wishlist: updatel channel.scm

[kristianlm]

Hi,

I noticed your channel.scm is using the gochan implementation. Cool!

I've recently rewritten gochan from scratch and, while this hasn't been tested out-in-the-wild yet, it might be worth considering to upgrade. The new version has some big improvements (you can select on send channels, timeouts are now ordinary channels).

Cheers!

[alvatar]

Awesome @kristianlm !!
Thanks for this!
One question: do you have an idea of any potential limitations we might encounter with Gambit's new multithread system (real OS threads + green threads)

[kristianlm]

I'm afraid I have no experience with Gambit so I can't comment on its multithreaded behaviour. But I suppose that if Gambit's mutex-lock! and mutex-unlock! still behave the same way, I don't see a reason why this shouldn't still work. gochan is using mutexes to coordinate reads/writes to a channel's memory, and condition-variables to notify/unblock sender-/receiver-threads.

[alvatar]

sorry for the delay on this @kristianlm ! I'll get to it ASAP

[alvatar]

Hey @kristianlm. You seem to be doing a lot of updates recently on gochan. Do you plan on freezing a version anytime soon?

[kristianlm]

Hi Alvatar,

Yes, I've been active trying to clean up the API and making it better. The problem is that I haven't had much time to use gochan in practice! Since version 2.1, there's only been 1 API change though: gochan-select will now no longer evaluate your clause body if you don't specify an ok flag. Otherwise, the API is relatively stable.

However, I just realized that version 2.0 introduces a matchable egg dependency, and that's a dependency you may not have access to?

[alvatar]

Thanks @kristianlm
It'd need to be ported, but we certainly have matching.

[kristianlm]

It's a great pattern-matching scheme library, so it's definitely not a bad library to include :) I suspect Alex Shinn has done a great job already of making matchable easily portable.

[alvatar]

Hi @kristianlm, I've started porting your new version. I think it's feasible and not too hard. I'm going to wait on you to freeze a version if that's ok. Would you be able to let me know when you have a version that you consider relatively stable, please?

[kristianlm]

I will keep you updated on the gochan status. The problem is, however, that it's not in production use anywhere yet. So I don't know what you've got in mind when you say "relatively stable" :D

[alvatar]

I just mean that the API isn't expected to change for a while. How's that sound?

[kristianlm]

Oh of course, you mean api stability. I'll give it a week and see if I can think of any changes that need to be done and get back to you then if the API still covers the use-cases I've run into!

Cheers and thanks for dealing with this!

[alvatar]

Cool!
No rush at all. Thanks!