🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm
Set by lizmat on 8 June 2022.
[Coke] what does Mu.sink do? (it's defined to do nothing, but does it somehow do something?) 02:38
Operator was added to rakudo and to S02-types/WHICH.t, but is not otherwise tested in t/spec. Does this make it part of the spec, or is this an accidental inclusion that should be reverted. 02:50
m: Operator.new
camelia Cannot make a Operator object using .new
in block <unit> at <tmp> line 1
[Coke] I ask because there's a raku/doc ticket about it. 02:51
m: new Proc::Async::Pipe 02:57
camelia ===SORRY!=== Error while compiling <tmp>
Unsupported use of C++ constructor syntax. In Raku please use: method
call syntax.
at <tmp>:1
------> new Proc::Async::Pipe⏏<EOL>
[Coke] m: Proc::Async::Pipe.new
camelia Too few positionals passed; expected 5 arguments but got 1
in block <unit> at <tmp> line 1
[Coke] ^^ this isn't in roast. is it an internal detail? 02:58
releasable6__ Next release in ≈6 days and ≈15 hours. There are no known blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 03:00
[Coke] any PRs to roast should trigger a check for documentation on that feature.
finally closed a doc ticket opened by niner in 11/2019 03:20
07:21 finanalyst joined 08:33 sena_kun joined
lizmat [Coke]: Operator is a bit in limbo at the momen 08:54
it is used in RakuAST, my plan was to use it in the legacy grammar as well, but ran into issues 08:55
it should become a full-fledged citizen in 6.e, but might still have some changes until thenb
so I'd rather treat it as an implementation detail until then, like the rest of RakuAST
08:58 sena_kun left
finanalyst lizmat: I found .join.naive-word-wrapper in RakuDoc::To::Text, but cant find where naive-word-wrapper is documented. 09:21
lizmat it's an implementation detail
github.com/rakudo/rakudo/blob/main...3588-L3595 09:22
m: sub a(uint $a) { dd $a }; a -1 # meh, a buglet 10:25
camelia -1
lizmat m: my uint $a = -1; say $a
camelia 18446744073709551615
lizmat m: Q|sub a(uint $a) { dd $a }; a -1|.AST.EVAL 10:26
camelia 18446744073709551615
lizmat ah... but fixed in RakuAST *phew* :-) 10:27
ab5tract: ^^ maybe a test *nudge* *nudge* :-)
ab5tract :) 10:53
11:39 MasterDuke left
ab5tract lizmat: any thoughts on R#5576 12:22
linkable6 R#5576 [open]: github.com/rakudo/rakudo/pull/5576 Give META_REDUCE_RIGHT behavior for 1-arg case
ab5tract I’m not sure that one-arg == no-op ever made sense, but if we want it for left we should do it for right 12:24
lizmat actually no, apart from your thought "should we even be doing this"
perhaps nine has more ideas
ab5tract am I misremembering or did we once have a version of camelia / evalable that could process a gist from github? 13:21
committable6: HEAD gist.githubusercontent.com/ab5trac...r2000.raku 13:23
committable6 ab5tract, Successfully fetched the code from the provided URL
ab5tract, ¦HEAD(a26b756): «:contoto("Any")␤:slot(Any)␤:type("Any")␤assign requires a concrete object (got a Any type object instead)␤ in block <unit> at /tmp/YrFu5B6BIh line 10␤␤ «exit code = 1»»
ab5tract evalable6: gist.githubusercontent.com/ab5trac...r2000.raku 13:24
evalable6 ab5tract, Successfully fetched the code from the provided URL
(exit code 1) :contoto("Str")
:slot("b")
:type("Str")
Cannot assign to an immutable value
in block <unit> at /tmp/3qYdNTQxmo line 10
ab5tract so... we use `op.count` to determine when to do the reduction: github.com/rakudo/rakudo/blob/a26b...kumod#L190 13:27
however, that feels unforunately inadequate in light of multi-ness :( 13:28
lizmat it could well be improved if @args where an IterationBuffer and |$args.List 13:30
ab5tract hmm 13:32
would that adress "when" to call the provided op? or do we need to adjust user expectations that it always call the longest-arg candidate(s)? 13:35
lizmat sorry, I was just talking about performance... :-) 13:36
ab5tract :)
well from my first pass at R#2000, fixing the issue of maintaining containerization involves removing binds and adding a `return`... 13:37
linkable6 R#2000 [open]: github.com/rakudo/rakudo/issues/2000 [data types][metaop] `reduce` needlessly deconts
ab5tract gist.github.com/ab5tract/7b6eda846...econt-diff 13:38
so not a net-win on the performance side :/ 13:39
but that does fix the second example provided in the issue
maybe this slower path could be restricted to only cases where the first argument is in a container to begin with? 13:43
lizmat possibly... I must admit I'm a bit distracted atm 13:44
ab5tract no worries :) 13:45
thanks for the heads up!
