»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.
00:10 canopus_ is now known as canopus 00:11 pierre_ joined 00:28 ifdog joined 00:29 rindolf left 00:32 ifdog left
Xliff Does anyone know if cglobal still needs the full library name as stated here: docs.perl6.org/language/nativecall..._variables 00:34
00:34 wisti left
Xliff Nope! Don't look like it. 00:37
00:45 AlexDaniel left
Xliff Also what's the best way to handle FILE* passing using NativeCall? 00:46
timotimo just use Pointer without anything specific, i guess
Xliff Hrm. 00:48
timotimo they aren't really pointers, but when you use Pointer you get the right size for pointers on your platform, right?
leont In C89, not all pointers need to be equally sized AFAIK 00:50
Xliff Yeah.
timotimo oh jesus christ ...
what platforms do we support that have that? 00:51
Xliff Hmm... lemme look up IO::Handle, again.
leont I don't think any reasonable platforms have differently sized pointers, but it is legal and has existed
geekosaur timotimo, none I suspect. variable pointer sizes were a thing on old mainframes 00:52
timotimo OK
leont Old DOS also had a long/short pointer concept
geekosaur like on PDP10s a (char) pointer could index part of a machine word, since it packed from 4 to 6 characters depending on representation into a 36-bit word
leont but that wasn't type dependent like some platforms had them -_-
timotimo oh yeah, lptr, i remember hearing of those 00:53
geekosaur also there used to be tagged architectures, where the tag size depended on the type 00:54
Xliff OK. So --> sub fdopen(int32 $fd, Str $mode) is native returns Pointer { * }; 00:55
That way I can use $iohandle.native-descriptor()
The only trick is guessing $mode. 00:56
leont That problem isn't perl related :-p
01:01 ocbtec left
Xliff leont: Actually, it is. 01:01
IO::Handle knows the mode, so why can't it be retrieved from the object?
01:02 girafe left, cdg left
Xliff Nothing in the docs. 01:02
01:03 setty1 left
Xliff I mean it really isn't the problem as long as the person opening the IO::Handle is the person converting it to a pointer, but that's not always the case for modules. 01:03
leont Ah like that
01:04 setty1 joined 01:05 skink joined
skink viki, @_@ 01:07
For the thing earlier, say $buf works after doing Buf.new($buf[..]), but only if you previously said $buf
01:11 aborazmeh joined, aborazmeh left, aborazmeh joined
geekosaur Xliff, I don't think IO::Handle knows the mode /per se/. the underlying libuv descriptor probably (but not certainly) does, but may not make it available or moarvm may not provide a way to query it 01:16
and not all of them are opened in a way where you can know for sure (if you inherit a handle as stdin/stdout/stderr, it is not necessarily open for read/write/write respectively: ttys/ptys, network sockets, and on BSDs pipes can be open read+write) 01:17
Xliff geekosaur: Hence my comment. 01:18
So I have to hold my nose and use "w" for my calls to fdopen() 01:19
But you are correct, my assertion that IO::Handle always knew the mode was incorrect. 01:20
Well... first .h file conversion to module complete. On to the actual classes.
First step in a looong journey.
timotimo have you had a look at gptrixie yet? 01:22
because that'll turn a .h file into a nativecall-using module for you
Xliff Yes, I am very familiar with gprixie. I used it for Audio::OggVorbis. 01:24
It has its flaws.
timotimo good to know
Xliff Particularly, it has a maximum limit before it will crash. 01:25
timotimo oh, wow
Xliff It will *not* handle libxml2
Nor, by extension, libxslt
timotimo i wonder what exact part causes the crash? 01:26
skink Speaking of nativecall stuff
I am heavily limited as to what methods I can put on a repr('CPointer'), yeah?
Xliff timotimo: Don't know. I haven't had the patience to complete my trace.
skink: You can have methods on a repr('CPointer') as long as you don't try to use self. 01:27
timotimo skink: shouldn't be, no
you can totally use self
Xliff Hmmm...
That's changed since my XML::LibXML days.
timotimo it could be i'm full of crap
Xliff (ie... the summer)
timotimo but it shouldn't be a problem
skink I can use self yeah, for self.free() { handle_free(self) } 01:28
Xliff skink: What are you trying to do?
skink But I can't have a method that asks for in this case a size_t
Xliff skink: Yeah, that's legit
timotimo "asks for"?
skink >Cannot look up attributes in a type object
Xliff Thaaat's the one
timotimo that sounds like you're calling a method on an undefined instance of something
Xliff skink: Can you pastebin some code? 01:29
timotimo actually, that error could totally get more output
skink sec
Xliff s/pastebin/fpaste/
skink gotta update this gist anyway
Xliff Or gist.
I need to stop using pastebin and start using more gist.
skink gist.github.com/skinkade/0a1a83adf...47b2224ed2 01:30
I copy-pasted that codeblock from the nativecall docs, apart from CCRNG.buffer(size_t $len) 01:31
01:31 wisti joined
Xliff skink: Yeah. repr('CPointer') can't have attributes. 01:32
skink que lastima
Guess I just need a buffer(ptr, len) then? 01:33
timotimo i just pushed a moarvm commit that also tells you what exact type it's trying to access
Xliff skink: Why do you think you need $!initialized? 01:34
skink Like I said, copy-paste :)
Xliff LOL
Fair enough.
skink Without buffer though, it works
timotimo i wonder why we allow attributes to be put onto an "is repr('CPointer')" class at all 01:35
01:35 leont left
Xliff Yeah, it's confusing. 01:36
skink So we can do hacky crap? :)
Xliff Could be useful as long as only the P6 side uses it though.
skink: You could have an inner class that handles the repr('CPointer') stuff. 01:37
Then a regular P6 class.
If you really want to do it the way I think you are trying to do it.
MasterDuke IO::Handle had a .mode, it returns the .mode of its $!path 01:38
skink I mean this is just a POC that I was seeing if I could make look nice.
MasterDuke *has
doesn't seem to be in the docs though
Xliff MasterDuke: Is that the mode it was opened with or the file permissions?
MasterDuke file permissions
Xliff HAH! Knew it. 01:39
Not the mode I'm looking for. Thanks, though.
MasterDuke tbrowder was also looking for the opened mode, Zoffix suggest how he could add it
01:40 mst joined, BinGOs left, BinGOs joined, Ulti joined
MasterDuke irclog.perlgeek.de/perl6/2016-11-02#i_13505117 01:41
01:42 pierre_ left
Xliff m: class A { class B {}; }; my $a = A.new 01:42
camelia ( no output )
Xliff skink: I've commented on your gist. See if that makes any sense. 01:43
01:44 mst left, mst joined, aborazmeh left 01:46 aborazmeh joined, aborazmeh left, aborazmeh joined, pierre_ joined
Xliff Aaand another. ;) 01:46
*sigh* 01:48
viki is Zoffix?
skink You cannot declare attribute '$!rng' here; maybe you'd like a class or a role?
for the last one
first one got the same 'you cannot have attributes' thing
Xliff That man has more nicks than random alias generator. 01:49
Maybe he is a random alias generator.
skink: o_O
skink: What language is rust written in? 01:53
That is not C
skink what :D
Xliff gcc barfed when I tried to compile the first part of your gist. 01:54
Do you know how hard compiler barfage is to get out of cotton?!?!?
]8-D
skink Oh, well this is good 01:55
I am now getting inconsistent sigsegvs :D
script'll work once or twice then die 01:56
Xliff headdesks
Rust is a language. Thanks skink!
:P
skink I thought you were joking :)
Xliff Nope. No alcohol, tonight.
Inconsistent SEGVs is something you will always have to worry about with NativeCall 01:57
But rakudo is no longer complaining, right?
About the only thing you can do is to turn on debugging information in your rust-build lib 01:58
Then use perl6-gdb-m
Just for shits and grins... Comment out $!rng.free
skink I reverted back to some earlier non-OO code 01:59
Xliff It's possible that DESTROY isn't being called when you think it should be.
And with that.... I must go *poof* 02:00
Good luck, skink!
02:03 cdg joined
japhb Xliff: People seem to assume a lot regarding $Zoffix.nicks. For one, that they are merely nicks and not distinct personas. Or that Zoffix is the root persona. Or that all $Zoffix.nicks are male-gendered (and that Zoffix itself is). I see no particular reason to assume any of these, unless Zoffix has specifically stated otherwise .... 02:04
seatek even if he/she/* stated otherwise, he/she/* might be lying 02:08
yoleaux 3 Nov 2016 14:01Z <gfldex> seatek: have you considered to include the following module into the tutorial? github.com/jnthn/grammar-debugger
geekosaur assumes they are nicks, and Zoffix is approximately human
seatek i see no reason to assume Zoffix is human, or even a mammal 02:09
japhb
.oO( On the Internet, why would anyone care if you're a dog? )
 02:10
seatek i would be unreasonbly swayed to the points any dog was trying to make 02:11
gfldex: i didn't even know about that grammar debugger myself! I'll need to play with it for a while before i do so i can sound like i know something 02:13
02:22 Su-Shee joined 02:25 Su-Shee_ left
Xliff I still want to learn grammars. 02:29
I just haven't had the time.
japhb: So the rule about Zoffix is "Do not assume you know Zoffix" 02:30
seatek Xliff: new tutorial for grammars in the docs now that takes you from knowing nothing about them rather than having to work it out for yourself
Xliff I got that, already. :)
seatek: docs.perl6.org/language/grammars ?? 02:31
seatek Xliff: would be curiouss if you found it helpful
Xliff Will take a look, then 02:32
seatek Xliff: docs.perl6.org/language/grammar_tutorial
skink japhb, Can we actually have a $Zoffix.nicks for one of the bots to list all his known aliases?
