is the planned `sockets` module going to be async?...

# io

[kevin.cianfarini]

is the planned sockets module going to be async? If so, is it planned for these async sockets to support TCP/IP backpressure via suspending functions?

[altavir]

As far as I know, plans for suspended io are, well, suspended, at the moment. It actually requires a lot of work, so it will probably come after regular io is more or less complete.

[kevin.cianfarini]

ah gotcha. IO core is planned to be released with 1.4, right?

[altavir]

No, as far as I know, it is not attached to any mayor kotlin release. It requires some work and testing right now. I would work on it, if I had a litle bit of time. I hope I will be able to do it soon.

[kevin.cianfarini]

I'd like to work on it too

[kevin.cianfarini]

specifically the async bit

[kevin.cianfarini]

I've been reading a lot about kqueues and java nio and have a working implementation of reading async with suspend via java nio

[altavir]

Async not implemented at all yet. We probably need to discuss it more. I think that @elizarov has some idea about how he wants to organize it.

[kevin.cianfarini]

finding time to contribute is hard, through

[kevin.cianfarini]

ah okay, gotcha

[altavir]

Currently it is more important to test and find use-cases and better API.

[altavir]

I have a fork in our lab repository, but it is out of date and contains some solutions of my own. Some of them I like, some proved to be not so good after all.

[kevin.cianfarini]

Gotcha

[kevin.cianfarini]

Well, I check up on the repo pretty much daily. If anyone knows some bite sizes pieces of work I can contribute to, let me know

[kevin.cianfarini]

(along side testing)

[jimn]

even windows is basically netbsd networking. it is my hope that whatever comes out of kotlin for async networking is more relevant to posix than java NIO was when it was conceived. the landscape is different now. posix has more or less prevailed over redmond.