»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_log/perl6 | UTF-8 is our friend! 🦋
Set by Zoffix on 25 July 2018.
MasterDuke timotimo: github.com/facebook/zstd/releases/tag/v1.4.0 00:40
MasterDuke blog post i don't think i'd seen before about perl6 solutions to the weekly challenge #3 perl6.eu/regular-pascal.html 00:47
ugexe i tried to do the pi challenge polyglot... it didnt turn out so great 01:44
gist.github.com/ugexe/c0c447738bd9...8db7178dd2
finishes quickly in perl 5, but i'm not sure it'll ever finish running in perl 6 lol 01:45
its taken 5 minutes to print ~40 of the 1000+ digits so far 01:46
need a quicker way to do polyglot +</<< than eval 01:47
MasterDuke are many different values getting shifted by many different values? if not could you cache the results and then use those? 01:52
ugexe hey that did the trick 01:57
its still slow as shit but it finishes after a minute of doing nothing
$g = $bitshift_cache{$b} //= (bitshift($b, 1) - 1); 01:59
MasterDuke huh, that gives more digits when run with perl5 02:03
ugexe try again. i must have made some not-so-superficial changes 02:09
MasterDuke ah, now it works with both 02:11
ugexe im not really familiar with this algorithm at all either. i used what was on rosetta code 02:13
MasterDuke whoops, segv when trying to profile 02:15
ugexe i only write code that cannot be profiled 02:16
MasterDuke sounds like a great way to preserve your anonymity 02:18
Kaiepi so while debugging an issue with perl 6 on openbsd i may have accidentally found a vulnerability in the os that lets me read anything and everything in memory, completely disregarding anything that'd otherwise stop me from being able to do that 02:29
i'm really excited to write a proof of concept for it since it's not every day you get to hack what claims to be the most secure os
MasterDuke congrats? 02:30
timotimo: around?
Kaiepi dw i'm not doing anything malicious with this
ugexe i wonder if there is a way to create an infix << 02:39
#`() sub infix:<< << >>($a, $b) { $a +< $b }; or some such 02:40
ugexe m: sub infix:«<<»(Int:D $a, Int:D $b) { $a +< $b }; say 1 << 1 02:42
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/KdgISISZTO
Unsuppo…
ugexe, Full output: gist.github.com/4d24c5ef730c30cb79...3e3c61114f
ugexe seems like i can't avoid the compile time error suggesting to use +<
should that error really be happening without p5isms? 02:46
then again its not exactly a p5ism so it also kinda makes sense 02:49
MasterDuke maybe an `is tighter/looser` would make it work? 02:50
yeah, that's really a c++ism
ugexe github.com/rakudo/rakudo/blob/7ee0...4017-L4019 03:22
while the error makes sense, i'm still unsure it should error when the infix has been defined by the user. 03:24
ugexe those are the only lines calling .sorryobs 03:25
MasterDuke m: use isms; multi sub infix:«<<»(Int:D $a, Int:D $b) { $a +< $b }; say 1 << 1 03:57
evalable6 2
MasterDuke maybe the error should suggest `use isms`? 03:59
SmokeMachine m: PROCESS::<$*RED-DB> = 42; say PROCESS::<$*RED-DB> 09:22
evalable6 42
SmokeMachine m: $*RED-DB = 42; say PROCESS::<$RED-DB> 09:24
evalable6 (exit code 1) Dynamic variable $*RED-DB not found
in block <unit> at /tmp/gMjyHsuuH9 line 1
SmokeMachine m: my $*RED-DB = 42; say PROCESS::<$RED-DB>
evalable6 (Any)
timotimo ugh, i hate it when a command segfaults unless run under gdb 09:26
Geth doc: 48e6e1820e | (JJ Merelo)++ | doc/Language/statement-prefixes.pod6
Adds quietly as statement prefix refs #2034
 09:39
synopsebot Link: doc.perl6.org/language/statement-prefixes
jmerelo Don't want to get thick, but we really need help in the documentation. If you have a few minutes, or even an hour, to spare, take a look at the issues, and do whatever you can. 09:40
