»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_logs/perl6 | UTF-8 is our friend!
Set by moritz on 22 December 2015.
lookatme morning 00:30
steve-o total perl 6 noob question alert 00:37
can i use Tk with perl6
?
jdv79 i don't think there's any bindings 00:38
there's some gtk ones
lookatme buggable: eco gtk
buggable lookatme, Found 4 results: GTK::Simple, GTK::Simpler, GTK::Scintilla, Inform. See modules.perl6.org/s/gtk
lookatme buggable: eco tk 00:39
buggable lookatme, Found 4 results: GTK::Simple, GTK::Simpler, GTK::Scintilla, Inform. See modules.perl6.org/s/tk
lookatme steve-o, ^^^ 00:39
steve-o yeah i was just looking at that site. so if i want to make an extremely simple desktop app with perl6 is GTK a tool that will work? 00:40
jdv79 gtk is the gnome gui stuff 00:41
lookatme IDK
thought gtk support win
though
jdv79 yeah, it does 00:42
lookatme you can see, the appveyor test is failed
steve-o ok thanks i will look into GTK 00:43
pompomcrab my $resp = HTTP::UserAgent.new.get: $url; 01:41
Segmentation Fault
i have no idea how to debug this, but i guess it's time to learn. what should i start with? 01:42
in perl5 i would use B:: to see what line it was executing where i croaked. what is the p6 equiv of B::?
pompomcrab the $url is 'wikileaks.org/macron-emails//get/' LOL 01:44
maybe HTTP::U::A doesn't like https
and yes, this is just a simple scraper to pull all the emaisl
pompomcrab err i mean $url is 'wikileaks.org/macron-emails//get/2' 01:45
MasterDuke pompomcrab: try with per6-valgrind or perl6-gdb 01:49
lookatme and may try to update your rakudo if you run old version 01:50
pompomcrab MasterDuke: thx 01:51
lookatme buggable: eco io 01:54
buggable lookatme, Found 235 results: p6doc, Bailador, BioPerl6, v5, Web. See modules.perl6.org/s/io
AlexDaniel m: for ‘foo bar baz’.match: :g, / (\w+) / { say ~$0 } 02:09
camelia foo
foo
foo
AlexDaniel why is it always foo? I do understand that I was supposed to write it like this:
m: for ‘foo bar baz’.match: :g, / (\w+) / { say .[0] } 02:10
camelia ｢foo｣
｢bar｣
｢baz｣
AlexDaniel but still, why?
geekosaur it doesn't know how to re-alias $0 for each returned Match object 02:14
really, it can't (what if you save into an array and then random-access, should it know which index you were working with to alias $0?) 02:15
so $0 always gets aliased to the first Match object
probably also $/
I mean, perl 5 would probably have some special case recognizing that you used for ...match(:g) and aliasing stuff inside the loop 02:17
that's not really perl 6 style magic though
raschipi Can it do symbolic references with a number to access $0, $1, $2, etc? 02:18
AlexDaniel I'm thinking why would it be aliased to anything at all 02:22
perlawhirl match populates $/, which is what you're accssing with $0 (which is sugar for $/[0])... instead you really want to access $_ inside you're loop. 02:25
*your
and while it seems more consistent to treat things that way, i think it could be confusing for people, particularly those familiar with p5 semantics
AlexDaniel um… 02:27
geekosaur AlexDaniel, $/ is an alias to the result of the most recent regex match. in this case, there is no single "most recent", there is a list of 3. it uses the first, pretending there was no :g. (I could argue $/ should be Match in that case...) 02:28
I do see what you want it to do; I do not see why you think it is something that should just automatically happen without help 02:29
there is *one* match operation. it has completed, producing 3 Match objects.
AlexDaniel no, I'm thinking if it does nothing useful with :g, then why put anything into $/ at all
geekosaur it is too late for $/ to be each one in turn
unless that knowledge is coded into for
*that* is what I meant by it being Match (that is, the Match type object) 02:30
AlexDaniel m: for ‘foo bar baz’.match: :g, / (\w+) / { say $0 }
camelia ｢foo｣
0 => ｢foo｣
｢foo｣
0 => ｢foo｣
｢foo｣
0 => ｢foo｣
perlawhirl AlexDaniel: it's not equivalent to p5: while ( $foo ~= /(\w+)/g ) { say $1 }
AlexDaniel: It doesn't match progressively, it's all at once
m: for 'foo bar baz' ~~ m:g/(\w+)/ { say $/; exit }
camelia (｢foo｣
0 => ｢foo｣ ｢bar｣
0 => ｢bar｣ ｢baz｣
0 => ｢baz｣)
geekosaur this is for not while. .match is not a generator that gets tickled with each step through the loop; the whole result has been produced and for is now iterating over that result 02:31
this is perl, not icon or haskell :p
mst *perl6 02:32
I've implemented such generator behaviour in perl5 just fine
geekosaur sure, can be done in python too. but it's not inherent behavior 02:33
mst sure, just let's not casually call perl6 perl, masak and I went to some effort to get it clear that it wasn't to avoid stupid internecine warfare
AlexDaniel oh… I think I see it now 02:34
I should've printed $/ to see what's going on 02:35
geekosaur hm. I was thinking more language family, it isn't a perl.*-ish thing to magically turn anything into lazy or generator expressions automatically
it is a perl.*-ish thing to do it in specific cases though (different cases in perl 5 and perl 6)
perigrin perl.*-ish turned *me* into a lazy generator 02:35
so I present myself as a contradictory example
AlexDaniel thanks! 02:36
mst geekosaur: "this is a perl" would express the same POV and I'd agree 100% 02:37
(and sorry for pedanting it but I've spent too long squelching silly arguments so obviously I'm (?:over)?sensitive) 02:38
geekosaur anyway it probably would make sense to have $/ be something useless in that case to catch thinkos like that. unless @Larry would prefer that for recognize this case and rewrite it ... but I am recalling that older Perls did that at one point and it was discarded as encouraging for/while confusion 02:39
(possibly recalling incorrectly though) 02:40
well, sometihng like that, not this specific (equivalent) case 02:41
(I .. have had a few cases where a perl programmer tried to use a similar idiom in not-perl and was utterly mystified at the result, because other languages didn't quietly treat your for as a while.) 02:42
AlexDaniel geekosaur: ok, wait, but why? It is useful 02:43
raschipi In P6 for is quietly treated as a map instead.
AlexDaniel m: for ‘foo bar baz’.match: :g, / (\w+) / { say $/ }
camelia (｢foo｣
0 => ｢foo｣ ｢bar｣
0 => ｢bar｣ ｢baz｣
0 => ｢baz｣)
(｢foo｣
0 => ｢foo｣ ｢bar｣
0 => ｢bar｣ ｢baz｣
0 => ｢baz｣)
(｢foo｣
0 => ｢foo｣ ｢bar｣
0 => ｢bar｣ ｢baz｣
0 => ｢…
AlexDaniel geekosaur: here you have all of your results
geekosaur oh, I just remembered what I was thinking of, it's the behavior of the each iterator 02:44
and they were doing the opposite thing: while (each ... )
which, no, very no 02:45
AlexDaniel m: ‘foo bar baz’.match: :g, / (\w+) /; for @$/ { say .[0] } 02:46
camelia ｢foo｣
｢bar｣
｢baz｣
lookatme m: for @(‘foo bar baz’.match: :g, / (\w+) /) { say .[0] } 02:52
camelia ｢foo｣
｢bar｣
｢baz｣
AlexDaniel that does nothing 02:53
tyil[m] Perl conf is pretty cool so far 02:59
You guys are missing out
mst geekosaur: using 'each' in perl5 is invariably an error 03:00
geekosaur: it only exists that way to iterate DBM files that only allowed a single handle
geekosaur: I've banned it from all codebases where I can do so :) 03:01
perlawhirl speaking of missing out... i'm in australia and don't really have a chance to go to all the cool confs... is there some otherway I can get some nice merch, like those plushie camelia's?
