| 15 Jan 2022 |
|  Not_A joined the room. | 21:22:11 |
| 16 Jan 2022 |
|  PatAPizza joined the room. | 23:26:13 |
| 17 Jan 2022 |
 Yorick Peterse | Making some solid progress with the type checker. Currently working on defining/checking method signatures, which is a bit tricky due to different types of methods having different rules | 16:00:19 |
| Yorick Peterse | e.g. module methods can't yield, but instance methods can, and for instance methods the type of self/Self differs based on the presence of a when clause | 16:00:45 |
| Yorick Peterse | e.g. for class Array[T] { fn to_string when T: ToString }, self is basically Array[T: ToString] rather than just Array[T] | 16:01:20 |
| Yorick Peterse | * e.g. for class Array[T] { fn to_string when T: ToString { ... } }, self is basically Array[T: ToString] rather than just Array[T] | 16:01:30 |
| Yorick Peterse | I think the biggest obstacle at the moment (and probably this week as a whole), is that the various method types to implement are similar enough that you want to reuse code, but different enough that this isn't easy | 19:16:39 |
| Yorick Peterse | I think I got instance method signatures done, just not entirely sure yet if I'm happy with the code, but that's a problem for tomorrow 🙂 | 22:56:02 |
| 18 Jan 2022 |
|  ライアン changed their display name from Ryan to ライアン#2827. | 00:24:21 |
| ライアン changed their profile picture. | 00:24:23 |
| ライアン changed their display name from ライアン#2827 to ライアン. | 00:24:24 |
|  injuly#6820 changed their profile picture. | 06:09:33 |
|  Arnaud[Madlib] changed their display name from Askero to Arnaud[Madlib]#7886. | 15:15:16 |
| Arnaud[Madlib] changed their display name from Arnaud[Madlib]#7886 to Arnaud[Madlib]. | 15:15:17 |
| Yorick Peterse | Now for the tricky bit: disallowing ref T as input/output for async methods | 15:22:19 |
| Yorick Peterse | Verdagon#9572: How does Vale handle refs and concurrency, if at all? The challenge is 1) not allowing you to send refs across threads 2) not allowing sending of owned values across threads if they still have refs. I've been looking into this, and it seems you basically can't handle that unless: 1) you insert runtime checks (which can get expensive) 2) disallow moving when refs exist (limiting) 3) use a different approach | 17:36:38 |
| Yorick Peterse | For that last one: an option is to disallow moving when a ref exists, then introduce weak refs that do allow moving. You still can't send them across threads, but you can remove the need for ref counting at runtime | 17:37:16 |
| Yorick Peterse | The problem is that you may still need to insert a check when sending to clear any weak refs | 17:37:35 |
| Yorick Peterse | Actually that check is very similar: either you check ref counts and error when there are any, or you clear weak refs and potentially error when you try to load one later | 17:38:03 |
| Yorick Peterse | I think the weak ref approach is pretty similar to Vale's generational references, except it relies on indirection opposed to fat pointers | 17:38:27 |
| Yorick Peterse | Generics make this annoying too, because if you have e.g. async fn some_message[T](values: Array[T]) { ... }, the compiler has to check every call-site to make sure that whatever is passed to values doesn't contain a ref or weak | 17:39:40 |
| 20 Jan 2022 |
|  [cthulhu] changed their profile picture. | 04:44:33 |
| [cthulhu] changed their display name from ApacheActual#6945 to [cthulhu]. | 04:44:35 |
|  Futuresight 𝄢᮷⁂⁘♫∞ joined the room. | 21:54:03 |
| 22 Jan 2022 |
|  redninja106 joined the room. | 04:07:35 |
|  arwn#2335 changed their profile picture. | 22:32:46 |
| 23 Jan 2022 |
|  Jackojc#5426 changed their display name from Jackojc [wot++/klaxon] to Jackojc#5426. | 15:44:33 |
| 24 Jan 2022 |
|  ForeignSasquatch#6574 changed their display name from ForeignSasquatch to ForeignSasquatch#6574. | 03:46:29 |
|  Nikos Plugachev the Piccolist changed their display name from Nikos Plugachev the Piccolist to Nikos Plugachev#2827. | 04:39:56 |
| Nikos Plugachev the Piccolist changed their display name from Nikos Plugachev#2827 to Nikos Plugachev the Piccolist. | 04:39:57 |