sena_kun jmerelo, are there any urgencies or a massive breakage ongoing? 09:48
jmerelo sena_kun: it's documentation, it's not like it's not going to compile or anything. But there's a massive backlog of 6.d features, for instance. 10:17
sena_kun: the issue for doing that is pinned. In many cases, it's only checking that the feature is mentioned correctly. 10:18
Geth doc: e42f3287a9 | (Zili Chen)++ (committed using GitHub Web editor) | resources/i18n/zh/README.zh.md
Sync README.zh.md with README.md
 10:24
jmerelo sena_kun: there are also parts of the language that are not covered. Just look for the issues < 1000 10:29
sena_kun jmerelo, well, I am doing a full shift, including overtimes, for Comma almost every day, and while I am not, I am working on ldap. :| There are certainly a lot of tickets for docs, but I have a feeling that people are working on what they find fun and however time allows. 10:33
and I see people doing various improvements on their free time too 10:34
jmerelo sena_kun: it's quite clear that Perl6 is -Ofun 10:38
sena_kun: but it must be -Ofun for everyone, specially for people that arrive at Perl 6 for the first time through documentation. 10:39
sena_kun: all personal reasons for not working in the documentation are valid (or the ecosystem, or MoarVM for that matter)
sena_kun: but, all personal reasons considered (including mine), we still need help in the documentation, even if checking if the use of the ¿ sigil in Junctions in 6.d is documented is not fun at all. 10:40
sena_kun jmerelo, sorry, I've took a wrong approach from the start with your words. Yes, I guess you are right.
jmerelo sena_kun: maybe I have done that a bit too, so sorry about that. But anyway, solving issues helps, but spreading the word helps too. 10:43
Geth ecosystem: kalkin++ created pull request #450:
Update [email@hidden.address]		 10:54
ecosystem: a6cd7bb519 | (Bahtiar `kalkin-` Gadimov)++ | META.list
Update [email@hidden.address]
ecosystem: bb0cc6d346 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | META.list
Merge pull request #450 from kalkin/master

Update [email@hidden.address]
patrickb .hug sena_kun 11:30
.hug jmerelo
no huggable?
patrickb hugs sena_kun
patrickb hugs jmerelo 11:31
Thanks for your work, both of you!
lizmat seconds 11:33
jmerelo hey, thanks! 11:38
notable6: twitter.com/Tux5/status/1118459720623435776
notable6 jmerelo, I cannot recognize this command. See wiki for some examples: github.com/perl6/whateverable/wiki/Notable
jmerelo weekly: twitter.com/Tux5/status/1118459720623435776
notable6 jmerelo, Noted!
sena_kun gets timeout for tux.nl/Talks/CSV6/speed4.html 11:58
sena_kun oh, seems like it is just me. 11:58
lucs Any way to tell which version of a native lib a module is using? 15:03
jnthn It'd appear in the strace output, I guess 15:05
lucs I'll try that, thanks. 15:06
lucs jnthn: Yep, perfect. 15:11
jmerelo weekly: Solution to the 4th Perl Weekly Challenge, by Joelle Maslak github.com/jmaslak/perlweeklychall...slak/perl6 15:31
notable6 jmerelo, Noted!
jmerelo releasable6: status
releasable6 jmerelo, Next release in ≈3 days and ≈3 hours. 12 blockers. 0 out of 218 commits logged (⚠ 51 warnings)
jmerelo, Details: gist.github.com/0c894aa0e9f62b5375...fe1573decb 15:32
Wahnberger hi there: I have a file which i read with IO.lines .. inside i call a database method for adding some data.. psql server runs into "sorry, too many clients already" 15:56
i assume that IO.lines is doing some concurrency by default? If so .. how can i reduce that? 15:57
i have tried stuff like that: "race for $path.IO.lines(:batch(1))" but nothing happens 15:58
jmerelo Wahnberger: how do you call the database method? 16:01
sena_kun Wahnberger, can you show your code? Do you use DB::Pg or DBIish?
