00:50 <FromGitter> <Timbus> From what I can see, it only accepts fixed offsets, using `%z`, `%:z`, and `%::z` ⏎ I can't see it in the docs, but it's here: https://github.com/crystal-lang/crystal/blob/master/src/time/format.cr

00:52 <FromGitter> <Timbus> If it's a string zone like 'Luna/Copernecus', you'll have to parse the location out first, then use `Time::Location.load` and pass it as a third argument to `Time.parse`

00:53 <FromGitter> <Timbus> I guess =/

08:23 <Jenz> Is something wrong with crystalshards.xyz?

08:23 <Jenz> I just get loads of errors

08:23 <Jenz> Oh there it works again, nvm

08:41 <hightower4> Jenz, you can also always use crystalshards.org

08:42 <Jenz> Oh I didn't know, thanks

11:02 <Groogy> so I am having an unexpected nil type

11:02 <Groogy> https://gist.github.com/Groogy/7685df05cd137b8442cc8a2ea85986f7

11:02 <Groogy> I don't get why it is telling me gui.remove_root window on line 10 that the type can be nil?

11:02 <Groogy> I tried adding the unless gui.nil? but it still says that

11:05 <Yxhuvud> strange. What happens if you cahnge it to `if gui` ?

11:06 <Groogy> same

11:07 <Groogy> I am working around it by just passing a method for it to call

11:07 <Groogy> but yeah this is a very unexpected behaviour

11:07 <Groogy> unless I'm missing something

11:07 <Yxhuvud> right, the if statement in the block shouldn't be necessary as you check it outside

11:07 <Groogy> yepp, and the type shouldn't magically disappear if @gui is set to nil for instance

11:08 <Groogy> and the context shouldn't be collected by the gc as logn as I have a reference so I am a bit puzzled as of why

12:21 <crystal-gh> [crystal] j8r closed pull request #6423: Create an abstract Any for JSON/YAML (master...any_abstract_struct) https://git.io/fNWXs

12:43 <Groogy> if I have a @type.subclasses in a macro, is there a way to check if one of the subclasses are abstract and skip it?

12:45 <Groogy> just .abstract?

13:09 <crystal-gh> [crystal] ysbaddaden opened pull request #6426: Fix: fork and signal child handlers (master...fix-fork-and-signal-child-handlers) https://git.io/fNlmO

13:17 <crystal-gh> [crystal] r00ster91 opened pull request #6427: Add :default to colorize and document ColorRGB, Color256 (master...patch-3) https://git.io/fNlm8

15:31 <literal> do I need anything more than "make spec" to run the test suite? I get this failure on master: https://gist.github.com/hinrik/747587c37c2119c2815b7bf729a93cda

15:36 <literal> kind of strange because that spec passes just fine if I do "bin/crystal spec/std/file_spec.cr"

15:52 <crystal-gh> [crystal] Exilor opened pull request #6428: Fix Atomic#swap with reference types (master...master) https://git.io/fNlGp

16:33 <crystal-gh> [crystal] hinrik opened pull request #6429: JSON: Handle non-standard UTF-8 string escapes (master...json_utf8_escapes) https://git.io/fNln0

17:49 <FromGitter> <bew> Groogy still there?

17:50 <Groogy> yepp

17:53 <FromGitter> <bew> Because the block is captured, the captured variable `gui` gets referenced for this block, and currently the compiler will keep all the types of the var (without current type restriction/simplification) when it's passed to a captured block.

17:54 <Groogy> but the type should not have nil? right? @gui is the one that has nil

17:55 <Groogy> and even if I did test for nil inside the block it didn't work

17:56 <FromGitter> <bew> @literal that looks weird.. I think the explanation is that some other spec uses a `base` variable, and the compiler thinks that you're trying to use it..

17:58 <FromGitter> <bew> Groogy because you have a reference inside the block to the variable, you have the handle it like an ivar, so assign it to a local var first

17:58 <Groogy> so if I do gui = gui it would work?

17:58 <Groogy> that's... weird

17:59 <FromGitter> <bew> And `if gui = @gui` is basically rewitten as `gui = @gui` and `if gui` so the full type has Nil too, but we never see it in the if

18:01 <FromGitter> <bew> Groogy yeah should work (maybe, with another local var name though), I agree it's weird.. I know there's an open issue to fix this but it's not for now..

18:01 <Groogy> ah okay, it's not by design at least

18:01 <Groogy> I needed the code to be in a method anyway because it needed to be used by other stuff anyway

18:01 <Groogy> so it solved itself

18:04 <FromGitter> <bew> Ahah nice

18:04 <FromGitter> <bew> Not sure now if it's by design or not..

18:05 <FromGitter> <bew> #6040

18:05 <DeBot> https://github.com/crystal-lang/crystal/issues/6040 (Bug: Only typeof in proc body does not account for closure var)

18:05 <FromGitter> <bew> But there is no question of `if` hefe

18:05 <FromGitter> <bew> Here*

18:39 <Groogy> I have a function defined like this: def internal_marshal(tuple : Tuple(*U)) forall U

18:39 <Groogy> how do I write the signature to make it work with named tuples?

18:40 <FromGitter> <bew> `NamedTuple(**U)`?

18:40 <Groogy> "expecting token 'CONST', not '**'

18:44 <Groogy> hmm how is named tuple actually defined

18:45 <Groogy> hmm it is defined like that

18:46 <Groogy> maybe double splatting isn't supported in a generic method?

18:59 <FromGitter> <rishavs> Any ideas what am i doing wrong here? ⏎ ⏎ ```code paste, see link``` [https://gitter.im/crystal-lang/crystal?at=5b54d3f4f477e4664ab55d44]

19:01 <FromGitter> <bew> Groogy looks like a bug, it should be supported I think

19:01 <FromGitter> <bew> @rishavs you forgot to call `new` for the hash

19:02 <Groogy> yeah, you are assigning the type to the tree variable

19:04 <FromGitter> <rishavs> @bew thanks, as always

19:07 <FromGitter> <bew> Yw :)

21:40 <FromGitter> <DanilaFe> Hi! I've got a question about C bindings in Crystal. What would be the best way to bind C code that isn't installed system-wide? Say, I have a git submodule with a C library in it, how should I proceed with linking it to Crystal?

21:42 <oprypin> for cross platform: compile a shared library and always add it to LIBRARY_PATH when compiling and LD_LIBRARY_PATH when using

21:44 <oprypin> for posix: example here https://github.com/Papierkorb/qt5.cr/tree/master-ready-to-use/src/qt5/binding `@[Link(ldflags: "#{__DIR__}/../../../ext/binding_linux-gnu-x86_64-qt5.10.a -lstdc++ -lQt5Core -lQt5Gui -lQt5Widgets")]`

21:45 <FromGitter> <DanilaFe> Thanks a lot!

21:46 <oprypin> well my first message is also posix-specific, i guess, but that's just because on Windows you don't need to do this, current directory is in library path

21:47 <oprypin> also windows is not really supported lol

21:49 <FromGitter> <DanilaFe> Not yet, anyway

21:49 <FromGitter> <DanilaFe> hopefully Crystal gets there

22:08 <FromGitter> <DanilaFe> hold on, inside lib, i can't create a struct with a lowercase name?

23:49 <crystal-gh> [crystal] asterite opened pull request #6432: Fix: private aliases at top-level not considered private (master...bug/5752-private-alias-top-level) https://git.io/fNlgo