tyil[m] perlawhirl: I can ask wendy if I dont forget 03:03
She's the one at the merch stand
perlawhirl thanks 03:04
snarkyboojum oO(plushie camelias!) 04:19
dorothyw Sorry to intrude. I am wondering if there is a perl6 repl that is sutiable as a shell 05:27
*suitable
dorothyw Even if there was just a really good repl for perl6 I think it could be hacked into being a suitable shell 05:29
dorothyw Like when I type perl6 into the terminal and enter a repl I get similar issues as the perl repl where when I use arrowkeys for example I get these weird escape codes printed 05:35
dorothyw Going to reconnect just a moment 05:36
dorothyw b 05:40
ryu0 dorothyw: option 1. install the Readline or Linenoise packages. 05:41
ryu0 dorothyw: option 2. use rlwrap to wrap perl6 command. 05:42
dorothyw: it's not really a shell like BASH but it will give you a complete REPL.
dorothyw I've seen these mentioned on stack overflow and I am on the Linenoise github page though it eludes me how to install Linenoise
I am very new to perl though I am perl6 all the way because macros and because better than bash etc
ryu0 do you have zef alreay installed? 05:43
dorothyw no though I have rakudo installed
ryu0 you need to install zef first.
ryu0 most instructions assume use of zef for installing modules for perl6. 05:43
dorothyw installing zef as per github instructions 05:44
ryu0 i've been working on making it more convenient to install on ubuntu, but it'll take a bit. there's a lot of modules in the rakudo star distribution. 05:45
36ish. 05:46
dorothyw Getting an undeclared names error 05:47
ryu0 dorothyw: paste it somewhere please.
dorothyw hastebin.com/raw/nitebizoya
ryu0 ... How old is your rakudo? 05:49
dorothyw I performed sudo apt-get install rakudo from ubuntu 16.04 05:50
ryu0 ... that's why.
dorothyw Want me to compile?
ryu0 I have a better idea then. 05:51
uninstall the repo rakudo. it's very ancient and unlikely to work with current perl6 modules.
Here is my Perl 6 PPA. It's only for xenial right now as it's still a WIP. But it should be enough for you to get started. 05:52
launchpad.net/~ryu0/+archive/ubuntu/perl6
xenial is the current ubuntu you're using.
dorothyw moment 05:53
ryu0 i'm hoping this will be an asset to novices who at least know enough to use PPAs. 05:54
dorothyw do I add it as xenial main or xenial universal? 05:56
ryu0 Uh... xenial main I assume, but normally people just use the commands given at that page. It takes care of those details for you.
# add-apt-repository ppa:ryu0/perl6
It imports the PPAs public key and adds an APT repo for it. 05:57
dorothyw Should have just igven me that command :p
ryu0 Sorry, thought you could find it from the page.
dorothyw So now I apt-get install rakudo
right? 05:58
ryu0 after an apt update.
or apt-get
either works.
apt-get install rakudo perl6-readline zef
that should get you started.
you can use zef to install anything else you need, either into your private perl6 stuff or system-wide. 05:59
dorothyw installing zef
ryu0 i'm just packaging everything that ships with rakudo star so it's a complete set of the baseline packages.
just not there yet, but what is there should be quite usable. 06:00
does this solve your problem?
dorothyw Yeah that gets rakudo installed then I can try to install linenoise 06:01
ryu0 ok.
i was suggesting perl6-readline as it does the same thing as linenoise
and is already packaged by my PPA.
dorothyw ah
ryu0 i'm using it to substitute for Linenoise. 06:02
dorothyw apt-get installing it now
ryu0 If you were using perl6 on another platform, then you'd use Linenoise.
But really, I only see Linenoise being of interest if you want to use it as a module, but that's not your use case right now.
You just wanted a complete REPL for perl6, yes? 06:03
dorothyw Yeah i want to try to get a perl shell going
ryu0 rakudo automatically has one if you have Readline or Linenoise installed.
dorothyw I just typed perl6 and pressed enter and it seems to be working better now 06:03
ryu0 Ok.
lookatme perl6intro.com/#_installing_perl_6 06:15
dorothyw What's wrong with this line? sub term ($arg1, $arg2){q:x/$arg1 "$arg2"/;} ; term("echo", "hello world") 06:17
lookatme m: sub term ($arg1, $arg2){q:x/$arg1 "$arg2"/;} ; term("echo", "hello world") 06:20
camelia qx, qqx is disallowed in restricted setting
in sub restricted at src/RESTRICTED.setting line 1
in sub QX at src/RESTRICTED.setting line 11
in sub term at <tmp> line 1
in block <unit> at <tmp> line 1
dorothyw I'm pretty new to perl error messages 06:23
lookatme dorothyw, you need qq:x I think
and sub term ($arg1, $arg2){print qq:x/$arg1 "$arg2"/;} ; term("echo", "hello world") 06:24
ryu0 lookatme: cool. i was just referring them to my PPA since it's easier for novices to get running, especially if a release happens to fail to build for some reason. 06:25
ryu0 lookatme: seems like a good idea, especially if we want perl6 to expand. 06:26
:-)
dorothyw So now I've got my term function right
lookatme :) 06:27
dorothyw but what I need to do is take a string and parse up to the first space and make that two values, then plug value 1 into $arg1 and value 2 into $arg2
lookatme oh, I thought you want something visual
but I 06:28
but I'm sure qq:x is what you want
ryu0 all looks like jibberish to me. i'll find out what it means some time. 06:29
but it sounds like they're trying to partially reproduce the effects of the POSIX function wordexp. 06:30
lookatme qq allow variable interpolation, and q not 06:31
dorothyw I am trying to ge tthe same functionality as system() from perl
ryu0 Uh, it'd probably be easier to just make a NativeCall reference. 06:31
lookatme dorothyw, if you want system, you can use QX
ryu0 Or w/e is more appropriate. 06:32
hm. 06:33
m: sub foo(Int $arg1) { print $arg1; }
camelia ( no output )
ryu0 m: sub foo(Int $arg1) { print $arg1; }; print foo(1); 06:33
camelia 1True
ryu0 O_o
m: sub foo(Int $arg1) { print $arg1; }; print foo("a");
camelia 5===SORRY!5=== Error while compiling <tmp>
Calling foo(Str) will never work with declared signature (Int $arg1)
at <tmp>:1
------> 3 foo(Int $arg1) { print $arg1; }; print 7⏏5foo("a");
ryu0 \o/ 06:34
I'm liking this.
I always hated how dynamic languages made static typing conventions tedious to implement.
dorothyw I am basically just fed up with bash and so I did alot of research to try to replace bash and I think perl6 is a good way to do that 06:36
ryu0 you do realize perl6 isn't meant as a replacement for a system shell? 06:36
though if you want to try, be our guest. 06:37
lookatme replace bash ?
ryu0 it's a better programming language for sure.
lookatme you can try some other thing like `zsh`
I use zsh
dorothyw Well I think an issue with bash is that every single line of code has got to begin with a filename
bash is not really built for programming in general just accessing files and executing them with arguments
ryu0 system scripting, basically. 06:38
it's very good for its niche but not much else.
dorothyw perl6 also has extensible macros so afaik a user can extend perl6 to be whatever they want, unlike bash which has a limited and punishing syntax
ryu0 that's just the nature of POSIX shells. 06:38
ryu0 it was never meant to be a true programming language. 06:39
dorothyw This is the issue I hate with bash is that it is not meant to be a true programming language
ryu0 BASH is fine in its own time and place.