jnthn It won't do any concurrency unless you stick in `race` or `hyper` explicitly 16:04
Wahnberger I use DB::Pg 16:08
Wahnberger connection, single inserts and querys woring well so far 16:10
when the loop is running (IO.lines...) i check the connections in psql (pg_stat_activity) and they are exeeding the limit 16:11
so i am assuming that i this loop runs with concurrency 16:12
uhm..have to drop off.. will ask later, sorry 16:13
moritz which limit? number of connections limit? 16:22
moritz if so, that sounds like a bug 16:28
ugexe grumbles something about posting blatantly incorrect answers to stackoverflow 16:41
sena_kun ugexe, you can bash me directly for this one. However, the post is already deleted. 16:42
ugexe not yours
sena_kun erm, really? Takao == my account, just in case
ugexe yeah. stackoverflow.com/a/55731976/1772220 16:43
frankly this answer is just babbling
sena_kun well, the question is not particularly clear regarding feature X or feature Y, but rather in the area of "X doesn't work, there are like 10 or something reasons why it can be broken, but how do I fix that?" 16:44
which is a really gray area, imho
ugexe i understand. it does not help anyone to list a bunch of unrelated problems
because then anytime some precomp on install issue comes up someone will just regurgitate that list of unrelated problems 16:45
"oh did you bump the version?" well why would this even make sense to suggest? 16:49
jmerelo ugexe: you've got the privs to edit. You can edit the answer to make it better. 16:49
ugexe i had already posted an answer 16:50
jmerelo ugexe: well, thanks for the babbling bit. I'm not going to counterattack.
sena_kun it is also a gray area of where is stackoverflow and where is a bug tracker. I'd be much happier with it being narrowed down to particular issues, as I've saw questions that was, in fact, bug reports of Cro issues. So there can be no other answer than "Well, that's a bug" or "Well, that's a bug, I've fixed it".
jmerelo ugexe: did you *read* the answer? 16:51
sena_kun s/issues/questions/
ugexe obviously i read it
jmerelo ugexe: your answer was not helpful at all. I voted it up anyway.
ugexe: Main thing is that it's not an actionable answer. There's nothing the OP can do about it. 16:52
ugexe: you keep telling the OP things do not make sense. That's not really helpful.
ugexe you are being dense
jmerelo ugexe: you mention "module if" and don't link to it. It's not helpful in any way, might take you in the wrong direction, and does not solve the problem either.
ugexe because i dont expect the author to debug a transitive dependency 16:53
sena_kun wanders off to look at inline refactoring implementing
jmerelo ugexe: if you read it, you don't understand it. What I was saying is that effectively, it looks like the module was installed, and probably with the same version. We have to take the OP at face value: it does not find the module
ugexe: that means that the module was installed without that particular file, which is missing. 16:54
ugexe you frankly dont kow what you are talking about
so i'm done discussing this with you
jmerelo ugexe: as always, it's a pleasure to interact with you. Not. 16:55
ugexe: you can mention me by name, also. Not a problem to defend my blatantly incorrect views. 16:57
fatguy i have a TWEAK submethod on parent class, whenever i create an obj to child class, it will run the TWEAK from parent class. i don't want that, how can i achieve this ? 17:43
timotimo are you allowed to change the parent class? 17:45
tony-o m: role A { submethod TWEAK(|) { "A.TWEAK".say; }; }; class B does A { submethod TWEAK(|) { "B.TWEAK".say; } }; B.new; 17:46
evalable6 B.TWEAK
tony-o if you're not allowed to change the parent, you can override it (and just not call nextwith or similar) 17:47
ugexe they would need to change the parent from a class to a role still 17:48
fatguy what is ( | ) 'pipe' for ? 17:50
i can change the parent class. so my idea is to check the config file on parent class only 17:51
i will spawn the child based on certain condition
the parent class is called 'instance' for example and the child will be instance1...10 17:52
ugexe its an anonymous capture (since its not getting used in that example). if you wanted a non anonymous capture you'd use e.g. TWEAK(|c) { say c.perl } 17:53
timotimo the parent class can compare what type "self" has
fatguy i called the parent class to initiate a config check and log at the beginning of main script
tony-o ah, if you didn't have access you could wrap the TWEAK method also to check the .mro
guess, that wouldn't work either 17:54
not with mro, at least
timotimo m: class Parent { submethod TWEAK { say "parent's TWEAK has $(self.^name)" } }; class Child is Parent { submethod TWEAK { say "child's TWEAK has $(self.^name)" } }; Parent.new; say "and now child:"; Child.new
evalable6 parent's TWEAK has Parent
and now child:
parent's TWEAK has Child
child's TWEAK has Child
timotimo you'd then self.WHAT === Parent, for example
tony-o should a .wrap TWEAK have access to `self` ? 17:56
m: class A { submethod TWEAK(|) { "A.TWEAK".say; }; }; class B is A { submethod TWEAK(|) { "B.TWEAK".say; } }; say A.^can("TWEAK")[0].wrap(-> (|) { self.^name; }); B.new;
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/ay830bJfR1
'self' …
tony-o, Full output: gist.github.com/6c00c758d775b4a0b4...4b6ed16c57
ugexe why does declaring the TWEAK of both parent and child as `method` instead of `submethod` call both TWEAKs?
