04:18 <SeanTAllen> Pony 0.22.0 has been released: https://www.ponylang.org/blog/2018/05/0.22.0-released/

06:37 <hmans> \o/

14:02 <SeanTAllen> Pony-stable 0.1.2 has been released: https://www.ponylang.org/blog/2018/05/pony-stable-0.1.2-released/

16:14 <strmpnk> Is there a reason Pony requires the other exec bit to be set here?

16:14 <strmpnk> https://github.com/ponylang/ponyc/blob/master/packages/process/process_monitor.pony#L245

16:15 <strmpnk> Shouldn't we rely on the system to determine if something is permissible to execute?

16:17 <strmpnk> I see the commit, where we add the authority to the monitor, but it seems like this is at least an incomplete check (we ignore user and group bits) and I'm not sure what this check adds to the security model in this case if we have the authority (which would allow us to change that bit).

16:19 <strmpnk> I've been setting this bit for now but digging into this code a little further has confused me on this particular point.

16:47 <jbk> it does seem like it could be problematic if a platform is using ACLs

16:55 <strmpnk> If we want to keep this, I'd also switch the error we return from ExecveError. I spent some time very confused how I was getting an error but saw no such call in my strace output.

16:55 <strmpnk> Either way we chose, I'm happy to patch it. I'll open an issue for discussion.

17:20 <SeanTAllen> patch it!

17:27 <strmpnk> SeanTAllen: Which way? Just want to know what the core team thinks is best here.

17:28 <strmpnk> SeanTAllen: Granted, I believe it's your code too ;-) so if you don't have feelings, perhaps I'll send that PR in then.

17:30 <SeanTAllen> is it?

17:30 <SeanTAllen> i fixed some issue in there.

17:30 <SeanTAllen> ¯\_(ツ)_/¯

17:31 <SeanTAllen> when i added that, it might have been in haste...

17:31 <SeanTAllen> so i probably "did it wrong"

17:31 <strmpnk> It makes sense in some "being very careful" sense but I can't think of why we need it. I wasn't certain if there was an actual concern or security model that was being implemented beyond the authority addition.

17:32 <SeanTAllen> i dont remember. was there an issue associated with it?

17:33 <strmpnk> https://github.com/ponylang/ponyc/issues/1180

17:37 <strmpnk> I'll look at it again after I find some lunch here. I'll look at what other languages do and try to mirror that since I think the authority object is more of what we want anyway.

17:39 <SeanTAllen> strmpnk: i need to do another release to test a change i made to release builds so... maybe your fix will the the "all by itself" release issue

17:52 <strmpnk> hahah. Sounds good to me.

17:53 <strmpnk> bougyman: Hey, I was looking at the voidlinux packages today. Do you mind if I update ponyc for you?

23:12 <pzel> Hi, I ran across a situation that I don't understand today: https://gist.github.com/pzel/fef1416c2d29532257898e4640579283

23:12 <pzel> I'm trying to return the keys of an internal Map[String, ..] from an actor, via a promise

23:13 <pzel> So if I fist construct an Array[String] iso, push each of the keys() onto it, and `recover val` on it, it works as expected.

23:14 <pzel> However the one-liner fails to compile(https://gist.github.com/pzel/fef1416c2d29532257898e4640579283#file-example-pony-L18)

23:15 <SeanTAllen> yup

23:15 <SeanTAllen> that makes sense

23:16 <SeanTAllen> only sendable items are allowed to be used in a recover

23:16 <pzel> okay, so then I tried to extract that collect() outside, by first making an Array[String] ref and collecting on it

23:18 <pzel> (updated gist: https://gist.github.com/pzel/fef1416c2d29532257898e4640579283#file-example-pony-L31)

23:19 <SeanTAllen> yeah you cant use a ref inside a recover

23:19 <SeanTAllen> so...

23:19 <pzel> okay, so only iso + manual loop, as Iter works on refs

23:20 <SeanTAllen> try this...

23:20 <SeanTAllen> https://www.irccloud.com/pastebin/VerzMi68/

23:20 <SeanTAllen> that might work

23:21 <SeanTAllen> nope that is angry

23:21 <pzel> yeah, Iter want ref

23:21 <pzel> I thought trn is the pattern to use here, but Iter won't have any of it

23:22 <SeanTAllen> theodus does a lot of work with the itertools package

23:22 <SeanTAllen> id open an issue an ping him on it

23:23 <SeanTAllen> i dont use itertools so i cant speak to it really

23:23 <pzel> will do.

23:23 <SeanTAllen> maybe there is a reason for ref

23:24 <SeanTAllen> and actually ya, you cant use trn or iso there

23:24 <SeanTAllen> unless collect were to return the value too

23:24 <SeanTAllen> interesting

23:34 <jemc> what we're missing here is paul's RFC about a notation for isolated regions

23:34 <jemc> same thing I'm really wanting in my CRDT library

23:35 <jemc> a way to tell the compiler "I want to be able to use two `ref`s from different isolated regions alongside eachother in the same code, but guarantee that they don't actually touch eachother"

23:36 <jemc> or at least, that's what it boils down to

23:37 <jemc> it would dramatically expand the category of safe code that pony can prove as safe, and from what I've done thought-experiments on so far, it seems to handle a lot of the cases where you want "borrowing" I think

23:38 <jemc> (it also adds another layer of stuff-you-have-to-think-about when it comes to capabilities, so that's the downside)

23:39 <SeanTAllen> im looking forward to seeing his RFC