... |
| 18:55:58
| <rwaldron> | One thing, I can feedback here or on the PR? |
| 18:56:47
| * Fishrock123 | quit (Ping timeout: 252 seconds) |
| 18:58:00
| <jwalden> | I'm watching both places :-) |
| 18:58:59
| <rwaldron> | Just this: |
| 18:59:08
| <rwaldron> | get [Symbol.iterator] |
| 18:59:53
| <jwalden> | that's mostly just a fencepost to be sure the operation doesn't accidentally proceed past the @@asyncIterator get/test, and if it does things will Die |
| 19:01:03
| <rwaldron> | Sorry, I just meant that it should be: "[Symbol.iterator]()", not "get [Symbol.iterator]()" |
| 19:01:35
| <jwalden> | rwaldron: no, that was intentional -- even the mere act of *getting* is too far, shouldn't proceed past that to calling |
| 19:01:48
| <jwalden> | I mean, sure, either would work |
| 19:01:55
| <jwalden> | throw-on-get is just more restrictive |
| 19:02:01
| <rwaldron> | Oh, because with get, you'll have this result: TypeError: fakeiter is not iterable |
| 19:02:15
| <rwaldron> | I assumed you wanted that Test262Error to come through |
| 19:02:32
| * Fishrock123 | joined |
| 19:03:20
| <jwalden> | hum, really? GetIterator's operation is "? GetMethod(obj, @@iterator)" so would just throw the T2E, and I think that would just propagate outward uncaught |
| 19:04:30
| <rwaldron> | Can I try to walk through this with you? Maybe I've misunderstood something, so that might help |
| 19:04:37
| <jwalden> | sure |
| 19:05:02
| <jwalden> | https://tc39.github.io/proposal-async-iteration/#sec-runtime-semantics-forin-div-ofheadevaluation-tdznames-expr-iterationkind last step is what would make that GetIterator call, and seems like that would throw |
| 19:06:57
| <rwaldron> | The normal execution of an asyncIter would first look for [Symbol.asyncIterator]() {} on an iterable, if it found that, it would call it (and so on); if it did not find one, it would then look for [Symbol.iterator]() {}, and if it found that, it would try to call it. |
| 19:07:32
| <rwaldron> | In this case, it finds [Symbol.asyncIterator], but it's not a function, it's this other icky thing and that should end the whole show right there |
| 19:08:31
| <jwalden> | rwaldron: yes, it'll call the icky thing and throw because the icky thing throws |
| 19:09:06
| <jwalden> | rwaldron: the [Symbol.iterator] part is in case there's an implementation bug or something where the bad asyncIterator isn't called |
| 19:09:13
| <rwaldron> | great, then my "concern" is downgraded to a "nit" and because it doesn't matter, nothing should ever get that far anyway |
| 19:09:26
| * rwaldron | nods |
| 19:09:29
| <rwaldron> | Cool |
| 19:11:05
| <rwaldron> | Excuse me bterlson |
| 19:11:13
| <rwaldron> | While we have your attention |
| 19:11:14
| <rwaldron> | https://github.com/bterlson/eshost/pull/41 |
| 19:11:35
| <rwaldron> | That's relevant for supporting this ickiness that jwalden has uncovered |
| 19:11:42
| <rwaldron> | thx in advance |
| 19:11:53
| <bterlson> | rwaldron: I am investigating presently, IIRC ch.exe has some API to get the actual ULEO |
| 19:12:01
| <bterlson> | will comment once I find it |
| 19:12:01
| <rwaldron> | Oh rad |
| 19:12:57
| <rwaldron> | I'm still trying to figure out how we might use %GetUndetectable() with d8 (as Bakkot mentions here https://github.com/bterlson/eshost/pull/41#issuecomment-337349144) |
| 19:14:20
| <rwaldron> | I kind of want to do some fancy footwork with eshost and sniff out calls to `uncallableAndIsHTMLDDA()` in the source as a way to signal adding "--allow-natives-syntax" to hostArgs |
| 19:14:50
| <rwaldron> | (in the d8 runtime) |
| 19:15:12
| <rwaldron> | And by that I mean d8 agent. Obviously. |
| 19:15:24
| <rwaldron> | ;) |
| 19:16:09
| <bterlson> | rwaldron: gah looks like the API is not available in ch.exe :/ |
| 19:22:23
