»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.
hoelzro sortiz: I'm having trouble breaking your change, which is good =) 00:07
I'm wondering, though, if we shouldn't override FAILGOAL in rakudo to throw a typed exception that we can look for, kind of like X::Syntax::Missing 00:08
sortiz hoelzro, That was my first approach, but there are tests in roast that expect X::comp::AdHoc :( 00:11
hoelzro =/ 00:13
maybe those tests need to change? I don't know how comfortable that makes me in a post-6.c world, though 00:14
dinner &
sortiz CY
Bon Appetite! 00:15
sevvie sortiz: s/te/t/
masak .oO( J grand, a petit ) 00:16
sevvie voulez vous coucher avec moi, monsoir? 00:19
sortiz tnks sevvie. :)
sevvie <3 00:21
I've been out of the loop a week or so, how grows the butterfly?
masak thought it was "ce soir" 00:22
timotimo are we proposing cuddles?
sevvie masak, it might be. It's been some time since I've heard the song and couldn't remember the words too well; I guessed based on scansion. 00:23
timotimo: whY would you think that wE are propoSing cuddles? 00:24
masak :P
masak commutes 00:27
timotimo <3 00:28
leont Indeed it's ce soir. It's quite an interesting song to analyze, actually (for various reasons)
sevvie ... evernoted. 00:29
timotimo i'm going to bed alone now. i'm even leaving the cats behind 00:33
but i actually have an easier time sleeping when alone 00:35
sortiz What fault have cats?
timotimo they get super annoying when it's time for them to get fed in the morning :) :)
leont You mean, they love you intensely and without reservation ;-) 00:36
timotimo yep. they'd love me to my face and chest
anyway, i'm heading out :) 00:37
sortiz till tomorrow timotimo, sleep well without cats :) 00:39
timotimo i spent some more time letting the cats adore the light spots that my glasses project onto the walls 00:58
sortiz .tell hoelzro Overriding FAILGOAL the tests that fail are in S02-literals/char-by-number.t and S02-literals/quoting-unicode.t. In roast there are 37 tests that explicitly check for X::Comp::AdHoc, convert them to typed exceptions is for another project, though 00:59
yoleaux sortiz: I'll pass your message to hoelzro.
timotimo nothing terrible about throwing an X::Comp::AdHoc enhanced with an extra role 01:02
sortiz Can be, but the cleaner approach is turn them to typed exceptions, case by case. Some a inherited from nqp/Grammar and some are from naked 'die's in rakudo. 01:04
All of them should be X::Syntax::<something> 01:05
Right now I'm simply regexing the message, and works well. 01:09
In fact, some tests said '# no exception type yet' :) 01:11
skids m: class X::Syntax::Foo is X::Comp::AdHoc { }; use Test; throws-like { X::Syntax::Foo.new(:message<foo>).throw }, X::Comp::AdHoc 01:14
camelia rakudo-moar a7ec54: OUTPUT« 1..2␤ ok 1 - code dies␤ ok 2 - right exception type (X::Comp::AdHoc)␤ok 1 - did we throws-like X::Comp::AdHoc?␤»
skids Not that it should remain that way forever. 01:15
sortiz Sure, I known.
sortiz Are you up to volunteer? ;) 01:19
sortiz Surprised and delighted that native methods works!! 01:27
timotimo how else would we be able to claim c++ support?
sortiz This is only C, but had not tried. :) 01:29
For Pg DBDish: class PGconn is export is repr('CPointer') { method PQstatus (PGconn: --> int32) is native(LIB) { * } }; 01:31
Herby_ Evening, everyone! 03:06
yoleaux 24 Feb 2016 05:27Z <MadcapJake> Herby_: i'd suggest trying Crust if you do want to peruse webdev in Perl 6
Herby_ MadcapJake: I'll take a look, thanks 03:07
Herby_ \o 03:14
skids o/ 03:15
skids m: my ($n, $base) = 120973, 100; ($n ~~ m:ex/^[$<dig>=[(0|<[1..9]>\d*)<?{ +~$0[*-1] < $base }>]]*$/).map( { [+] (((+«$_<dig>).reverse) Z* (1, * * $base ... *)) }).sort».say 05:40
camelia rakudo-moar a7ec54: OUTPUT«1200973␤1209703␤12000973␤12009703␤102000973␤102009703␤120090703␤1200090703␤10200090703␤»
skids (solution to reddit daily programmer)
A bit slow.
Probably due to the match results not being a lazy uncached iterator. 05:42
dalek Iish: 5288b6b | (Salvador Ortiz)++ | lib/DBDish/Pg (4 files):
Pg now uses typed handlers and native methods

DBDish::Pg revamped to use 'is native' methods of typed CPointer classes at the native call interface level.
A lot of needless 'return' removed and general cleanup.		 05:58
Iish: f042c03 | (Salvador Ortiz)++ | lib/DBDish/Pg/ (2 files):
Two types made NativeCall unneeded in Pg::SH
nwc10 lizmat++ # setting compilation seems to be a few percent faster 07:01
[Tux] test 22.043 07:28
test-t 11.828
csv-parser 51.126
lizmat, any progress/mews on that new where bug? (just being curious) 07:29
AiHasBeenSolved www.sourcecodeonline.com/details/gh...ng_ai.html 07:42
AiHasBeenSolved Ghost AI coder leaves this message FYI -- For Your Information. Bye
FROGGS wow, that's an impressive code style: ai.neocities.org/perlmind.txt 08:16
sad that (s)he has left 08:17
moritz FSVO "impressive" :-)
FROGGS I mean, *every* line has its coment 08:18
that's probably what REXX hackers are used to
RabidGravy EUGH
FROGGS I wonder what that actually does 08:19
gfldex the comments didn't tell you? 08:19
TEttinger in soviet russia, ai solves you! 08:24
Hotkeys What's going on here 08:25
> f(@z)
([141 214 314 495 767] [908 709 314] [1617 314] [1931])
> f(@z)».join(' ').say
(141 214 314 495 767 1931 1931 1931)
gfldex Hotkeys: some stuff is joined (and separated by ' ') and some stuff is stringyfied
Hotkeys but that should join each list in that superlist 08:26
gfldex Hotkeys: also, you may want to step away from the dangerous say method if you work with lists 08:27
Hotkeys it does the thing even without say
gfldex Hotkeys: could you golf that a little better so we can play with it? 08:28
Hotkeys do you want f?
FROGGS we want an example that fits in a single line here, with e.g. less data 08:29
gfldex the result of f(@z) would do
Hotkeys evaling f(@z) in the repl gives ([141 214 314 495 767] [908 709 314] [1617 314] [1931]) 08:30
gfldex please .perl.put it 08:31
Hotkeys > f(@z).perl.put
([141, 214, 314, 495, 767], [1931], [1931], [1931])
gfldex m: @a = [141, 214, 314, 495, 767], [1931], [1931], [1931]; @a.perl.put; 08:32
Hotkeys however > f(@z)[1].perl.put 08:32
[908, 709, 314]
gfldex camelia: you alright?
FROGGS ahh 08:36
"In the early stages of the construction of the Mind.pl program, you may type in a word such as "boys", then "play" and then "games" to tell the AI that "boys play games." The program will go about storing the words in auditory memory and at the end will show you the contents of the auditory memory, along with associative tags that connect words as concepts instantiated in a mind. "
"In the more mature AI Minds such as MindForth or the JavaScript AiMind.html program, once you tell the AI that "boys play games" and you enter a fact such as "John is a boy", the InFerence module will engage in machine reasoning and ask you, "Does John play games?" The Perl AI is on its way to the same level of reasoning. "
that's what I wanted to know
gfldex Hotkeys: the first element (that is a list) of the list is joined as asked by you, then the other 3 elements are then assumed to be lists with one element. Joining a list of one element is kinda hard. The resulting list of strings is then stringyfied (and join by a single space). 08:39
Hotkeys why are they assumed to be one element though 08:40
gfldex because .join demands a list
any item that is called with a listy-method is turned into a list on the fly 08:41
lizmat good *, #perl6! 08:42
lizmat [Tux] sorry, didn't get a round tuit yet 08:42
Hotkeys > (.WHAT for f(@z))
((Array) (Array) (Array) (Array))
they seem to already be listy though
gfldex Hotkeys: the interface to any object is the set of methods you call at runtime. The "type" of the object doesn't matter (unless you force a type-check).
in this case they are arrays with one element 08:43
Hotkeys okay but
> (.join(' ') for f(@z)).perl
("141 214 314 495 767", "908 709 314", "1617 314", "1931")
Why does this work then 08:44
what's different?
gfldex ».foo will call foo and take each return value to form a new list
that list is then stringyfied
in your initial example you either call .join to early or to late, depending on what output you would like to see. 08:45
Hotkeys I was under the impression that @foo».bar would call .bar on each element of @foo 08:46
which is also what the for loop there does essentially
what am I missing
gfldex what output would you like to see? 08:47
moritz doesn't >> descend further into sub-structures?
moritz has lost track of current hyper semantics
Hotkeys I would like to see the output the for loop produced
("141 214 314 495 767", "908 709 314", "1617 314", "1931")
that is
each element of each sub list joined
moritz @list.map(~*)
@list.map(*.join(' ')) 08:48
Hotkeys I know
but
gfldex ».join should actually do that 08:49
Hotkeys postfix ». / postfix >>.
