| 26 Nov 2020 |
 yorickpeterse | At some point I definitely want multi-lingual support for the manual, but I think that may take a while | 23:20:40 |
| 27 Nov 2020 |
 kiti_nomad | Ok | 03:59:33 |
| 3 Dec 2020 |
| yorickpeterse | I'm planning on overhauling how processes are defined and type-checked, some details (for those curious) can be found in https://gitlab.com/inko-lang/inko/-/issues/227#note_458744618.
Basically the idea is to make actors proper types (like objects), and turn message passing more into an RPC system, instead of being able to send arbitrary types at any point in time | 01:07:47 |
| yorickpeterse | There's some other work that needs to be done first though, such as reworking the object memory layout (without this, the proposed setup for actors wouldn't be efficient) | 01:08:14 |
| yorickpeterse | Plus I need some more time anyway to think this through | 01:08:26 |
| yorickpeterse | * Plus I need some more time anyway to think this through | 01:08:29 |
| 7 Dec 2020 |
|  erlend.sh changed their display name from erlend_sh to erlend.sh. | 01:00:24 |
| erlend.sh set a profile picture. | 01:00:48 |
| yorickpeterse | I managed to (finally) make the Object type a trait, instead of it being an object all other objects are an instance of | 23:44:41 |
| yorickpeterse | That in turn finally opens the way of fully getting rid of inheritance at the VM level | 23:44:50 |
| yorickpeterse | Which will massively simplify method lookups | 23:44:58 |
| yorickpeterse | Downside is that Hash/Equal are no longer implemented for objects by default, but I actually like that | 23:46:00 |
| yorickpeterse | There are plenty of objects for which implementing Equal is nonsensical | 23:46:10 |
| yorickpeterse | e.g. performing structural equality (A == B) on two sockets makes no sense | 23:47:55 |
| yorickpeterse | referential equality (now A.same_object?(B)) is totally fine however, because it's only True if A and B are the exact same objects | 23:48:19 |
| 8 Dec 2020 |
| kiti_nomad | Like table in lua? | 20:30:02 |
| yorickpeterse | I'm not very familiar with Lua, but it indeed seems to only provide referential equality by default | 20:43:08 |
| yorickpeterse | (for tables) | 20:43:12 |
| 14 Dec 2020 |
| yorickpeterse | And optional types are a goner too: https://gitlab.com/inko-lang/inko/-/commit/a276a68c79b8744ef4714bad03d8a684fdbb89bf | 21:44:57 |
| yorickpeterse | Took a few days to fix the hundreds of errors this caused, as a substantial portion of the runtime had to be updated | 21:45:48 |
| 23 Dec 2020 |
| yorickpeterse | Inko 0.9.0 has been released: https://inko-lang.org/news/inko-0-9-0-released/ | 20:48:37 |
| 1 Jan 2021 |
| kiti_nomad | Fine | 14:54:13 |
| 11 Jan 2021 |
| kiti_nomad | I think inko needs to run on the arm platform | 16:41:35 |
| kiti_nomad | iot will explode in 2021 | 16:42:06 |
| kiti_nomad | Maybe you need to develop a virtual machine for arm architecture | 16:45:11 |
| 12 Jan 2021 |
| yorickpeterse | I don't have any ARM hardware, so I can't test if Inko works on it. In theory it should work fine on ARM64. If anything doesn't work, contributions would be greatly appreciated :) | 22:47:12 |
| yorickpeterse | My current priorities are:
- Reducing the instruction set by using builtin/extern functions
- Adding an API for spawning OS processes
- Change panic unit tests to spawn a new process
- Remove panic handlers
- Add a new way of defining/using actors (similar to how Pony does things, but less annoying)
- Replace the GC with single ownership/move semantics (roughly based on https://researcher.watson.ibm.com/researcher/files/us-bacon/Dingle07Ownership.pdf)
| 22:53:53 |
| yorickpeterse | This alone is enough work for at least 3 months | 22:54:01 |
| yorickpeterse | So ARM testing isn't going to happen any time soon :) | 22:54:09 |
| yorickpeterse | I've basically reached a point where some old ideas/internals no longer work, and I have to change them now before I continue with the self-hosting compiler (don't want to support these crappy internals there) | 22:54:59 |
