»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.
00:00 labster joined 00:02 RabidGravy left 00:05 pdcawley_ joined 00:07 pdcawley left 00:08 woodruffw left, woodruffw joined, woodruffw left 00:09 tmch left, woodruffw joined, tmch joined 00:11 perlawhirl left 00:14 ggoebel16 left, woodruffw left
dalek kudo/nom: b243a96 | timotimo++ | src/Perl6/Metamodel/ (2 files):
"setdebugtypename" lives in metamodel Naming now instead of ClassHOW
 00:15
00:26 SCHAAP137 left 00:27 BenGoldberg joined
rudi_s sortiz: Sorry for the late reply. Awesome, thank you. our works just fine. 00:30
Btw. is there a way to ignore the "is export" from a module. Like perl5's use Foo::Bar (); ?
00:30 musca` is now known as musca 00:34 rindolf left, pdcawley joined
sortiz rudi_s, That is the difference between 'use' and 'need': 'use' imply 'need' *and* 'import' 00:35
rudi_s sortiz: Just to be sure. need Foo::Bar in perl6 works like use Foo::Bar () in perl5? 00:36
sortiz rudi_s, Exactly. 00:37
00:37 pdcawley_ left
rudi_s sortiz: Perfect, thank you. - I didn't realize this when reading the docs and was confused that such an important could be missing, but now it's pretty clear (and I've no idea how I could miss that). 00:39
sortiz rudi_s, yw
00:41 spider-mario left 00:45 cbk joined, ggoebel16 joined 00:48 dfcarpenterak left 00:53 skids joined 00:59 dfcarpenterak joined 01:06 labster left, yeahnoob joined 01:07 dfcarpenterak left 01:09 labster joined
timotimo i think it'd probably be easier to write the "dump all QAST nodes to json" as a single method on QAST::Node that knows all about the individual nodes we have and such 01:18
rather than splitting it out into multiple methods like the rest of the code already does, with its "dump extra info" and such
individual nodes, but especially individual roles
which is, i assume, where the majority of the complexity comes from
01:29 yeahnoob left 01:31 yeahnoob joined, ZhangGong|2 joined, yeahnoob left 01:42 justine joined 01:52 kid51 left 01:53 Actualeyes joined 01:59 snarkyboojum joined 02:05 snarkyboojum left 02:07 kalkin-_ joined 02:11 kalkin- left 02:24 vendethiel joined 02:31 arcetera left 02:32 arc__ joined 02:33 arc__ is now known as arcetera
zostay m: my $c = -1; (4 * [+] (1, 3 ... *)[0 .. 100_000].map({ ($c = -$c) / $_ })).say; # happy approximately pi day 02:33
camelia rakudo-moar b243a9: OUTPUT«3.14160265348972␤»
02:33 arcetera left, arcetera joined 02:34 arcetera left 02:35 arc__ joined, arc__ is now known as arcetera, arcetera left, arcetera joined 02:45 ilbot3 left 02:47 ilbot3 joined
MadcapJake` how can I do the FILE C type in nativecall? 02:47
02:49 vendethiel left
BenGoldberg class FILE is repr('CPointer') { }; 02:52
02:52 TimToady left 02:53 TimToady joined
BenGoldberg Note that in C, FILE is a struct, and you'd be using pointer-to-FILE all over the place, whereas 'class FILE ...' creates a class which behaves like a pointer... it's more like a c-level pointer-to-FILE. 02:54
02:57 BenGoldberg left 03:03 toddr joined
toddr Hi, so perl6 doesn't have . in INC right? 03:04
03:07 toddr left
skids not by default 03:09
Juerd I keep wishing that . or ./lib was in there
Or at least ~/.lib
(Or something like it)
skids Better than having things hijacked because you didn't notice a .pm file in ./ IMO. 03:11
03:14 toddr joined
Juerd I guess 03:15
But what would be the reason not to have any homedir based thing in there by default?
toddr I missed the answer when my IRC client crashed didn't I? 03:16
Juerd Yes. The answer was yes.
Or, actually: <skids> not by default
There's PERL6LIB=. or perl6 -I. or use lib '.'; 03:17
toddr So I'm putting a presentation together. I'm wondering if there's any documentation on why this decision was made
Juerd <skids> Better than having things hijacked because you didn't notice a .pm file in ./ IMO.
MadcapJake` is that how it works in Perl 5?
toddr don't get me wrong. I applaud it.
skids I could see a subdir of homedir. Probably just NYI given the state of package installers, different systems have different conventions e.g. ~/.local 03:18
geekosaur yes, perl5 doesn't include . in @INC
toddr ?
geekosaur or ~ or any subdir thereof, by default
toddr no . is at the end of @INC in perl5
Juerd MadcapJake`: No, . is in @INC in Perl 5.
geekosaur sighs, too tired... 03:19
skids It is on my system.