02:32 cdg left
seatek Xliff: if you stumble on something or it doesn't make sense somehow along the way... 02:32
Xliff seatek: I will drop you a line or something. 02:33
Or write a gist and pass that through yoleaux 02:34
skink ... is yoleaux pronounced yolo?
seatek Xliff: sweet thanks Xliff :) I really want it to work for people. I had a heck of time trying to understand them myself at first 02:36
Xliff skink: Yes. Using a bad french accent.
Well, I get the grammar creation part. It's the Actions that I don't get. 02:37
Which I guess we'll find out.
seatek :)
Xliff LOL 02:41
"But not really." 02:42
seatek hehe :)
Xliff Not a good sign in a dry piece of documentation, but makes for great reading.
seatek well, they're a little weird...
Xliff Good documentation should make for moderately decent reading. 02:43
And this is a tutorial, so less formal language is usually helpful.
seatek gfldex: love the ::Tracer! that would have come in handy had i known about it. have a feeling i'll need to get some inheritances built up to really effectively learn the debugger element 02:44
Xliff Hmmmm.....
seatek Xliff: thanks! yeah, that's what I was kinda going for... i wish i would have had that. i was struggling trying to figure out what was going on for a long time
Xliff seatek++: I now understand what "make" and "made" do after hearing about them for so long. 02:45
I still don't see the reason for it yet....
seatek Xliff: ha!! yeah. such a simple thing once it clicks eh?
keep going you will :)
Xliff seatek: Wow! That was a quick and easy read for a simple grammar, and it makes sense. 02:47
seatek Xliff: yay! :)
Xliff The next thing that would be helpful is a "step 2" or "Intermediate" tutorial.
seatek now do you know what actions are? :)
Xliff Go from parsing single lines to a multi-line object... ala a simple language parser. 02:48
Oh, I've always known they were a class, but I have the gist of them.
seatek multi-line wouldn't be any different -- newlines are just whitespace 02:49
Xliff At least now I may be able to get some idea as to how SLANG::SQL works.
Nice job, seatek!
Now I have to get ready for sleep. My eyelids are getting heavy. 02:50
seatek Xliff: thank you!
Xliff yw
seatek well, good dreams
02:50 Rawriful left
dalek c: de536c7 | seatek++ | doc/Language/grammar_tutorial.pod6:
mention Grammar::Debugger

and also format break a couple lines at 80 cols. Maybe 100 cols. Maybe 1 80 and the other 100.		 03:08
synopsebot6 Link: doc.perl6.org/language/grammar_tutorial
03:10 PerlJam joined 03:23 skink left 03:32 atweiden joined 03:33 djbkd joined 03:51 atweiden left, wisti left
samcv i'm trying to alter the value of just one value in a multi dimensional array. doing %hash{'key'} = thing => 'value'; removes the other things from the hash that are there 03:52
well that are in 'key'
geekosaur well, yes, that is exactly what you told it to do 03:54
%hash{'key
er
03:54 noganex_ joined
geekosaur %hash{'key'}{'thing'} = 'value'; 03:55
samcv that's what i thought, and i tried that. but i get an error, let me find it
geekosaur wonder if you need a . in there (which would be ugh)
samcv maybe it's because it's %!hash in a method, but i can do what i said above that removes values because it replaces the whole key fine 03:56
03:56 pierre_ left
samcv Type check failed in assignment to %!channel-event-data; expected Hash but got Array hmm 03:57
geekosaur m: my %h = (a => {b => 'c', d => 'e'}); %h<a><d> = 'f'; dd %h
camelia rakudo-moar 5b2bde: OUTPUT«Hash %h = {:a(${:b("c"), :d("f")})}␤»
03:57 noganex left
samcv but i'm doing now.Str 03:57
m: my %h = a => now.Str;
camelia ( no output )
samcv m: my %h = a => now.Str; say %h; 03:58
camelia rakudo-moar 5b2bde: OUTPUT«{a => Instant:1478231921.060494}␤»
geekosaur ...then why pair syntrax there
samcv %.channel-event-data{$e.nick}{'spoke'} = now.Str; this is what i have in my file
so maybe it's the now.Str getting something weird
m: now.Str.WHAT.say
camelia rakudo-moar 5b2bde: OUTPUT«(Str)␤»
samcv hm
geekosaur oh, I read that error backwards... 03:59
03:59 pierre_ joined
samcv ok i tested and %.channel-event-data{$e.nick}{'spoke'} = '1000'; also removes the other values in it 04:00
geekosaur oh... it's an attribute. that gets trickier
samcv not sure why
yeah maybe i should not do that
geekosaur it's calling an accessor, which may not be doing what you expect
04:08 cyphase left 04:12 cyphase joined
samcv hmm looks like it was getting a multidimensional array not a hash 04:21
what's the best way to convert from an array to a hash?
gfldex m: dd <a b c d>.Hash 04:22
camelia rakudo-moar 5b2bde: OUTPUT«Hash % = {:a("b"), :c("d")}␤»
gfldex depends on the structure of your array
04:22 wisti joined
gfldex m: dd do for <a b c d> { $^a => $^b } 04:23
camelia rakudo-moar 5b2bde: OUTPUT«(:a("b"), :c("d"))␤»
gfldex that's actually a list of Pair tho
samcv i'm trying to use json::tiny {name => [{spoke => 0} {join => 1477091584} {part => 1477090875}], here is one of the things in the array 04:26
when i print it
when i do .Hash it says typcheck failed. i specified my Hash %hash 04:27
gfldex can you gist the code in question? 04:28
samcv here you go gfldex vpaste.net/Esirp 04:30
%channel-event-data = from-json($!channel-event-fh.slurp-rest).Hash; this line is what's causing the typecheck error
gfldex i'm asking for a gist because of packages.debian.org/jessie/gist btw 04:31
also, a gist would show line numbers. Makes talking about the code easier. 04:33
from-json can return all kind of things. So just calling .Hash on it is dangerous in any case 04:37
samcv yeah there's not much info on the github page for it
gfldex ohh, my Hash %foo is a Hash of Hash and from-json will never return that 04:39
04:40 khw left
gfldex as I see it from-json can return Int, Str, List (or Array) and Hash 04:40
Xliff What does the JSON look like? 04:42
gfldex i would leave the data returned as it is and either handle it with a given/when block (if it's simple) or a recursive multi walker if it's complex
04:42 Sgeo_ joined 04:43 Tonik joined, Su-Shee_ joined 04:45 Sgeo left
gfldex samcv: also: gfldex.wordpress.com/2016/08/01/wa...the-flats/ and gfldex.wordpress.com/2015/12/20/i-...-them-now/ 04:46
04:46 Su-Shee left, Su-Shee_ is now known as Su-Shee 04:47 skids left 04:49 pierre_ left
samcv ah so leaving it as a list would be best... 04:54
that sounds sub optimal. well i created the json from a hash at least
but unfortunately tiny json isn't returned it as one i guess 04:55
04:55 pierre_ joined 04:58 cpage_ left 05:00 pierre_ left
gfldex m: sub type-dump(\any = Empty){ any ~~ Iterable ?? any».&type-dump !! any.WHAT }; say type-dump([a=>1, b=>2, {:c(3)}, 42 ]) 05:01
camelia rakudo-moar 5b2bde: OUTPUT«[(Pair) (Pair) {c => (Int)} (Int)]␤»
gfldex samcv: ^^^ simply type dumper that will walk a tree 05:02
Perl 6 is awesome :)
samcv it is
05:15 Su-Shee_ joined 05:18 Su-Shee left 05:32 Cabanossi left 05:33 wamba joined 05:35 Cabanossi joined 05:43 Tonik left 05:56 cpage_ joined 06:01 wisti left 06:11 Tonik joined 06:13 Su-Shee joined 06:16 Su-Shee_ left 06:18 Khisanth left 06:23 domidumont joined 06:25 ufobat joined 06:27 domidumont left 06:28 domidumont joined 06:29 canopus left 06:30 djbkd left, djbkd joined 06:32 Khisanth joined 06:33 perlawhirl left 06:35 djbkd left 06:36 domidumont left 06:38 canopus joined 06:39 aborazmeh left 06:41 andrzejku joined, cgfbee left 06:42 aborazmeh joined, aborazmeh left, aborazmeh joined 06:45 bjz joined 06:48 RabidGravy joined 06:49 pierre_ joined 06:50 cgfbee joined
japhb .tell tadzik I sent you a PR to add 256-color and 24-bit RGB color support to Terminal::ANSIColor. Enjoy. :-) 06:51
yoleaux japhb: I'll pass your message to tadzik.
06:56 domidumont joined 07:02 andrzejku left, bjz left, jameslenz left, andrzejku joined 07:04 sivoais left 07:05 sivoais joined 07:14 darutoko joined 07:16 bjz joined, neuraload joined
llfourn_ m: my $foo = "bar"; for |$foo { $_ = "baz" }; # is |$foo on a non-iterable meant to decont? 07:17
camelia rakudo-moar 5b2bde: OUTPUT«Cannot assign to an immutable value␤ in block <unit> at <tmp> line 1␤␤»
07:19 jameslenz joined
Xliff gfldex++: Awesome code! 07:21
07:21 bjz left 07:22 bjz joined 07:27 eliasr joined 07:28 RabidGravy left 07:29 perlawhirl joined 07:31 firstdayonthejob joined 07:35 perlawhirl left 07:42 firstdayonthejob left 07:46 wamba left 07:47 bjz left 07:48 TeamBlast left 07:49 sivoais left 07:50 bjz joined 07:51 sivoais joined 07:54 Su-Shee_ joined 07:55 Su-Shee left 07:58 Su-Shee_ left 07:59 Su-Shee joined 08:11 TeamBlast joined 08:13 bjz left 08:14 andrzejku left 08:18 bjz joined 08:23 noganex_ left, brrt joined 08:25 noganex joined 08:32 maybekoo2 joined
masak morning, #perl6 08:35
DrForr Mornin'. 08:36
08:37 zakharyas joined 08:42 Su-Shee_ joined 08:43 Su-Shee left, Su-Shee_ is now known as Su-Shee
lizmat good morning, #perl6! 08:43
masak ahoj lizmat 08:44
08:44 neuraload left
brrt good *, lizmat, DrForr, masak 08:44
08:45 bjz_ joined, bjz left
lizmat brrt: maybe it's time for a blog post on the JIT progress ? :-) 08:46
08:46 neuraload joined
brrt :-$ 08:46
yes... it probably is...