Use it as an asset to your other stuff ideally. 06:40
I don't expect my command line shell to do a lot honestly.
lookatme yeah, also there some non-POSIX shell implementation
dorothyw The issue I have with bash is that bash is often one of the first languages new programmers learn and it is really not fully satisfying as a programming language 06:40
ryu0 yes. zsh is a common choice for people that hate BASH. 06:40
... It is?
jast I don't exactly hate bash but I like zsh more
ryu0 I thought the cool kids all learned Python these days. 06:40
jast sure, or JavaScript 06:41
dorothyw When i was learning linux bash was one of the first things I learned
lookatme I remember there are some guys want make a perl shell replace system shell, and also a perl OS :)
you can try to do something like that
jast I think it's probably quite different when your primary goal is "learning linux"
ryu0 Not many people know it but it's also possible to extend BASH through its C API.
jast but my guess is most people learning to program start out on windows or os x
andrzejku there is kind of Perl Shell already 06:42
jast two, as I recall
dorothyw I like perl6 better because of the macros
andrzejku and a long time ago there was a guy which tried to make Perlix
ryu0 You can write BASH builtins as long as it exposes a specific C ABI.
andrzejku a perl operating system
jast one is called Zoidberg, it gets extra points for naming, even if I've never actually tried it 06:42
andrzejku but he didn't finish it as someone rushed him 06:43
lookatme andrzejku, yeah, I heard Perlix from you :)
ryu0 i did that once to allow dumping the shell variables to JSON.
jast I don't think building an OS on top of an interpreted language is the way to go
ryu0 it was for a distro that wanted to collect metadata about their shell script source packages.
lookatme dorothyw, what macro ?
andrzejku the legend says that if you start to do Perl OS that bad things will happend
jast that's almost like putting a bytecode-based VM with GC and above-average memory use inside embedded devices, like mobile phones
... oh.
andrzejku :<
ryu0 That's Java.
LOL
dorothyw lookatme: afaik perl6 has fully extensible macros meaning you can invent your own syntax
lookatme dorothyw, I know the slang 06:44
ryu0 dorothyw: so do many Lisps.
dorothyw Lisp has really failed to make a good repl outside of emacs though
jast with lisps it's debatable that they create 'new syntax', though :}
ryu0 haha
lizmat dorothyw: that's the theory, but most slangs are not based on macro's but instead on subclassing Perl6's grammar
ryu0 dorothyw: you can also look at TCL, it's closer to a true language and tends to follow some shell syntax conventions. 06:45
includes a shell-like interpreter too
jast Tcl is... different
dorothyw I have been looking at gforth for a long time
jast bit of a fringe language these days, too
dorothyw cause gforth is very small and shell-like
ryu0 indeed it is.
jast for 'very small', check out lua
dorothyw lua is nice though I shy away from it's syntax. Plus no fully extensible macros I believe 06:46
ryu0 puts [exec ls *] 06:46
:)
jast true, and I don't like its syntax too much, either
dorothyw The reason I want fully extensible macros is this: we started with shells built into the os in the 1940s and rewrote those for bourne shell and rewrote that as ksh and rewrote that as bash and now we are rewriting that as zsh. So I am hoping for a language that will not need to be rewritten 06:47
One shell to rule them all that will last forever
ryu0 dorothyw: ... uh zsh is a separate project from bash.
jast well, I'm not so sure that's actually achievable
lookatme How about fishshell ?
dorothyw fish and zsh are both good but they lack fully extensible macros so if I have a syntax feautre that I want I can't get it. 06:48
jast I've thought about language design extensively back when I thought I was much more inventive than I actually am :)
ryu0 ok, then TCLs out.
there's not many languages that let you remodel the language itself.
jast the problem with fully extensible macros is that you'll inevitably run into conflicts between various extensions
dorothyw There's ways around that
lookatme and also I know a non-POSIX shell Elvish written with GO
jast as long as each syntax extension is isolated, e.g. by putting it into a block of sorts with an identifier of sorts, it's all easy 06:49
but if you want to essentially turn one language into another, that's a different story
dorothyw Basically whenever you walk down a syntax hallway you have to have a way to go back to back your way out of a syntax corner
You don't want escalators that are one way where you add a syntax feature and it's stuck with you 06:50
jast I'm talking about a set of extensions that make your input ambiguous
it could have two different meanings depending on the order in which you try the extensions (if we're talking about recursive descent parsing, but the ambiguity is independent of the parsing emthod)
dorothyw My thinking is that you create a non ambiguous way to hop into and out of a specific macro. So like macrox***>??? or something really specific like that 06:51
jast that backtracking might be necessary at some point is kind of straightforward :}
yeah, that's what I meant by confining into a block
that's not quite "fully extensible macros", though, if you ask mne
dorothyw It's kind of like how quotations marks work in code. You have an entry point and an exit point, both the quote mark.
True that's just how I would do it 06:52
jast and one thing that gets very difficult with this is properly displaying errors if there's a delimiter mismatch
the best macro system is worthless if the error messages from the parser are meaningless
dorothyw I don't know a good solution to that other than walking through what your macro evaluates to 06:53
jast I have a design for a language in mind that does this kind of thing, but with some extra restrictions
dorothyw Basically there are some key rules for being a good macro writer. 1. If you have a language with full macros don't write a macro that blocks your full macro abilities. 2. Make your macros unambiguous and clear. 3. Have your macros evaluate to good code 06:54
ryu0 jast: i've used C macros extensively. you have to take care with them.
err
jast it's a lisp at heart, but with syntax (gasp)... and by forcing all syntax extensions to respect proper balancing of parentheses, it's not too complicated to translate all of it back into s-expressions internally
ryu0 dorothyw: ^
jast C's macros are a dirty hack, though 06:55
ryu0 hell GNU C has some Lisp-like extensions for their macros.
({ }) blocks for one.
dorothyw So I am a lisper I'll admit but I am starting to be convinced that at it's current state lisp is not ready to be a shell language.
ryu0 most languages weren't meant to be a shell language.
jast sure is, if you're a die hard lisp fan :)
just make sure to have a few spare parens keys 06:56
dorothyw I am not really die hard but I think common lisp is an example of a language that can fit nearly all applications.
I am more of a macro fan.
ryu0 Did you look at other lisp implementations?
dorothyw I see common lisp as the most technologically superior lisp but I think scheme and elisp are also very good lisps 06:57
ryu0 I think I've seen more Lisp derivatives than any other languages.
dorothyw It is very easy to write a lisp
That's sort of off topic though I am talking about macros in general
jast yeah, anyone with a stack and a basic understanding of ASTs can write a lisp
ryu0 I actually liked the ML family more myself but any.
anyway 06:58
dorothyw When I started programming in c++ I always assumed there would be something like macros available to me where I could make syntax that did exactly what I wanted and I kept looking and looking and found out few languages actually do this
ryu0 Used mixins before?
TEttinger forth?
dorothyw Yes I am a huge forth advocate
Basically top tier languages: common lisp, perl6, forth 06:59
TEttinger I don't find forth readable (yet), but I like a lot of its qualities
same with perl6, which I doubt I could ever read
ryu0 dorothyw: did you look at D? i've heard it's a nice hybrid of different stuff.
dorothyw ryu0: macros are the key 07:00
forth nees to be cobol-ized
TEttinger Nim goes hard into imperative/procedural but emphasizes compile-time programming as key
nadim m: role R{}; my @a = [] but R ; my $a = [] but R ; dd @a, $r ;
camelia 5===SORRY!5=== Error while compiling <tmp>
Variable '$r' is not declared
at <tmp>:1
------> 3 = [] but R ; my $a = [] but R ; dd @a, 7⏏5$r ;
nadim m: role R{}; my @a = [] but R ; my $a = [] but R ; dd @a, $a ;
camelia Array @a = []
Array+{R} $a = $[]
ryu0 i never did much with common lisp, so I can't really know to what extent you want to use macros.
i just know macros and mixins tend to provide similar functionality.
dorothyw Basically I think writing a programming language where you can't change the syntax however you want is a sin
ryu0 Ok, then most languages are out then. 07:01
TEttinger fixed syntax is easier to make IDE integration actually work with
it also can help avoid cases where users make the grammar ambiguous
I think perl6 has a solution for that last one 07:02
nadim Can someone tell me with "but" does not add the role to the @a but does to $a? m: role R{}; my @a = [] but R ; my $a = [] but R ; dd @a, $a ;
ryu0 dorothyw: i once tried common lisp but i couldn't get past the crap the #lisp community threw at me.
dorothyw Lisp community sux but support question tho
jast communities... who needs those anyway 07:03
TEttinger yeah I've definitely seen emacs supremacy stuff in lisp communities
jast that's stupid, notepad++ is obviously way superior :P 07:03
ryu0 i decided to move onto something less hostile. common lisp wasn't worth it to me.
jast for most things I try to avoid interacting with the communities, that works too 07:04
ryu0 shrugs.
