| 07:44 | <annevk> | The inability to moderate review spam on GitHub is rather annoying |
| 08:00 | <annevk> | andreubotella: https://github.com/whatwg/html/issues/3223 |
| 08:02 | <andreubotella> | annevk: I see there' |
| 08:03 | <andreubotella> | I see there's a PR with missing tests, and the linked wpt PR was merged |
| 08:03 | <andreubotella> | I'll take a look later to see if the tests are enough |
| 08:04 | <annevk> | andreubotella: from https://github.com/whatwg/html/pull/3276/files I wonder if we need to invoke the percent-encoding algorithm in the URL Standard for some stuff |
| 08:05 | <annevk> | andreubotella: although at some point I guess we'd be getting really close to replacing the RFC completely |
| 08:05 | <annevk> | which maybe we should just do, it's not a very complicated format |
| 08:09 | <andreubotella> | I was thinking this would probably need its own percent-encode set, but it's quite an odd one, what with not including \x1B or the non-ASCII bytes |
| 08:11 | <andreubotella> | well, we couldn't use percent-encode after encoding anyway because the entity references wouldn't have to be percent-encoded |
| 08:11 | <andreubotella> | I'll think about it |
| 08:12 | <annevk> | andreubotella: well, that's what percent-encode after encoding does, it doesn't percent-encode when the encode or fail returns fail |
| 08:12 | <annevk> | or fail algorithm* |
| 08:13 | <annevk> | Those other bits are indeed weird though, so maybe it needs its own thing |
| 15:54 | <croraf> | I'm checking this example: https://streams.spec.whatwg.org/#example-rs-push-backpressure . There is a note in the code snippet |
| 15:55 | <croraf> | "This is called if the internal queue has been emptied, but the stream’s consumer still wants more data." |
| 15:55 | <croraf> | I don't understand why this part after the comma (,) is present. Seems like it is irrelevant. |
| 15:55 | <croraf> | What would it mean "the stream's consumer still wants more data" technically? |
| 15:56 | <andreubotella> | if the queue is empty and the consumer isn't gonna read anymore, you don't need to push more data into the queue |
| 16:21 | <croraf> | How you know that the consumer isn't gonna read anymore? |
| 16:23 | <croraf> | The pull is gonna be called as soon as the queue has been emptied below the highWaterMark, without any other condition. |
| 16:23 | <croraf> | andreubotella, |
| 16:23 | <Domenic> | If the consumer doesn't call read(), then they don't want more data |
| 16:24 | <Domenic> | I.e. if the consumer doesn't call read(), then the queue is higher than the highWaterMark, so pull() won't be called. |
| 16:25 | <croraf> | Let's assume the hwm is 100 and the queue is at 50. The pull() won't be called if reader.read() is not called? |
| 16:27 | <Domenic> | If the queue is at 50, then the consumer wants more data |
| 16:27 | <Domenic> | It's only when the queue is above the HWM that the consumer doesn't want more data |
| 16:28 | <croraf> | The "consumer wants more data" is kind of ambiguous. |
| 16:28 | <Domenic> | Yes, it's a comment in an example; it's not an algorithm in a spec. |
| 16:29 | <croraf> | I think this comment is just confusing the matter. As far as I see it now. |
| 16:29 | <croraf> | As soon as queue is bellow hwm the pull is going to be called repeatedly. This is the only thing that is relevant, I think. |
| 16:30 | <Domenic> | You're probably best off ignoring it then :) |
| 16:31 | <croraf> | What does it mean that the pull() is called repeatedly. |
| 16:31 | <Domenic> | Ultimately, I think you're best off reading the algorithms to answer these questions. |
| 16:32 | <benjamingr__> | https://streams.spec.whatwg.org/#queuing-strategies read how queuing works |
| 16:35 | <croraf> | Thanks benjamingr__ I read it, I don't think it answers my question. |
| 16:36 | <benjamingr__> | then I recommend you do what domenic suggested and go through the algorithm carefully |
| 16:43 | <croraf> | Thanks benjamingr__ , can you point me to which algorithm to consult for clarifying what it means "be called repeatedly" from https://streams.spec.whatwg.org/#dom-underlyingsource-pull |
| 16:53 | <andreubotella> | I think I had a somewhat wrong impression of how the hwm and pull worked |
| 18:39 | <croraf> | Why isn't in here https://streams.spec.whatwg.org/#rs-class-definition the underlyingSource of the constructor, defined as being of type UnderylingSource? |
| 18:43 | <croraf> | OK, I just ran into a Note explaining why. |