[Coke] added github.com/Raku/doc/issues/4475 for Operator and others. 15:30
Would like to make sure that we have a good 6.e experience with the docs when 6.e goes live.
lizmat [Coke]++ 15:43
[Coke] There's no specific timeframe on that, yes? Just "when RakuAST lands" atm? 15:50
no pressure, just planning.
lizmat Indeed, "when RakuAST lands"
jdv will there be a party? 16:37
lizmat that would be very nice indeed :-) 16:38
ab5tract m: use nqp; my %h; my $p := %h<a>; dd :$p, :desc(nqp::getattr($p.VAR, Scalar, q|$!descriptor|).^name) 17:58
camelia :desc("VMNull")
:p(Any)
lizmat m: use nqp; dd nqp::null 17:59
camelia Mu
lizmat m: use nqp; dd CMNull
camelia ===SORRY!=== Error while compiling <tmp>
Undeclared name:
CMNull used at line 1
lizmat m: use nqp; dd VMNull
camelia ===SORRY!=== Error while compiling <tmp>
Undeclared name:
VMNull used at line 1
ab5tract According to: github.com/rakudo/rakudo/blob/a26b...akumod#L29
IIUC I should be seeing `ContainerDescriptor::BindHashKey` 18:00
or.. hmm.. `$p` itself is the `ContainerDescriptor::BindHashKey` 18:07
huh, this is weird: 18:08
m: use nqp; my %h; my $p := %h<a>; dd :$p, :desc(nqp::getattr($p.VAR, Scalar, q|$!descriptor|).^name), :next($p ~~ ContainerDescriptor::BindHashKey)
camelia Bool::False
ab5tract (dd fails quite curiously)
lizmat m: my $a; use nqp; dd nqp::getattr($a.VAR, Scalar, /$!descriptor) 18:10
camelia ===SORRY!=== Error while compiling <tmp>
Variable $!descriptor used where no 'self' is available
at <tmp>:1
------> p::getattr($a.VAR, Scalar, /$!descriptor⏏)
lizmat m: my $a; use nqp; dd nqp::getattr($a.VAR, Scalar, q/$!descriptor) 18:11
camelia ===SORRY!=== Error while compiling <tmp>
Couldn't find terminator / (corresponding / was at line 1)
at <tmp>:1
------> :getattr($a.VAR, Scalar, q/$!descriptor)⏏<EOL>
expecting any of:
/
lizmat m: my $a; use nqp; dd nqp::getattr($a.VAR, Scalar, q/$!descriptor/)
camelia Mu
lizmat m: my $a; use nqp; say nqp::getattr($a.VAR, Scalar, q/$!descriptor/).raku
camelia No such method 'raku' for invocant of type 'VMNull'. Found 'raku' on
type 'Mu'
in block <unit> at <tmp> line 1
lizmat m: my $a; use nqp; say nqp::getattr($a.VAR, Scalar, q/$!descriptor/).^name 18:12
camelia VMNull
lizmat ab5tract: ^^ golfed
m: use nqp; say nqp::getattr((my $).VAR, Scalar, q/$!descriptor/).^name
camelia VMNull
ab5tract ah, I wasn't initially raising any issues regarding dd 18:13
it only produced surprising results when I added a smartmatch on `ContainerDescriptor::BindHashKey` 18:14
nine m: say [∈] "a" 18:58
camelia True
nine obviously
ab5tract :) 18:59
I think I might even have seen a report or at least a conversation about this recently
seems like a reasonable place to use a Failure to me 19:01
nine m: say [∈] set(), set(set()), set(set(set())) 19:02
camelia True
nine m: say [∈] set(), set(set())
camelia True
nine m: say [∈] set() 19:03
camelia True
nine Actually that does make some sense
m: say [∈]
camelia True
ab5tract m: sub infix:<gg>($a, $b --> Set) { set($a, $b) }; dd [gg] "a" 19:04
camelia "a"
ab5tract does that make sense though?
I think this one is probably a pretty tight corner, I respect that there was some thought that went into treating it this way in the design 19:05
m: say [∈] Mu 19:06
camelia True
nine I have my answer fully typed up but I'm still reluctant to send it. 19:13
ab5tract no rush 19:17
nine I don't think this is something where just more time helps. It needs concrete real-life use cases 19:18
See, that's exactly why I tend to stay out of language design discussions. It's sooo easy to have a gut reaction (in this case I have actually typed two answers in opposite directions, both claiming that they are "clearly" right) and so hard to make an actually informed, far looking decision. 19:20
ab5tract indeed :/ 19:27
it would be interesting to read both of those, FWIW :)
nine Well I posted the second one
ab5tract do you reckon it deserves a problem solving thread? 19:33
lizmat wonders where nine posted 19:36
ab5tract R#1705 19:41
linkable6 R#1705 [open]: github.com/rakudo/rakudo/issues/1705 [metaop] META_REDUCE_RIGHT doesn't check for 1-arg case like META_REDUCE_LEFT
lizmat thanks :-) 19:43
ab5tract m: use nqp; my %h; my $p := %h<a> = Any; dd nqp::getattr($p, Scalar, q|$!descriptor|).^name; $p = 5; dd %h 19:54
camelia "ContainerDescriptor::Untyped"
{:a(5)}
ab5tract just when I think I'm starting to understand :) 19:55
lizmat m: mh %h; my $p := %h<a>; say $p; say %h<a>; $p = 42; say %h<a> # ab5tract do you grok this? 19:59
camelia ===SORRY!=== Error while compiling <tmp>
Variable '%h' is not declared. Perhaps you forgot a 'sub' if this was
intended to be part of a signature?