TEttinger "you need this version of emacs with this mode" "no that mode has been superceded by blingerblang" "really? I thought the blingerblang guy had moved to an off-the-grid commune"
ryu0 ... lol 07:05
could have just said foobar
jast "this mode only works with this third-party fork of emacs that doesn't do releases and offers no binaries"
TEttinger it sounds more like a CL lib name
jast not sure that actually exists, though :)
ryu0 guys, anyone that can answer nadim? he did ask something.
jast I'd love to but I have no idea :) 07:06
lookatme m: role R{}; my @a = [] but R ; my $a = [] but R ; dd @a, $a ; 07:07
camelia Array @a = []
Array+{R} $a = $[]
lookatme m: role R{}; my @a = [] but R ; my $a = [] but R ; dd @a[0], $a ;
camelia Any @a = Any
Array+{R} $a = $[]
lookatme role R{ method r() { say "CALL ME"; }; }; my @a = ([] but R); my $a = [] but R ; dd @a, $a ; say ([] but R).r; 07:10
m: role R{ method r() { say "CALL ME"; }; }; my @a = ([] but R); my $a = [] but R ; dd @a, $a ; say ([] but R).r;
camelia Array @a = []
Array+{R} $a = $[]
CALL ME
True
lookatme m: role R{ method r() { say "CALL ME"; }; }; my @a = ([] but R); my $a = [] but R ; dd @a, $a ; $a.r; ([] but R).r; @a.r;
camelia Array @a = []
Array+{R} $a = $[]
CALL ME
CALL ME
No such method 'r' for invocant of type 'Array'
in block <unit> at <tmp> line 1
lookatme and how about use does 07:11
m: role R{ method r() { say "CALL ME"; }; }; my @a does R = []; my $a = [] but R ; dd @a, $a ; $a.r; ([] but R).r; @a.r;
camelia Array+{R} @a = []
Array+{R} $a = $[]
CALL ME
CALL ME
CALL ME
lookatme nadim, ^^ don't know why that not work 07:12
m: role R{ method r() { say "CALL ME"; }; }; my @a does R = []; my $a = [] but R ; dd @a, $a;
camelia Array+{R} @a = []
Array+{R} $a = $[]
ryu0 lookatme: eh? I thought Roles couldn't provide an implementation. I thought they only described a class API.
lookatme ryu0, it can 07:14
ryu0 it appears to act more like an abstract class then.
lookatme you can defined the default implementation
ryu0 so it behaves more like an abstract class. Can Roles themselves be instantiated?
lookatme It's quite helpful when you need a default behavior 07:15
ryu0 Or can roles only be instantiated by derivatives?
lookatme yeah, perl6 generate same name class for u 07:16
ryu0 oh, so it's more like a general purpose base class.
assuming inheritance follows.
lookatme they call that autopun
ryu0 trying to bridge Perl6 to the general OOP concepts they use in C++, C#, and Java. 07:17
jast a more common name for the same concept is 'mixin' 07:19
ryu0 Oh. 07:20
jast it's not too uncommon: en.wikipedia.org/wiki/Mixin#Progra...use_mixins
ryu0 I thought it was inheritance related because i assumed you could also use other classes instead of just the base class in place of that role's type. 07:22
ryu0 So a role then is an interface of sorts that provides a default implementation if the implementing class doesn't define its own. 07:23
err that can provide
jast yes, and that implementation can use methods from the target class 07:24
ryu0 sounds like a hybrid of interfaces and abstract classes.
lookatme docs.perl6.org/type/auto-punning 07:26
auto-punning is not mixin
docs.perl6.org/language/objects#in...try-Mixins
stmuk larry streaming live 07:39
snarkyboojum there's a public live stream? 07:47
stmuk www.facebook.com/pg/theperlconference/videos/ 07:48
snarkyboojum thank you!
lookatme how long it will last ? stmuk 07:50
stmuk its about 20 mins in 07:51
lookatme oh 07:52
lookatme ugexe, zef crash when run install: gist.github.com/araraloren/fdb91f8...ae41f42d8f 08:20
grondilu are methods "my" or "our" by default? 08:24
grondilu I'm wondering because I'm not sure I should write class Foo { my Foo method clone {...} } or class Foo { our Foo method clone {...} } 08:25
lookatme grondilu, my or our is use for declare static member 08:26
grondilu oh 08:27
lookatme docs.perl6.org/language/classtut#Static_fields?
grondilu I had seen this syntax somewhere, I thought it was a fancy way of specifying the return type. 08:28
so I guess I should stick to the method clone(--> Foo) {...} syntax
lookatme I prefer `method clone(*%_) of Foo { ... }` 08:30
El_Che ugexe: is Zef OK to be used a a module or is it meant as internal for the zef command only? E.g. to use it to download the chain of sources to an specific dir without installing them?
lookatme grondilu, if you want some deep clone feature. I can give you some advice, if you need 08:34
El_Che (sorry for the silly questions all-around, just trying stuff obout distribution) 08:35
grondilu lookatme: no it was just a dummy example here 08:37
lookatme grondilu, oh :) 08:38
stmuk "perl is a four letter word" :) 08:54
grondilu m: role Foo { method f(--> Num) { rand } }; class :: does Foo { has $.f = 1.2e0 }.new.f 08:55
camelia ( no output )
grondilu m: role Foo { method f(--> Num) { rand } }; class :: does Foo { has $.f = 1.2e0 }.new.f.say
camelia 0.26245213534034
grondilu m: role Foo { method f(--> Num) { rand } }; class :: does Foo { has $.f = "Bar!" }.new.f.say
camelia 0.791408320845822
grondilu :/ 08:56
I guess that makes sense
grondilu m: role Foo { method f(--> Num) {...} }; class :: does Foo { has $.f = "Bar!" }.new.f.say 08:57
camelia Bar!
grondilu but that^ bothers me
lookatme m: role Foo { method f(--> Num) {...} }; class :: does Foo { has $.f = "Bar!" }.new.Foo::f.say 09:03
camelia Stub code executed
in method f at <tmp> line 1
in block <unit> at <tmp> line 1
lookatme grondilu, what do you mean ? 09:07
grondilu I was expecting a type error 09:09
thing is : I would like to create two types of classes implementing "Foo" : one would do it with a f method, and one would do it with a f attribute. I wanted both to obey the type constraints defined in the role Foo. 09:10
lookatme Hmm, I don't know about that 09:11
jnthn grondilu: Methods are has-scoped by default, not my or our
grondilu m: class Foo { has Foo method clone {...} } 09:12
camelia ( no output )
grondilu I see
thanks
lookatme grondilu, but, method or accessor in `::` has its own return type constraint
m: role Foo { method f(--> Num) { rand } }; class A does Foo { has $!f = "Bar!"; method f() of Str { $!f; }; }; A.new.f.say 09:13
camelia Bar!