| <jwalden> | rwaldron: whoops, was just looking at the test one last time to verify its failure in unpatched SpiderMonkey and saw a typo in it (plus an opportunity to put a bad-exception error message into the assert.sameValue message) -- mind taking a look at/merging the last micro-change on the same getiterator-async-funky branch? |
| 19:24:22
| <jwalden> | rwaldron: https://github.com/jswalden/test262/commit/fe90a3f0db4f80114f10a4e4c1f05d817bea7b30 |
| 19:24:56
| <jwalden> | looks like when you push to a merged branch, Github switches to claiming the particular referenced commit is what was merged, loses recognition of what the branch was that was asked to be merged originally |
| 19:25:00
| <rwaldron> | jwalden nope, don't mind |
| 19:29:43
| <rwaldron> | jwalden https://github.com/tc39/test262/commit/99ee383d3fc2e10342842cad40bad89db123fdc1 |
| 19:29:55
| <jwalden> | woo |
| 19:29:58
| <jwalden> | gentleman and a scholar |
| 19:30:21
| <rwaldron> | lol 👍🏽 |
| 19:58:17
| * bathos | quit (Quit: bathos) |
| 20:04:21
| * Fishrock123 | quit (Remote host closed the connection) |
| 20:04:48
| * jwalden | hopes other impls will expose ULEOs in their shells, bets there are widespread bugs of this nature across implementations |
| 20:04:59
| * Fishrock123 | joined |
| 20:06:03
| <Bakkot> | don’t necessarily need it in the shells! I will add the host hook to my web runner sometime soonish and just use the actual document.all |
| 20:06:49
| <Bakkot> | in fact, I’m on lunch break right now, one sec… |
| 20:08:25
| <Bakkot> | oh, wait. jwalden: uncallableAndIsHTMLDDA isn’t actually document.all? |
| 20:09:09
| <Bakkot> | because document.all is callable without throwing a type error |
| 20:09:12
| * Fishrock123 | quit (Ping timeout: 258 seconds) |
| 20:13:38
| * bathos | joined |
| 20:15:09
| * Fishrock123 | joined |
| 20:18:15
| <jwalden> | Bakkot: it *may* return document.all, but it isn't required to |
| 20:18:29
| <Bakkot> | document.all doesn’t meet the requirements, though, does it? |
| 20:18:42
| <jwalden> | Bakkot: Call(document.all, ..., <<>>) throws a TypeError |
| 20:18:56
| <jwalden> | Bakkot: ^ that's the peculiar requirement of the return value of uncallableAndIsHTMLDDA |
| 20:19:59
| <Bakkot> | … does it? |
| 20:20:02
| <jwalden> | having an [[IsHTMLDDA]] internal slot is an optional add-on behavior that is permitted by uncallableAndIsHTMLDDA's interface |
| 20:20:14
| <jwalden> | Bakkot: yes -- you're invoking it with too few arguments |
| 20:20:42
| <jwalden> | Bakkot: uncallableAndIsHTMLDDA() { return {}; } is also an entirely permissible implementation, because it meets the uncallability requirement |
| 20:20:56
| <Bakkot> | Ah. It does not throw when given no arguments in either chrome or safari |
| 20:21:00
| <Bakkot> | Hence my confusion. |
| 20:21:12
| <jwalden> | that...sounds wrong |
| 20:21:19
| * Fishrock123 | quit (Remote host closed the connection) |
| 20:21:39
| <Bakkot> | Domenic would know the spec here, I expect |
| 20:21:48
| <jwalden> | data:text/html,<script>try { document.all(); } catch (e) { document.write(e); }</script> |
| 20:21:52
| <jwalden> | TypeError: Not enough arguments to HTMLAllCollection.__legacycaller. |
| 20:21:55
| <jwalden> | ^ Firefox behavior |
| 20:22:15
| <jwalden> | and given we autogen the binding stuff for this I'd be very surprised if that's not the required behavior |
| 20:22:18
| * jwalden | consults specs |
| 20:22:46
| <jwalden> | oh, ugh |
| 20:22:48
| <jwalden> | https://html.spec.whatwg.org/multipage/common-dom-interfaces.html#HTMLAllCollection-call |
| 20:23:02
| <Domenic> | Yeah... document.all is callable.... |
| 20:23:18
| <Bakkot> | Domenic: the question is, does calling it with no arguments throw or not |
| 20:23:21
| <jwalden> | agreed it's callable, the question is its behavior on attempt to call |
| 20:23:22
| <Bakkot> | Firefox thinks it does, for some reason |
| 20:23:24