tony-o m: class A { submethod TWEAK(|) { "A.TWEAK".say; }; }; class B is A { submethod TWEAK(|) { "B.TWEAK".say; } }; A.^can("TWEAK")[0].wrap(-> (|) { say self.^name; }); B.new; # dies with self not available in the wrapped TWEAK 17:57
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/2HvwV476qk
'self' …
tony-o, Full output: gist.github.com/296a38e467d7941a9a...3f287d54b2
timotimo because the buildplan maker goes through the whole MRO and collects everything named TWEAK or BUILD
and it stashes away the actual code objects in the buildplan if i'm not mistaken
ugexe right. but is it this way because it was not considered to e.g. check for non submethods, or because the optimizations force this particular design? 17:58
fatguy thanks, i can work with if self.^name eq 'instance' for this 18:01
timotimo i'm not sure what the semantics should be if a method would be picked up every step of the class hierarchy
you'd have the tweak method of one class be called for any child that doesn't have its own
that seems like a very strange thing to want to have
sena_kun Can I use test data from a Perl 5 module? It uses artistic license, can't I get sued or something? Will mentioning it in the readme be enough? 18:08
Not like I've heard of a lot of such cases, just don't want to be rude to the author. 18:10
timotimo i think it should be fine? you're just not allowed to call it the same thing or claim you made the original i think?
tony-o i don't think you can redistribute if you've modified the code under artistic
without modification, you need to include license info 18:11
You may Distribute verbatim copies of the Source form of the Standard Version of this Package in any medium without restriction, either gratis or for a Distributor Fee, provided that you duplicate all of the original copyright notices and associated disclaimers. At your discretion, such verbatim copies may or may not include a Compiled form of the Package.
sena_kun no, I am not modifying the code, I just want to take test data to not re-implement the same foo-bar stuff.
ugexe turn your clock back 10 years. commit your code. claim they stole the tests from you
tony-o lol 18:12
the ol' windows95 registration hack
sena_kun Net::LDAP is probably older than 10 years. :p
tony-o he's working on metric years 18:14
100 seconds in 2 10 hour segments of a day with 20 days per month and 10 months in the year
and years are twice as long 18:15
sena_kun p-probably... ok, I guess I'll peek at the original just a little, considering that the actual implementation code is drastically different, and test data always can be edited after
tony-o sena_kun: the 1.0 license has different but similar requirements 18:18
if you're looking at an old set of tests, that might be relevant
opensource.org/licenses/Artistic-1.0 18:19
sena_kun metacpan.org/source/MARSCHAP/perl-...canon_dn.t <- I am looking at this
sena_kun I don't want to re-invent those `O=Test,C=GB` into `O=Test2,C=EnoughDifferenceForEveryone` 18:20
the code is not modified, and my code itself is pretty original
Xliff \o 19:20
m: role A { method new(1) { say 'Hi' }; }; class B does A { }; B.new(1) 19:21
evalable6 Hi
Xliff m: role A { method new(1) { say 'Hi' }; }; class B does A { }; A.new(1)
evalable6 Hi
Xliff m: role A { has $.a; submethod BUILD(:$!a) { }; method new(1) { self.bless(a => 1) }; }; class B does A { }; A.new(1).a 19:24
evalable6
Xliff m: role A { has $.a; submethod BUILD(:$!a) { }; method new(1) { self.bless(a => 1) }; }; class B does A { }; A.new(1).a.say
evalable6 1
Xliff m: role A { has $.a; submethod BUILD(:$!a) { }; method new(1) { self.bless(a => 1) }; }; class B does A { }; B.new(1).a.say
evalable6 1
Xliff m: role A { has $.a; submethod BUILD(:$!a) { say 'Building role' }; method new(1) { self.bless(a => 1) }; }; class B does A { submethod BUILD { say 'Building class'; }; B.new(1).a.say