lookatme m: role Foo { method f(--> Num) { rand } }; class :: does Foo { has $.f = "Bar!" }.new.f.say
camelia 0.441744955363787
lookatme jnthn, why it choose f in Foo ? not access in :: ? 09:14
nadim m: role R{}; my @a = [] but R ; my $a = [] but R ; dd @a, $a ; 09:17
camelia Array @a = []
Array+{R} $a = $[]
nadim jnthn: do you have an explanation for why the array does not do the role when role is assigned to @var but it works when it is a $var 09:18
jnthn lookatme: Because an attribute only generates a method if there isn't already one
nadim: = is assignment, not binding 09:19
jnthn Assignment is a copying operation 09:19
lookatme oh, thanks, that make sense
jnthn So my @a = blah means "iterate blah and put the things from it into @a" 09:20
nadim jnthn: so the lhs being an array does not get the extra role
lookatme so when I defined a f method in ::, it will call f first
nadim m: role R{}; my @a := [] but R ; my $a = [] but R ; dd @a, $a ;
camelia []
Array+{R} $a = $[]
jnthn lookatme: Yeah, in fact if you write class C { has $.a; method a() { } } then your method wins, by the same "only generate if it's missing" rule
lookatme jnthn, got it! :) 09:21
m: role R{}; my @a does R := [] but R ; my $a = [] but R ; dd @a, $a ; 09:22
camelia []
Array+{R} $a = $[]
lookatme m: role R{}; my @a does R := ([] but R) ; my $a = [] but R ; dd @a, $a ;
camelia []
Array+{R} $a = $[]
jnthn nadim: Yeah, assignment always means "put into". That's why things like my @pairs = %h; can work; it iterates %h (getting pairs) and puts them into @a
lookatme m: role R{}; my @a does R = ([] but R) ; my $a = [] but R ; dd @a, $a ;
camelia Array+{R} @a = []
Array+{R} $a = $[]
suman___ Hi everyone 09:31
Greetings from Nepal
I need help here in this question stackoverflow.com/questions/456096...ith-perl-6 09:32
moritz hi suman___
nadim lookatme: your example may not do what you want, you already have an array but R on the lhs. the interesting thing would be to assign is a normal array not one that doe R 09:33
lookatme m: role R{}; my @a does R = [] ; my $a = [] but R ; dd @a, $a ; 09:34
camelia Array+{R} @a = []
Array+{R} $a = $[]
lookatme m: role R{}; my @a = ([] but R) ; my $a = [] but R ; dd @a, $a ;
camelia Array @a = []
Array+{R} $a = $[]
lookatme suman___, you can simple use substitutions replace \begin{...} like s/'\begin{verbatim}'/'\begin{minted}{perl6}'/ 09:41
suman___ lookatme Sounds good, will try 09:42
lookatme you welcome 09:43
off work
bye
suman___ lookatme thanks bye 09:43
nadim m: role R{has $.x}; my @a does R = [1] ; @a.x =7 ; 09:45
camelia Cannot modify an immutable Any ((Any))
in block <unit> at <tmp> line 1
jnthn m: role R{has $.x is rw}; my @a does R = [1] ; @a.x =7 09:47
camelia ( no output )
nadim yes rw
nadim m: my role R { has $.max_lines is rw = 0 } ; my @d2 does R = [1..2] ; @d2.max_lines = 1 ; 09:50
camelia ( no output )
nadim Ha! ^^ does not compile here, error is: No such method 'max_lines' for invocant of type 'Array' 09:51
timotimo if the invocant is "of type 'Array'", then the role mixin got lost
tyil perlawhirl: I asked Wendy, I am to give you her email address so you can discuss it directly with her 09:53
nadim timotimo: my @array does R is indeed an Array only 09:54
timotimo oh, huh 09:54
nadim timotimo: no I am wrong, well no I am right, this is actually funny. 09:57
timotimo: I had an error in my code, @a is redeclared a few lines below but it makes it fail a few lines up. I need to golf this.
timotimo ah, heh. 09:58
timotimo facebook video player has no speed option :\ 09:59
nadim it seems that when a variable is redeclared in the same scope, the latest declaration is the one that is taken for the whole scope. Makes sense, is there a way to make a variable redeclaration a fatal error. 10:01
timotimo it already warns, doesn't it? 10:03
nadim yes it does 10:05
nadim very rightly so, so this is a mess from me but I was surprised to see that the type of an object at line 10 is change by a redeclaration on line 20. It makes perfectly sense, I made the error 10:07
nadim but I would't mind having it as an error 10:07
timotimo hm, the general case is you accidentally have my $x; and later my $x; again, and it's no problem. so you see the warning, the program runs fine anyway, but you can immediately pull it out 10:14
timotimo but i'm not sure what our stance on warning vs error is 10:16
stmuk timotimo: youtube-dl web.facebook.com/theperlconference...852932049/
nadim is it possible to make it an error, specially if the types are different, which is the case I had here.
timotimo stmuk: oh i hadn't thought of that! 10:16
it is, with a bit of coding
stmuk the video url comes from right clicking the original 10:18
stmuk there is a perl 6 talk starting now 10:20
timotimo i'm still watching larry's keynote 10:21
actually, i'll defer that to download 10:22
nadim then I'll go watch it too :) 10:25
nadim woops that's long 10:31
timotimo can't read the slides properly
i imagine the url at the top would let me see the slides
but i can't read the url either
tyil stmuk: does that work for all presentations that were posted? :o
I want to share the High End Unicode to some folk online 10:32
stmuk tyil: I think so
tyil pretty cool 10:32
I'll check it after this talk
people are dying to see it :p
timotimo why don't we have a kind of Mix that can hold Complex Numbers? :) 10:43
timotimo ab5tract++ # good talk 11:02
nadim good to know, I've see another one but it was more a social type and little content, except the social content which is also of value 11:15
timotimo: shall I open a request for the redeclaration as error or there is so little chance that it is not worth it? 11:16
raschipi Did you guys see that bioinformatics software now need input sanitation? 12:02
nadim don't all software need that? 12:08
raschipi Somone wrote an exploit in DNA. 12:09
ryu0 raschipi: now you can say that viruses run in the genes. 12:17
lol
AlexDaniel raschipi: source? 12:30
raschipi AlexDaniel: www.wired.com/story/malware-dna-hack 13:05
stmuk more unicode p6 at www.facebook.com/theperlconference/ 13:55
mr_ron In docs.perl6.org/language/regexes#Ba...er_classes word boundary is given as <|wb> but I don't think you need the '|' here to make it zero width. <wb> and <|w> are both word boundaries. Can I take out the leading '|' in <|wb>? 13:56
moritz I prefer <?wb> actually 13:56
<wb> captures, but capturing a word boundary doesn't make all that much sense 13:57
geekosaur basically, the use of specifically | is commentary (telling the reader it's a zero width assertion), the more important part is that since it doesn't start with an alphanumeric, it is non-capturing 13:59
do not underestimate the value of signalling your intent to a future reader of the code
mr_ron m: my $m = "The quick brown fox" ~~ /(.)(<wb>)quick/; say $0.chars; say $1.chars 13:59
camelia 1
0
mr_ron Why does <|wb> need the pipe and not not <ww> ? 14:01
raschipi_ Here for me the pipe has the same effect for both. 14:04
mr_ron m: say "The quick brown fox" ~~ /(.)(<wb>)(quick)/; say "The quick brown fox" ~~ /(.)(<|wb>)(quick)/; say "The quick brown fox" ~~ /(.)(<?wb>)(quick)/; 14:10
camelia ｢ quick｣
0 => ｢ ｣
1 => ｢｣
wb => ｢｣
2 => ｢quick｣
｢ quick｣
0 => ｢ ｣
1 => ｢｣
2 => ｢quick｣
｢ quick｣
0 => ｢ ｣
1 => ｢｣
2 => ｢quick｣
raschipi_ m: say "The quick brown fox" ~~ /(.)<?wb>(quick)/; 14:12
camelia ｢ quick｣
0 => ｢ ｣
1 => ｢quick｣
mr_ron The doc only says zero-width and not non-capturing - introducing the extra concept at that point in the doc seems more confusing. At least both <wb> and <ww> should be same. I think better to leave (non-)capturing out but if you want to leave then maybe <|wb>... Word Boundary (zero-width assertion, non-capturing with '|'). 14:20
raschipi_ mr_ron: The best way to explain it is to send a patch to the docs. 14:23
mr_ron thanks
japhb ab5tract: What was your talk about? (I saw timotimo mention it above) 15:34
timotimo "rakudo has cool builtins, which is a good thing. also, you can read the source of these builtins because it's in perl6"
timotimo is a mostly wrong executive summary 15:36
kybr i'm trying to fix what i experience as a broken link in the docs. gist.github.com/kybr/074966feb5da5...5a8762186e 15:54
geekosaur kybr, it's a relative link 16:10
geekosaur looks like someone copy-pasted text from type/Baggy to routine/roll 16:11
so the L<pick|#method pick> turned into <a href="#method_pick"> which will only work on the Baggy page. the browser shows you not the relative link but what the link resolves to based on the page URL 16:12
geekosaur also, you seem to be looking for the source backwards. the link is from roll to pick, you cited a link in pick not in roll 16:14
github.com/perl6/doc/blob/master/d....pod6#L150 is the link
oh, I see, this is actually the generated docs pulling in stuff from elsewhere. it needs to fix relative links 16:15
not sure who needs to do this, likely file a docs bug
it should be adjusting relative links based on the "From <place>" url 16:16
moritz anyone can do that who can read and write Perl 6 code
it might not be trivial, though 16:17
geekosaur right, in this case the doc builder needs to (a) save the original source page (b) delete an existing anchor (c) add in the new one --- but that handles only the case L<text|#anchor> 16:17
geekosaur other kinds of relative links would be harder 16:18
moritz or rewrite the links from relative to absolute when transplanting the POD data structures 16:27
user3 m: say +("aaa \n bbb \n ccc \n" ~~ m:g/\n/) 16:33
camelia 3
user3 is there a better way than that to count the number of newlines in a string?