I'm overengineering my blog posts, usually
moritz brrt: it would also be a awesome if you wrote something about the JIT project for the perl 6 advent calendar 08:50
DrForr Well, I announced the Perl 6 Pocket Reference (not the actual name) at Cluj.pm - Going to get started this weekend.
moritz \o/ 08:51
brrt it's time for perl6 books, yes :-)
moritz DrForr++
brrt moritz: but *what*?
moritz has prepared three blog posts so far for his book, but will wait a bit with publishing them
brrt: whatever you find interesting. Or others could find interesting 08:52
brrt usually perl6 advent calendars are 'treats' of things that can be practically applied by readers
well... what i find interesting
moritz doesn't have to be this will
*way
brrt this morning I realized that there is an isomorphism between recursion and queueing
is isomorphism even the correct word?
moritz you could distill a progress report from the grant reports so far; or write about the tiler
brrt recursion is a subset of queueing?
moritz if it's a subset, it's not isomorphism :-) 08:53
brrt true
hmm
08:55 wamba joined
DrForr Amusingly one of my slides last night had people asking about transfinite arithmetic :) 08:55
08:56 Su-Shee left
brrt what is transfinite arithmetic.... is it what it sounds like? and how would that ever work 08:56
DrForr Arithmetic involving aleph-null and above :) 08:57
08:58 rindolf joined 09:01 wamba left 09:02 Actualeyes left 09:06 labster left
lizmat afk for a bit& 09:13
09:14 Actualeyes joined 09:17 seatek left 09:18 wamba joined, labster joined 09:23 rindolf left 09:26 KevinYang joined 09:29 KevinYang left 09:30 espadrine joined 09:31 araujo joined 09:39 pierre_ left 09:54 pierre_ joined
viki m: sub type-dump(\any = Empty){ any ~~ Iterable ?? any».&type-dump !! any.WHAT }; say type-dump([a=>1, b=>2, {:c(3)}, 42 ]) 10:04
camelia rakudo-moar 050cf7: OUTPUT«[(Pair) (Pair) {c => (Int)} (Int)]␤»
viki m: say deepmap {.WHAT}, [a=>1, b=>2, {:c(3)}, 42 ]
camelia rakudo-moar 050cf7: OUTPUT«[(Pair) (Pair) {c => (Int)} (Int)]␤»
viki Xliff: Rust is written in.. Rust :) 10:05
10:13 cpin joined 10:17 nadim joined 10:21 rindolf joined 10:24 ocbtec joined
viki japhb: interesting hypothesis :) What if there is no Zoffix persona and the different nicks represent the ever-changing and ever-new personas assumed by the persona-less Zoffix entitity? Either that, or Zoffix is just trolling everyone :} 10:24
.oO( the zoffix is a lie )
 10:25
10:28 leont joined 10:30 AlexDaniel joined, pmurias joined 10:33 djbkd joined 10:34 labster left 10:38 djbkd left 10:40 espadrine left 10:42 mephinet left, eliasr left, notbenh_ left 10:43 timeless left 10:44 chansen_ joined, avalenn left, damnlie_ joined, JimmyZ joined, damnlie left, notbenh_ joined, mephinet joined 10:45 avalenn joined, ilmari joined, AlexDaniel left, mrsolo joined 10:46 pierre_ left 10:47 clkao joined, timeless joined 10:48 Hotkeys joined 10:49 robinsmidsrod joined 10:52 benchable6 left, committable6 left 10:54 dylanwh left 10:55 dylanwh joined 10:57 ugexe joined 10:59 jnap_ joined
moritz the conitnuity (or lack thereof) or identity is an interesting research topic, IMHO 11:04
11:05 rodarmor left
brrt continuity of what? 11:05
11:05 cdg joined
moritz s/or/of/ 11:05
of identity
11:06 user9 left 11:07 user9 joined, rodarmor joined 11:11 perlawhirl joined 11:12 zakharyas left
perlawhirl hi perlers. 11:13
i have a question that's kinda hard to explain over irc, so i've put it in a gist: gist.github.com/0racle/2c29dc66413...a795767a08 11:14
I'm not sure if I can do it by returning an explicit Seq.new... 11:16
brrt I think the thing you are running into is that you want lazy-indexed behaviour rather than lazy-sequence behaviour 11:18
the AT-POS solution seems good to me
perlawhirl yeah, it's good for something simple, but means you can't really pass additional args to .thing(). 11:20
11:21 leont_ joined
perlawhirl my other though was just to return an anon class that implements AT-POS 11:21
might have a play with that idea
11:21 nadim_ joined 11:24 leont left 11:25 nadim left 11:29 mephinet left 11:30 mephinet joined 11:32 cgfbee left 11:33 TEttinger left 11:37 cdg_ joined 11:39 cdg left 12:09 perlawhirl left 12:15 pierre__ joined 12:19 zakharyas joined 12:21 cgfbee joined 12:23 yoleaux left, yoleaux joined, ChanServ sets mode: +v yoleaux 12:24 wamba left
dalek c: 8e6f20a | coke++ | doc/Language/grammar_tutorial.pod6:
remove trailing whitespace
 12:27
synopsebot6 Link: doc.perl6.org/language/grammar_tutorial
12:39 raiph left
pmurias node.js-- # v8-profiler is broken on node 7.0 and the pull request to fix it is unmerged :( 12:40
masak pmurias: oh, so it's not just me -- node.js seems to be a bit of a wild west of "move fast and break things" 12:43
DrForr AFAIK TauStation's node.js stack is still technically broken because there's a broken module in npm. 12:46
masak DrForr: does it prevent them from doing work? 12:47
perlpilot masak: that's been my experience as well.
moritz I hope they run a local npm mirror
masak this year I read an article whose point was "if you really care about stability, make a snapshot of node_modules"
I went looking for it the other week, but I couldn't find the article again :/ 12:48
it was in connection to the pad-left thing
moritz we have a nexus running, which acts as a caching proxy for npm modules
masak moritz: it's not just about them being available though, it's also about them being updated under your feet with breaking changes
moritz masak: can't you pin versions? 12:49
DrForr No, we just pin module versions using .. gulp?
masak .oO( from the department of "OMG dependencies are tricky" )
moritz: yes -- provided you do that for everything I guess it's the same
moritz masak: well, there's something to be said for rebuilding the environment on each build, just to make sure you still can
gfldex m: class C { method subscript(|c){ say "subscript of: ", c.perl } }; multi sub postcircumfix:<[ ]>(C:D \SELF, \pos){ SELF.subscript(pos) }; C.new.[1,42]; 12:50
camelia rakudo-moar b1c41b: OUTPUT«subscript of: \((1, 42))␤»
gfldex .tell perlawhirl objects can be made aware of being subscripted with class C { method subscript(|c){ say "subscript of: ", c.perl } }; multi sub postcircumfix:<[ ]>(C:D \SELF, \pos){ SELF.subscript(pos) }; C.new.[1,42]; 12:51
yoleaux gfldex: I'll pass your message to perlawhirl.
12:52 mscha joined
mscha m: say e**(π*i) + 1; 12:52
camelia rakudo-moar b1c41b: OUTPUT«0+1.22464679914735e-16i␤»
12:55 R_I_D joined
masak m: my %h = a => 1, b => 2; subset mh of Str where %h{$_} :exists; my mh $x = "b"; %h<b> :delete; say $x 12:56
camelia rakudo-moar b1c41b: OUTPUT«b␤»
masak ta-daa!
(not a but, just something to keep in mind)
bug*
R_I_D do modules need to be explicitly linked when perl6 is called? 12:58
masak no
R_I_D: the whole compilation stage is largely hidden and implicit -- you indicate a source file, and perl6 goes and figures it out from there 12:59
R_I_D okay. I have a User.pm6 file which delcares a user object, and a test-user.pm6 file which invokes a new() one.
the test file has User.pm6 at line one, but when I run perl6 test-user.pm6, it says cannot find User 13:00
timotimo '.' isn't in the search path by default 13:01
masak I tend to do `export PERL6LIB=lib`
(that's assuming that the User.pm6 file is in lib/)
R_I_D You put that in your .bashrc?
masak no, but that could totally be done
13:02 cdg_ left
masak I just re-run it when I have a fresh tab that I do perl6 development in :) 13:02
13:02 brrt left
DrForr I just do 'alias 6="perl6 -Ilib"' and put my libraries in lib/. 13:02
viki R_I_D: you can also include paths with -I switch
13:02 cdg joined
DrForr (which is pretty much perl6 standards.) 13:03
13:04 rindolf left
R_I_D ah! it's alive! thanks everyone 13:06
13:07 cdg left
masak no problem! bring it on! :D 13:07
13:08 rindolf joined
mscha m: my \ℕ = 1..∞; say 42 ∈ ℕ; 13:13
13:14 lichtkind joined
camelia rakudo-moar b1c41b: OUTPUT«(timeout)» 13:14
13:14 R_I_D left 13:20 wamba joined
viki m: my \ℕ = 1..∞; say 42 ~~ ℕ; 13:20
camelia rakudo-moar b1c41b: OUTPUT«True␤»
viki mscha: the ∈ is a set operator and converts args to sets, which causes the hang. The smartmatch against a range is the Right Way to do it.