Hyper method call operator. Will call a method on all elements of a List out of order and return the list of return values in order.
gfldex camelia where are you?
Hotkeys is a bit misleading if it doesn't actually do that
jast it does do that
your confusion comes from using .say on its result
try this: f(@z)>>.join(' ').perl.say
Hotkeys it also fails with .perl.put
s/fails/does the weird/ 08:50
<Hotkeys> > f(@z).perl.put
<Hotkeys> ([141, 214, 314, 495, 767], [1931], [1931], [1931])
moritz gfldex: I can't even ping the server anymore that it runs on
jast there's no >>.join in that
Hotkeys without the join
something weird is happening
and idk what
even without hyper
gfldex /o\ the internet got a flat
moritz last traceroute point is somewhere in a hetzner DC 08:51
Hotkeys > (.perl for f(@z))
([141, 214, 314, 495, 767] [908, 709, 314] [1617, 314] [1931])
gfldex i prefer dd to loot at things with structure
s/look/loot/ (i better take of this pirate hat) 08:52
Hotkeys I'm wondering why f(@z).foobarbaz is turning everything but the first list into a single value
whether it be join or perl or whatever
jast I just tried the same thing (@x>>.join.perl) with a constant array and it does the expected thing
Hotkeys here's f 08:53
my &f = {my@b=@^a.sort;my@c=[@b];@b=flat(@b [email@hidden.address] while +@b>1))}
it's a messy one liner cause I'm golfing
jast I get the same behaviour with that 08:56
Hotkeys it just seems weird 08:57
Hotkeys because iterating over it works fine 08:57
but regular methods do the weird collapsing
jast I think you're going to have to take this to people who know way more about Perl 6 than I do 09:05
might actually be some kind of bug
nine m: say "I'm up" 09:30
camelia rakudo-moar a7ec54: OUTPUT«I'm up␤»
nine excellent
lizmat afk for a few hours& 09:38
kjs_ good morning. How would one test whether a hash element is an array 09:39
timotimo having the same value a bunch of times at the end looks a lot like you're accidentally re-using a single array where you'd expect multiple different ones, or multiple arrays that share the same scalar container in their first slot
but i'm not messing with that blob of code that has not a single space in it :P 09:42
timotimo also, the thing about hypers that you probably don't know and that's potentially undocumented, is the "nodal" semantics 09:43
when a method that's marked "is nodal" is called on stuff, it won't descend into sublists
it will operate on the list itself instead
llfourn kjs_: %hash<foo> ~~ Array 09:44
timotimo m: say (+<<[[1, 2, 3], [5, 6], [9], [9, 9, 9, 9]]).perl
camelia rakudo-moar a7ec54: OUTPUT«[[1, 2, 3], [5, 6], [9], [9, 9, 9, 9]]␤»
timotimo m: say ([[1, 2, 3], [5, 6], [9], [9, 9, 9, 9]]>>.elems).perl
camelia rakudo-moar a7ec54: OUTPUT«(3, 2, 1, 4)␤»
timotimo ^- see the difference here
without the nodal semantics, the second invocation would have given [[1, 1, 1], [1, 1], [1], [1, 1, 1, 1]] 09:45
kjs_ llfourn: many thanks - that did the trick. 09:48
llfourn yw :)
lizmat m: use nqp; my num @a; say nqp::islist(@a) # how can we differentiate between a num and int here ? 10:22
camelia rakudo-moar a7ec54: OUTPUT«1␤»
lizmat afk again&
psch m: my num @a; @a.of.say 10:25
camelia rakudo-moar a7ec54: OUTPUT«(num)␤»
psch not sure that captures the intent of the question, though...
jnthn It's how I'd do it, and you can use nqp::objprimspec on the result 10:28
ab6tract o/ #perl6 10:45
timotimo hey you
ab6tract so quick question: how can i test whether a sub was successfully imported into the package scope? 10:47
i thought it would look like this:
m: module S { sub s is export { say "hi!" } }; import S; s(); say OUR::{'&s'}:exists
camelia rakudo-moar a7ec54: OUTPUT«hi!␤False␤»
abraxxa is the 2016.02 a bugger release? I have only troubles with it
timotimo "bugger" release? 10:48
abraxxa nothing I touch works
timotimo :o
stop touching things!!!
psch m: module S { sub s is export { say "hi!" } }; import S; s(); say MY::{'&s'}:exists
camelia rakudo-moar a7ec54: OUTPUT«hi!␤True␤»
jnthn abraxxa: Import is lexical.
abraxxa can't install Net::AMQP on a Debian 8 box whereas it worked on my Ubuntu 15.10 notebook
psch OUR:: is what is shared from the current scope, isn't it?
jnthn psch: OUR:: is the current *package*
psch ahh 10:49
jnthn Orthogonal to current lexical scope
So MY:: is the right thing for checking about imports, yes.
ab6tract jnthn: ok. i guess i would have thought that checking OUR would be as sensible 10:50
but yeah, now that i think about it.. a test file is not really a package
jnthn OUR:: in the mainline is just another way to spell GLOBAL:: 10:51
ab6tract i love the scoping in perl 6
timotimo abraxxa: what kind of errors are you getting? 10:51
ab6tract jnthn: gotcha!
jnthn in module Foo { ... } then OUR:: is short for Foo::
etc.
Well, not short there, but... :)
ab6tract that's clear :D
llfourn m: our $foo = "foo"; say GLOBAL::<$foo>; say OUR::<$foo> 10:52
camelia rakudo-moar a7ec54: OUTPUT«foo␤foo␤»
llfourn and if you're outside of a declared package your package is GLOBAL 10:53
El_Che hi jnthn, has your grant been voted yet? I saw results for rperl in my rss feed
ab6tract llfourn: but imported thingies don't show up there anyway
jnthn El_Che: Not as far as I know. 10:54
El_Che: I've heard nothing on it yet.
El_Che ok, no new is good news then :)
jnthn El_Che: Hopefully soon...I'm about ready to get to work :)
ab6tract is trying his hands at a Perl 6 port of Sereal 10:54
El_Che fingers crossed
susmus hello 11:28
psch hi susmus 11:29
DrForr o/ 11:29
timotimo ohai 11:36
psch hrm, this is odd 11:50
something gets confused in the JavaHOW
specifically with the MethodContainer role
as in, a Java class that i successfully load can dispatch to its Java methods
but those don't show up in .^methods
on the other hand, methods added with .^add_method show up in .^methods but can't be dispatched to 11:51
lizmat ab6tract++
moritz fwiw dispatching uses find_method, not .^methods
which likely uses method_table
psch moritz++ 11:52
that's a hint at least
nine psch: in case you wondered. The "# $jtype.WHO{$_.key} := $_.value for @existing;" line is for when there's already a Foo::Bar::Baz package and you are creating a wrapper class called Foo::Bar 11:53
jnthn Generally, meta-objects can keep whatever internal state they want
psch nine: it's commented because it NPEs afair. probably something else going on there that's not quite right...
jnthn And can transform it into what's needed
In the Perl 6 MOP classes we do need both an array and a hash 11:54
Hash for fast lookup, array because we need to know the order methods were added in
psch hm, it probably has something to do with 7432d6202ede8569cac13ecbfae78afafc915fa4
no wait 11:55
that's impossible cause that's a really old commit
well, 2 years and change, which is older than any jvminterop work i did :) 11:56
psch hm, find_method directly fetches %!methods, which only has methods added with .^add_method, if any 12:04
which means dispatch for Java classes is completely weird and i have no idea what i'm doing..? /o\
right, nqp::findmethod finds the original ones 12:06
moritz it just means that it doesn't call methods from parent classes 12:07
so, no inheritance
FROGGS I thought you .^add_method all methods that exist on java classes?
psch FROGGS: no, github.com/rakudo/rakudo/blob/nom/....java#L896 12:08
dalek kudo/nom: 817b0b9 | lizmat++ | src/core/Buf.pm:
Streamline Blob creation

  - when creating from Blob:D or int array, use nqp::splice (1.5x faster)
  - pre-size buffer if we can know number of elements to be added beforehand
  - add private method for failing with element number info
FROGGS ahh 12:09
psch the curious bit is why the bind_attribute_boxed calls don't seem to take 12:09
a few lines below the linked one
afaict i'm not messing with the HOW anywhere either... 12:10
ZoffixW .tell nemo I'm not deathly sick :) While still not fully well, the reason I'm not around is because I've just moved and (a) I don't have Internet at my new place yet and won't until next week. (b) I'm too busy with unpacking and settling in :) 13:08
yoleaux ZoffixW: I'll pass your message to nemo.