kybr unfortunately, i'm not going to be able to file a bug beyond what i already posted on that gist. i'm happy to file the content of my gist and geekosaur's response. is that enough or does someone else want to take that on? 16:34
user3 or in general to count the number of instances of a given character in a string
Geth doc: 5a2cd9da98 | (Steve Mynott)++ | doc/Type/Iterable.pod6
remove trailing whitespace
 16:35
ckraniak Ok trying out perl6, can't get nativecall to find a dll that's in the cwd 16:37
Haven't done lots with dlls but so far as I can tell there's nothing up with the dll itself 16:38
I have tried to pass the full path in and that doesn't work 16:39
raschipi_ m: say +"aaa \n bbb \n ccc \n".comb.grep("\n") #user3 16:41
camelia 3
user3 raschipi_: hmm, looks even more complicated...
lol
raschipi_ .comb divides the string into a list of charachters, then grep separates the ones we want, then casting to numeric returns the number in the list 16:42
user3 yes
timotimo hold on 16:44
comb *also* matches stuff
m: say "foo bar baz quux lol omg wtf".comb("o").perl
camelia ("o", "o", "o", "o").Seq
timotimo m: say "foo bar baz quux lol omg wtf".comb(/<[aeiou]>/).perl
camelia ("o", "o", "a", "a", "u", "u", "o", "o")
raschipi_ m: say +"aaa \n bbb \n ccc \n".comb("\n");+"aaa \n bbb \n ccc \n \n".comb("\n") 16:46
camelia WARNINGS for <tmp>:
Useless use of "+" in expression "+\"aaa \\n bbb \\n ccc \\n \\n\".comb(\"\\n\")" in sink context (line 1)
3
raschipi_ m: say +"aaa \n bbb \n ccc \n".comb("\n");say +"aaa \n bbb \n ccc \n \n".comb("\n")
camelia 3
4
user3 say +"aaa \n bbb \n ccc \n \n".comb("\n") 16:47
evalable6 4
user3 wow
it's better
mspo one thing I like about fancy quotes is that there is an explicit open and an explicit close 16:48
so there's no real «balancing» to be done
raschipi_ m: say +"aaa ccc".comb("c");say +"aaa\n ccc\n ccc".comb("c")
camelia 3
6
raschipi_ They're so much better that they're used everywhere except where a small number of characters was needed. 16:51
mspo: But balancing issues remain, just like with bracing. 16:52
mspo raschipi_: easier to detect, I think is what I'm saying 16:53
raschipi_ I see your point.
mspo raschipi_: or maybe there's no real advantage and I'm just making it up :)
ckraniak where is the search path for nativecall on Windows documented?
ugexe docs.microsoft.com probably 16:55
ckraniak As in "is native([stuff])" 16:56
Can't point it at cwd apparently 16:57
raschipi_ ckraniak: You might try to explore this stuff on Unix, we will be able to help better.
ugexe its wonky. i think it does something different if its a relative path vs an absolute path and str vs io::path
kybr geekosaur: is this the same problem that i described? github.com/perl6/doc/issues/1144
raschipi_ Especially because it's better defined there. MS Windows is... problematic. 16:58
ckraniak Can't argue with you there 16:58
raschipi_ Do you have a motive to this on Windows? 16:59
ugexe you probably have to do './' for $*CWD
github.com/rakudo/rakudo/blob/nom/...names.t#L9 so much for windows 17:00
raschipi_ I didn't say that it's simple, just that it's predictable. Not because of the OS in itself but because of conventions regarding the libraries. 17:01
ugexe I tried to bring some sanity to it here github.com/rakudo/rakudo/pull/730 17:02
ckraniak I have tried ./
Windows is more available
ugexe ckraniak: what do you get for 17:03
perl6 -e "use NativeCall :TEST; say guess_library_name('./foo')"
ckraniak Same as $*CWD basically 17:06
ugexe depends how you look at it. note the difference between ./foo and foo 17:08
"foo" and "foo".IO also return different things (guess_library_name is a mess) 17:10
ckraniak guess_library_name("$*CWD/foo") == guess_library_name("./foo") 17:11
ugexe m: use NativeCall :TEST; say guess_library_name("./foo"); say guess_library_name("foo"); say guess_library_name("foo".IO) 17:12
camelia /home/camelia/libfoo.so
libfoo.so
/home/camelia/libfoo.so
ckraniak "foo".IO evaluates to full path, "foo" does not 17:14
Should say
ckraniak In guess_library_name 17:15
ugexe right, and thats weird because 17:16
m: say "foo".IO.Str
camelia foo
ckraniak Dropping the dll in system32 and leaving off ./ does not work apparently 17:19
Other dolls in system32 do work, like opengl32
*dlls
Gotta be an environment variable or a registry key or something 17:20
kybr does rand 17:41
does srand set the seed for pick and roll?
geekosaur kybr, yes, that's the sam eissue 18:15
ckraniak May be that it can find it but error message is decceptive 18:20
Python cannot load it either but can definitely locate iy 18:21
Guess: trying to load 32 bit dll into 64 but exe
ryu0 If Linenoise and Readline are both installed, which will rakudo choose to use for its REPL? 18:22
ckraniak so moar is 64 but then 18:23
*bit
jdv79 can anyone else install JSON::Tiny on the latest rakudo? 18:30
jdv79 This is Rakudo version 2017.07-144-gec7bc25 built on MoarVM version 2017.07-365-gc0f7a3b 18:31
jdv79 is everyone at yapc or watching it? 18:36
ugexe jdv79: fails for me 18:40
jdv79 does anything weird happen < 1m afterwards?
anywhere from immediately to ~30s later all my gnome-terminals vanish without a trace 18:41
ugexe i think unicode is getting corrupted, and your terminals couldnt handle what it output 18:42
gist.githubusercontent.com/ugexe/9...tfile1.txt
jdv79 ah. gist.github.com/anonymous/5facb80b...f0a2569e38 18:45
fun
ckraniak GOT IT 18:49
was 32 bit dll in 64 bit exe
mingw-m64 fixed it
ckraniak god bless cmake 18:50
mspo am I in bizzaro world? :) 18:50
ckraniak Perl6 apparently has the world's least helpful error message for this, you see 18:51
mspo sounds like a bug
having good messages is a goal
ckraniak Its sort of passing through the message from LoadLibrary though 18:52
Actually wait I lied that was Python
ckraniak The error message is "cannot locate native library", but the problem is actually the dll was compiled to the wrong architecture 18:54
timotimo isn't that a quirk of how windows does that?
ckraniak Maybe
timotimo like, if you're a 32bit process, dlls that were made for 64bit don't actually exist?