13:23 bjz joined
viki m: multi infix:<∈> (\n, Range \r) { n ~~ r }; my \ℕ = 1..∞; say 42 ∈ ℕ; 13:23
camelia rakudo-moar b1c41b: OUTPUT«True␤»
13:23 cpage left 13:24 cpage joined
awwaiid though then 12.5 ∈ ℕ will still not work out very well :) . or will it... 13:24
13:24 bjz_ left
awwaiid m: my \ℕ = 1..∞; say 12.5 !~~ ℕ 13:24
camelia rakudo-moar b1c41b: OUTPUT«False␤»
awwaiid AMAZING
13:24 cpage_ left, cpage__ joined, cpage__ is now known as cpage_
viki :) 13:25
m: my \ℕ = 1..∞; say 12.5+1i ~~ ℕ
camelia rakudo-moar b1c41b: OUTPUT«True␤»
viki heh
m: multi infix:<∈> (\n, Range \r) { n ~~ UInt and n ~~ r }; my \ℕ = 1..∞; say 42+9i ∈ ℕ; 13:26
camelia rakudo-moar b1c41b: OUTPUT«False␤»
viki m: multi infix:<∈> (\n, Range \r) { n ~~ UInt and n ~~ r }; my \ℕ = 1..∞; say 42.5 ∈ ℕ;
camelia rakudo-moar b1c41b: OUTPUT«False␤»
viki m: multi infix:<∈> (\n, Range \r) { n ~~ UInt and n ~~ r }; my \ℕ = 1..∞; say 42 ∈ ℕ;
camelia rakudo-moar b1c41b: OUTPUT«True␤»
viki There we go. 13:27
awwaiid m: multi infix:<∈> (Int \n, Range \r) { n ~~ r }; my \ℕ = 1..∞; say 42 ∈ ℕ
camelia rakudo-moar b1c41b: OUTPUT«True␤»
awwaiid er
13:28 leont_ left
awwaiid m: multi infix:<∈> (Int \n, Range \r) { n ~~ r }; my \ℕ = 1..∞; say 42+9i ∈ ℕ 13:28
camelia rakudo-moar b1c41b: OUTPUT«Cannot resolve caller infix:<∈>(Complex, Range); none of these signatures match:␤ (Int \n, Range \r)␤ in block <unit> at <tmp> line 1␤␤»
viki m: my \ℕ = subset :: of UInt where 1..∞; say 45+5i ~~ ℕ
camelia rakudo-moar b1c41b: OUTPUT«False␤»
viki m: my \ℕ = subset :: of UInt where 1..∞; say 45 ~~ ℕ 13:29
camelia rakudo-moar b1c41b: OUTPUT«True␤»
viki :)
awwaiid nice
is the 1..∞ adding any value over UInt there? 13:30
viki Oh, I guess not :D
awwaiid++
awwaiid maybe zero?
viki Oh, right
13:31 wamba left 13:37 bjz left 13:39 user9 left 13:42 JimmyZ left, JimmyZ joined 13:46 neuraload left 13:48 skids joined 13:50 nicq20 joined
nicq20 Hello \o 13:51
pmurias masak: yes, node.js seems to have a very wild west approach in general 13:52
DrForr o/
moritz pmurias: would you be willing to write a blog post on nqp-js/rakudo-js for the perl 6 advent calendar? 13:53
perlpilot moritz++ that would be excellent IMHO 13:58
14:00 aborazmeh left 14:03 wisti joined
awwaiid what... zoffix became viki now? You just like to reset my context all the time! I need state damnit to shortcut assumptions -- purely functional conversations are not efficient! 14:04
viki tehee 14:06
awwaiid I would have taken yesterday's conversation about disabling flicker failing tests quite differently, for example. On the one hand I would have given much more weight to someone I recognized as an active contributor and non-stranger, on the other hand I would have felt much more comfortable being direct about my observatons and experiences. Lame I tell you. LAME. But that's fine. Break my world. BREAK IT GOO 14:08
D
awwaiid switches over to paying attention at $day-job
mst awwaiid: when I'm not sure I /whois since he's still @perl6.party 14:09
though at this point I've got a fairly good mental index of which nicks turn out to be tie()ed to Zoffix
awwaiid makes sense. My irc skillz and instincts could be improved :)
mst I've been an op for over ten years, my alt spotting spidey senses have been built up by a decade of idiocy
awwaiid hehe
viki :D 14:10
mst in this case, they're being used for something more pleasant
but it's still the same basic set of instintcs
14:10 khw joined
perlpilot mst: don't give him hints about what to change though. Next thing you know we'll be talking 5 different nick that are all Zoffix and we won't be able to tell. 14:10
:-) 14:11
14:11 ufobat left
awwaiid I want to see those 5 talk to themselves in a constructive way. Imagine the collaboration potential! 14:11
I of course expect 5x the PRs
moritz you want a DDoS on github? 14:12
nicq20 How do I know that everyone here isn't Zoffix already?
awwaiid indeed. How do any of us know.
moritz nicq20: maybe you are Zoffix!
perlpilot Any nick I don't recognize, I immediately assume is Zoffix. It's worked well so far :)
awwaiid haha
mst perlpilot: many years ago when we were both younger and stupider I played hide and seek with zoffix over a channel ban; I'm reasonably confident I can spot him even if he's trying to stay cloaked ;) 14:13
14:14 kurahaupo__ joined
mst new nick + similar speech patterns + similar areas of knowledge is generally quite sufficient 14:14
moritz
.oO( cogney accept helps! )
 14:15
nicq20 mst: Plus the UL may help
*URL
mst I meant even if he'd changed the hostmask
perigrin moritz: cockney? 14:22
14:22 kurahaupo__ is now known as kurahaupo 14:24 zacts left, pierre__ left 14:27 pierre__ joined, mscha left
Xliff mst: So you are "on" to The Zoffix Collective? 14:38
Heck! We might as well just cave and call it what it really is. 14:39
mst hah
Xliff Resisting The Zoffix is futile.
nicq20 Xliff: Hivemind?
arnsholt had actually forgotten that Zoffix is the base name
Xliff nicq20: Close. Up the geek index by 20.
arnsholt At this point it's just "that guy"
Xliff arnsholt: Look, if one of his nicks is "locutus", I'm gone. 14:40
14:42 yoleaux left 14:44 yoleaux joined, ChanServ sets mode: +v yoleaux, pierre__ left 14:45 pierre__ joined
[Coke] loves the code archeology of knowing when someone did a cut and paste because of the shared typos. 14:45
14:45 user9 joined
japhb Xliff: Patrick Stewart is a 6er? I'm all for it! 14:53
Although at this point I've seen the "mild shock" meme template so many times it's about the second thing I think of when I think of him now. 14:54
14:57 sufrostico left
Xliff japhb: To reduce such an illustrious career to a simple image.... sad.... 14:59
15:00 Lucas_One left 15:01 Lucas_One joined, huajie-hu joined
masak [Coke]: I once saw phylogenetic techniques being applied to a bunch of chain emails. you could basically topo-order them by looking at where typos and changes were introduced. 15:06
ilmari is it solipsistic in here or is it just zoffix? 15:07
yoleaux 3 Nov 2016 12:23Z <tbrowder> ilmari: please check my WIP module at github.com/tbrowder/Misc-Utils-Perl6 for proposed module names shown in the table in the README. comments, please.
15:08 huajie-hu left 15:09 kyclark joined, cdg joined
viki
.oO( zofflipsism )
 15:09
The theory that only Zoffix exists :P 15:10
15:10 cdg left
ilmari and we are all a figment of their imagination 15:10
15:10 cdg joined
ilmari s/a figment/figments/ 15:10
i english gud
viki Zoffix cogitet, ergo sum 15:11
15:13 Rawriful joined 15:17 huajie-hu joined
huajie-hu zzz 15:18
viki xxx 15:19
nicq20 yyy? 15:20
viki 🦋🦋🦋
pmurias moritz: I'll think about it, if I get rakudo.js to a (semi-)presentable state I'll try to do that
nicq20 Oops, ccc
15:21 brillenfux joined
moritz pmurias: ok, thanks 15:24
15:24 brillenfux left
moritz pmurias: IMHO it would be worth it writing about it even if it's not semi-presentable, just to get more publicity for the project 15:24
15:25 huajie-hu left 15:29 huajie-hu joined 15:30 zacts joined
kyclark How can I turn off any output buffering? In Perl 5, it’s $| =1. 15:31
15:31 ufobat joined
[Coke] multis or multies ? (I prefer the former) 15:31
viki [Coke]: the latter. And it's the correct English. 15:32
[Coke] docs.perl6.org/language/5to6-perlvar - "currently autoflush is not implemented in Perl 6"
15:32 huajie-hu left
kyclark Ah, thanks. 15:32
[Coke] viki: except that "multi" is a "word" in Perl 6 lingo.
but ok, multies is fine. one less thing to change here. Danke. 15:33
viki [Coke]:oh, i confused the rule for -y ending nouns. 15:34
kyclark: I'm not seeing any buffering, at least when printing to STDOUT 15:36
kyclark I’m working on a little game that refreshes the entire screen. It seems laggy.
I’ll post in a minute
bouncy balls!
15:38 wamba joined
[Coke] ponders adding more code words from perl 6's src file to the doc spell checker to a) future proof it, and b) allow us to spell check the src comments. Probably just not worth it. 15:39
s/src file/src dir/
kyclark OK, here’s a bouncy ball game: pastie.org/10955904 15:50
It seems shaky because of output buffering (I think). Overall, it’s working well, but I’d love feedback.