ZoffixW waves and submerges under the shadows
ab6tract hmmm.. does anyone know how to map the @EXPORT_OK pattern to p6? 13:20
ie, to make an import "rule" for a whole bunch of things dynamically? 13:21
El_Che ab6tract: doc.perl6.org/language/modules#EXPORT ? 13:23
ab6tract El_Che: yeah, i've been up and down that doc a few times..
moritz I guess the weird thing is that EXPORT in that example doesn't take any arguments 13:27
FROGGS ab5tract: just mark the subs you want to export "is export(:MY_TAG)"
moritz which makes dynamism harder
dalek c: 11ebd2b | (Claudio Ramirez)++ | doc/Language/modules-extra.pod:
META6.info -> META6.json
 13:28
FROGGS moritz: but EXPORT can take params, no? 13:29
I mean, the other examples show it 13:30
though, that not what I think what the original question was 13:31
the idomatic way of exporting stuff on demand is to use custom tags
ab6tract FROGGS, moritz: github.com/Sereal/Sereal/blob/mast...nstants.pm 13:38
i think i will just go the "long form" route and do:
constant SRL_HDR_ALIAS is export(SRL_HDR_ALIAS) = 46; 13:39
FROGGS hmmm
hoelzro o/ #perl6
yoleaux 00:59Z <sortiz> hoelzro: Overriding FAILGOAL the tests that fail are in S02-literals/char-by-number.t and S02-literals/quoting-unicode.t. In roast there are 37 tests that explicitly check for X::Comp::AdHoc, convert them to typed exceptions is for another project, though
FROGGS why not export all constant with the tag :constants or so? 13:40
ab6tract FROGGS: sure, but that doesn't match the original behavior
FROGGS does it have to?
ab6tract not necessarily :)
FROGGS see :o)
and I would probably put these constants into five enums 13:42
SRL_HDR, SRL_MAGIC, SRL_MASK, SRL_SIZE, SRL_PROTOCOL
then you can use the enum type object as a constraint in signatures
ab6tract docs.perl6.org => 'enum' not found 13:44
FROGGS like: github.com/FROGGS/p6-XML-LibXML/bl...s.pm#L5-L9
is used here: github.com/FROGGS/p6-XML-LibXML/bl...14N.pm#L11
hoelzro now that 6.c has been released, how flexible can we be in changing AdHoc exceptions to another typed exceptions? sortiz and I wanted to make FAILGOAL raise a X::Comp::FailGoal (or something like that) instead of X::Comp::AdHoc, but that breaks roast 13:46
ab6tract hrrmm
m: enum XS { X1 => 1, X2 => 2 }
camelia rakudo-moar 817b0b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/JqnzurrrTG␤Undeclared name:␤ XS used at line 1␤Undeclared routine:␤ enum used at line 1␤␤»
moritz then roast should be fixed to only check for X::Comp (at least in the 6.c-errata branch) 13:47
jnthn hoelzro: AdHoc => typed is fine
FROGGS ab6tract: dont use curlies
ab6tract ah :D
jnthn I went through the test suite and removed all mentions of X::AdHoc, just making them say Exception
So, no changes should be needed to 6.c-errata
(To be clear, I did that before 6.c) 13:48
moritz jnthn: you seem to have overlook X::Comp::AdHoc
*overlooked
jnthn Grr
moritz jnthn: I dind't think of that easier when we discussed that topic :/ 13:49
jnthn OK, then we should s/X::Comp::AdHoc/Exception/ in 6.c-errata
But don't put into 6.c-errata exception types that never existed in 6.c
moritz m: say X::Comp::AdHoc ~~ X::Comp
camelia rakudo-moar 817b0b: OUTPUT«True␤»
ab6tract FROGGS: thanks!
moritz jnthn: or replace it with X::Comp
jnthn Or that, yes
moritz which is a little bit more specific
hoelzro jnthn: ah, great! so I should replace X::Comp::AdHoc with X::Comp? on which branch of roast? 13:51
sortiz \o #perl6 13:53
sortiz JIT
moritz hoelzro: on 6.c-errata 13:54
hoelzro *nod*
moritz hoelzro: though you can do it too on master
hoelzro master is what's bound for 6.d, I take it?
moritz yes
hoelzro what's the policy for propagating changes between 6.c-errata and master? merging, cherry-picking, etc? 13:56
moritz doesn't know of the existance of any policy 13:57
lizmat only errors should be merged back ?
hence the "errata" name ?
jnthn hoelzro: Don't do merges between them
hoelzro I would assume that 6.c-errata would be merged into master on a regular basis, but I don't want to step on toes
jnthn I'd prefer commits in master, then cherry-pick what's needed into errata 13:59
hoelzro ok
sortiz If I understand well, in 6.c-errata only the tests that check X::Comp::AdHoc should be changed for Exception, no? 14:00
jnthn sortiz: Well, X::Comp instead of Exception, it looks like 14:01
But yeah, that'd give us the flexibility we shoulda had in the first place.
I expect there to be less errata with time
hoelzro one hopes =) 14:02
dalek kudo/nom: fe0768b | lizmat++ | src/core/Buf.pm:
Introducing Blob.allocate

Create your Blob-like objects with a given number of elements, optionally with values other then 0 to initialize with.
  $ 6 'dd Blob.allocate(11,1,2,42)'
Blob.new(1, 2, 42, 1, 2, 42, 1, 2, 42, 1, 2)
lizmat jnthn sortiz ^^^ 14:03
sortiz lizmat++ 14:04
gfldex what reminds me of a question i wanted to ask
how is the transition from 6.c -> 6.d planned for docs?
or better, are there any plans? 14:05
lizmat gfldex: if anything, I think there are too many plans 14:05
gfldex i shall plan to make a plan then :->
lizmat :-)
sortiz hoelzro, I can resurrect the override of FAILGOAL from my branch, what exception should I use? 14:07
hoelzro sortiz: I made a X::Comp::FailGoal locally; we can just use that if you don't mind 14:08
brian__ How do I append one Buf[uint8] to another? (gist.github.com/bduggan/e740964547406f1c856d) 14:09
hoelzro right now I'm working on converting X::Comp::AdHoc → X::Comp in roast
jnthn brian__: ~ 14:09
sortiz hoelzro, no problem. 14:10
gfldex joking aside, we will need a way to keep as many different versions of the docs around as we keep versions of the language. At least for reference. Or we might end up with docs littered with: "but in 6.d you do it that say". Or unmarked stuff that simply wont work in 6.c and resulting confusion.
brian__ jnthn: thanks! 14:11
lizmat brian__: alternately, you can do $buf.append($buf2) nowadays
jnthn Only in bleeding edge though. :) 14:12
But yeah, that'll be in the next lang version :)
brian__ ok, yes that didn't seem to work in 2016.01
jnthn Well, provided we have tests for it :)
lizmat jnthn: there are some append tests nowadays
lizmat jnthn: also, are you happy with the Blob.allocate interface ? 14:13
lizmat looks outside to enjoy the snow
jnthn lizmat: Not quite sure about the initialization-with-sequence part... 14:14
lizmat well, it can be a single value as well
jnthn lizmat: Will ponder a little on that. May be fine. Something feels a tad off, but I don't quite know what yet. Maybe nothing. :) 14:15
lizmat I was more thinking allowing something like a 0xdeadbeef initialization option
jnthn My guts are messed up enough at the moment that gut feeling isn't worth much :P
lizmat jnthn: sorry to hear you're not feeling well :-( 14:16
jnthn: related: I was thinking of allowing Buf:D.allocate as well, for changing the size of the buffer (larger *and* smaller) 14:17
any thoughts on that ?
jnthn Think I'd rather it were called something else...
lizmat reallocate ?
jnthn Yeah
Would match-ish the C-ish naming without the shortening :) 14:18
lizmat but allowing shortening would be ok, right ?
jnthn I'd rather go with reallocate, if we have allocate 14:19
perlpilot lizmat: fwiw, combining fixed size allocation and initialization seemed weird to me too.
jnthn perlpilot: calloc does it ;) 14:20
No, wait :)
calloc doesn't it :)
You'd typically memset
perlpilot also ... Perl isn't C ;)
jnthn But yeah
lizmat perlpilot: the thing is, that Blob's are immutable, so you cannot have an .initialize method later
jnthn I don't think I might an initial value in there
*I mind
lizmat again, Blob's are considered immutable, so initializing later is not really an option, hence the initialization values with the allocate 14:21
dalek kudo/nom: 9f74027 | coke++ | README.md:
Note current pumpking.