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/ovgX7bmzbC
Missing…
Xliff, Full output: gist.github.com/a4727c1f8ebbdbcddf...4e48c85e2a
Xliff m: role A { has $.a; submethod BUILD(:$!a) { say 'Building role' }; method new(1) { self.bless(a => 1) }; }; class B does A { submethod BUILD { say 'Building class'; }; }; B.new(1).a.say 19:25
evalable6 Building class
(Any)
Xliff m: role A { has $.a; submethod BUILD(:$!a) { say 'Building role' }; method new(1) { self.bless(a => 1) }; }; class B does A { }; B.new(1).a.say
evalable6 Building role
1
fatguy i had config file, read using "filename".IO.lines, need to ignore '#' at the beginning and empty line, i can use starts-with for '#' but for empty line i use unless .Stringy to make it work. tried using starts-with(' ') and regex \s+ but it didn't work. is there any better way ? 19:28
sena_kun you can just `next unless $line` 19:29
m: for ('a', '', 'b') -> { next unless $_; say $_ } 19:30
evalable6 (exit code 1) Too many positionals passed; expected 0 arguments but got 1
in block <unit> at /tmp/oH9Nv5HaeR line 1
sena_kun m: for @('a', '', 'b') -> { next unless $_; say $_ }
evalable6 (exit code 1) Too many positionals passed; expected 0 arguments but got 1
in block <unit> at /tmp/zDIAyLjL_Z line 1
sena_kun erm
m: for @('a', '', 'b') { next unless $_; say $_ }
evalable6 a
b
sena_kun fatguy, ^ 19:31
fatguy ok thanks
sena_kun m: for @('a', '', 'b') { next if .chars == 0; say $_ }
evalable6 a
b
sena_kun also this
unclechu m: for @('a', '', 'b') { say $_ unless $_ } 19:32
evalable6
unclechu m: for @('a', '', 'b') { say $_ if $_ }
evalable6 a
b
sena_kun m: for ('a', '', 'b') { say $_ if $_ }
evalable6 a
b
sena_kun m: for 'a', '', 'b' { say $_ if $_ }
evalable6 a
b
sena_kun m: .say if $_ for 'a', '', 'b' 19:33
evalable6 a
b
unclechu m: say $_ if $_ for @('a', '', 'b')
evalable6 a
b
Xliff m: role A { has $.a; submethod BUILD(:$!a) { say ::?CLASS.^name.say; say 'Building role' }; method new(1) { self.bless(a => 1) }; }; class B does A { }; A.new(1).a.say
evalable6 A
True
Building role
1
Xliff m: role A { has $.a; submethod BUILD(:$!a) { say ::?ROLE.^name.say; ::?CLASS.^name.say; say 'Building role' }; method new(1) { self.bless(a => 1) }; }; class B does A { }; A.new(1).a.say 19:36
evalable6 A
True
A
Building role
1
ugexe MasterDuke: using "#`() use isms; sub infix:«<<»(Int:D $a, Int:D $b) { $a +< $b };" instead of the previous bitshift sub (both using the cache still) from 45s to 10s 20:33
the cache is no longer needed even 20:34
guifa kinda feels like if should set up an implict context. So you can do “return $_ if %foo<bar>:exists” 21:22
masak m: my %foo = bar => "OH HAI"; if %foo<bar> -> $value { say $value } 21:28
evalable6 OH HAI
masak if you don't insist on using :exists and the inverted `if`, you have your wish
guifa Interesting that it requires the pointing block though 21:33
jnthn `.return with %boo<bar>` is definedness is good enough, which it usuall is 21:34
*usually
*if
Apparently too tired to type :)
guifa ooh I like that
tyil not sure if this is interesting enough to put in the p6 weekly pleroma.tyil.nl/notice/9hrPeqI82uP3W3i8aO 21:35
guifa I had no idea you could just call .return on a value
tyil same, but that looks p cool 21:36
guifa I actually have to call another method first before the routine but it works fine 21:37
.format( … ).return with %foo<bar>
sjn tyil: does your image contain the rakudo star modules too? 21:41
tyil no, bare perl6
no zef either (and the python one doesnt contain pip on alpine i believe) 21:42
sjn a nightly R* based on that one would be cool, I think
tyil I could probably just put RUN a zef install on a list of modules that R* also includes and get there 21:43
sjn btw, isn't adding a build-date an anti-pattern?
