05:00 <dynarr> why can HashMap.insert() return an error? my first though would be resize fails due to OOM, but then... wouldn't any function that allocated memory be partial? is that actually the case and i missed it?

05:04 <njbartlett> dynarr: It looks like in practice it never does return an error. Lines 136-139 of map.pony

05:04 <njbartlett> else

05:04 <njbartlett> // This is unreachable, since index will never be out-of-bounds.

05:04 <njbartlett> error

05:04 <njbartlett> end

05:10 <dynarr> :S hmm

05:11 <dynarr> i created a similar scenario using an iterator (not in a for loop), where i switch on .has_next() but still have to wrap .next() in a try

05:12 <dynarr> i would have expected both of those APIs to return (None | V)

05:12 <dynarr> but i guess a match for that isn't really any better than a try block :p

05:25 <njbartlett> Oh I think it is

13:58 <emilbayes> Im trying to build pony-sodiym, but get this error: http://pastebin.com/rvPk4CCX

13:58 <emilbayes> does anyone know what it means?

13:59 <emilbayes> Do I have an outdated LLVM or something?

14:13 <Praetonus> emilbayes: This is most likely something wrong in the code generator. Could you open an issue on github with your code (if possible reduced to a minimal case), your version of ponyc and your version of LLVM?

14:14 <emilbayes> Praetonus: I can try. It's my first time doing anything in Pony though, and I wanted to contribute to pony-sodium, so not sure if I can reduce it

14:15 <Praetonus> And one last thing, can you rerun the compiler with the --extfun flag and post the result here?

14:18 <emilbayes> Praetonus: get exactly the same output

14:18 <emilbayes> Should I try updating my llvm? I think it's fairly recent, but does pony require a certain version?

14:18 <emilbayes> arr will try with one of the versions in the readme

14:19 <Praetonus> I don't think it's LLVM. From your gist you're using 3.8.0, which we support

14:19 <emilbayes> Hmm, I just recompiled HEAD of ponyc too

14:20 <Praetonus> The assertion failure talks about AVX. We have unsolved issues with AVX, it's possible you're encountering these

14:21 <emilbayes> So should I open an issue? I'm can have a look at the other issues but I'm not sure I'd have a clue if I'd be making a duplicate

14:23 <Praetonus> Yes, please open one

14:24 <emilbayes> Done :) https://github.com/ponylang/ponyc/issues/1176

14:25 <Praetonus> Thanks

14:26 <emilbayes> I'll try with the docker version in the meantime

15:10 <SeanTAllen> travisgriggs: you can cross compile, we do, its just a little bit of a pain right now.

15:11 <SeanTAllen> njbartlett: the problem with (None | V) vs error is that you have to pay the cost of the match on every usage of the None | V which especially in hot paths can be a real perf hit vs error which will almost never happen. A single match is less expensive than a single error but when you start looking at it in bulk, it becomes an issue.

22:52 <runehog_> I'm trying to understand why _TimingWheel is package-private. It's a solid bit of technology (and I'd like to use it for audio-rate timing, i.e. triggered when a certain # of samples have passed rather than via asio events...)

22:53 <runehog> if it's just a matter of missing test coverage, I'd be happy to take that on.