| <Domenic> | Why can't these just be WPT testing actual document.all, not this made up Mozilla JS she'll thing? |
| 20:23:29
| <jwalden> | ^ that URL seems pretty unambiguous |
| 20:23:29
| <Domenic> | *shell |
| 20:24:16
| <Bakkot> | Domenic: see discussion on https://github.com/tc39/test262/pull/1287 |
| 20:24:18
| <jwalden> | because WPT tests aren't run nearly as easily/frequently by people who primarily hack on an engine as an engine, not as an engine in a browser |
| 20:24:38
| <jwalden> | sure, we write for browsers, but the majority of our work is browser-agnostic |
| 20:24:40
| * bathos | quit (Quit: bathos) |
| 20:24:51
| <Domenic> | How frequently does this test need to be run? |
| 20:24:59
| <jwalden> | and for these tests, the vast bulk of the tests as written are ECMA functionality, and it's not really practical for WPT people to review them |
| 20:25:45
| <jwalden> | depends how often we modify the relevant code, or even just write it |
| 20:25:46
| <Domenic> | Isn't it normatively disallowed for other objects to have the IsHTMLDDA slot? |
| 20:26:27
| <jwalden> | I don't believe it is -- spec heavily discourages having objects have it, but it doesn't/can't really restrict it to just document.all because that's beyond its purview |
| 20:26:52
| <Bakkot> | Technically it’s a NOTE which I think implies not normative |
| 20:27:08
| <Domenic> | Meh, I'm gonna give up on this, I'm burned out on test262 discussions. Sorry, gotta duck out. |
| 20:27:08
| <Bakkot> | the “implementations should not create any [others]” bit, that is |
| 20:27:14
| * bathos | joined |
| 20:27:38
| <Bakkot> | aaaanyway jwalden I think the hook in test262 should be fixed to match the spec’d behavior of document.all |
| 20:28:44
| <jwalden> | definitely it's desirable/useful that a test here not require having a full browser build to test it, 'cause building a little JS engine versus a big browser is a very different degree of time requirement |
| 20:28:46
| <jwalden> | anyway |
| 20:28:50
| * jwalden | goes to fix document.all() |
| 20:35:17
| <jwalden> | now, it *may* be the case that .return === null or .@@asyncIterator === null causes a subsequent TypeError |
| 20:35:27
| <jwalden> | or rather, when invoking either returns null causes one |
| 20:37:27
| <jwalden> | "Let innerReturnResult be ? Call(return, iterator, « received.[[Value]] »)." "If Type(innerReturnResult) is not Object, throw a TypeError exception." |
| 20:37:46
| <jwalden> | well, ugh |
| 20:38:00
| <jwalden> | so in that case .return === document.all will cause a TypeError to be thrown |
| 20:38:49
| <jwalden> | well, won't *necessarily* |
| 20:39:20
| <jwalden> | on the other hand, if you can get received.[[Value]] to be "", then you would I think hit step 1 of https://html.spec.whatwg.org/multipage/common-dom-interfaces.html#concept-get-all-named and return null and cause that TypeError |
| 20:39:30
| <jwalden> | bah, this is all very sadmaking |
| 20:44:47
| * Fishrock123 | joined |
| 20:49:03
| * caitp | quit (Ping timeout: 248 seconds) |
| 20:53:16
| * Fishrock123 | quit (Remote host closed the connection) |
| 20:53:50
| * Fishrock123 | joined |
| 20:55:20
| * caitp | joined |
| 20:57:32
| * not-an-aardvark | joined |
| 20:58:16
| * Fishrock123 | quit (Ping timeout: 258 seconds) |
| 21:04:47
| * Fishrock123 | joined |
| 21:09:03
| * Fishrock123 | quit (Remote host closed the connection) |
| 21:42:53
| * bathos | quit (Quit: bathos) |
| 21:44:08
| * bathos | joined |
| 22:09:53
| * Fishrock123 | joined |
| 22:15:12
| * Fishrock123 | quit (Ping timeout: 260 seconds) |
| 22:20:32
| * bathos | quit (Quit: bathos) |
| 22:51:58
| * Fishrock123 | joined |
| 23:33:36
| * bathos | joined |
| 23:34:09
| * bathos | quit (Client Quit) |
| 23:34:47
| * bathos | joined |
| 23:38:10
| * Fishrock123 | quit (Remote host closed the connection) |
| 23:38:38
| * Fishrock123 | joined |
| 23:48:22
| * bradleymeck | joined |