00:09 <doublec> Perelandric: looking now

00:13 <doublec> Perelandric: in the call T.test(), T is the trait. It needs to be an instance of the trait.

00:14 <doublec> hmm, that doesn't work either

00:15 <doublec> I have to head out - will look again in a bit

01:10 <doublec> Perelandric: this is a cut down version of your issue http://pastebin.com/k25RqxTe

01:10 <doublec> Perelandric: unfortunately I don't know why it doesn't work

01:11 <doublec> The non generic equivalent works fine

01:11 <doublec> http://pastebin.com/ZenqAfU6

01:14 <doublec> huh, in the first paste it says Foo iso is not a subtype of T ref. Where did it get iso from?

02:18 <Perelandric> doublec: Thanks for the review. Yeah, I'm confused.

02:19 <Perelandric> Just so I understood, I thought `T.test()` would basically call the constructor with no args, and then call `.test()`

02:23 <doublec> Perelandric: you're right it does

02:25 <Perelandric> OK, good. I double checked with `T.create().test()` just to be sure.

02:26 <Perelandric> So maybe I should just file an issue on this one.

02:26 <doublec> Perelandric: Yes, I think that's a good idea

02:28 <Perelandric> doubled: Super, will do in a few hours. Just about to crash (again).

02:28 <Perelandric> Thanks again for the review and confirmation.

02:30 <doublec> np

02:55 <SeanTAllen> so to anyone who sees the above. There's a problem in that code. You have some type T that you want to be able to call a constructor for and then test() the existence of test() is guaranteed. There's no guarantee that `T` will have a 0 arg constructor.

03:09 <SeanTAllen> in the original version there's another issue that I see... MyTrait test returns a MyTrait

03:11 <SeanTAllen> I could create a GenericTest of Foo. And the Array would expect a Foo but, Foo can return a Bar or any other MyTrait and be legal, but that wouldn't be legal in GenericTest.

03:21 <doublec> SeanTAllen: good catch!

03:24 <doublec> SeanTAllen: this adds a constructor but same error http://pastebin.com/zuDKNZKA

03:28 <doublec> Oh I'm dumb. I see it.

03:29 <doublec> In "let b: T = a", because it's generic code T might not be something that Foo can assign too.

03:30 <doublec> The joys of being a polygot programmer - forgetting what works where

09:50 <posnea> not sure if this is the right place to report this, but I am following the ponylang tutorial and it compiled from source fine, but when I try to compile the hello world example I get a "Pass 'Move heap allocations to the stack' is not initialized. Verify if there is a pass dependency cycle. Required Passes: Segmentation fault (core dumped)" error

10:31 <Candle> doublec: heh, This weekend, I was writing Java, [xml!], Ruby, [graphviz], bash, pony, javascript, [html], [css]. Not the same project mind you...

10:37 <doublec> Candle: I remember going from doing a lot of lisp to doing c++. I was constantly writing (if (do this) ...)

10:42 <Candle> The java I was writing was a maven plugin to handle pony. Mostly because I thought I knew Maven pretty well, and I was fed up with trying to do pacjage management some other way.

10:42 <Candle> package*

10:43 <Candle> There are issues with Maven, it doesn't solve every problem, but it has to be better than nothing.

11:13 <doublec> Candle: did you get it working?

12:14 <Candle> doublec: Mostly, i need to re-discover my ability to publish to maven central.

12:15 <Candle> I'm also hoping that the --path option to ponyc can be specified multiple times!

12:17 <Candle> Given I was going to package a library as a zip/tar.gz (decision as to which is to be made) I need to work out where I'm going to unpack said archive so that it can be passed to --path.

12:19 <Candle> Finally, I need to document a whole load of decisions I've made, and my rational for choosing that option!

12:20 <Perelandric> SeanTAllen: Thank you, that makes sense. Though it would seem like making the trait generic would solve it...

12:21 <Perelandric> ...by making sure those that implement it return an instance of the same class from `test()`...

12:21 <Perelandric> ...but it doesn't seem to help. https://gist.github.com/Perelandric/19501e1d1e93271468077157bbf9be24#gistcomment-1849633

12:42 <SeanTAllen> posnea if you come back and see this, can you report on github?

12:45 <SeanTAllen> Perelandric: this is what you want... https://gist.github.com/SeanTAllen/ba0ce66d6396ed9c2e5b2538c1297904

12:46 <SeanTAllen> Your array has to have a lower bounds of MyTrait so it can hold any MyTrait

12:58 <Perelandric> SeanTAllen: Thank you, though in this case, I actually want it to be an Array of `T`.

12:58 <Perelandric> I see that it'll work if I do this: `let a = Array[MyTrait[T]]`

12:59 <Perelandric> ...with my generic version of `MyTrait`, though I can't quite comprenend why `T` alone doesn't work since it's already constrained.

13:00 <Perelandric> But for now, I'm just happy to get it to compile! :-)

13:04 <SeanTAllen> what do you mean by `T` alone Perelandric?

13:05 <Perelandric> I mean that if `T` is `Foo`, then the `a` array should be `Array[Foo]`

13:06 <SeanTAllen> Can you gist some code to make sure I understand exactly what you mean?

13:07 <Perelandric> Absolutely. I'll update my previous gist, and add what I think it should look like expanded.

13:10 <Perelandric> Here's the generic version that compiles and creates an instance of `GenericTest[Foo]`

13:10 <Perelandric> https://gist.github.com/Perelandric/19501e1d1e93271468077157bbf9be24#gistcomment-1849666

13:12 <Perelandric> And how I imagine it would look expanded: https://gist.github.com/Perelandric/19501e1d1e93271468077157bbf9be24#gistcomment-1849669

13:14 <SeanTAllen> The expanded on has the problem I originally mentioned. That isn't equiv to what you put in the previous gist.

13:15 <SeanTAllen> your array there would be Array[MyTrait] not Array[T] for it to work

13:15 <SeanTAllen> T is constrained to being of type MyTrait but T is not my trait, it is the type that you supply, Foo, Bar etc

13:16 <SeanTAllen> There's definitely some funkiness in there though

13:16 <Perelandric> I probably am just not represention the expansion correctly

13:16 <Perelandric> In the generic version, I basically just want the array to be whatever `T` was set as.

13:17 <Perelandric> So if `T` is `Foo`, as in the main actor, then the array should be `Array[Foo]`

13:18 <Perelandric> but whatever uses that generic does need to be constrained by `MyTrait`.

13:22 <Perelandric> I guess maybe `let a = Array[MyTrait[T]]` makes sense instead of `Array[T]` since that matches the return value of `test()`.

13:23 <Perelandric> ...though I'd have expected `MyTrait[T]` to be equivalent to `T: MyTrait`. Perhaps not.

13:26 <Perelandric> Ignore the last line. I didn't express that clearly.

13:30 <Perelandric> ...

13:31 <Perelandric> And come to think of it, in my actual code, I don't think `var a = Array[MyTrait[T]]` will work.

13:31 <Perelandric> Or at least it'll force me to unpack `T` from `MyTrait`.

13:32 <Perelandric> I'd rather just be able to have an Array of `T`.

13:44 <Perelandric> ...duh

13:44 <Perelandric> In `MyTrait`, the `test()` method needs to return `T` instead of `MyTrait[T]`

13:46 <Perelandric> https://gist.github.com/Perelandric/19501e1d1e93271468077157bbf9be24#gistcomment-1849669