tyil R* does not do any patches on mainline perl6 does it?
a build-date tag, you mean?
sjn reproducible builds ftw
I see a RUN date "+%FT%TZ" > /var/docker/meta/build-date 21:44
tyil oh
that's not in the perl6 docker image, that's in the cpan-p6 one
.bots
cpan-p6 03cpan-p6:121.0.1 (2019-04-12T16:14:21Z) reporting for duty! [Perl 6] 02git.tyil.nl/perl6/app-cpan-uploadannouncer-irc
tyil that's how it gets that datetime stamp
it allows me to easily check if I'm looking at the right docker image, in case k8s wasnt able to pull a newer image yet 21:45
sjn but it makes it impossible to re-build an image from the same commit and end with the same build sha
tyil docker isnt a reproducible build to begin with, though 21:46
sjn really? 21:47
ctilmes_ tyil: alpine testing/edge has moarvm/nqp/rakudo installed in them.
you can run echo dl-cdn.alpinelinux.org/alpine/edge/testing >> /etc/apk/repositories && apk -U add rakudo
tyil ctilmes_: I heard, but this is latest master branch every night, so slightly more up to date
sjn well, even if that's true, it's a bad argument for not making the effort to make your own build reproducible
tyil and easier to try out custom patches (like one from samcv to check if memory consumption improved) 21:48
ctilmes_ My latest pull was rejected ;-( but I believe that was due to known rakudo bug: github.com/alpinelinux/aports/pull/6845
tyil sjn: well, I don't intend cpan-p6 to be reproducible here, because I intentionally want it to be different so I can easily see if I have a newer image
the perl6 one does not have this timestamp
sjn: git.tyil.nl/docker/perl6/src/branc...ker/alpine this is the perl6 docker image 21:49
sjn tyil: I think you might be misunderstanding the point of having reproducible builds 21:50
tyil no, I do like reproducible builds
but first of all, docker is not a reproducible build system
and second of all, the perl6 image doesn't have that timestamp 21:51
sjn what makes you say docker isn't a reproducible build system?
tyil what makes you think it is?
Nix is a reproducible build system, it makes sure that settings are always correct, including system datetime, pins on specific versions of other packages and validates them all via hashing 21:52
docker doesnt do any tricks to make sure the system state is consistent across builds, and there's no validation on whether RUN commands every have the exact same effect 21:53
sjn hmf
tyil while it may sometimes accidentally make a rather reproducible thing, that doesn't make it a reproducible build system
sjn thought NixOS used docker...
tyil I believe they do build in a container, but additionally they ensure system state is always the same for every build as well 21:54
not sure if they specifically use docker for builds, though
(fwiw, this is what I learned from a NixOS talk, I don't use NixOS myself) 21:56
cpan-p6 New module released to CPAN! JSON::Path (1.6) by 03JNTHN 22:02
tyil damn, python:latest is almost 1gb 22:06
python latest 59a8c21b72d4 2 weeks ago 929MB
hythm_ p6: my $i = 0; loop { last if $i == 7; $i++; LAST say $i } 22:09
evalable6 7
hythm_ p6: sub s { my $i = 0; loop { last if $i == 7; $i++; LAST say $i } }; s
evalable6 (exit code 1) 0
No such method '!capture_phasers' for invocant of type 'Code'
in block <unit> at /tmp/roVf2vPCkV line 1
sjn that looks like a bug to me 22:14
hythm_ i don't get the above error error on repl, instead i get "0" as output, which seems wrong anyway.