  pmichaud++		 14:22
jnthn lizmat: tbh, I mostly expect people to encounter Blob when they receive one from, say, I/O.
Anyway, on naming, I think either we need to have allocate/reallocate, or alloc/realloc, but not allocate/realloc :) 14:23
lizmat ok, I'll implement a reallocate for now 14:24
jnthn Yeah
Well, it's all changeable until it's in an actual lang release. :)
lizmat doesn't feel to me that the name needs to be huffmanized
jnthn No, I'd rather spell it out. It's what we've tended to do, except when the prevailing culture pushes us the other way.
Though, strictly speaking, I think the name "reallocate" *is* huffmanized 14:25
Huffmanized doesn't mean short, it means matching commonality of use. :)
[Coke] (docs 6.c vs. 6.d) there's a ticket about being able to track which language versions things were introduced or removed in. 14:26
lizmat well, doesn't need to be huffmanized further then :-)
jnthn Aye :)
ab6tract should i expect "\xF3" to produce the same value between p5 and p6?
because at first glance this is not the case
lizmat well, inasofar "" generates a Str in P6, and is untyped in P5 ? 14:27
ab6tract lizmat: true.. that's a definite difference. i meant more in terms of the character... to be honest i'm not at all familiar with these underlying guts (pack/unpack, protobufs, encodings, and the like) 14:28
jnthn m: say "\xF3"
camelia rakudo-moar fe0768: OUTPUT«ó␤»
ab6tract jnthn: that's what p6 gives me locally
jnthn ab6tract: Right, which is correct...
ab6tract p5 outputs a question sign
jnthn .u ó
yoleaux U+00F3 LATIN SMALL LETTER O WITH ACUTE [Ll] (ó)
lizmat ab6tract: I think P5 has the trouble with displaying it 14:29
ab6tract ok
jnthn ab6tract: I'm certainly the wrong person to ask about what Perl 5 does. But I'm guessing you may have to opt in to...something...:)
lizmat probably something about STDOUT not being UTF-8 ? 14:29
ilmari -CSAD
jnthn Anyway, in Perl 6, \xABC refers to a Unicode codepoint number.
lizmat $ perl -E 'say "\x61"'
a
LizyPro:rakudo.moar liz$ 6 'say "\x61"'
a
lizmat looks the same for me for less troublesome chars 14:30
ab6tract cool
ilmari $ perl -CS -E 'say "\xf3"'
ó
ab6tract this is all because "In protocol version 3 the magic string has been changed to "=\xF3rl", where \xF3 is "s" with the high bit set. "
nice! thanks liz
jnthn That sounds like a protocol that wants byte level processing... :) 14:31
lizmat ab6tract: ilmari showed how to make STDOUT be utf-8 :-) 14:31
ilmari that probably means it's the byte 0xf3, not the character U+F3
ab6tract in their Sereal::Decoder::Constants module, they are then using a different escape scheme, doing "\363"
ilmari that's octal
ab6tract oops.. sorry for the misattribution ilmari :)
ilmari m: say 0x363 == 0xf3 14:32
camelia rakudo-moar 9f7402: OUTPUT«False␤»
ilmari m: say 0o363 == 0xf3
camelia rakudo-moar 9f7402: OUTPUT«True␤»
abraxxa timotimo: sorry, busy day...again 14:33
ab6tract ilmari: sorry i'm so dense, but aren't the byte 0xf3 and U+F3 functionally the same, in terms of what the byte looks like?
abraxxa timotimo: all kinds of, mostly test failures on install, hangs on panda/p6doc start without params
ilmari ab6tract: can be any number of byte sequences, depending on the encoding U+F3 14:34
ab6tract but for utf8... 14:35
ilmari m: say <latin-1 utf-8 utf-16le utf-32le>.map: "\xf3".encode($_)
camelia rakudo-moar 9f7402: OUTPUT«Cannot call encode(Str: Any); none of these signatures match:␤ (Str $: Str:D $encoding = { ... }, Bool:D :$replacement, *%_)␤ (Str $: Str:D $encoding = { ... }, Str :$replacement, *%_)␤ in block <unit> at /tmp/TaoE8jtQuw line 1␤␤»
ilmari m: say <latin-1 utf-8 utf-16le utf-32le>.map: "\xf3".encode(*)
camelia rakudo-moar 9f7402: OUTPUT«Cannot call encode(Str: Whatever); none of these signatures match:␤ (Str $: Str:D $encoding = { ... }, Bool:D :$replacement, *%_)␤ (Str $: Str:D $encoding = { ... }, Str :$replacement, *%_)␤ in block <unit> at /tmp/vQm2u3juFC line 1␤␤»
ilmari m: say <latin-1 utf-8 utf-16le utf-32le>.map: { "\xf3".encode($_) }
camelia rakudo-moar 9f7402: OUTPUT«Unknown string encoding: 'utf-16le'␤ in block <unit> at /tmp/Uf7AQBsri9 line 1␤␤»
ilmari m: say <latin-1 utf-8 utf-16 utf-32>.map: { "\xf3".encode($_) }
camelia rakudo-moar 9f7402: OUTPUT«Unknown string encoding: 'utf32'␤ in block <unit> at /tmp/uQR5039n5x line 1␤␤»
ilmari m: say <latin-1 utf-8>.map: { "\xf3".encode($_) }
camelia rakudo-moar 9f7402: OUTPUT«(Blob[uint8]:0x<f3> utf8:0x<c3 b3>)␤»
ab6tract so rather than this: constant SRL_MAGIC_STRING_HIGHBIT is export(:Constants) = "=\x[F3]rl"; 14:36
i want: constant SRL_MAGIC_STRING is export(:Constants) = Buf.new('='.ord, 0xf3, 's'.ord, 'r'.ord, 'l'.ord);
?
ilmari or "=\x[F3]rl".encode('latin-1')
since codepoint == byte in latin1
ab6tract bam 14:37
thanks a lot!
ilmari you probably want Blob, not Buf, since it's constant 14:39
Buf is mutable, so someone ciould do SRL_MAGIC_STRING[0] = 0xff; # boom
ab6tract gotcha 14:40
looks like thats what encode returns anyway
and 'eq' is still a valid comparison operator between two Blobs? 14:41
or is there a more appropriate comparison op?
ilmari you want eq 14:42
== is numeric comparison, which compares the length 14:43
dalek ast: 6198907 | hoelzro++ | S (10 files):
Change X::Comp::AdHoc → X::Comp

We don't need to be *that* specific when checking compilation error types, because otherwise we are restricted in how far we can refine those types down the road
ast/6.c-errata: 409ee89 | hoelzro++ | S (10 files):
Change X::Comp::AdHoc → X::Comp

We don't need to be *that* specific when checking compilation error types, because otherwise we are restricted in how far we can refine those types down the road
ilmari m: say Blob.new(0x42) == Blob.new(0x37)
camelia rakudo-moar 9f7402: OUTPUT«True␤»
ilmari m: say Blob.new(0x42) == Blob.new(0x42, 0x37) 14:44
camelia rakudo-moar 9f7402: OUTPUT«False␤»
hoelzro I've made the change, and rakudo passes, but this concerns me: github.com/perl6/roast/blob/619890...isc.t#L317
dalek DBIish/oracle: 037c377 | (Salvador Ortiz)++ | lib/DB (9 files):
DBIish/oracle: Rationalize 'use' and 'need' in modules
DBIish/oracle:
DBIish/oracle: With module 'require' fixed, now DBDish.pm loads roles and
DBIish/oracle: other modules only needs DBDish.