Juerd MadcapJake`: And metacpan.org/pod/local::lib is sort-of popular. I wish something somewhere in ~ was in Perl6's include paths by default 03:20
Speaking of this. What is that array called in Perl 6?
MadcapJake` I'm guessing it has to do with CUR 03:21
$*REPO?
skids hrm. So that's why I couldn't find the precomps in ~/.local where I thought I put them with --prefix. Hardcoded to ~/.perl6 apparently or something. 03:23
MadcapJake` Seeing as technically modules are located in a separate precomped/mangled directory, it's more about maintaining true cross-platform unicode support
yeah precomps does not follow a make --prefix flag 03:24
skids is there a corresponding flag?
MadcapJake` skids, github.com/rakudo/rakudo/blob/nom/...ompilation 03:25
skids huh. But my hacked-up rakudobrew should be setting NQPs --prefix as well. 03:29
MadcapJake` --prefix wasn't properly passing to nqp until recently, i believe 03:30
skids nqp --show-config | grep prefix shows moar::prefix=/home/me/.local, but my precomps are still in ~/.perl6. 03:31
03:33 lostinfog left
MadcapJake` prefix only handles installs, precomp doesn't honor it 03:34
nqp dir only deals with prefix inside Configure.pl (iiuc) 03:35
03:35 geekosaur left
MadcapJake` I think the only way you can adjust precomp location is with RAKUDO_PREFIX 03:36
skids Then I don't get github.com/rakudo/rakudo/blob/nom/...try.pm#L69 03:37
03:37 geekosaur joined 03:38 cognominal left, noganex_ joined
MadcapJake` maybe RAKUDO_PREFIX is being set somewhere? I'm really an NQP noob but a glance at Configure.pl doesn't really show prefix going anywhere special (must be wrong though...) 03:39
03:39 cognominal joined
skids well, it's not in my shell's env. Too tired to figure this out now though. 03:40
timotimo github.com/rakudo/rakudo/blob/nom/...Config.nqp - could this be interesting?
MadcapJake` maybe it's this: github.com/perl6/nqp/blob/master/C...ure.pl#L93 03:41
03:41 noganex left
MadcapJake` yeah i think it's that and then it's stored here: github.com/perl6/nqp/blob/master/C...#L235-L239 03:43
(for MoarVM)
oi, makefiles are totally brick walls for me :P 03:44
03:44 toddr left
MadcapJake` honestly it looks like prefix is just used there to install to the right location, I'm not really understanding how nqp::backendconfig attains its data 03:45
skids It looks like the 'home' repo doesn't have a configurable to alter it. 03:50
03:52 cognominal left, buharin joined 03:53 cognominal joined 03:54 justine left 04:06 cpage_ left 04:07 cpage_ joined 04:28 wamba joined 04:29 skids left 05:09 buharin left 05:10 yqt left 05:12 khw left, buharin joined
saaki is there a nicer way to write the following? (2..22).grep({not $_ % 6}).say 05:31
basically want to skip values in a range 05:33
05:43 cpage_ left
Hotkeys m: say (2..22).grep(*!%%6) 05:43
camelia rakudo-moar b243a9: OUTPUT«(2 3 4 5 7 8 9 10 11 13 14 15 16 17 19 20 21 22)␤»
Hotkeys Like that saaki ?
m: say (2..22).grep(*!%6) 05:44
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/NXsrKJ38xo␤Cannot negate % because multiplicative operators are not iffy enough␤at /tmp/NXsrKJ38xo:1␤------> 3say (2..22).grep(*!7⏏5%6)␤ expecting any of:␤ infix␤ infix stoppe…»
Hotkeys Okay
So yeah the first one
I assume
05:45 cpage_ joined 05:46 Cabanossi left
saaki p6: (2..22).grep({not $_ % 6}).say 05:46
camelia rakudo-moar b243a9: OUTPUT«(6 12 18)␤»
05:47 tmch left
saaki Hotkeys: not the right sequence :| 05:48
05:50 Cabanossi joined
go|dfish p6: (2..22).grep(* %% 6).say 05:50
camelia rakudo-moar b243a9: OUTPUT«(6 12 18)␤»
saaki let's say i want values in increments of 1_000_000, just doesn't make sense to use grep on a sequential range. guess the range class doesn't have a provision for skipping 05:51
maybe a gather-for-take i guess 05:53
05:55 adeeb joined
adeeb Hello guys I am adeeb. I need help in extracting email address along with first name and lastname using perl software from any given website. I would be grateful to y99ou 05:57
05:57 ZhangGong|2 left 05:58 darutoko joined
Hotkeys saaki: oh 05:58
saaki i gave a bad example on the range side since i don't care about 2
jdv79 adeeb: maybe p5 is better at this task. i think there may be some cpan modules to so stuff like that already. 05:59
Hotkeys m: say (6, 12 ... 100) # all multiples of 6 to 100
camelia rakudo-moar b243a9: OUTPUT«(6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96)␤»
Hotkeys Like that?
jdv79 s/so/do/
Hotkeys if you want more general
adeeb thank you can you guide me a link 06:00
jdv79 ask in #perl
saaki Hotkeys: perfect!
