00:04 <FromGitter> <drosehn> they are all needed, because they are related to each other.

00:07 <FromGitter> <drosehn> a @-variable is an instance variable tied to one instance of some class. Often you'll want to initialize that variable when you create the object. So you'll say `myobj = SomeClass.new(17)`, where the `17` is the initial value you want for some @-variable in that object. So the `initialize` method will pick up the value you specified on `SomeClass.new`, and put that in the @-instance variable.

00:09 <FromGitter> <drosehn> Now that's great if the @-variable is never used outside of the object you've created (and certainly that happens). But if you **do** want to reference it or set it from somewhere outside that object, then the object must provide methods which access (or set) the @-variable.

00:10 <FromGitter> <drosehn> if you create a `property` for some @-variable, that creates a "getter" and a "setter" method for the variable.

00:11 <FromGitter> <drosehn> I hope that helps to show how these things are related to each other.

00:18 <miketheman> drosehn: It does, thanks for the explanation. I think I'll try to challenge the assumption that the original implementation uses an `options` hash and merges things around. Instead I'll try using a Config class that gets passed into the initialize

01:13 <crystal-gh> [crystal] dukex opened pull request #3374: Add HTML.unescape [Closes #3107] (master...add-html-unescape) https://git.io/vPmsn

05:42 <crystal-gh> [crystal] dukex opened pull request #3375: Add `expect` syntax to spec (master...add-expect-syntax) https://git.io/vPm8D

08:12 <crystal-gh> [crystal] ysbaddaden opened pull request #3376: Add Process.executable_path? to access current executable file path (master...std-process-executable-path) https://git.io/vPmux

11:21 <vasilakisfil> hey! has anyone spotted any serialization library in crystal ?

13:43 <FromGitter> <Val> `YAML::Generator` disappeared without any line in `Changelog` until 0.19.2 ?

13:52 <FromGitter> <Val> Ok I saw #3353

13:52 <DeBot> https://github.com/crystal-lang/crystal/issues/3353 (Bug in round tripping YAML with "*"?)

14:08 <crystal-gh> [crystal] asterite closed pull request #3375: Add `expect` syntax to spec (master...add-expect-syntax) https://git.io/vPm8D

14:11 <miketheman> I'm having trouble wrapping my head around how to best write a udpsocket server that gets an unknown amount of data. Ruby has a convenience method on Socket: http://ruby-doc.org/stdlib-2.3.1/libdoc/socket/rdoc/Socket.html#method-c-udp_server_loop

14:15 <miketheman> the `.gets` method will either block until `\n` or a predetermined number of bytes

14:16 <miketheman> the `.receive` method appears to interact with binary messages

14:31 <asterite> miketheman: I think you can use receive and then create a string out of the slice, or create a string and read into its slice (there's a String.new overload for that)

14:37 <miketheman> asterite: I'll look at that, I guess I have to allocate a size of the slice to the max size of the expected message?

14:38 <Papierkorb> What's needed to port Crystal to a new operating system? I'd be happy with the BoehmGC already

14:43 <miketheman> asterite: this appears to work: https://carc.in/#/r/1b7h

14:44 <miketheman> (side note, Ruby has `.to_f` to Time.now for higher precision, Crystal doesn't have that yet)

14:46 <miketheman> is there any detriment to sizing the slice to much larger, i.e. max udp packet size? `message = Slice(UInt8).new(65_507)`

14:49 <miketheman> or would I benefit from moving the `message` declaration outside of the loop?

14:51 <miketheman> ah, if I do that I get weird overwriting behaviro

15:08 <miketheman> Anyhow, it looks like it's doing the thing I wanted, thanks @asterite

16:26 <leafybasi> I want to reduce an array to an array of named tuples, if I call "reduce([])" I am told to provide a type, "reduce([] of NamedTuple)" says its not specific enough, but "reduce([] of NamedTuple(key: String, value: Int32))" it complains when trying to add a tuple to the array within the block, I'm a little cofnused as to why...

16:26 <leafybasi> Sorry reduce an array of strings to an array of named tuples*

16:27 <leafybasi> The error is long so I've gisted it: https://gist.github.com/barisbalic/9fe8eec829ca34973be27c651d3f4f4c

16:28 <BlaXpirit> leafybasi, are you sure you understand what 'reduce' does?