as ridiculous as that sounds 18:55
geekosaur yes, windows cannot cross-load dlls
timotimo well, of course it can't cross-load them
ckraniak The error message isn't somethi bc you could pull help from though
geekosaur if you rin a 32 bit process on 64 bit windows, it starts up a cut-down hyperv to run it in a 32-bit container
timotimo but it pretends the file doesn't exist, no?
geekosaur 64 bit processes cannot load 32 bit dlls
ckraniak Yes and it took a few hours to realize that was the problem instead of some weird path resolution issue 18:56
geekosaur hm. I thought recently someone was getting a "bad file format" type error from that
timotimo that's definitely what you get on linux
ugexe El_Che: yes you can use zef's modules to do things without the cli - see github.com/ugexe/zef/blob/33c18c90...tory.t#L45 19:01
ryu0 ugexe: is there any reason to package panda? i read zef supersedes it. 19:02
ugexe the only reason would be that a few packages may have panda listed as a dependency (see: `zef rdepends panda`) 19:08
ryu0 \o/ 19:13
12 modules left until i have rakudo star packaged completely.
raschipi_ ryu0++ 19:18
Are you submitting them to Debian?
ryu0 raschipi_: ubuntu PPA.
raschipi_ So?
raschipi_ My question remains... 19:19
ryu0 not really planning on it per say. debian has their own packaging policies that i don't really know too well.
and ultimately they'll be out of date long before they make it into one of their major releases. 19:20
ryu0 it may make more sense later when rakudo is more mature that having an older release isn't a big deal. 19:20
raschipi_ Nice, give me a link and I will do it myself. 19:21
ryu0 launchpad.net/~ryu0/+archive/ubuntu/perl6
debian already has the 3 core packages of rakudo though, just a heads up.
i'm not done packaging just yet, but you can certainly reuse how i figured out to package the modules. 19:22
ryu0 raschipi_: why? do you actually use debian? 19:23
raschipi_ Just looking at that page, I can already tell they're too out of policy to be useful. 19:23
ryu0 heh. too many?
raschipi_ No, Perl/perl6 ackages have to follow the debain perl policy, which means that "perl6" goes at the end of the name of the package. 19:24
ryu0 Oh.
raschipi_ Are you open to renaming them?
ugexe github.com/ugexe/zef/issues/117 # detailed discussion on some issues with packaging for Debian (mostly precomp related) 19:25
ryu0 raschipi_: a lot of work when i've already come so far. lol 19:25
hm. 19:26
raschipi_ For example, the package for File::Directory::Tree has to be named libfile-directory-tree-perl6 19:27
ryu0 raschipi_: i don't really feel like renaming them, but i can share what i learned if you feel like writing debian packages. 19:28
raschipi_ I will ask if I need anything. If you want to write, can you please do it in the link ugexe posted above? 19:29
ryu0 it already documents the main point.
ryu0 packaging a certain script from rakudo package for installing the modules into the vendor directory. 19:30
it works quite well. the only thing i added was to symlink if anything gets installed into the bin directory.
it's also how the stuff is packaged on ARCH. 19:31
raschipi_: i advise writing a skeleton package when doing the modules. it helps a lot. 19:32
i have only had to make minor adjustments for depends and such. 19:33
raschipi_: there's ~35 modules to package. most only depend on rakudo, others depend on one or more module packages. 19:35
raschipi_ I know the debian Perl6 team is already working on it, I will go trough the team channels, thanks. 19:36
ryu0 ok.
sorry i can't really help you, but i felt it best to just let debian do it their way. 19:37
raschipi_ Just FYI, soon their packages will be pulled by Ubuntu and there might be conflicts, you should check to add appropriate conflicts to your packages.
ryu0 once it makes into a ubuntu LTS i'll probably discontinue this PPA.
raschipi_: that'll probably be awhile. 19:38
raschipi_ Why do you say that? 19:39
ryu0 because debian stable tends to take years to get a new release.
raschipi_ Ubuntu doesn't pull from stable, it pulls from sid.
ryu0 Oh. 19:40
raschipi_ And Ubuntu will release in October, they try to pull multiple times and as close as possible to the release date if they don't detect problems. 19:41
ryu0 ... has debian even packaged a single perl6 module yet? all i've seen is rakudo. 19:43
mst ryu0: pleae don't discontinue the PPA 19:44
ubuntu are completely useless at shipping perl stuff
ryu0 mst: ok.
mst: you actually want to use this PPA i'm building?
mst I want people to be able to do so 19:45
my solution is to only use ubuntu when shadowcat customers are paying me to support perl on it
at which point their "snapshot debian testing at random and ignore it for six months" policy is a billable hour generator ;)
raschipi_ ryu0: did you understand my point about conflicts? 19:46
ryu0 raschipi_: yes, if someone's trying to use both at once. 19:47
i did distribution packaging for 8 years. i know what a conflict is.
just not for debian really.
i'll deal with it if/when it arises. 19:48
ryu0 raschipi_: any idea why debian chose to split rakudo and its library into different packages? it seemed pointless. 19:50
raschipi_ You mean the "rakudo" and "rakudo-lib" split? 19:50
ryu0 yes.
ryu0 afaik, nothing except rakudo even links to that library. 19:52
raschipi_ I 'm not aware of their possible motivation. 20:04
moritz probably a matter of policy, in case anybody wants to link to it later 20:05
much like Inline::Perl5 now links to libperl.so
raschipi_ Policy requires splitting if more than one package links to it. But since just rakudo links against it, it's not required. 20:06
moritz ok
raschipi_ According to the description, the lib is an ELF-handling lib, useful for general use, so they might have split preemptively in case someone does want to use in the future. It isn't much work anyway, all handled by the automatic tools. 20:10
timotimo hey everybody 20:18
bikeshedding time
moritz sheds his bike and his skin 20:19
timotimo what methods on a IO::Socket::Async would you expect to give you access to the ip of the local and remote sockets?
stmuk ryu0: I think you need to set NO_NETWORKING_TESTING for perl6-lpw-simple 20:23
ryu0 stmuk: Ah. I had disabled the tests because they failed due to no networking access. 20:24
i'll put that in the next time I have to rebuild it.
mspo getsockname and getpeername ? 20:26
those are the c functions, right?
timotimo uv also names them that, yeah 20:32
i do split it into ip and port, though
and also i only support ipv4 and ipv6
ryu0 ah, so no unix sockets. 20:33
timotimo we don't support them at all yet ;( 20:34
ryu0 not too surprising. they're not universally portable. 20:35
namely windows has nothing comparable.
stmuk ryu0: if you are using the official release tarballs for nqp 2017.07 rather than the nqp in R* you might want to consider this patch as well 20:36
build.opensuse.org/package/view_fi...f?expand=1
ryu0 why? what's it even do? 20:37
stmuk it fixes --ll-exception 20:38
ryu0 i don't even know what that does.
Ah.
timotimo normally exceptions are abbreviated, skipping over frames inside the compiler and builtins 20:39
with --ll-exception those are also shown, which is much more useful if you're reporting bugs in the compiler and/or builtins
ryu0 was this fixed in git?
don't really want to have to rebuild everything just for this :| 20:40
stmuk well maybe considering checking opensuse for patches in future .. it's likely R* will use them 20:41
ryu0 R*? 20:42
stmuk rakudo star 20:43
ryu0 oh.
stmuk this fix is in star too
timotimo yup, we used a non-release version of nqp in rakudo star to get this fix 20:46
ryu0 either way i'll get it next time i bump nqp so i don't think i'm going to worry about it at this time. 20:47
i just checked and it's been patched in git.
ryu0 timotimo: are the monthly releases of moarvm, nqp, and rakudo recommended for actual use? 20:50
stmuk if you fix it this time you will have a mechanism for next time a similar problem occurs 20:52
ryu0 stmuk: i just wish the precomp crud didn't require a recompile of every module package. 20:53
stmuk ryu0: I think there has been talk of making precomp more friendly to binary packagers in irclog.perlgeek.de/perl6-toolchain/ 20:55
ryu0 ideally there'd be some ABI and old compiled bytecode would still work, though maybe less than optimally. 20:56
just the way it is right now is like having static linkage. 20:57
timotimo the monthly releases are good, but sometimes we do the terrible thing where the rakudo star release has a patch that the monthly release doesn't
the second-last month we did like six point releases in a row 20:58
that wasn't good, either
ryu0 timotimo: mostly i wanted to package the same feature set as rakudo star, maybe a bit more depending on what i get requests for or find useful. 20:59
timotimo i see
ryu0 hm.
but i'll work this patch in. 21:00
i just assumed that the separate releases were exactly the same. -_-
stmuk there was some talk of removing the time stamps from the SHA1 type precomp files
ryu0 SHA1? i thought they would be using SHA256 by now. 21:01
stmuk ryu0: the monthly releases are like Linus kernel releases .. I don't think a bit of patching is always bad 21:01
most linux distros don't use a Linus kernel 21:02
ryu0 considering how they've found collisions with SHA1 now. 21:03
stmuk its great more people are experimenting with packaging rakudo
well git still uses SHA1 as well .. ]
ryu0 stmuk: frankly rakudo star should be packageable directly but i found that to be next to impossible currently. 21:04
raschipi_ ryu0: They can find collisions if they control both inputs.
