02:05 <FromGitter> <Blacksmoke16> @asterite ⏎ ⏎ ``` def to_s ⏎ colorize texts[1].downcase ⏎ end``` ⏎ ⏎ did you mean to define a non io overload? [https://gitter.im/crystal-lang/crystal?at=60078fd1410c2214400debab]

07:10 <FromGitter> <Blacksmoke16> repo raz @jrei:matrix.org https://github.com/athena-framework/config/issues/9 if you guys have any specific thoughts/suggestions/feedback based on that convo we had a while ago, and/or your experiences/opinions

07:11 <FromGitter> <Blacksmoke16> (scroll to the last section if you want the tl;dr on what I'm leaning towards)

11:26 <raz> yup the described approach looks good to me. just a singleton (or something i can access from everywhere) that can be populated from YAML, env, etc.

11:30 <raz> in practice i think most people prefer env for everything anyway (i certainly do). config files are just always a headache when dealing with multiple envs/devs, preventing stuff from leaking into git, multi-file deploys etc.

11:43 <repo> hey web people! do you know if it's possible to tell a client with some response headers not to send specific headers to the url specified in the location header in a temporary redirect response?

11:43 <repo> i want to prevent leaking of authorization headers

11:58 <raz> hmmm

11:58 <raz> i'm surprised it sends the header at all. shouldn't that normally we stripped when following a redirect?

11:59 <raz> maybe, if it's the same domain, browsers are more lax about it. i don't think there's really a response-header to control it. (possibly it could be done indirectly with CORS, but... don't think i'd want to rely on such)

12:33 <repo> raz: hm yeah it's the same domain

12:33 <repo> (localhost even)

12:34 <repo> i wouldn't care because both services are owned by me but sending the header results in an invalid request from minio's perspective

12:44 <repo> hm https://curl.se/docs/CVE-2018-1000007.html

12:45 <repo> seems like it actually checks the host

12:51 <FromGitter> <asterite> @Blacksmoke16 I program in Ruby during the day so sometimes I mix things, hehe :-)

13:35 <FromGitter> <Blacksmoke16> 😉

14:02 <hightower3> is there an off the shelf class I can use for a circular buffer? Like if I want it to contain 200 elements, replacing oldest ones as new ones come in?

14:03 <FromGitter> <Blacksmoke16> https://crystal-lang.org/api/master/Deque.html?

14:05 <hightower3> but I would have to remove excess elements, right?

14:05 <FromGitter> <Blacksmoke16> ```# Remove old ⏎ d.pop ⏎ ⏎ # Add new ⏎ d << new_value``` [https://gitter.im/crystal-lang/crystal?at=600838c36b20cf7730c7b6dd]

14:05 <hightower3> I'm not clear from the description if it is just a basis for implementing a circ buffer or it actually behaves as one

14:06 <FromGitter> <Blacksmoke16> its deff not going to auto remove items over some max capacity

14:06 <FromGitter> <Blacksmoke16> but thats easy enough to do

14:06 <hightower3> yes, yes, just asking, ok great thanks

14:46 <hightower3> I am writing to an IO::Memory and at some point I need to check if the last thing written was ';'. Is there a nicer way to do it than seeking -1, peeking, and then leaving it at that or again seeking by +1 if it wasn't ';'?

14:52 <hightower3> actually found a way to solve it by comparing size rather than checking the char, but still interested in the answer in general

14:58 <repo> Blacksmoke16: "A description may also be included to describe what the query param is used for. In the future this may be used for generating OpenAPI documentation for the related parameter."

14:58 <repo> holy shit!

14:59 <repo> that would be a dream come true!

14:59 <FromGitter> <Blacksmoke16> 😆 yea, we're at a point where doing something like that would be possible as you can access all the defined routes and the data associated with them

15:00 <FromGitter> <asterite> hightower3: you can call `to_slice` on it and check that

15:28 <repo> Blacksmoke16: hm not sure if i found a bug or if it's my own stupidity again, but should be easy to reproduce either way: I'm missing a VirtualProperty from the serialized json. I use groups and have defined them on the method as well (also set ASRA::Expose)

15:28 <repo> i'll set up a minimal example, one sec

15:32 <FromGitter> <Blacksmoke16> i think its a bug

15:33 <FromGitter> <Blacksmoke16> doesnt look like im including groups on virtual properties

15:37 <FromGitter> <Blacksmoke16> not including*

15:38 <repo> https://p.jokke.space/eqE/

15:38 <repo> ah ok

15:41 <repo> also not since_version and until_version

15:41 <repo> is it just a matter of adding them here?

15:41 <repo> https://github.com/athena-framework/serializer/blob/c73901d2540b8b08dea9b3a5b21937758a5bf8fc/src/serializable.cr#L151

15:42 <FromGitter> <Blacksmoke16> pretty much, i can grab it in a few

15:42 <repo> <3

15:52 <FromGitter> <Blacksmoke16> serializer component is my least favorite 😆

15:54 <repo> Blacksmoke16: how come?

15:54 <repo> it's super useful :)

15:54 <repo> fwiw i like it a lot :)

15:55 <FromGitter> <Blacksmoke16> lets just say it could use another pass, not a fan of the implementation and some stuff i did 🙈

15:56 <repo> hm

15:56 <FromGitter> <Blacksmoke16> always feel like its going to explode ha

15:57 <repo> do you already know what you would do differently now?