[Coke] kyclark++ 15:51
kyclark I wanted to write a Pong, but I have no idea how to capture keystrokes
viki You need to turn off shell buffering and then capture via $*IN 15:52
15:52 vike joined
viki recalls doign something like that but totally forget the codes 15:52
moritz $*IN.getc?
or does that still require an enter/return? 15:53
[Coke] getc is still buffered, someone just wrote that up for the docs.
viki moritz: yeah, it's $*IN.getc, but you still have to tell the shell to not buffer 15:54
kyclark: your game works fine on my screen.
kyclark: by "lag", do you mean the jerky motion of the ball? That's just 'cause you're moving it by character and not by pixel
moritz viki: some nasty stty invocation, maybe? 15:55
kyclark Yes, I guess that is what I mean. Just jumpy. Of course my kids think it looks really primitive, but they gave me a thumbs-up just the same.
I took inspirado from the Perl 6 Game of Life and Forest Fire games.
[Coke] docs.perl6.org/routine/getc#Buffering_terminals 15:56
kyclark Can anyone advise on how to turn off the shell buffering to capture single keystrokes? It was a question I posed the other day, too.
[Coke] section right after that talks about the issue with combiners.
kyclark Is it a call to “stty” to unbuffer? 15:57
viki fiwiw, there's a program called unbuffer 15:58
sudo apt-get install expect-dev
tbrowder morning #perl6 15:59
viki noon, tbrowder
mspo that's only for pipes
tbrowder good noon, viki
viki oh, OK, they I may have used something elese 16:00
mspo set stty_init "-opost" 16:01
it's a sh + tcl script you can probably borrow from
viki Didn't work
tbrowder i would like to have a hash inside a sub that is static for the life of the program, but i don't think that's currently possible. i can have a file scoped hash, but is there a better way to minimize its exposure? 16:02
viki tbrowder: state
m: sub foo { state $x = 42; dd $x++ }; foo; foo; foo; foo ;
camelia rakudo-moar 2ef2fd: OUTPUT«42␤43␤44␤45␤»
tbrowder ah, thanks, i was looking for static...great! 16:03
mst m: { my %x := (foo => 'bar'); sub foo ($x) { %x{$x} } } dd foo('foo')
camelia rakudo-moar 2ef2fd: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Strange text after block (missing semicolon or comma?)␤at <tmp>:1␤------> 3foo => 'bar'); sub foo ($x) { %x{$x} } }7⏏5 dd foo('foo')␤ expecting any of:␤ infix␤ infix stopper…»
viki Hm... stdbuf -i0 -o0 -e0 perl6 -e 'my $c = $*IN.getc; dd $c' also doesn't work for some reason :/
mst m: { my %x := (foo => 'bar'); sub foo ($x) { %x{$x} } }; dd foo('foo')
camelia rakudo-moar 2ef2fd: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Undeclared routine:␤ foo used at line 1␤␤»
viki mst: subs are lexical
by default
mst oh for crying out loud
yes
I realised that as soon as I saw the error 16:04
viki :)
16:04 cdg left 16:05 cdg joined
nicq20 Is it possible to grab the name of a token alternation? Basically get 'bar' from `token foo:sym<bar> { ... }` 16:06
viki nicq20: <sym>
Well, <sym> would match 'bar' 16:08
m: say grammar { proto token foo {*}; token foo:sym<bar> { .+? <sym> .+ } }.parse: "meow bar moo", :rule<foo>
camelia rakudo-moar 2ef2fd: OUTPUT«｢meow bar moo｣␤ sym => ｢bar｣␤»
nicq20 Hmm. I 'm probably just doing something wrong then. :/ 16:09
viki: Is it possible to use that outside the grammar? Like in an actions class for it? 16:10
16:10 skids left
viki nicq20: "that" being what? 16:10
nicq20: in actions you'd declare multi method foo:sym<bar> ($/) {} that would be called when sym is <bar> 16:11
16:12 Sound joined
viki m: say grammar { proto token foo {*}; token foo:sym<bar> { { make first *, &?ROUTINE.name.comb: /"<"<(.+)>">"/ } } }.subparse("meow bar moo", :rule<foo>).made 16:13
camelia rakudo-moar 2ef2fd: OUTPUT«bar␤»
viki ahahaha :)
16:13 cdg left 16:14 cdg joined
nicq20 viki: What I'm trying to accomplish is grab the "name" of the rule/token from inside an actions class. If I need to make a method for each alternation, that's fine. I'm just being lazy really. 16:14
viki nicq20: what's in $<sym> ? 16:15
nicq20 viki: From within TOP
timotimo right, if you match <sym>, you should also have a $<sym> 16:16
viki nicq20: I kinda figured the :sym stuff was precicely so you could have multiple separate action methods. So I don't think I udnerstand your laziness approach. If you aren't using multi action methods, why use :sym at all and not an alternation? 16:17
nicq20 From in TOP, I was doing `method TOP($/) { make: $<thing><sym>.Str }` but that does not return anything. 16:18
viki nicq20: did you match for <sym>?
nicq20 viki: I am using multiple tokens of the same base-name, but I wanted to be "lazy" about it because there are only three alternations. 16:19
viki nicq20: so why are there three tokens instead of one alternation with three variants? 16:20
m: say grammar { rule TOP { $<a>=\d+ $<sign>=<[+-]> $<b>=\d+ }; }.parse("4 + 4", :actions(class {method TOP($/) { make $<sign> eq "+" ?? $<a> + $<b> !! $<a> - $<b> }})).made 16:21
camelia rakudo-moar 2ef2fd: OUTPUT«8␤»
viki m: say grammar { rule TOP { $<a>=\d+ $<sign>=<[+-]> $<b>=\d+ }; }.parse("4 - 2", :actions(class {method TOP($/) { make $<sign> eq "+" ?? $<a> + $<b> !! $<a> - $<b> }})).made
camelia rakudo-moar 2ef2fd: OUTPUT«2␤»
nicq20 viki: Sorry, bad with terminology. :/ Let my try and paste the code somewhere.
16:23 maybekoo2 left
nicq20 viki: pastie.org/10955917 16:24
viki nicq20: just use make ~$/
nicq20: oh, wait 16:25
16:25 pmurias left
viki nicq20: you want the command to be one of "status"/"check"/"add"? 16:25
nicq20 viki: Right.
viki nicq20: I don't see that information preserved anywhere when I dump the match object. 16:27
In fact, your method command isn't even called 16:28
nicq20 viki: I can make a new method for each one of the `command:sym<stuff>` tokens that returns 'status'/'check'/'add' appropriately. I wanted to know if there was a better way to do it first.
viki m: say grammar { rule TOP { $<a>=\d+ <sign> $<b>=\d+ }; proto token sign {*}; token sign:sym<plus> { "+" }; token sign:sym<minus> { "-" } }.parse("4 + 4", :actions(class {method command ($/) { dd [ "Here", $/ ] }})).made 16:29
camelia rakudo-moar 2ef2fd: OUTPUT«(Any)␤»
16:29 cosimo joined
viki Oh, duh, I'm using <sign> :P 16:29
m: say grammar { rule TOP { $<a>=\d+ <sign> $<b>=\d+ }; proto token sign {*}; token sign:sym<plus> { "+" }; token sign:sym<minus> { "-" } }.parse("4 + 4", :actions(class {proto method sign ($/) { dd [ "Here", $/ ] }})).made 16:30
camelia rakudo-moar 2ef2fd: OUTPUT«(Any)␤»
viki Nope, still ain't called
nicq20 I can't find anything in the docs about if it is supposed to be or not. 16:31
16:31 araujo left
timotimo it wouldn't surprise me if it actually only calls the one with the "correct" "full" name 16:32
i.e. including the :sym<...> part
viki Yeah, I'd expect it to work that way
16:32 araujo joined
nicq20 Makes sense. Only call the proto if none of the others match. 16:33
timotimo actually, i think the :sym mechanism is only a thing inside grammars
or is that wrong?
viki m: say grammar { rule TOP { :my $*SYM; $<a>=\d+ <sign> $<b>=\d+ }; proto token sign {*}; token sign:sym<plus> { { $*SYM = "plus" } "+" }; token sign:sym<minus> { { $*SYM = "minus" } "-" } }.parse("4 + 4", :actions(class {method TOP($/){ make "Sym is $*SYM" }})).made
camelia rakudo-moar 2ef2fd: OUTPUT«Sym is plus␤»
viki nicq20: another way ^ Though IMO adding each :sym methods in actions is cleaner 16:34
nicq20 Oh, well. I'll just do it manually. :)
viki m: class { method you:are<wrong> { say "yep" } }."you:are<wrong>"() 16:35
camelia rakudo-moar 2ef2fd: OUTPUT«yep␤»
timotimo viki: you misunderstood me :) 16:36
m: class { method you:are<wrong> { say "yep" } }."you"()
camelia rakudo-moar 2ef2fd: OUTPUT«No such method 'you' for invocant of type '<anon|70393936>'␤ in block <unit> at <tmp> line 1␤␤»
timotimo ^- that's what i meant by that
not the syntax, but the mechanism that'll call those symmed variants for you
16:37 domidumont left
viki In actions? It doesn't call the one without the full name. 16:37
nicq20 I originally got the idea that it would work from this: docs.perl6.org/language/grammar_tutorial . At the end of it, the `<sym>` bit is used in the actions class. 16:39
viki nicq20: that works because there's a <sym> token being matched in the grammer 16:40
Whereas yours doesn't.
And <sym> is the value of the :sym<...> bit, so that's how you end up being able to retrieve that information.
token foo:sym<whatever> { <sym> } matches text "whatever" 16:41
And captures it into the $<sym> capture
nicq20 Oh, ok. That makes sense.