hoelzro X::Comp on its own doesn't have a payload attribute; that's provided via X::AdHoc 14:44
perlpilot lizmat: on reflection, I think what was bothering me about your original allocate() example was the "flatness" of the args and the auto-repeat of the sequence of values. My brain wanted some separation between the size of allocation and the initialization and a way to tweak the initialization (what if I wanted 1,2,42,0,0,0,0,0,0,0,0 ? how would I do that? )
hoelzro so I'll probably go back later and revisit taht 14:44
perlpilot lizmat: anyway, I don't have any specific suggestion or "answer", those are just my thoughts. 14:45
perlpilot lizmat++ for allocate/reallocate btw :) 14:45
lizmat perlpilot: Blob.new(1,2,42, 0 xx 8)
m: dd Blob.new(1,2,42, 0 xx 8) # just checking 14:46
camelia rakudo-moar 9f7402: OUTPUT«Blob.new(1, 2, 42, 0, 0, 0, 0, 0, 0, 0, 0)␤»
perlpilot lizmat: then why do we need allocate when I can just say something like that anyway? 14:47
lizmat because it's relatively slow 14:48
a repeated list is nqp::spliced into the blob, which is much faster
Buf/Blob's slowness, specifically wrt to NC, has been a complaint that I heard a lot on the channel 14:51
dalek rakudo/more_ml_repl: 47bd65f | (Salvador Ortiz)++ | src/Perl6/Compiler.nqp:
rakudo/more_ml_repl: Multi-line REPL: Detect grammar's failed goals.
rakudo/more_ml_repl:
rakudo/more_ml_repl: That allows the following session:
rakudo/more_ml_repl: > my $a = <
hoelzro sortiz: ^ REPL changes, feel free to have a look 14:51
perlpilot aye
hoelzro now I need to hurry up and get ready for work o_O
sortiz hoelzro, Thanks! 14:52
perlpilot it's still bothersome in my head. Either I'll get used to it or I'll think of something "better" :)
lizmat m: dd Blob.allocate(10,42)
camelia rakudo-moar 9f7402: OUTPUT«Blob.new(42, 42, 42, 42, 42, 42, 42, 42, 42, 42)␤»
nine Is there really no way to prevent dalek from getting kicked?
lizmat m: my @int @a = ^10; dd Blob.allocate(20,@a)
camelia rakudo-moar 9f7402: OUTPUT«5===SORRY!5=== Error while compiling /tmp/LF_vOVnurt␤Two terms in a row␤at /tmp/LF_vOVnurt:1␤------> 3my @int7⏏5 @a = ^10; dd Blob.allocate(20,@a)␤ expecting any of:␤ infix␤ infix stopper␤ statement end␤ …» 14:53
lizmat m: my int @a = ^10; dd Blob.allocate(20,@a)
camelia rakudo-moar 9f7402: OUTPUT«Blob.new(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9)␤»
lizmat m: dd Blob.allocate(20,Blob.new(^10))
camelia rakudo-moar 9f7402: OUTPUT«Blob.new(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9)␤»
moritz would have expected the end to be zero-filled 14:54
perlpilot could .new be made to do this? Blob.new( :size(11), init(1,2,42), :repeat ) # something like this 14:57
dalek kudo/nom: 60a187d | lizmat++ | src/core/Buf.pm:
Introducing Buf:D.reallocate($elems)

Simple and straightforward without initialization options, as it's a Buf and you can do with you want with it.
perlpilot er, :init
lizmat wonders whether jnthn was aware he's also the pumpking now ? 14:58
nine This sounds like a reasonable strategy: github.com/sopel-irc/sopel/issues/...-162449543
lizmat perlpilot: the :init would be superfluous, as that's already any positionals that you pass 15:00
Blob.new(*@init, :size(x), :repeat) # comments anyone? 15:01
basically add 2 nameds to Blob.new
sortiz The oracle branch of DBIish is now travis-ci clean and can be merged, objections anyone?
moritz lizmat: somehow that feels like over-engineering. What's the use case for :repeat? 15:02
buharin hello :) 15:02
my friends
lizmat moritz: repeating a non-null pattern for debugging, for instance
buharin o/
moritz lizmat: is that debugging sufficiently performance sensitive that it deserves its own parameter+ 15:03
s/\+/?/
you could use (flat @list xx *)[^$size] or so
lizmat moritz: indeed you could, and it would be slow 15:04
Juerd I think it's a weird use case
perlpilot moritz: repeating values is quite useful ... If Blob.new(1, :size(20), :repeat) == PDL::ones(20) (ones and zeroes got used all the time when I was doing PDL stuff) 15:05
arnsholt :repeat would be similar to circular lists in Lisp, wouldn't it?
arnsholt An infinite list in finite space 15:05
moritz arnsholt: no, just initialization
lizmat arnsholt: except it would be limited to :size parameter
moritz perlpilot: that convinces me more
Juerd The one exceptional time you need something like this, you can also choose between a less performant list of integers or just reusing the buf, I think 15:06
lizmat Juerd: this was mostly about immutable blobs
Juerd perlpilot: I can see why you'd want to init the buf with a single repeated value, because you may want to match something that exists already (for example, eeproms are often shipped with every byte set to 0xff) 15:07
lizmat m: dd Buf.allocate(100,0xff)
camelia rakudo-moar 9f7402: OUTPUT«Buf.new(255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255, 255…»
lizmat m: dd Blob.allocate(10,0xff) 15:08
camelia rakudo-moar 9f7402: OUTPUT«Blob.new(255, 255, 255, 255, 255, 255, 255, 255, 255, 255)␤»
Juerd So perlpilot's example would be Buf.allocate(20, 1), right?
perlpilot Juerd: multiple repeating values can be used in convolutions (for example).
Juerd Or Blob
lizmat Juerd: yes
yes
Juerd perlpilot: I'm not saying it can't be useful
lizmat and since the "1" is repeating, 1,2,3 should also be repeating 15:09
moving this functionality to Blob.new with :size and :repeat optionals, I can live with 15:10
I'll just hold off from doing anything here until we have consensus
and if we don't have that by 6.c,1, I'll just rip it out again
.oO( wouldn't be the first time)
 15:11
dalek kudo/nom: c9d4650 | PerlJam++ | CREDITS:
update pumpking in CREDITS
 15:13
mst lizmat: got to keep your revert count up to stay ahead of everybody else's :D
moritz lizmat: fwiw perlpilot convinced me that :repeat probably isn't over-engineering 15:17
ab6tract maybe it's not worth it (since the language itself guarantees it), but i don't see a decent way to test a compile time error (in this particular case, a sub sig constraint error) 15:39
jnthn ab6tract: throws-like takes a string, I think...it the EVAL's it 15:41
masak m: use MONKEY-SEE-NO-EVAL; try EVAL q[sub foo($x, [$y, $z where * < 5]) { say $z }; foo(1, [2, 10])]; say "hi"; say $! 15:42
camelia rakudo-moar c9d465: OUTPUT«hi␤Constraint type check failed for parameter '$z' in sub-signature␤ in sub foo at EVAL_0 line 1␤ in block <unit> at EVAL_0 line 1␤ in block <unit> at /tmp/pEEhJjomKu line 1␤␤»
masak ab6tract: ^^
ab6tract ahh 15:43
thanks
masak wow, we've gone eval-paranoid recently...
all-uppercase *and* a pragma. :)
geekosaur it 15:44
it's eeeeeeeval
masak but I guess I shouldn't complain
'cus it seems every time I do, another layer of paranoia is added :P 15:45
I know some languages where eval isn't strictly curtailed, but more like a fundamental building block of the whole theory of the language...
along with another thing called `apply`, which we don't have in Perl 6 15:46
lizmat masak: that's 6.c, so around for ~ 2 months now ?
masak yes, that sounds right
ab6tract m: use Test; sub f { fail 'because' }; throws-like f(), Exception, message => 'because'; 15:51
camelia rakudo-moar c9d465: OUTPUT« 1..3␤Use of uninitialized value $desc of type Any in string context␤Any of .^name, .perl, .gist, or .say can stringify undefined things, if needed. in sub proclaim at /home/camelia/rakudo-m-inst-2/share/perl6/sources/C712FE6969F786C9380D643DF17E85…»
ab6tract any idea why proclaim is complaining here? 15:52
lizmat m: use Test; sub f { fail 'because' }; throws-like &f, Exception, message => 'because';
camelia rakudo-moar c9d465: OUTPUT« 1..3␤ ok 1 - code dies␤ ok 2 - right exception type (Exception)␤ ok 3 - .message matches because␤ok 1 - did we throws-like Exception?␤»
ab6tract beyond the obvious ;)
lizmat: thank you :D 15:53
lizmat ab6tract: f() vs &f :-)
ab6tract lizmat: hrrmmm...
