02:07 <inanc> hi! i wanted to learn about that is microsoft backing or supporting or financing ponylang? what's relationship of ponylang with microsoft?

02:09 <SeanTAllen> there is no relationship between microsoft and pony. Sylvan works at Microsoft Research but there is no official relationship.

13:04 <ro6> Let's say I have an `let arr = Array[Bool](1)` shouldn't I then be able to update that first reserved spot with `arr(0) = true` or `arr.update(0, true)`? Every time I try this in a test program it errors on the update attempt.

13:07 <ro6> Essentially what I want is a sparse array. Is there another way to get that besides reserving sufficient space then using `update()` to set specific indexes?

13:13 <SeanTAllen> Can you explain what you expect the Array[Bool](1) to do in that scenario ro6 ?

13:13 <ro6> I was expecting it to allocate an empty array with space for at least 1 Bool value

13:17 <SeanTAllen> ok, and how is that connected to the sparse array comment? i don't see the connection between the two.

13:17 <SeanTAllen> can you send a gist of your code that isn't working for you?

13:20 <ro6> Sparse as in I might allocate an array of size 5, then progressively update index 4, then 3, then 1

13:21 <ro6> Writing the gist now

13:23 <SeanTAllen> So the "Essentially what I want is a sparse array. Is there another way to get that besides reserving sufficient space then using `update()` to set specific indexes?" question is... is there a way other than how you just described "sparse" to achieve the same effect?

13:24 <ro6> https://gist.github.com/ro6/325f3d9013ab57b43cd760798d0a4dee

13:25 <ro6> that gist prints `fail` when run

13:40 <SeanTAllen> i left a comment on your gist ro6

13:41 <SeanTAllen> If you want to manage to not allocate memory for the union type I used, then you can't use Array and would need to create your own class. I think there would be interest in a "sparse array" class being part of the standard library. That would be something that would go through the RFC process

13:49 <ro6> Ok, thanks for the info. I just took a look at the Array source and I see that it's throwing the error because the "bounds" it checks against depends on the "_size" variable, which is the amount of elements in the array, not the allocated size (https://github.com/ponylang/ponyc/blob/master/packages/builtin/array.pony#L129).

13:50 <SeanTAllen> you're welcome

13:54 <ro6> be functionally problematic.

13:54 <ro6> In my case I can probably get away with filling the "empty" spaces with false to begin with. However, in general I wonder why the Array implementation doesn't check against the "_alloc" variable instead. Granted it would be weird to be able to ask for an array of size 5 then update at position 6, which the current implementation would allow, but it wouldn't

13:59 <ro6> If you wanted to hide the implementation detail of space being reserved in a particular doubling pattern, you could store another field called eg _user_reserved to keep track of the size the user should be able to depend on . The class invariant would be _user_reserved <= _alloc

14:00 <ro6> Does that sort of change need to go through an RFC process, or could I just submit a PR for consideration?

14:02 <SeanTAllen> I don't understand what you mean by implementation detail of space being reserved.

14:02 <SeanTAllen> Having to reallocate memory has performance considerations. To me that means its not an implementation detail. Its something I very much need to know about.

14:03 <SeanTAllen> perhaps "implementation detail" has a different meaning to you than it does to me

14:03 <ro6> No, I agree

14:04 <ro6> I'm just saying that if the strategy were to change in the future for whatever reason, and a user of the Array class had depended on the prior allocation strategy, it would break the correctness of their code

14:04 <SeanTAllen> I think an RFC for a SparseArray that has different performance characteristics than Array would be more warmly received than such a change to Array.

14:05 <Rishav> does pony have any good web frameworks?

14:05 <SeanTAllen> Rishav: same answer as yesterday: https://www.dropbox.com/s/uomb8vp7i30ge37/Screenshot%202017-02-26%2009.05.41.png?dl=0

14:06 <SeanTAllen> even if we had any web frameworks (which we don't), "good" is rather subjective and would be hard to answer.