16:43 MilkmanDan left, MilkmanDan joined 16:44 [particle] joined
viki m: class Foo { proto method speak {*}; method speak:cat<meow> { say "yep" } }.speak 16:44
camelia rakudo-moar 2ef2fd: OUTPUT«Use of uninitialized value of type Any in string context.␤Methods .^name, .perl, .gist, or .say can be used to stringify it to something meaningful.␤ in block <unit> at <tmp> line 1␤Cannot resolve caller speak(Foo); none of these signatures match:…»
viki tehe 16:45
16:46 nicq20 left 16:47 setty1 left, setty1 joined 16:49 bisectable6 left, evalable6 left 16:50 evalable6 joined, bisectable6 joined, ChanServ sets mode: +v bisectable6, kyclark left, seatek joined 16:51 [particle] left 16:53 [particle] joined 16:55 Tonik left
dalek : 6dcf602 | (Zoffix Znet)++ | misc/perl6advent-2016/schedule:
Book Dec 24.
 16:57
17:00 silug left 17:01 zakharyas left 17:05 Sound left, hankache joined
hankache yello * 17:07
viki lo lo
hankache hiya viki
japhb 255,255,127
17:09 cdg left 17:10 mattp_ left 17:13 mattp_ joined 17:17 silug joined 17:21 Tonik joined
viki TIL a magpie is not a pie :o 17:21
.oO( a magpie pie )
 17:23
17:25 xinming left
timotimo magpies are the kind of bird who'd say "pickle pea! pump-a-rum!" 17:26
17:27 pierre__ left, pierre_ joined 17:32 pierre_ left
huf what 17:32
timotimo www.youtube.com/watch?v=7MlHmCpnJIU - watch out, memes 17:33
huf wow. what is this wonderful thing please 17:35
timotimo darksouls3.wiki.fextralife.com/Pick...a-Rum+Crow 17:36
viki huuu www.wired.com/2016/11/googles-chro...rity-model 17:37
.seen pmuchad
17:37 Rawriful left
yoleaux I haven't seen pmuchad around. 17:37
viki .seen pmichaud 17:38
yoleaux I saw pmichaud 15 Sep 2016 01:34Z in #perl6: <pmichaud> m: my $input = '(\d\d\d)'; my $m = 'a 123' ~~ /$0=<$input>/; say $m
viki We need to stort out HTTPS on rakudo.org
"Starting in January, Chrome will flip the web’s security model: Instead of warning users only about HTTPS-encrypted sites with faulty or misconfi­gured encryption, as Chrome currently does, it will instead flag as “not secure” any unencrypted sites that accept a username and password or a credit card." 17:39
timotimo well, we don't ask for username/password on rakudo.org 17:41
viki We actually do.
"Meta" column, bottom right., 17:42
17:42 nowan left
timotimo we know the risks when we log in there 17:42
DrForr letsencrypt is decent to set up.
viki What does us knowing anything have to do with Chrome actively listing our site as unsecure? 17:43
17:45 nowan joined
DrForr I woukd at one time hace been more circunspect, but now I see it as simple scsre tactics. 17:46
viki Because no one MIMed your traffic with your knowledge, not because of how easy it is to do :) 17:48
18:02 ale1ster joined 18:14 labster joined
viki Challenge for the day :) www.wechall.net/challenge/training.../index.php 18:22
A tiny 1-liner in Perl 6 :)
18:22 khw left 18:27 Rawriful joined 18:29 Rawriful left 18:31 canopus left 18:35 nicq20 joined, huf left, wamba left, huf joined
nicq20 Hello o/ 18:35
18:39 canopus joined 18:40 grondilu joined
viki \o 18:43
18:43 xinming joined, firstdayonthejob joined 18:47 domidumont joined
viki wow 18:48
m: my @l = "A".."Z"; dd @l[(1, 2, (10, 11), 3), (3, 4, 5)]
camelia rakudo-moar 2ef2fd: OUTPUT«(("B", "C", ("K", "L"), "D"), ("D", "E", "F"))␤»
viki TIL you can shape results using indexes like that
timotimo viki: i'd think it'd only warn as soon as it asks for username/password 18:49
viki: and no normal user will try to log in without having credentials
18:50 cdg joined 18:56 ocbtec left
timotimo we're still bursting a whole bunch of cpu usage every few minutes to update testers.perl6.org 18:59
nicq20 Huh, didn't know that was a thing. 19:01
timotimo i'd love someone to write up a tiny thing that only does the hard work when there has been new stuff since the last update 19:02
also, the reports seem kind of b0rked? all show up as N/A 19:04
viki m: say "meow" ~~ 0 19:09
camelia rakudo-moar 2ef2fd: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '3⏏5meow' (indicated by ⏏)␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
harmil_wk m: class Foo is Hash { has $.blah = 0 }; say Foo.new.blah.perl
camelia rakudo-moar 2ef2fd: OUTPUT«Any␤»
harmil_wk I have a feeling I'm doing something I should know better than to do...
19:09 firstdayonthejob left
viki Was that meant to explode? I kinda would want it to return False. 19:09
[Coke] viki: seems like that's missing a try somewhere, yah. 19:10
viki harmil_wk: Hash has .perl method
[Coke] m: say pi ~~ 0
camelia rakudo-moar 2ef2fd: OUTPUT«False␤»
[Coke] m: say "pi" ~~ 0
camelia rakudo-moar 2ef2fd: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '3⏏5pi' (indicated by ⏏)␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
harmil_wk viki: I'm not sure I understand...
viki s: Hash.new, 'perl', \()
SourceBaby viki, Sauce is at github.com/rakudo/rakudo/blob/2ef2...sh.pm#L206
viki harmil_wk: oh, wait, neverf mind :} 19:11
harmil_wk I wasn't calling .perl on hash, though
viki Yeah, I've just realized that
s: Hash, 'new', \()
SourceBaby viki, Sauce is at github.com/rakudo/rakudo/blob/2ef2...Map.pm#L12
viki Ah, there we go :)
s: Mu, 'new', \() 19:13
SourceBaby viki, Something's wrong: ␤ERR: Cannot resolve caller sourcery(Mu, Str, Capture); none of these signatures match:␤ ($thing, Str:D $method, Capture $c)␤ ($thing, Str:D $method)␤ (&code)␤ (&code, Capture $c)␤ in block <unit> at -e line 6␤␤
19:14 bstamour joined
harmil_wk I'm sorry, but it has to be said: you can't Mu new review... 19:14
:-)
19:14 firstdayonthejob joined
viki m: use nqp; class Foo { has $.blah = 0; method new { self.bless: :42blaz } }.new.blah.say 19:15
camelia rakudo-moar 2ef2fd: OUTPUT«0␤»
viki m: use nqp; class Foo { has $.blah = 0; method new { self.bless: :42blah } }.new.blah.say
camelia rakudo-moar 2ef2fd: OUTPUT«42␤»
viki m: use nqp; class Foo { has $.blah = 0; method new { nqp::create(self) } }.new.blah.say
camelia rakudo-moar 2ef2fd: OUTPUT«(Any)␤»
harmil_wk oh really...
viki harmil_wk: what I'm seeing is nqp::create(self) doesn't set the defaults and that's what the Map's .new method is doin' 19:16
[Coke] all the default logic is in the bootstrap file in rakudo, so that makes sense.
harmil_wk Yeah, that seems highly counterintuitive. At the very least, a warning (probably error) should result if your defaults are going to be ignored.
viki We can't warn about every possible usage without slowing the runtime down to a screetching halt 19:17
m: use nqp; class Foo { has $.blah; submethod BUILD (:$!blah = 42) {} }.new(:72blah).blah.say
camelia rakudo-moar 2ef2fd: OUTPUT«72␤»
viki m: use nqp; class Foo { has $.blah; submethod BUILD (:$!blah = 42) {} }.new.blah.say 19:18
camelia rakudo-moar 2ef2fd: OUTPUT«42␤»
viki m: class Foo is Hash { has $.blah; submethod BUILD (:$!blah = 42) {} }.new.blah.perl.say
camelia rakudo-moar 2ef2fd: OUTPUT«Any␤»
viki mother... 19:19
19:19 sufrostico joined
lizmat resists the urge to answer 19:19
[Coke] is there rakudo only code that demonstrates the issue? or do you have to (as viki did above), "use nqp" and then an nqp opcode?
harmil_wk My original code that derived from Hash
m: class Foo is Hash { has $.blah = 0 }; say Foo.new.blah.perl
camelia rakudo-moar 2ef2fd: OUTPUT«Any␤»
19:19 wisti left
viki lizmat: what's the answer? 19:19
I mean, answer to what
lizmat to your cry for help: "mother..." :-) 19:20
viki m: class Foo is Hash { has $.blah; method new { self.bless: %_ } }.new.blah.perl.say
camelia rakudo-moar 2ef2fd: OUTPUT«Too many positionals passed; expected 1 argument but got 2␤ in method new at <tmp> line 1␤ in block <unit> at <tmp> line 1␤␤»
viki Ah :)
harmil_wk lizmat: heh
viki m: use nqp; class Foo { has $.blah; method new { self.bless: |:42blah, |%_} }.new(:72blah).blah.say 19:21
camelia rakudo-moar 2ef2fd: OUTPUT«72␤»
viki m: use nqp; class Foo { has $.blah; method new { self.bless: |:42blah, |%_} }.new.blah.say
camelia rakudo-moar 2ef2fd: OUTPUT«42␤»
viki There. I win. 19:22
harmil_wk My point is that "class Foo is ::($x) { has $.blah = 0 }" will only work if $x is a class that doesn't do what Hash does... that feels like a very difficult-to-detect bug from the user's side.