16:28 <BlaXpirit> "reduce to an array" sounds like a paradox to me

16:28 <leafybasi> You can reduce to any single value

16:28 <leafybasi> My value happens to be a collection

16:29 <leafybasi> I could map and flatten, but reduce should allow me to do what I want

16:29 <BlaXpirit> please show code and maybe say what you want to do

16:29 <leafybasi> Sure thing

16:34 <BlaXpirit> leafybasi, well your error shows that you're trying to do Array(Array(T)) << T

16:37 <leafybasi> BlaXpirit: something like this: https://gist.github.com/barisbalic/3aac81e4dcd288d31a897dda04683a94

16:37 <leafybasi> edited, please refresh

16:40 <leafybasi> Okay

16:40 <leafybasi> I see it, my bad

16:40 <leafybasi> Thanks for talking it through BlaXpirit

16:42 <miketheman> is there a way to control a Fiber? I want a long-lived spawn/loop until a `.stop` message is sent, and am not sure how that might look.

16:42 <miketheman> Kind of like Ruby's Thread.kill

16:44 <leafybasi> miketheman: if the loop responds to incoming messages you would use channels

16:44 <leafybasi> But if it can't be blocked (while waiting) then I'm not sure

16:51 <Papierkorb> miketheman: if it has to be a long-running fiber, you could make it check every iteration or so if a "running" bool has become `false`. Later, when Crystal supports threading, you can simply use an AtomicBool instead of a Bool for thread-safety.

16:53 <BlaXpirit> aren't bools atomic anyway

16:55 <Papierkorb> It should if you don't care about the memory barrier

17:10 <miketheman> would this approach work for network/socket traffic? https://github.com/askn/crystal-by-example/blob/fec1c8e388f735d5212e08e47e4667a2e46956ef/closing-channels/closing-channels.cr

17:11 <miketheman> it doesn't appear so - since without channels I use `sleep` to control the Fiber's long-life, and this has a jobs max

17:11 <miketheman> unless I'm reading it wrong

20:12 <crystal-gh> [crystal] ysbaddaden opened pull request #3379: Fix: use __iconv on FreeBSD to skip invalid bytes (master...core-iconv-freebsd-skip) https://git.io/vPOIt

20:45 <crystal-gh> [crystal] asterite pushed 1 new commit to master: https://git.io/vPOY2

20:45 <crystal-gh> crystal/master 4733360 Ary Borenszweig: Compiler: fixed crash related to cast between unions having tuples. Fixes #3377

21:00 <crystal-gh> [crystal] faustinoaq opened pull request #3380: Fix named tuple (master...patch-1) https://git.io/vPO3p

21:16 <travis-ci> crystal-lang/crystal#4733360 (master - Compiler: fixed crash related to cast between unions having tuples. Fixes #3377): The build passed. https://travis-ci.org/crystal-lang/crystal/builds/164753280

21:16 <DeBot> https://github.com/crystal-lang/crystal/issues/3377 (Nil assertion failed on cast between union types)

22:26 <crystal-gh> [crystal] ysbaddaden pushed 1 new commit to master: https://git.io/vPOEY

22:26 <crystal-gh> crystal/master a8eef9b Faustino Aguilar: Fix named tuple (#3380)

22:29 <FromGitter> <anicholson> potentially silly question: when I use `puts` in my programs, who's the receiver of that message, and what's its type?

22:41 <FromGitter> <anicholson> oh I see, they belong inside the Top-Level Namespace

22:41 <FromGitter> <anicholson> that confuses this Rubyist ;)

22:43 <FromGitter> <anicholson> especially when the docs say "Everything is an Object"!

22:43 <FromGitter> <anicholson> Lies! :)

22:57 <travis-ci> crystal-lang/crystal#a8eef9b (master - Fix named tuple (#3380)): The build passed. https://travis-ci.org/crystal-lang/crystal/builds/164779610

22:57 <DeBot> https://github.com/crystal-lang/crystal/pull/3380 (Fix named tuple)