15:57 <FromGitter> <Blacksmoke16> but seems to be doing fairly well for you, so thats a good sign

15:57 <repo> yeah it's a joy to use :)

15:58 <FromGitter> <Blacksmoke16> i'm not sure, just feel like there is a better implementation come be achieved

15:58 <FromGitter> <Blacksmoke16> could be*

15:58 <repo> mhm

15:58 <repo> i think the visitor pattern works quite well here

15:59 <FromGitter> <Blacksmoke16> main 2 things i'm not a fan of are like 1) How easy it is to get into a state where your obj just returns `null` just because something couldnt be serialized

15:59 <FromGitter> <Blacksmoke16> and 2) how i implemented discriminator maps for deserialization

16:01 <FromGitter> <Blacksmoke16> ill save it for `0.3.0` :P

16:02 <repo> yeah ok

17:18 <FromGitter> <Blacksmoke16> repo https://github.com/athena-framework/serializer/releases/tag/v0.2.3

17:40 <repo> Blacksmoke16: sweet \o/

17:41 <repo> hm i'm struggling with https://tools.ietf.org/html/rfc5987 a bit.. is this basically just an URI.encode(...) for the UTF-8 extension?

17:48 <FromGitter> <watzon> Is there anyone that would be happy to see a Couchbase SDK library in Crystal?

17:49 <FromGitter> <watzon> Because I'm working on one, but it's hard and I need motivation lol

18:34 <FromGitter> <Blacksmoke16> is that the same thing as couchdb?

18:37 <FromGitter> <phillvancejr> Hello, will Crystal garbage collect everything at program exit? Specifically I'd like to know if finalize methods are called. I'm wrapping a C++ class in a Crystal class and I clean up the C++ with a Crystal finalize method. The C++ destructor is called if I set the Crystal object to nil then call GC.collect, however if I don't do this I don't see my destructor message. Is this memory leaking? Do I have

18:37 <FromGitter> ... to call the garbage collector manually to clean this up?

18:42 <FromGitter> <RespiteSage> If you're allocating something in program memory, the OS should clean it up when the program ends, regardless of whether it was leaking within the program while it was running.

18:43 <straight-shoota> When the program exits, there's no need for garbage collection because all memory is freed regardless

18:46 <FromGitter> <phillvancejr> thank you @RespiteSage @FromIRC !

18:46 <FromGitter> <phillvancejr> thanks @straight-shoota !

18:48 <FromGitter> <RespiteSage> I'm curious, though: how are you wrapping a C++ class?

18:52 <FromGitter> <phillvancejr> I'm writing a C wrapper first

18:53 <FromGitter> <phillvancejr> So I suppose I should have said binding to a C wrapper for a C++ class

19:02 <FromGitter> <RespiteSage> Okay, gotcha. I think it's fine you didn't say that, but I was just wondering if you were trying to bind directly to C++ somehow.

20:23 <FromGitter> <phillvancejr> is it possible to get the name of the current method in Crystal? In ruby I can do this: ⏎ ⏎ ```def func ⏎ name = __method__.to_s ⏎ name ⏎ end``` [https://gitter.im/crystal-lang/crystal?at=600891333855dd07fd694fd5]

20:23 <FromGitter> <Blacksmoke16> `{{@def.name.stringify}}`

20:24 <FromGitter> <phillvancejr> Thanks @Blacksmoke16 !!

20:26 <FromGitter> <asterite> TIL :-)

20:45 <FromGitter> <watzon> @Blacksmoke16 not the same as couchdb, though it is similar in that it's a nosql database.

20:45 <FromGitter> <Blacksmoke16> thats not confusing ha

20:46 <FromGitter> <watzon> Yeah idk why they chose that name lol

20:46 <FromGitter> <watzon> https://www.couchbase.com/comparing-couchbase-vs-couchdb

20:47 <hightower2> Hey btw, does crystal work only on systems supporting utf-8?

20:48 <FromGitter> <oprypin:matrix.org> hightower2, what does it mean for a system to support utf-8

20:51 <hightower2> I mean the String class "represents an immutable sequence of UTF-8 characters". Does doing that depends at all on the OS/environment features?

21:03 <FromGitter> <jrei:matrix.org> no

21:05 <FromGitter> <jrei:matrix.org> thankfully

21:07 <hightower2> great, thanks

21:08 <FromGitter> <jrei:matrix.org> and it is not like Python with `# -*- coding: utf-8 -*-`

21:11 <hightower2> haha

23:09 <repo> Blacksmoke16: do you know if composition can somehow work with athena-serializer? As in: define properties/methods in a module, add ASR annotations and include the module in a class?

23:16 <FromGitter> <Blacksmoke16> pretty sure that should work

23:16 <FromGitter> <HertzDevil> til you can use `*_` in type restrictions

23:17 <FromGitter> <Blacksmoke16> splat of any type?

23:43 <FromGitter> <HertzDevil> i really should start formalizing crystal's type system even if nobody else does

23:53 <FromGitter> <HertzDevil> (https://files.gitter.im/5588134315522ed4b3e2629f/6bTJ/Screenshot_20210121_075304.png)

23:53 <FromGitter> <HertzDevil> this is what i was talking about `NamedTuple`s