[Coke] Yup. I would say that should either work or, if it can't, throw an error asap that it can't work.
harmil_wk I'll rakudobug it for the record. 19:23
19:23 darutoko left
viki I forcee that will result in severe de-optimization. 19:23
And what about classes that don't even do the .new business? 19:24
IMO, if you're going to base a class off a gut class, it's your job to familiarize yourself with its workings.
m: class Foo is DateTime { has $.blah = 0 }.new.blah.say 19:25
camelia rakudo-moar 2ef2fd: OUTPUT«Cannot call Foo.new with no parameters␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in any at gen/moar/m-Metamodel.nqp line 3096␤ in block <unit> at <tmp> line 1␤␤»
viki m: class Foo is DateTime { has $.blah = 0 }.now.blah.say
camelia rakudo-moar 2ef2fd: OUTPUT«0␤»
19:25 wisti joined
viki m: class Foo is Set { has $.blah = 0 }.new-from-pairs("x").blah.say 19:26
camelia rakudo-moar 2ef2fd: OUTPUT«(Any)␤»
viki m: class Foo is Nil { has $.blah = 0 }.new.blah.say 19:27
camelia rakudo-moar 2ef2fd: OUTPUT«Nil␤»
viki m: class Foo is Array { has $.blah = 0 }.new.blah.say
camelia rakudo-moar 2ef2fd: OUTPUT«(Any)␤»
viki m: class Foo is Complex { has $.blah = 0 }.new.blah.say
camelia rakudo-moar 2ef2fd: OUTPUT«(Any)␤»
viki etc
Here's a good one: 19:31
m: class Foo is SetHash { has %.elems = :42meows }.new.elems.say
camelia rakudo-moar 2ef2fd: OUTPUT«{}␤»
viki m: class Foo { has %.elems = :42meows }.new.elems.say
camelia rakudo-moar 2ef2fd: OUTPUT«{meows => 42}␤»
viki So you need to know what you're doing when you're inheriting from core stuff.
samcv hmm from-json($json-string).WHAT.say => (Hash). but if i do: my Hash %hash = from-json($json-string), i get typecheck failed
viki Seems a flimsy idea too. 19:32
samcv Type check failed in assignment to %hash; expected Hash but got Array
viki samcv: is it a constant by any chance?
samcv which one? 19:33
harmil_wk Here's the problem I have: EVERYTHING is derived from something core. What happens when I use a module and derive from a class in that module that happens to derive from Hash or Set or whatever... Should I ever use defaults or are they just too uncertain to function?
viki samcv: constant %hash or my %hash?
samcv it's a my
19:34 sivoais left
viki star: use JSON::Fast; my %hash = from-json Q|{"foo": true}|; dd %hash 19:34
camelia star-m 2016.04: OUTPUT«Hash %hash = {:foo}␤»
ugexe you are typing your hash
samcv i'm sing json tiny
viki star: use JSON::Tiny; from-json Q|{"foo": true}|.WHAT.say;
camelia star-m 2016.04: OUTPUT«(Str)␤Input (4 characters) is not a valid JSON string␤ in sub from-json at sources/9B467EEF9267A777BB53BAA2F19BE2C9D756BEED (JSON::Tiny) line 59␤ in block <unit> at <tmp> line 1␤␤»
ugexe %hash is already a hash, Hash %hash is not the same thing
viki star: use JSON::Tiny; from-json(Q|{"foo": true}|).WHAT.say;
camelia star-m 2016.04: OUTPUT«(Hash)␤»
samcv ah ok ugexe
so i don't need to specify a type? 19:35
viki You can if you want to.
timotimo Hash %foo is "a hash where the elements are Hash"
ugexe depends on your requirements, but it *sounds* like you don't need to
viki my Int %hash = :42foo 19:36
m: my Num %hash = :42foo
camelia rakudo-moar 2ef2fd: OUTPUT«Type check failed in assignment to %hash; expected Num but got Int (42)␤ in block <unit> at <tmp> line 1␤␤»
viki m: my Int %hash = :42foo
camelia ( no output )
samcv thanks timotimo
so i could do my Str %hash if i wanted to make sure all of them were strings or something? 19:37
viki harmil_wk: that's a good question and I don't think there's a certain answer. Guaranteeing some internal layout of core code would restrict development severely, so we can't do that, but without that guarantee, yes, you can't assume defaults. 19:38
m: class Foo does Setty { has %!elems } 19:39
camelia rakudo-moar 2ef2fd: OUTPUT«5===SORRY!5=== Error while compiling <tmp>␤Attribute '%!elems' already exists in the class 'Foo', but a role also wishes to compose it␤at <tmp>:1␤»
viki harmil_wk: ^ like stuff like this. If we need to create %!whatever attribute in Setty, that does have potential to break someone's code who does Setty and has that attribute defined. 19:40
samcv if i try and do this: %channel-event-data{$e.nick}{'spoke'} = now.Str; i get: Type Array does not support associative indexing.
viki Also, to note, I wouldn't exactly claim "EVERYTHING is derived from something core", I'd think 95% of classes derive from Any. 19:41
19:41 sivoais joined
viki samcv: sounds like %channel-event-data{$e.nick} has an Array in it and not a hash 19:41
19:41 seatek left
samcv ah so a hash of arrays 19:41
viki prolly 19:42
ugexe m: sub validate(%hash (Str :$name, Int :$ver)) { %hash }; my %valid = :name<xxx>, :ver<1>; my %invalid = :name(Nil), :ver<fff>; say validate(%valid); say validate(%invalid) # its probably easier to validate your hash like this
camelia rakudo-moar 2ef2fd: OUTPUT«{name => xxx, ver => 1}␤Type check failed in binding to $name; expected Str but got Any (Any)␤ in sub validate at <tmp> line 1␤ in block <unit> at <tmp> line 1␤␤»
samcv i guess that makes some sense
19:43 RabidGravy joined
viki harmil_wk: though the impact can probably be ameliorated by some sort of convention... like all core attributes and internal methods have some sort of prefix and that would guarantee they don't clash with user code 19:43
viki contemplates the amount of work that would entail
ugexe prepend Rakudo::Internals:: to eerything
harmil_wk viki: Well, that addresses the issue you brought up, but doesn't really address the one I was concerned about. I do agree, though, that using some convention there is wise (and generally practiced in other languages) 19:44
lizmat re subclassability of Hash: we could so something like with Date.new 19:45
which *is* subclassable
viki s: Date, 'new', \()
SourceBaby viki, Sauce is at github.com/rakudo/rakudo/blob/2ef2...ish.pm#L28
viki s: Date, 'new', \("2016-11-11")
SourceBaby viki, Sauce is at github.com/rakudo/rakudo/blob/2ef2...ate.pm#L36
viki m: say class Foo is Date { }.new === Date 19:49
camelia rakudo-moar 2ef2fd: OUTPUT«False␤»
viki I rescind my previous comments. It does make sense for some behaviour to be expected and the non-setting of attribute defaults with Hash is LTA. 19:50
19:50 firstdayonthejob left
viki And on that note... 'tis quitting time \o/ 19:50
19:51 girafe joined
masak I just realized that with looping over a hash/dictionary/map, Perl 5, Perl 6, Python 2/3 and Java all have different behaviors. 19:51
Perl 5 loops over zipped keys/values in a flat list. Perl 6 loops over pairs as reified objects. Python loops over keys. Java doesn't allow it, and forces you to be more specific.
samcv hmm it looks like %hash has keys which point to arrays whith only one element, which hold a hash 19:53
er wait, no. a hash of arrays, which hold multiple elements, each of which is a hash 19:54
agh 19:55
harmil_wk viki: have a good one
19:56 firstdayonthejob joined
viki samcv, just dump your JSOn on this site and then use the Viewer tab to explore its structure: jsonviewer.stack.hu/ 19:56
masak I'll also note that of the four languages, only Java and Perl 6 could be said to have reified pairs. though Python compensates well with 2-tuples.
RabidGravy Anyway I went someway to absolution for spending several days being annoyed by Scala at work by making a Vagrantfile with rakudo 19:57
so arguably I have actually been paid to do Perl 6 stuff today :) 19:58
timotimo neat 19:59
20:00 Rawriful joined
harmil_wk I think that in general, my feeling is that defaults on attributes isn't a good idea. I hate to be all "there's only one way to do it", but it feels as if defaults on BUILD parameters is the more reliable tool, at least as things stand. 20:01
RabidGravy it depends what the requirement is, no point being absolutist about it 20:03
20:11 nicq20 left 20:13 ale1ster left
masak m: class C { has $.x = 18; submethod BUILD($!x = 5) {}; }; say C.new 20:14
camelia rakudo-moar 55b9e9: OUTPUT«C.new(x => 5)␤»
masak harmil_wk: would you say that your preference has anything to do with the fact that BUILD parameter defaults *override* attribute defaults? :) 20:15
RabidGravy I've always wondered whether that's right, that if there's a BUILD at all the defaults are ignored 20:16
masak here's how I remember it 20:17
I was also always wondering/skeptical about whether that's right 20:18
but tadzik was the one who put up the fight, and clearly expressed his dislike over the state of affairs
at which point TimToady spoke up, and make it abundantly clear that the way things are is not an accident, and he would *not* change it -- for reasons of performance -- just because it was deemed convenient to do so 20:19
tadzik Aww yiss
yoleaux 06:51Z <japhb> tadzik: I sent you a PR to add 256-color and 24-bit RGB color support to Terminal::ANSIColor. Enjoy. :-)
masak tadzik: did I get that approximately right?
