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 20 June 2015.
01:32 vendethiel joined 06:22 vendethiel joined 06:57 Ven joined 07:04 vendethiel joined
Ven \\o masak 07:53
masak o/
Ven masak: maybe you'll have more free $brane today :) 07:54
masak seems like it.
I'm doing some other stuff now in the morning, but I have your gist open in a tab :) 07:55
will get to it
Ven alright, good 07:56
.oO( I hope you close your tabs faster than I do, because mines have been piling up... )
masak I have that too
the ones on the far right tend to be quite active
Ven ...I guess I'm to blame for that as well *g*
masak partly, though maybe less than you think :P 07:57
Ven good.
The Comonad.Reader » Fast Circular Substitution, TYPE-DRIVEN DEVELOPMENT OF CONCURRENT COMMUNICATING SYSTEMS, Totality vs Partiality, Pluggable type position, Resource-dependent Algebraic Effects 07:58
i have around 15 like that open
masak haha
08:07 vendethiel joined 09:45 Ven joined 10:00 Ven joined 10:25 vendethiel joined 12:07 vendethiel joined 12:43 vendethiel joined 13:27 vendethiel joined 13:56 vendethiel joined 14:21 Ven joined 14:22 vendethiel joined 16:15 Ven joined
Ven masak: feeling more like macros now? :P 17:05
masak mood knocked weirdly sideways by $kid vomiting repeatedly for the past 90 minutes. 18:59
I'll have a look, though.
vendethiel uh, whoops 19:02
hope it doesn't look worse than slightly bad food :)
masak it doesn't, thankfully. 19:03
but there was a lot of crying involved (on his part), both before and during.
vendethiel rightfully
masak extra ')' on gist.github.com/vendethiel/53080d9...e1-6pl-L35 ? 19:04
vendethiel fixed
masak I can already see that I like what I'm reading 19:05
you seem undecided between `QAst` and `QAST`. the actual spelling is just `Q` :)
(hence "Qtree")
vendethiel fixed
masak nice 19:06
I'm looking at the way you handle unquotes in the quasi block, and I'm thinking "my, it'd be nice if they could actually do all that"
I wonder what it'd take for them to do that.
I wonder if typechecking ought to play a role in it. 19:07
vendethiel I somewhat wanted to ".subst" from blackcoffee for that gist
masak like, $param would have to be statically typed as Q::Var
vendethiel mmmmmmmmmhh. seems scary 19:08
you can't parse the quasi without the unquote. 19:09
masak that's what I mean. it feels like there has to be some restraint. 19:12
like, it's a parse error if what {{{ $param }}} gets is not a Q::Var
we actually don't have to statically know that... we have to know it at splice time. 19:13
no, wait. we have to statically know that for the reason you said. we have to parse the quasi.
vendethiel ...but we can't 19:15
masak I'm thinking I should focus on that question: "how can we make quasis that powerful?"
we can't?
vendethiel we can't parse the quasi
masak if we can't, then the parser must throw an error and abort the compile.
vendethiel my $q = quasi { sub infix:<+>($a, $b) is tighter(&infix:<,>) {} }; return quasi { {{{ $q }}}; say 3 + 4 / 5 } 19:16
masak ponders 19:19
I don't think the infix:<+> is supposed to be visible outside of its quasi without some kind of COMPILING:: shenanigans
vendethiel ..which solved my issue, opening yours (COMPILING) in the process :) 19:20
masak aye
masak asks TimToady 19:21
vendethiel well, it's simple, really 19:22
removing the ability to generate subs from macro would hinder its usefulness a lot :(
masak I think it might work. 19:25
or at least, suddenly I don't see the big problem. things seem to happen in the order they should. 19:26
vendethiel do you see my gist theoritically working? 19:27
re*
masak yes, I dub it the most concrete/fleshed-out macro gist ever. 19:33
I could see lots of small details needing re-work, but apart from that it seems like a reasonable thing to expect New Macros to be able to do.
(and a good example at that) 19:34
vendethiel tons and tons and tons of detais 19:38
details*
it's terrible code
masak ;) 19:57
let's aim to make it real, though.
maybe I can even convince you at some point to help prototype things in 007... :D 19:58
vendethiel oh, i don't think i'm anywhere near good enough to pull that feat off 20:01
masak it surprises me that you say that 20:07
you just implemented delimited continuations in hypothetical future Perl 6 macros
vendethiel that's easy 20:08
masak no, it bloody isn't 20:09
vendethiel of course it is. you just have to swap the inside of shift() with the inside of reset()
masak parsers and runtimes are easy. you just have to turn the source text into an intermediate form, then execute the intermediate form. 20:12
seriously, I think more people would be compiler writers if they realized (a) exactly what compilers are, and (b) that there's not that much to it 20:13
I know _why realized this at some point, with Potion
vendethiel oh, parsers and runtimes are "easy"
i'm writing one in haskell currently, and I have another project with the parser in 6 and the VM in rust 20:14
masak and 007 is a rather easy example of parsers and runtimes
vendethiel i never implemented macros! 20:16
masak as a point of reference, I found ops-as-routines tougher to get right in 007 than macros... 20:32
vendethiel ops-as-routines? 20:33
masak define an infix:<!>, and suddenly there is one in the parser's table of operators 20:35
vendethiel ooh
doesn't seem too hard, though? 20:36
I already have an array with a list of OPs
if I lex "operator infix §§", I just { 'operator' $<pos>=\\w+ $<op>=.+$ { @operators.push($<op>.Str) } } 20:37
no?
i only parse infixes tho... 20:38
masak except the definition is a `sub infix:<!>(l, r) { ... }`, so the installation has to happen as part of sub parsing 20:47
and then when a `!` is encountered in infix position, there has to be a lexical lookup for `infix:<!>`, in order to bind to the right definition. 20:48
each detail isn't all that tricky, but it's a lot of details.
vendethiel I see 21:00
oh yeah it means you need lexical scope at the parsing level. 21:01
masak it does. 21:04
but so does BEGIN
vendethiel true 21:18
alright 21:19
i'll think about this more still
but for now, 'night!
masak 'night
23:38 vendethiel joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!