timotimo oh the repl 22:17
sjn there's a 0 in that output here too 22:19
hythm_ oh, right, did not notice it 22:21
SmokeMachine m: class A { submethod TWEAK(|) { "A.TWEAK".say; }; }; class B is A { submethod TWEAK(|) { "B.TWEAK".say; } }; A.^can("TWEAK")[0].wrap(my method (|) { say self.^name; }); B.new; 22:22
evalable6 B
B.TWEAK
hythm_ updated rakudo issue #2836, seems related 22:25
timotimo R#2836 22:29
synopsebot R#2836 [open]: github.com/rakudo/rakudo/issues/2836 [phasers] `LAST` loop inside `sub` increments `$` for every iteration
MasterDuke ugexe: a couple more of those tricks and you're going to end up with a perl (5|6) -> polyglot perl compiler 22:31
ugexe get the worst of both worlds!
MasterDuke perl 6, the worst perl 5 *and* the worst perl 6 22:33
ugexe perl -1, 5-6 22:38
guifa Ugh. Mozilla released Fluent v1.0 today. I need to get datetime to work soon so I can get to 1.0 22:39
Juerd Oof, Perl6 got a lot slower on my test case since the last time I tried (months ago perhaps) 22:40
It used to take between 5 and 6 seconds, now 10+.
While I admit the circumstances of this benchmark are suboptimal, I can't think of any explanation for this except regression 22:41
ugexe is there a way to write this without conditional logic? sub polyslurp ($_) { "0" and (return "{$_.IO.slurp}") or (return do { open(my $fh, $_[0]); join("", <$fh>); }) };
MasterDuke Juerd: this is the mqtt regex test, right? 22:42
Xliff Is there any way to make an object's method override an implemented role's method? 22:53
ugexe have it mix-in an anonymous role at build-time that calls the method with the classname qualifier ala self.CLASSNAME::whatever 22:55
Xliff So.... 22:58
m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; A.new.a
evalable6 Role
Xliff m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = A.new but role { method a { self.A.a }; }; $a.new.a 22:59
evalable6 (exit code 1) No such method 'A' for invocant of type 'A+{<anon|1>}'. Did you mean 'a'?
in method a at /tmp/J6QZWNhDzo line 1
in block <unit> at /tmp/J6QZWNhDzo line 1
Xliff m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = A.new but role { method a { self.A::a }; }; $a.new.a 23:00
evalable6 Role
Xliff ugexe: ^^
m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = A.new but role { method a { "override" }; }; $a.new.a
evalable6
Xliff m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = A.new but role { method a { say "override" }; }; $a.new.a
evalable6 override
Xliff m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = A.new but role { method a { self.a }; }; $a.new.a 23:01
evalable6 (signal SIGHUP) «timed out after 10 seconds» 23:01
Xliff m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = A.new but role { method a { self.a }; }; $a.a
evalable6 (signal SIGHUP) «timed out after 10 seconds»
ugexe you are never using class B, so why are you including it?
Xliff m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = B.new but role { method a { self.B.a }; }; 23:02
evalable6
ugexe self.B::a
Xliff m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = B.new but role { method a { self.B::a }; }; 23:03
evalable6
ugexe not self.B.A
m: my role A { method a { say "Role"; }; }; class B does A { method a { say "Class"; }; }; my $a = B.new but role { method a { self.B::a }; }; say $a.a
evalable6 Class
True
Xliff That's a bit confusing, but OK.
/o\
Thanks.
OK. That's better. Thanks! 23:04
guifa jnthn: in doing some more tests, the .return with $foo while cleaner code takes more than 2x time :( 23:16
Probably because the with creates a code block (I knwo someone else had something else similar not long ago where that was the slowdown) 23:18
MasterDuke guifa: can you show two golfed code snippets so we can profile them? 23:20
guifa tio.run/##jY/dCoIwGIaP21V8xIIZOSyo...914ese6fgA 23:25
MasterDuke guifa++ 23:27
guifa MasterDuke: Results are actually even more in favor of the return foo if foo when you have more hits than misses (change the grep to !*.is-prime) 23:29
lucs If I change for example some code under ./src/core/... in my local clone of rakudo, will a 'make' be sufficient to update everything? 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!