then how do i test a sub that takes args? 15:54
lizmat m: use Test; sub f { fail 'because' }; throws-like { f() }, Exception, message => 'because';
camelia rakudo-moar c9d465: OUTPUT« 1..3␤ ok 1 - code dies␤ ok 2 - right exception type (Exception)␤ ok 3 - .message matches because␤ok 1 - did we throws-like Exception?␤» 15:54
ab6tract m: use Test; sub f($f) { say $f; fail 'because' }; throws-like &f('foo'), Exception, message => 'because';
camelia rakudo-moar c9d465: OUTPUT«foo␤ 1..3␤Use of uninitialized value $desc of type Any in string context␤Any of .^name, .perl, .gist, or .say can stringify undefined things, if needed. in sub proclaim at /home/camelia/rakudo-m-inst-2/share/perl6/sources/C712FE6969F786C9380D643D…»
ab6tract ok :) 15:55
lizmat ab6tract: put it in a block
aka Callable
lizmat afk for a bit& 15:56
ab6tract thanks everyone for the help 15:58
hoelzro is there a notion of anything like "test warnings" for Test.pm6? 16:02
I was thinking about that test I mentioned earlier, the throws-like 'X::Comp', ..., payload => '...';
hoelzro even though it works because the exception *happens* to have a payload method, I think it might be nice if throws-like checked X::Comp's method list and warned "hey, the exception has a payload method, but X::Comp doesn't. Something feels wrong" 16:03
dalek Iish/oracle: fb76c83 | (Salvador Ortiz)++ | t/ (3 files):
Fix the mysql problem, remove old tests
 16:25
dalek kudo-star-daily: 69fe09b | coke++ | log/ (9 files):
today (automated commit)
 16:30
dalek Heuristic branch merge: pushed 119 commits to DBIish by salortiz 16:38
MadcapJake what does it mean when a field in a C struct has a number in brackets? 16:41
e.g., «char reserved [32]» 16:42
sortiz It's a char array.
flussence sortiz++ # those Pg changes look really nice
MadcapJake oh ok, so 32 char slots? 16:43
psch i'd think about it as one char[32] slot vOv
MadcapJake o_O
sortiz MadcapJake, An inline array of 32 chars.
abraxxa who is salortiz on github?
geekosaur otherwise known as a string that can hold up to 31 chars (plus a terminating NUL) 16:44
MadcapJake how do i capture that with repr('CStruct')? CArray[int8] ?
sortiz abraxxa, I'm
abraxxa sortiz: thanks for the Oracle branch fixes and merging!
sortiz++ 16:45
nine jnthn: what do you think about adding "changing a CORE.setting class into a role" to the ok list in the versioning guidelines? I can't find any down sides, but you may have interesting insights there :)
MadcapJake CStruct doesn't recognize CArray for some reason
wait nvm 16:46
abraxxa sortiz: does that mean I can delete my fork on github and work directly in the DBIish repo (in a branch) from now on?
sortiz abraxxa, two misplaced $sth.finish in Test/DBDish drove my crazy! 16:47
abraxxa, Yes, sure.
MadcapJake CStruct is giving me a compilation error about the types I'm using but I only have 4 int64's and one CArray[int8] 16:49
abraxxa sortiz: I guess you can also delete the oracle branch 16:49
japhb (DBIish committers)++ # Lots of DBIish love of late, thank you all!
nine
.oO(20 years from now people are gonna wonder what's so "ish" about DBI)
 16:50
sortiz MadcapJake, need to see the case.
abraxxa, no yet. :) 16:51
MadcapJake sortiz, pastebin.com/QXk6ZmxH 16:52
abraxxa nine: MY plan is to use DBIish as playground for A DBI (whatever it will be called) 16:54
perlpilot MadcapJake: it's those where clauses 16:55
MadcapJake ohh, can't use those on cstruct fields eh? 16:56
abraxxa have to go now, bye!
sortiz MadcapJake, indeed no. :( 16:59
nine .tell abraxxa Makeshift solutions have the tendency to stick around much longer than people think when they implement them. Imagine people starting to use Perl 6 more and more, DBIish becoming better and better and the "real" DBI taking another 3-4 years to become usable. By then it will have a hard time catching up with a well established DBIish. 17:01
yoleaux nine: I'll pass your message to abraxxa.
lizmat perlpilot moritz: thinking more on Blob.new( :size, :repeat)
:repeat could also be a number, but then what does :size mean ? 17:02
flussence is cleaning up old github permissions I haven't used in years, apologies if it notifyspams anyone 17:04
perlpilot what would :repeat mean as a number (to the Blob)? repeat N times?
lizmat :5repeat without :size would just essentially be () xx 5 as initializer (but faster) 17:06
Blob.new((1,2,3) xx 5) vs Blob.new(1,2,3, :5repeat ) 17:07
it still feels better to me to have the special functionality in a separate allocate method, rather than trying to cram all functionality into .new 17:08
sortiz lizmat, totally agree.
TreyHarris huh, → isn't a valid synonym for -> ? Was it at one time (I'm sure it was; I mean, "was it for any length of time longer than it took to break something?"), or am I just dreaming? 17:12
psch .u right arrow
yoleaux U+02C3 MODIFIER LETTER RIGHT ARROWHEAD [Sk] (˃)
U+02F2 MODIFIER LETTER LOW RIGHT ARROWHEAD [Sk] (˲)
U+034D COMBINING LEFT RIGHT ARROW BELOW [Mn] (◌͍)
perlpilot lizmat: I can live with that I guess. (it kinda feels like .allocate will be used more than .new though) 17:13
skids → is the one that matches the texas under linux compose character.
TreyHarris thought I could be clever and use my Haskell texas to unicode bindings for Perl 6 as well
but every time I type -> I get → and that breaks 17:14
psch yeah, i was just wondering how many there are, but it's apparently too many for yoleaux, in any case
perlpilot .u →
yoleaux U+2192 RIGHTWARDS ARROW [Sm] (→)
psch but no, we only have '->' (and '<->') in token lambda currently 17:16
nine CompUnit::PrecompilationStore @stores = Array[CompUnit::PrecompilationStore].new( $precomp.store ),
I wonder if this ^^^ is really worth the type safety
perlpilot lizmat: though, would you change .allocate to be like .allocate(20, :init(1,2,42), :repeat) ? 17:17
lizmat again, why the :init, if Blob.new doesn't have that ? 17:20
also, I feel that :repeat should be the default, don't you think?
TreyHarris psch: yep, see that. <-> is a synonym for ->, or does it have a different function I can't immediately recall? 17:21
lizmat and having a named default to true is also a design smell, is it not?
RabidGravy sortiz++ # nice one on fixing up the DBIish
perlpilot lizmat: all I know is .allocate(11,1,2,42) feels wrong. What it does is slightly surprising. (i.e. same as Blob.new(1,2,42,1,2,42,1,2,42,1,2) ) 17:23
RabidGravy TreyHarris, <-> makes the parameters rw 17:24
lizmat m: dd Blob.allocate(10, (1,2,3)) # perhaps remove the slurpy candidate, so you would have to specify a list ? 17:25
camelia rakudo-moar c9d465: OUTPUT«Blob.new(1, 2, 3, 1, 2, 3, 1, 2, 3, 1)␤»
lizmat perlpilot: ^^
TreyHarris RabidGravy: oh, right, S06. forgot about that. so it wouldn't be as simple as just adding → to that token's alternation. 17:26
skids Maybe just stick with .allocate(:size, :ival) and leave the case of a repeating multivalue pattern to be sped up by future optimization?
I see less of a use case for filling with a pattern. 17:27
lizmat "Blob.new(0 xx 10000) vs Blob.new(:initial-elems(10000)) -> 200.2908/s vs 38052.1614/s" # I think we want it now ?
ugexe what about :contents(1,2,3)? there is already a Buf.contents method
perlpilot lizmat: that helps 17:27
dalek kudo/nom: 846d513 | lizmat++ | src/core/Buf.pm:
Remove the slurpy Blob.allocate candidate

If you want to initialize with a multi-byte pattern, you will need to specify it as a list, so:
   Blob.allocate(100, (1,2,3))
versus:
   Blob.allocate(100, 1,2,3)		 17:29
lizmat ugexe: we need something that can work on immutable blobs 17:30
perlpilot lizmat++
skids yes lizmat++
TreyHarris RabidGravy: would it be as simple as splitting it up into two sym tokens with token lambda:sym«->» { '->' | '→' } and token lambda:sym«<->» { '<->' | '↔' }, and then... hrm, I'm not seeing where the distinction between the two lambdas is made in nqp, it isn't in NQP/Actions.nqp where I'd expect it... 17:31
sortiz lizmat++ # consensus 17:31
TreyHarris (Sorry, I'm looking for something that will help me get up to speed and this seems like a very small bite to take a stab at.)