And this isn't a scurity critcal thing, they were talking just above about removing it...
ryu0 i just know it broke some subversion repos awhile back. 21:05
raschipi_ Rephasing: they can produce too documents that appear bona-fide but have the same sha1 hash.
Yes, if both documents are put into version control that relies on hashes, it will be a problem.
ryu0 i heard a mitigation is to also include the length in some respect. 21:06
raschipi_ But it's not something one expects to find in the wild. It could be used into a scam, though.
ryu0 or, just use a stronger hash.
raschipi_ Yeah, that's why it's only done with files that will hide the real lenght, like PDFs.
stmuk ryu0: yes it should be 21:07
raschipi_ Yeah, the recommendation to use a stronger hash is warranted if problems could arise. But in this case, they are talking about removing it.
raschipi_ If sha1 is used in other places, it should be changed because it's only a question of time before it causes problems. 21:08
Well, sha256 too, but it's a much longer time.
ryu0 might as well jump straight to sha512 to prolong the time. 21:09
stmuk pull requests welcome! :)
raschipi_ It makes sense to measure how much resources that takes.
People, I have to go now. Hugs for everyone. 21:10
ryu0 Can't... breathe...
lol
mspo timotimo: they return structures with a bunch of info, I thought 21:23
address and port at least 21:24
timotimo: erlang drops the "get" part 21:25
erlang.org/doc/man/inet.html#sockname-1
timotimo well, yeah, they do. but do i want to mirror that in perl6? 21:30
with my current code you get a string and an int for the local and remote address each
sockname stuff is just a flat struct, so i could also ask the user to pass in a size and just return a Buf of the bytes and have rakudo do whatever it wants with it 21:31
mspo timotimo: just giving another data point. The c names are probably the best :) 21:32
mspo timotimo: why joearms decided to drop "get" in 1987 or whatever is probably lost to history 21:32
mspo timotimo: I just like to think erlang is a good model for networking stuff in general. I could also be wrong completely :) 21:32
their binary encoding stuff is top notch, at least 21:33
mspo erlang.org/doc/programming_examples...ml#id66841 21:35
tbrowder is there 21:52
ff!
timotimo what i have right now are .peer-host, .peer-port, .socket-host, and .socket-port 21:53
tbrowder hi, #perl6. is there any way to capture a "&?ROUTINE.name" into a p6 var for later use? 21:54
timotimo &?ROUTINE.name is just a string, you can store strings in perl6 vars, of course
tbrowder i've tried that and haven't had it work, probably a coding error. see if i can demo it as you said... 21:56
timotimo sure
tbrowder m: sub f{ my $s = &?ROUTINE.name; say $s;}; f() 21:57
camelia f
tbrowder ok, i think i was trying to put it inside "" or something like that. 21:58
timotimo: thanks! 21:59
timotimo "it"? 22:02
tbrowder i was trying to do this inside a sub but it doesn't work: say "inside sub '&?ROUTINE.name'" 22:06
geekosaur I dont think & interpolates directly 22:07
m: say "inside sub '&?ROUTINE.name'"
camelia inside sub '&?ROUTINE.name'
geekosaur m: say "inside sub '{&?ROUTINE.name}'"
camelia 5===SORRY!5=== Error while compiling <tmp>
Undeclared name:
?ROUTINE used at line 1. Did you mean 'Routine'?
tbrowder but this does work: say "inside sub '{&?ROUTINE.name}'"
well, you actually have to be in a sub... 22:08
geekosaur yeh
tbrowder it doesn't stringify like, say, $*PROGRAM 22:09
m: say "in program $*PROGRAM"
camelia in program <tmp>
geekosaur yes, as I said, & sigil doesn't interpolate. more, it's a method call, so if it did interpolate & sigils then you'd get '<whatever &?ROUTINE stingifies to>.name' 22:10
tbrowder m: f{say "in sub '&?ROUTINE.name'"};f() 22:11
camelia 5===SORRY!5=== Error while compiling <tmp>
Undeclared routine:
f used at line 1
ugexe space bar is hot lava 22:12
tbrowder m: sub f{say "in sub '&?ROUTINE.name'"};f()
camelia in sub '&?ROUTINE.name'
geekosaur m: say "hi $*DISTRO.name"
camelia hi opensuse.name
»
geekosaur interpolation iirc only handles $ sigil with optional postcircumfix; not method calls or etc. which is why {} interpolation exists 22:14
recognizing the end of general expressions in a string is hard
tbrowder m: sub f{say "in sub '{&?ROUTINE.name}'"};f()
camelia in sub 'f'
tbrowder \o/ 22:15
MasterDuke m: say "inside $*PROGRAM.uc()" 22:16
camelia inside /TMP/EVALBOT-FILE-YVPTBWB1D6
MasterDuke m: say "inside $*PROGRAM.uc"
camelia inside <tmp>.uc
MasterDuke it interpolates method calls, but you have to end them with '()'
geekosaur yeh, that postcircumfix thing I mentioned I guess 22:17
MasterDuke ah, wasn't quite sure what you meant by that 22:18
tbrowder well, inside a script $*PROGRAM stringifies without the .uc or () 22:23
geekosaur I am still not sure what you are trying to get at, aisde from "must interpolate every possible sigil. sorry you intended that as literal text; interpolatin is clearly more important" 22:24
tbrowder m: say "inside program '{$*PROGRAM}'" 22:26
camelia inside program '<tmp>'
Geth Inline-Perl5: b26e3c4111 | (Stefan Seifert)++ | lib/Inline/Perl5.pm6
Speed up p5_to_p6 by avoiding smart match

when $some_enum_value calls Numeric.ACCEPTS which is very heavy weight when all we want is a numeric equality test which should be dirt cheap.
Geth Inline-Perl5: 83e2149596 | (Stefan Seifert)++ | lib/Inline/Perl5.pm6
Speed up p6_to_p5(Str) by only looking up the UTF-8 encoder once

Str.encode looks up the encoder in the encoder registry and handing of the actual encoding to that. As we always want UTF-8 encoding anyway we can fetch this Encoder object once and use it directly.		 22:43
jnthn Yay, somebody takes advantage of the encoder API :) 22:44
(I/O handles do exactly that also)
nine Ok, with the latest Inline::Perl5 commits, I'm now at 9.881s for csv-ip5xs (best of 5 runs, 100000 iterations). Used to be 13.042s earlier today and 17.483s last week.- 22:45
timotimo that is really good 22:56
mspo encoder api? 23:15
japhb mspo: The MoarVM/NQP/Rakudo stack has a pluggable string encoder/buffer decoder API, allowing you to orchestrate everything at the Perl 6 level, instead of having to write C code to alter encoding rules. 23:18
mspo docs? 23:24
jnthn mspo: These are quite recent additions, so the docs aren't all there yet; the Encoding role is: docs.perl6.org/type/Encoding 23:37

Powered by Cro and the Raku® Programming Language.

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

Thank you!