6macros: discussing the finer points of Perl 6 macros, Qtrees, and how to stay sane | irclog: irclog.perlgeek.de/6macros/today
Set by moderator on 28 July 2015.
01:34 vendethiel joined 06:37 FROGGS joined
masak oh? 07:35
08:02 FROGGS joined 15:51 Ven joined 16:12 Ven joined 16:33 Ven joined 16:47 Ven joined
Ven very small change though... 16:48
Ven admits to not being a huge fan of the README, because oftentimes, he just wants to know what the project is about 16:49
17:14 FROGGS joined 18:18 vendethiel joined
vendethiel "destructors"? 18:20
masak people seem to either love or hate the README. those who hate it seem to be a smaller group, and they say something much like that: "it's not to-the-point enough"
at this point, I'm guessing I'm confusing "many people" with "many complaints from one person", though :P
vendethiel: I don't know where I picked up that term as meaning "getters" -- probably reading about algebras or something 18:21
vendethiel ETOOOVERLOADEDINPROGRAMMINGINGENERAL?
masak "deconstructors" would've been more clear, but that's not what I've seen in the literature
vendethiel
.oO( what're you doing to my cons? )
 18:22
masak :) 18:23
vendethiel I'm not sure I "believe" in "value" 18:25
'specially if you want to add types on top of it
(there are some very nices papers about adding typing stuff *on top* of a language)
masak that sounds interesting. do you have the references handy, perchance? 18:37
also, I don't see the connection between value() and types. my first hypothesis is that you're confused as to what they do ;) 18:38
vendethiel I'm thinking of the "IntStr" and stuff in Perl 6 18:40
vendethiel realizes he still hasn't read "First-class macros have types (A. Bawden)" 18:43
aw, well, I don't know where I put that paper :[
masak I'm not suggesting anything like "IntStr", though 18:49
I'm just saying that it looks better if what's currently `my_int_literal.int()` instead were `my_int_literal.value()`. etc str and array 18:50
vendethiel oooh, yeah. 18:52
:P
yeah.
sounds like that'd be a role for the dynamic part (OO) of the language, though? 18:53
> class Q::Arguments: Array<Q::Expr>; 18:57
please. no non-matching brackes :[ 18:58
s/s\\b/ts/
masak "non-matching brackets"? where? 19:00
vendethiel <> 19:03
I mean.
sorry.
"brackets that are already used in a non-matching fashion"
"Tuple classes" = newtype?
masak I'm sorry, I can never learn what keywords like "newype" mean :( 19:09
I learn them over and over, in various Haskell tutorials, but they never stick
vendethiel it's just a wrapper "around" a type. Pure love. 19:10
literally just creates a container so that you can't, say, add Meter and Int
masak I just saw your email about it. let me ask a follow-up here.
vendethiel newtype Meter Int
masak `newtype Arguments = (Expr, Foo)`
vendethiel disabled github emails because he gets far too many
masak I only get emails from you :P
vendethiel that's one syntax. The other one would just be newtype Arguments (Expr, Foo) (or even without the parens)
oh, only from 007? 19:11
maybe I should enable it for specific repos
masak currently, yes.
vendethiel guessed it wasn't literally "just him" ;-)
masak I like getting them, so that's fine
no, I mean I get 007 stuff, and you're the one who is active right now ;)
vendethiel yes, got it :)
masak `newtype Arguments = (Expr, Foo)` seems like a good solution to tuple classes. I don't see how it solves what I want for array types at all. 19:12
but maybe you were talking about tuple classes?
vendethiel that seems to be what you said, yes? 19:13
I mean -- classes that just are a tiny wrapper around a few parts, right?
masak the sum total of my `class` proposal right now has three parts: "record" types (with attributes), "tuple types" (fixed tuple of heterogenously typed elements), and "array types" (array of unknown length but with same-typed elements) 19:29
record types would go well with attribute and method access; tuple types with the 'case' statement; array types with indexing and array operations
vendethiel tuple types with case statements? 19:32
"newtype", I believe, could go for the 2nd/3rd part 19:33
masak I don't see how it could do the 3rd part. 19:45
your `newtype Arguments = (Expr, Foo)` syntax above seems to completely miss the point -- unless I completely miss how it would do arrays
vendethiel newtype Arguments = [Expr]? :P 19:55
I don't understand the diff. between "typed arrays" and your "array types" 19:56
masak slight difference. "array types" are a part of my suggestions in github.com/masak/007/issues/32 -- they are core, and there's no relation to types as such. 20:05
"typed arrays" would be something like `my a: Array<Q::Expr>;` 20:06
"array types" would be what would enable me to write `my args = Q::Arguments [...];` -- that is, an array constructor, but with a type in front of it.
and I see I've used "typed arrays" wrong once in the comment. retconning -- hold on. 20:08
done.
the problem with your proposed `newtype Arguments = [Expr]` is that the `[Expr]` syntax clashes with a proposed/conjectural type syntax that means "array of exactly one Expr element" 20:09
something like `Array where { .elems == 1 && .[0] ~~ Expr }` 20:10
vendethiel ah, I see 20:14
that was a ml/haskell-ish "'a list" or "[] a"
masak nodnod 20:15
I'm not sure I'll need the array syntax just to say [Foo, Bar, Baz] -- it's a bit too much like a tuple.
but I also think `[Expr]` doesn't *look* like it means `Array<Expr>` -- to me, it looks more like the thing with the `where` clause above :) 20:16
vendethiel fair enough 20:34
if [] means a tuple type
then probably, [] a = a?
:P
masak they're certainly isomorphic to each other, if that's what you mean ;) 20:43
I wouldn't call them the exact same type
that's Perl-think :P
nice -- I love the test suite. 21:23
I had \\" and \\\\ in strings unescape in the parser. one test unexpectedly started failing. 21:24
turns out it's correct to put in the string as-is in the Q::Literal::Str
it gets unescaped as it gets evaluated
why? because the Q::Literal::Str represents what it says in the code, and in the code there's still a backslash. 21:25
without that test failing, I totally would've committed the wrong fix.

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!