tadzik I don't like BUILD eitger
Masak: yup 20:20
masak I never did ask follow-ups on the performance rationale, but I *do* trust that TimToady has thought this one through
tadzik The one from Moose may be improper, but it's useful
masak maybe the argument would go something like "running one BUILD is way faster than running lots of silly little attribute default thunks"
which I guess I buy 20:21
tadzik: what does the one from Moose do that Perl 6 doesn't? 20:23
tadzik japhb: <3, will check it out.after $movie
Bbl&
masak alright :)
RabidGravy I've got used to it now, but I still occasionally get bitten when I've put defaults in and then decide that I need a BUILD for some reason
masak assumes he could get tad zik's attention again by writing his nick, but chooses not to :)
RabidGravy: yes -- that's what bothers me too. in the "wrong code that looks right" department, those attribute defaults look like they would do something, but they all become sitting ducks the moment a BUILD shows up. 20:24
20:24 domidumont left 20:26 girafe2 joined
masak the most cheerful I can say about it is that at least the action-at-a-distance is limited to the same class, because (a) BUILD is a submethod, and (b) one tends to initialize !-twigil attrs, again because of (a) 20:26
20:27 girafe left
ugexe m: role Foo[$a] { has $.a = $a }; my %args = :a<2>; my $foo = Foo[1].new(|%args); say $foo.a # just use roles for all the things 20:27
camelia rakudo-moar 55b9e9: OUTPUT«2␤»
masak ugexe: I was in the "just use roles for all the things" camp for a long time 20:28
ugexe: the place where it falls down is when you start doing multiple inheritance
and you realize that you're dealing with a flat thing, not a hierarchy 20:29
I had this happen in a very real way in the 007 project
arnsholt One feature in SmallTalk I really like is the method call chain syntax
Not sure how it'd be transferrable to a more C-like syntax though
masak see github.com/masak/007/commit/ecb49f...459d3e9bba
ugexe all you have to do is add method and submethods of new, BUILD, some logic in the body of the role itself to... 20:30
heh yeah i know it starts to suck pretty fast 20:31
masak arnsholt: could you tell me a bit more about the method call chain syntax, and what you like about it?
arnsholt It's basically just a syntax to invoke several methods on the same invocant
masak ugexe: from then on, the rule in 007 was "roles for interface-y types, classes for concrete types"
ugexe: but 007 is now gearing up to do all the MOP stuff in 007 rather than in Perl 6, so that's going away too :) 20:32
arnsholt So in a sense similar to what I think is called "fluent interfaces" (or some such), where methods return the invocant
masak arnsholt: what's the syntax?
ugexe i still like class_name[...].new() for things that are a container though
arnsholt Semicolon instead of period as statement terminator, and statements don't start with an invocant (there being more or less only return statements and method call statements in SmallTalk) 20:33
So you can for example do, to populate a hash: "dict := Dictionary new; add: 'foo' -> 1; add: 'bar' -> 2; yourself" 20:34
masak re-vows to get through worrydream.com/EarlyHistoryOfSmalltalk/
arnsholt Where yourself is a method that returns the object it's called on. Sounds like a silly method, but this is why it's there =)
masak :) 20:35
arnsholt (For the particular example above theres Dictionary>>with:with: which does the same thing, but that'd be missing the point)
Another absolutely hilarious SmallTalk method is become:
Which takes mutates the instance it's invoked on into the instance passed as an argument 20:36
Which is kind of o.O
masak I've heard about that method
I think I read about it on ward's wiki
arnsholt Sounds entirely plausible 20:37
I haven't had occasion to use it yet, sadly =)
20:41 tushar joined 20:44 avalenn_ joined 20:48 AlexDaniel joined, avalenn left 20:52 TEttinger joined
konobi TimToady: I thought you might find this interesting: www.kickstarter.com/projects/23012359/minima 21:05
masak discovers a lot of other nice c2.com articles in looking for the one about `become:` 21:08
samcvtest having some issues with variable names with - in them 21:14
Variable '%channel-event-data' is not declared, and then before, said something like invalid use of '-' in sink context 21:15
21:17 hankache left
ugexe thats... impressively vague 21:17
21:18 Actualeyes left
samcvtest yeah sorry trying to find it. it could be related to maybe problems in another spot. so i'm investigating 21:18
i'm thinking maybe it's not matching a parenthisis or something and misinterpreting the variable.. 21:19
Useless use of "-" in expression "$fh-3.say( to-json( %.channel_event_data) )" in sink context, oh looks like
wasn't related to the variable, i renamed the variable to use underscores 21:20
so it's probably the to-json function
jnthn $fh-3 would be parsed as infix -
grondilu timotimo: that typeface is barely readable
ugexe not really a good idea to use -<number> in a variable name
samcvtest oh. the filehandle. ugh derp
grondilu timotimo: sorry meant to address that to TimToady
jnthn We required alpha after a -
*require
timotimo is ok :)
21:21 Actualeyes joined
jnthn (for it to be parsed as a variable name) 21:21
21:21 setty1 left 21:22 setty1 joined 21:26 bjz joined 21:32 bstamour left
RabidGravy oh dear I just looked at the Jenkins job definition xml 21:37
I really wish I hadn't 21:38
only a java programmer could have perpetrated "<com.coravy.hudson.plugins.github.GithubProjectProperty plugin="github@1.22.4">" 21:39
21:39 benchable6 joined 21:40 committable6 joined, ChanServ sets mode: +v committable6, MasterDuke_ joined
timotimo m) 21:40
21:44 MasterDuke left 21:46 ugjka left 21:47 tushar left 21:49 ugjka joined 21:50 Tonik left
RabidGravy it's not even a sane schema 21:58
22:06 pierre_ joined 22:08 seatek joined 22:16 wisti left 22:24 sufrostico left 22:28 BenGoldberg joined 22:33 ale1ster joined
tadzik masak: so, in Moose when I do BUILD I can put it there and not put it there and it doesn't impact the rest of the class ocde 22:35
in P6 when I do BUILD I break attribute initialization in some cases and not in some others
and when a new attribute comes in I need to modify the BUILD signature again for attrs to initialize properly 22:36
I wrote this stupid patch 5 years ago that introduced a method called AFTERPARTY that did all I wanted :P 22:37
timotimo just implement BUILDALL and put callsame at the top
tadzik I just want to Do Stuff When Everything Is Ready As It Should Be. Could do it in Moose, cannot in P6
'that could work; I assume there's a catch, and a reason why we usually go the overload-new/submethod-BUILD way instead 22:38
perigrin tadzik: obviously the answer is to port Moose to p6. 22:39
perigrin hides.
tadzik perigrin: fortunately there's Inline::Perl5 so we don't have to :P
perigrin :D
22:40 seatek left 22:43 pierre_ left 22:44 pierre_ joined 22:47 bjz left 22:48 pierre_ left 22:49 rindolf left 23:04 RabidGravy left 23:08 pierre_ joined
ugexe always trick the end user into doing all initialization so its their fault 23:10
konobi TimToady: any idea if there is something else already existing? 23:12
23:13 pierre_ left
masak tadzik: thank you -- that helped 23:17
now all I could ask for is for TimToady to put that into perspective somehow
tadzik :)
I remember talking/arguing with TT over this and coming out as wrong (duh :P) 23:18
masak but giving the TimToady genie any explicit instructions is a fool's errand... right :P
we'll... see what happens
mst everybody argues with TT and loses, especially when it comes to context
hence DBIC's _rs methods etc. 23:19
masak .oO( Club Of People Who Argued With TT and Lost )
MasterDuke_ i think mst had a bunch of comments/suggestions/something about the difference between Moose and Perl 6 a couple weeks ago 23:20
in regards to the initialization of attributes 23:21
mst right, basically, as soon as you define a BUILD submethod, perl6 goes "surprise! let's make you type more than raw perl5 OO" 23:22
tadzik yes!
mst if you duplicate all your attributes as named parameters in it, it goes back to doing mostly the right thing 23:23
when I complained about this, TimToady basically said that yes, people had noticed, and what really needed to be done was to have a subroutine for 'do the basic BUILD stuff' you could call anywhere you liked in your custom BUILD (thereby giving you the Moose functionality back -and- control over ordering), but nobody'd come up with a good name for it so it hadn't been implemented 23:24
the fact that as a result, for most of the cases I care about, perl6 OO is more typing and requires me to duplicate attribute names and hope I don't typo, is why I'm using perl6 via Inline::Perl6 for stuff but stick to writing my OO stuff in perl5+Moo 23:25
tadzik now that's just crazy
mst well, that and perl6 has nothing like lazy either
tadzik nodnod
mst so basically perl6 OO is inferior to Moo for 99% of my use cases
mst just wants grammars, I'll wait for the rest of the language to catch up to perl5 23:26
timotimo wow
masak I'd love for p6 to have a good lazy story
I 100% agree with mst that that's currently missing
mst basically, perl6's BUILD is actually (in my head) POPULATE, i.e. slot population 23:27
I'd like there to be two submethods
one for overriding slot population
and one for the much more common case that Moo(se) provide 23:28
23:28 ale1ster left, ttkp6 joined
mst given Moo/Moose's more featureful attributes, I've honestly never wanted to override slot population that way anyway 23:28
masak I don't know about all y'all, but this sounds like something that could be tried out in a branch 23:30
mst I'd be happy to play with it if somebody did
but when TimToady says "yeah, we don't have that, we didn't think of a name" and lizmat basically told me she preferred positional constructors and manual population in BUILD, I just went back to writing the OO parts in perl5 since that already has all the features I needed 23:31
esp. given p3rl.org/Function::Parameters to give me typed signatures
masak mst: could you perhaps provide some details of what you imagine the two submethods would do? 23:32
mst easy
POPULATE would to what BUILD currently does, i.e. override everything and make you do everything by hand
BUILD would do what BUILD in Moo(se) does, i.e. populate the slots and then let you add whatever logic you want
(names based on my mental model, not intended as recommendations or anything) 23:33
23:33 Actualeyes left
masak ok, at this point I think there's enough info for whoever feels motivated to try it out in a branch 23:33
mst this is, if memory serves, at least the third time I've had this conversation
masak :) 23:34
mst maybe I need to give a YAPC talk titled "why perl5 OO is better than perl6" to motivate people to change that ;)
masak heh 23:35
mst not serious. that would be almost as awful as lizmat's "why perl5 is the new perl4" talk in terms of fucking up community relations 23:36
a blog post called "things I miss from M*" might not be a terrible plan though
23:37 firstdayonthejob left
mst grumbles 23:41
"you stole most of the things Moose did better than your old metamodel, why couldn't you've stolen these too" :P
23:41 perlawhirl joined 23:45 wamba joined 23:51 cdg left 23:59 nadim_ left, nadim_ joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!