RabidGravy TreyHarris, I'm almost certainly not the right person to be asking about this as I try to avoid that part of the code ;-) 17:32
ugexe m: say Blob.new(0 xx 5).contents; # right, but the value of .contents here seems to match well enough to what you want to pass in to be repeated (.new(5, :contents(0))
camelia rakudo-moar c9d465: OUTPUT«(0 0 0 0 0)␤»
lizmat likes to reach consensus by removing code :-)
but not too much :-)
TreyHarris RabidGravy: understood. 17:34
can anyone else give me a pointer? or, if adding a digraph synonym is a lot harder than it looks (my fruitless attempt just now to find where texas pointies are suggests it may be), tell me this is not a good newbie project? 17:35
I haven't made any Perl 6 commits since Pugs days, and then I mostly just wrote tons of tests. 17:37
flussence adding bits to the core of the language is going to be much harder than, say, adding unicode multis for normal infix operators. definitely not an easy project to start off with 17:40
TreyHarris flussence: okay. it was trivial with Pugs (I need to clean out my ~src/p6 one of these days!), so I didn't know 17:42
i almost always learn by scratching my own itch, so I'll continue with my userland project and see if and when the next itch scratches. er... something like that. 17:43
bazzaar \o good evening all 17:45
lizmat bazzaar o/ 17:46
bazzaar perl6 really is a pleasure to utilise :) 17:48
bazzaar spurt ('testfile',@aa) .... is it possible to control the output delimiter? set it to "\n"? 17:51
ugexe @aa.join("\n") 17:52
if you mean set the newline character you can do that too, but you do that on a string, not an array 17:53
psch bazzaar: you can specify the output delimiter on a Handle with nl-out, see doc.perl6.org/routine/put#role_IO
eh
that's not what i wanted to link, the docs for IO::Handle explain it differently...
oh, no, it's method put on IO::Handle 17:54
doc.perl6.org/type/IO::Handle here
psch well. looks like i got a bit confused there. the docs for sub put state it uses $handle.nl-out, on the page for IO::Handle 17:55
psch ah, that's the same text as for put in class Mu... 17:55
dalek kudo/nom: 2afe79f | lizmat++ | src/core/Buf.pm:
Remove some superfluous returns
 17:56
MadcapJake can you create typedefs with NativeCall? 18:00
DrForr Structs and unions, yes. 18:01
timotimo you can create typedefs even without nativecall
MadcapJake what about for a callback?
timotimo you may have to spell it different from inside a routine, but it should work 18:02
MadcapJake how do you mean?
DrForr Too bad there isn't a project out there that creates a complex struct and returns a linked list of them from a callback.... Oh, wait, there is :) I do that from Inline::Scheme::Guile :)
timotimo m: sub foobar(&bar:(int32, int32)) { } 18:03
camelia ( no output )
timotimo m: sub foobar(&bar:(int32, int32)) { }; say &foobar.signature[0].perl
camelia rakudo-moar 846d51: OUTPUT«:(&bar where { ... })␤»
timotimo m: sub foobar(&bar:(int32, int32)) { }; say &foobar.signature[0].arguments[0].perl
camelia rakudo-moar 846d51: OUTPUT«Method 'arguments' not found for invocant of class 'Signature'␤ in block <unit> at /tmp/f2L7dzQkJt line 1␤␤»
timotimo what's it called
m: sub foobar(&bar:(int32, int32)) { }; say &foobar.signature[0].parameters[0].perl
camelia rakudo-moar 846d51: OUTPUT«Method 'parameters' not found for invocant of class 'Signature'␤ in block <unit> at /tmp/bF65PlfIdF line 1␤␤»
timotimo m: sub foobar(&bar:(int32, int32)) { }; say &foobar.signature[0].params[0].perl
camelia rakudo-moar 846d51: OUTPUT«&bar where { ... }␤»
timotimo huh, it's a where clause? i must have written this wrong, then
bazzaar ugexe : and psch : thankyou both for helping me. The documentation is really well laid out, I shall have to study it. usgexe your solution using .join is what I need I think. 18:04
psch timotimo: i think it translated that to a constraint on the sig for the &-param
timotimo could be
what's the correct spelling? i haven't done this in ages :)
psch timotimo: 'cause we don't have subtypes for Signatures with specific parameters
(which i think is perfectly fine... :) ) 18:05
nine timotimo: &bar (int32, int32) 18:06
The space is important
psch m: sub f(&b:(Int, Str)) { say "k" }; f( sub (Int $, Str $) {}); f( sub (Str $, Str $) {} )
camelia rakudo-moar 846d51: OUTPUT«k␤Constraint type check failed for parameter '&b'␤ in sub f at /tmp/1sfL_Dh66g line 1␤ in block <unit> at /tmp/1sfL_Dh66g line 1␤␤»
psch isn't the space for unpacking?
e.g. @a ($head, @tail) 18:07
MadcapJake so if I take foobar and place it in the signature of a native sub, it will "unbox" to the signature inside foobar?
psch isn't sure how all that interacts with NativeCall though vOv 18:08
nine psch: oh, indeed!
Intriguingly it still worked with NativeCall 18:09
lizmat m: sub byte() { ... }; byte() # warning appears inappropriate 18:11
camelia rakudo-moar 846d51: OUTPUT«WARNINGS for /tmp/_bsOqb4TPE:␤Useless use of constant value byte() in sink context (line 1)␤»
lizmat and it should die because of the yadayadayada?
psch m: sub Int() {...}; Int()
camelia rakudo-moar 2afe79: OUTPUT«WARNINGS for /tmp/JR7oQkRElb:␤Useless use of constant value Int() in sink context (line 1)␤»
lizmat I guess it's a case of DIHWIDT 18:12
timotimo what i was meaning is that you may have to spell it like "constant ThatCallback = Routine[Foo, (Bar, Baz)] or something
bazzaar spurt "testfile", @aa.join("\n") ; ...... this works a charm, would it be worth adding to the documentation of spurt? 18:16
FROGGS o/ 18:18
psch \o
jnthn nine: Well, it would work in so far as role punning means .new and inheritance still work. Introspection might look a bit different, though. And, of course, no adding requirements (... methods). 18:19
nine jnthn: introspection always means MOP, doesn't it? And the MOP is implementation specific. 18:20
psch public static SixModelObject findmethod(SixModelObject invocant, String name, ThreadContext tc) { 18:23
public static SixModelObject findmethod(ThreadContext tc, SixModelObject invocant, String name) {
nine Though I wonder how well we could keep up the latter if an alternative Perl 6 implementation came along.
psch that seems a bit... weird
i mean, i get that two candidates might make sense, but differntiating them by *order* of identical parameters..? :/
jnthn nine: I suspect the MOP will get increasingly test-covered over time. 18:28
lizmat nine: making Baggy inherit from typed hashes was also considered a nono 18:29
because of changed introspection
jnthn nine: Which type are you thinking of changing that way? 18:32
nine Distribution
jnthn Hm
I'd agree the risk of problems on that one is quite low. :) But curious about the rationale? 18:33
nine Not exactly one that's gonna be used in user code all that often :)
ugexe++ suggested having repository specific implementations of Distribution for better code structuring and better composability. 18:34
jnthn I think I can see it... :) 18:35
nine So a CompUnit::Repository::Installation::Distribution would know how to find its resources, source files, etc. and we'd no longer have to pass them separatly to $repo.install 18:35
Or rather that you could pass a Distribution::Tar/Distribution::GitHub or whatever object to .install while $comp-unit.distribution would return the CompUnit::Repository::Installation::Distribution object 18:37
jnthn *nod*
No objections.
nine \o/
lizmat fwiw, +1 from me too 18:38
jnthn bbl 18:39
[Coke] so, I may have jumped the gun putting jnthn down as pumpking - we don't really have a defined role for that in the project, but I was thinking of it as a combo of his role as lead dev and generator of a lot of the release docs that we're now trying to follow. so, if we end up changing that line or putting out a clarification or whatever, there wasn't a coup or anything. :) 18:43
jnthn++ pmichaud++
lizmat [Coke]++ # moving forward on the issue :-) 18:46
if I look at the role of the P5 pumpking, it is much more a manager's role in the background 18:51
[Coke]: one, I might add, has been played a lot by you already 18:52
El_Che with the customary "herding cats" experience in the resumé :)
dalek kudo/nom: 4f33801 | lizmat++ | src/core/ (3 files):
No longer show module compile errors twice

Fixes RT #127176
This is done by adding a VERBATIM-EXCEPTION capability to Exception, by means of a special Rakudo::Internals.VERBATIM-EXCEPTION semaphore which is set when precompilation dies and we need to exit		 18:56
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=127176
nine lizmat: what do we do with STDERR output when precomp succeeds? 18:59
lizmat nine: hmmm.. like warnings? hmmm... good point, they get eaten now 19:00
I guess the easiest solution would be to note() them ? 19:01
nine Think so, yes.