at <tmp>:1
------> mh ⏏%h; my $p := %h<a>; say $p; say %h<a>; $
lizmat m: my %h; my $p := %h<a>; say $p; say %h<a>; $p = 42; say %h<a> # ab5tract do you grok this?
camelia (Any)
(Any)
42
ab5tract that part makes sense. 20:00
how it's being done with a `ContainerDescriptor::Untyped` is the part that confuses me
m: use nqp; my %h; my $p := %h<a>; dd nqp::getattr($p, Scalar, q|$!descriptor|).^name; $p = 5; dd %h
camelia "ContainerDescriptor::BindHashKey"
{:a(5)}
ab5tract ^^ this I grok 20:01
lizmat ok, by assigning Any to the container, it changes its descriptor to whatever it was indicated to be
ab5tract but then how does it know to still bind into the hash?
lizmat it bound it to the hash on assignment 20:02
after that, the knowledge is no longer needed, so discarded
see role ContainerDescriptor::Whence 20:03
ab5tract but there is no ContainerDescriptor::Whence involved in the second example... this is what I'm confused about 20:07
m: use nqp; my %h; my $p := (%h<a> = Any); dd nqp::getattr($p, Scalar, q|$!descriptor|).^name; $p = 5; dd %h
camelia "ContainerDescriptor::Untyped"
{:a(5)}
lizmat there you are binding the result of the assignment, which already switched the descriptors 20:09
afk& 20:11
ab5tract I understand that part... it's what is causing me to as the question: how does the next assignment "know" that is bound into the hash?
anyway, enjoy your afk :) 20:12
lizmat that's because %h<a> returns a container that has ContainerDescriptor::Whence descriptor
really afk& 20:13
ab5tract cheers
20:15 finanalyst left 21:46 sena_kun joined
ab5tract so the issue I've been poking at in R#2000 boils down to this: 21:54
linkable6 R#2000 [open]: github.com/rakudo/rakudo/issues/2000 [data types][metaop] `reduce` needlessly deconts
ab5tract m: use nqp; my %h; sub s(%h) { return %h<a> }; my $p := s(%h); dd nqp::getattr($p, Scalar, q|$!descriptor|).^name
camelia "VMNull"
ab5tract m: use nqp; my %h; sub s(%h) { return %h<a> }; my $p := s(%h); dd nqp::getattr($p, Scalar, q|$!descriptor|).^name; $p = 42 21:55
camelia "VMNull"
Cannot assign to a readonly variable or a value
in block <unit> at <tmp> line 1
ab5tract Should have checked that earlier, would have saved me some trouble 21:56
m: use nqp; my %h; sub s(%h) { %h<a> }; my $p := s(%h); dd nqp::getattr($p, Scalar, q|$!descriptor|).^name; $p = 42 22:06
camelia Cannot look up attributes in a Any type object. Did you forget a '.new'?
in block <unit> at <tmp> line 1
ab5tract (in case anyone was wondering)
m: use nqp; my %h; sub s(%h) { my $ = %h<a> }; my $p := s(%h); dd nqp::getattr($p, Scalar, q|$!descriptor|).^name; $p = 42 22:07
camelia Cannot look up attributes in a Any type object. Did you forget a '.new'?
in block <unit> at <tmp> line 1
ab5tract I've tried just about every other combination (my $ :=, return-rw, and more combinations thereof and thereabove) but I don't see a reason to flood the channel with them. 22:16
Is it expected for container descriptors to get dropped from the return value of a callable? 22:17
lizmat yes 22:19
by default, return values are deconted on return, unless the Callable has "is raw" or "is rw" on it 22:20
In fact, that's how postcircumfix <> works :-)
m: my $a = 42; use nqp; say nqp::iscont($a); say nqp::iscont($a<>) 22:21
camelia 1
0
lizmat ab5tract ^^
ab5tract thanks :) 22:23
still a fair bit for me to comprehend wrt containers and their descriptors
lizmat it's a bit of a rite of passage indeed
ab5tract at the risk of taking too much of your time, what's the difference between `is raw` and `is rw` in this case? 22:25
I assume something that is already `rw` will remain so via `is raw` 22:26
lizmat Actually, on return values I don't think there's a difference 22:35
at least not functionally
anyways, time for some shuteye 22:37
afk again&
releasable6__ Next release in ≈5 days and ≈19 hours. There are no known blockers. Please log your changes in the ChangeLog: github.com/rakudo/rakudo/wiki/ChangeLog-Draft 23:00
23:00 sena_kun left

Powered by Cro and the Raku® Programming Language.

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

Thank you!