14:06 <SeanTAllen> ro6: sorry, i'm having a hard time following that. i think i'd need a concrete example.

14:07 <Rishav> sorry Sean. i went afk and then IRC logged me out

14:07 <Rishav> :D

14:08 <SeanTAllen> irrcloud is fairly inexpensive and great for that Rishav

14:09 <SeanTAllen> you were still online when i gave that answer yesterday, a change in IRC client might help there too. even it logs you out, it should show you messages you received before being logged out.

14:11 <Rishav> thanks anyway.

14:23 <ro6> SeanTAllen: I was just a bit surprised by the inability to write to an arbitrary location in an array that you know has been allocated. Coming from a lot of other languages it's pretty standard to be able to do that. I'll take a quick crack at rewriting the std Array class so people can see exactly what I'm referring to. If that leads on to a separate

14:23 <ro6> SparseArray implementation, that's fine too. I think there's even further memory optimizations you can make if you know an array will be sparse.

14:24 <ro6> The current implementation seems to be maintaining the invariant that elements (as well as allocated space) are contiguous

14:32 <SeanTAllen> if an Array is 1 element in size, as far as I am concerned, saying update element 2 is an error. that is out of bounds.

14:32 <SeanTAllen> it doesn't matter how much memory has been reserved for it.

14:32 <SeanTAllen> in the example i put on your gist, this is explictly handled.

14:32 <SeanTAllen> the Array is of Bool or None and is sized appropriately

14:34 <SeanTAllen> I dont think there is any such inability, you just need to type it correctly

14:35 <SeanTAllen> You could do Array[Bool] and as long as you initialize all the values our the where-ever you have allocated, you now have a valid array of that length

14:36 <SeanTAllen> But I'm interested to see what you come up with

14:39 <SeanTAllen> There are definitely space v performance trade offs that can be made data structures and having options available in the standard library to make those tradeoffs is a great thing

14:39 <SeanTAllen> ro6: if you need any help, get stuck on any pony-isms, etc let me know. either here or via the mailing list and time permitting, i'll help as best i can

14:40 <SeanTAllen> also are you aware of the development sync that happens almost every wednesday? we go over RFCS, PRs, issues and also have a discussion period. anyone in the community is welcome to join and encouraged to do so.

14:41 <ro6> oh, great. I wasn't aware of that. I just found Pony a week ago or so.

14:41 <SeanTAllen> if you join the pony-dev mailing list, you will get an email every week with info about the Sync.

14:41 <ro6> great, I'll do that. Thanks

14:42 <SeanTAllen> you're welcome

14:42 <ro6> Now that I'm looking at it from the type perspective, I can see why it's built as it is now.

14:46 <SeanTAllen> vaguely related to that

14:46 <SeanTAllen> Pony's union types are very nice and easy to use

14:47 <SeanTAllen> I appreciate that I can do (Bool|None) rather than having Some[Bool] or Maybe[Bool]

14:47 <SeanTAllen> its a small difference but in the end somehow has had a dramatic impact on how I approach things

14:47 <SeanTAllen> its hard to explain

14:48 <ro6> Yeah. This would be the first language I've actually built something in that has that. The other one I'm aware of is Ceylon.

14:48 <ro6> It's definitely convenient

14:50 <SeanTAllen> For a while, MaybePointer was called Maybe and that lead to all sorts of confusion from folks who know Haskell (and Scala folks who know that languages have Maybe to correspond to Scala's Some)

14:59 <TwoNotes> Can PONYPATH be a list of directories?

15:02 <ro6> SeanTAllen: Yep, that would have gotten me too.

15:25 <SeanTAllen> TwoNotes: yes, see how pony-stable manages it

18:17 <TwoNotes> SeanTAllen, where would I find that?

20:53 <SeanTAllen> TwoNotes: where would you find what?

21:02 <SeanTAllen> pony-stable?

21:02 <SeanTAllen> https://github.com/jemc/pony-stable