lizmat testing... 19:02
yeah, seems to do the trick, spectesting for issues 19:04
nine lizmat++ 19:05
lizmat hmmm... note is adding an extra \n .... $*ERR.print it is then :-) 19:08
dalek kudo/nom: ac88669 | lizmat++ | src/core/CompUnit/PrecompilationRepository.pm:
Make sure we don't eat warnings during precomp

Also, we don't need the else, as we're dying in the if before		 19:16
lizmat hmmm... I wonder if it would make sense to make Blob.list an iterator, rather than copying the whole Blob right away 19:34
sortiz I think it's a good idea. 19:36
lizmat, Especially by the saving in memory. 19:38
timotimo well, it's also about semantics 19:41
what happens when you iterate over the buf and change it at the same time
do you want to iterate over the new values when you reach them ("lazy" semantics) or do you want the original values to come out of the iterator? 19:42
[Coke] timotimo: didn't we already decide that during the GLR? 19:45
timotimo well, for arrays perhaps 19:46
potentially not for bufs :)
sortiz A more interesting question to me is, can I acquire a lock over an object for the current thread? 19:48
lizmat what would that give you ?
do you want to bar access from other threads ? 19:49
sortiz Yes, ban other threads for atomic changes, for example.
jnthn sortiz: OO::Monitor 19:50
jnthn Uh, OO::Monitors :) 19:51
You can't "ban other threads" with a lock
sortiz jnthn, I knew there was an answer! Thank you.
jnthn Locks require everyone to play ball :)
sortiz Sure, cooperative lock.
jnthn So, structure your code so that all access to the data goes through some point of control.
OO::Monitors is one way to achieve that. 19:52
timotimo well, you can "ban other threads" in general
you just have to turn off interrupts on the CPU
jnthn (You know that only one thread will ever be inside any of the object's methods at a time)
sortiz There is much to study ahead 19:54
jnthn Generally, you can get a fair way with concurrent design by doing OO design really well :) 19:56
(Following "tell, don't ask" and the like)
jnthn The other major strategy I tend to turn to being finding a way to look at the information as an event stream, then using supplies. 19:57
sortiz I understand the concepts. 19:58
sortiz Unfortunately the user documentation is incomplete, so I resort to the code. 20:01
jnthn sortiz: For what in particular? 20:02
If you're not already and have time, please file issues on doc.perl6.org for things you miss
skids gist.github.com/skids/fabd0304f40e24db33d0 is my list of Supply questions 20:03
sortiz Just an example: Yesterday I *discover* that 'is native' methods are possible. 20:05
And works like a charm! 20:06
sortiz jnthn, When I have time, I prefer fill the blanks ;) 20:07
*to fill
timotimo sortiz: you seem to think supplies are a method to introduce parallelism
they never introduce additional parallelism. that's their design, really
so different taps on some supply should not run in parallel 20:08
sjn \o
timotimo o/ sjn
sjn tries to search doc.perl.org for docs on ENV 20:09
getting no hits :-\
is there a way to fix that?
skids timotimo: were you actually referring to my gist? 20:10
moritz sjn: lots of ways to fix that, all of them involve quite some work :/
sortiz timotimo, What makes you think that? I haven't mentioned Supplies. 20:11
mst sjn: try goolging with 'site:' ?
sortiz %*ENV is mentioned in doc.perl6.org/language/variables#Dy..._variables 20:12
sjn mm, found it
wish it was simpler to find
sjn munches some cheese with his whine 20:13
jnthn skids: Writing you some answers. 20:14
lizmat
.oO( wite whine ?)
 20:15
lizmat m: my $b = Buf.new(1,2,3); for @$b { $_ = 42 } # so, do we want this to work for Buf? 20:21
camelia rakudo-moar ac8866: OUTPUT«Cannot assign to an immutable value␤ in block <unit> at /tmp/YAdS1J2uCk line 1␤␤»
timotimo sortiz: sorry, i meant skids
sortiz While installing OO::Monitors: ==> Testing OO::Monitors 20:23
t/basic.t ...... ok
Weird node visited: QAST::BVal
lizmat jnthn: my $b = Buf.new(1,2,3); for @$b { $_ = 42 } # do we want this to work ?
timotimo lizmat: the default is to have <-> $_, right?s o i'd expect this to work 20:25
lizmat ok, I'll see if I can make that work
not so for Blob, right ?
timotimo sounds like you want atposref_* for this?
no, not for blob
skids timotimo: Personally I'm more interested in managing parallel events than having blocking automatically added to my code -- I realize this is not the common use case. The serial supply changes left me wondering how to go about preserving the API for non-serial supplies since every api call transforms to serials. In any case it all needs doccing, so I'll unstash my S17 patch and look at jnthn++'s answers and do what I can.
jnthn sortiz: I'm aware of the warning; it's harmless, I think 20:26
timotimo skids: OK, i see you're on top of things :)
lizmat timotimo: looks like it
timotimo BBIAB
sortiz Building blocks like OO::Monitors deserve to be in core, IMO. 20:29
lizmat sortiz jnthn warnings confirmed 20:30
jnthn sortiz: There are dozens of such things I could make that we'd be able to classify as "building blocks" in the same way, though. 20:34
sortiz jnthn, They should at least be mentioned in a prominent place :) 20:35
Hardly I had found on my own. 20:36
jnthn sortiz: Yeah. When I wrote a .Net concurrency course, I made a few decision trees for folks.
sortiz: Maybe I should try and port them to Perl 6 :)
sortiz Well, by the time I leave. DBIish robbed me several hours of sleep ;) 20:40
o/ #perl6
lizmat sortiz o/ 20:41
jnthn o/ 20:44
jnthn skids: gist.github.com/jnthn/04999dd3af5c9df91a17 20:48
Probably a worthwhile read for anyone else interested in concurrency stuff too
skids Thanks jnthn++ 20:50
Hotkeys jnthn++
DrForr I'm having trouble tracing an error through a callback. 21:01
perigrin use thinner onion paper.
DrForr There's nothing useful being reported after 'Failed'. It does report the line where the function was un, but nothing beyond that. And eliminating the callback is not an option. 21:04
timotimo callbacks are really problematic, yeah 21:14
you mean NativeCall, right?
DrForr No, this is Crust.
timotimo oh 21:18
DrForr I'll worry about it tomorrow, I suppose. 21:21
dalek kudo/nom: f25d41c | lizmat++ | src/core/ (2 files):
Streamline Blob|Buf.list

  - use iterator, so that we don't need to copy the buffer when iterating
   element by element
  - split into Blob/Buf candidates
  - Buf candidate passes rw value when iterating ($_++ for @$buf works)
  - add BlobbyIterator role, for general usage in other situations		 21:22
FROGGS lizmat++ 21:24
lizmat enough hacking for today, good night #perl6! 21:27
FROGGS gnight lizmat 21:28
Hotkeys I've never really thought to ask: What exactly is 'sink context' 21:44
timotimo it's when something that produces a value doesn't get its value put into anything 21:45
it's mostly important for lazy lists and stuff that we want to have evaluated for side-effects
Hotkeys ah 21:47
why is it called sink context?
mst people often use source/sink to mean producer and consumer of values
moritz because the returned value just disappears into the kitchen sink
mst in this case it's a context that basically consumes the value itself 21:48
Hotkeys ah 21:49
okay, thanks
mst reasonably mnemonic once you understand it, I think 21:50
(in that I was basically able to reverse engineer it in my head just now :)
skids The etymology probably comes via electrical engineering where sinking circuits provide a return to ground. Of course, thanks to Ben Franklin, ground is actually the *source* of electrons, so we just have to pretend everything is made out of aluminum or other hole carriers to feel good about it. 22:04
(Or build all our circuits upside down with a positive ground) 22:06
skids decommute 22:07
Skarsnik sortiz, hm, why did you put Real and Num in Pg? I think the idea was to have all float in Rat x) 22:10
MadcapJake how would I represent a C struct field that holds a function? 22:15
Skarsnik Pointer 22:22
you can't call it, it's not implemented
RabidGravy MadcapJake, if you want to call them then you need to make a tiny C wrapper which you pass the pointer and the rest of the arguments to actually execute it 22:49

Powered by Cro and the Raku® Programming Language.

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

Thank you!