Hotkeys m: say {($^a, * + $a ... $^b)}(1_000_000, 10_000_000) 06:01
camelia rakudo-moar b243a9: OUTPUT«(1000000 2000000 3000000 4000000 5000000 6000000 7000000 8000000 9000000 10000000)␤»
Hotkeys There's essentially an anonymous function for multiples of a up to b
06:01 adeeb left, ZhangGong|2 joined
Hotkeys If you want the first n numbers 06:02
saaki Hotkeys: cool. also, i missed out on the *
Hotkeys m: (6, 12 ... *)[^10] 06:03
camelia ( no output )
Hotkeys er
saaki Inf
?
Hotkeys m: say (6, 12 ... *)[^10]
camelia rakudo-moar b243a9: OUTPUT«(6 12 18 24 30 36 42 48 54 60)␤»
Hotkeys First 10
* is hard to explain
saaki i see
Hotkeys It means 'whatever'
And does different things in different contexts
Like here's the Fibonacci sequence 06:04
m: say (0, 1, * + * ... 100)
camelia rakudo-moar b243a9: OUTPUT«(0 1 1 2 3 5 8 13 21 34 55 89 144 233 377 610 987 1597 2584 4181 6765 10946 17711 28657 46368 75025 121393 196418 317811 514229 832040 1346269 2178309 3524578 5702887 9227465 14930352 24157817 39088169 63245986 102334155 165580141 267914296 433494437 70140…»
Hotkeys Well I did that poorly
But you get the gist
saaki yes, thanks 06:05
Hotkeys In a sequence like that * + * acts like $^a + $^b
But at the end of a sequence it basically acts as inf
Unless you make it a conditional
like ?p6> say (1,2,4 ... * > 30) 06:06
Er
Wrong command
And no new line
But
It does what it looks like it does
saaki can't seem to find those uses of * in docs 06:07
Hotkeys design.perl6.org/S02.html#The_Whatever_Object 06:08
Here's some from the spec
saaki really like the docs for the most part, just tough to find some of these cool little patterns there 06:09
Hotkeys yeah 06:10
the docs aren't perfect but they'll improve with time
jdv79 you could grep the setting for * and Whatever maybe 06:11
Hotkeys Anything you can't find in the docs you can probably find in the specs with a little elbow grease
saaki have managed to dig up some things from design.* via google, bit surprising sometimes. should probably just read them through. 06:12
Hotkeys lol
be aware not everything in the spec is implemented yet
but whatevers implemented is probably in the spec
if that makes sense :p 06:13
saaki it does
would like to contribute, seems nqp needs some work 06:17
06:31 billj joined
billj p6: say 3; 06:32
camelia rakudo-moar b243a9: OUTPUT«3␤»
saaki may resolve to docos
06:35 billj left 06:51 jack_rabbit joined 06:57 ufobat joined 07:02 domidumont joined 07:05 jack_rabbit left 07:06 domidumont left 07:07 domidumont joined 07:09 FROGGS joined 07:23 sjoshi joined 07:26 abaugher left 07:27 abaugher joined 07:29 jack_rabbit joined 07:33 ChoHag_ is now known as ChoHag 07:35 CIAvash joined 07:36 atweiden joined 07:38 firstdayonthejob joined 07:39 nakiro joined 07:41 st_iron joined 07:42 el_amigo joined, st_iron left
masak morning, #perl6 07:43
atweiden in rakudo source, in the case of `multi sub infix:<eqv>(FatRat, FatRat)`
(github.com/rakudo/rakudo/blob/b243...ic.pm#L32)
is this the intended behavior?
m: my FatRat $f = FatRat(989898988898989898980909090939838302233848473393040383234234509484.3489523478234723847238432423); say $f eqv FatRat(989898988898989898980909090939838302233848473393040383234234509484.3489523478234723847238432423); say $f eqv FatRat(989898988898989940247844191134641988224436627880004840480824357399.83933482381524942767485346932);
camelia rakudo-moar b243a9: OUTPUT«True␤True␤»
atweiden or this
m: my FatRat $f = FatRat(989898988898989898980909090939838302233848473393040383234234509484.3489523478234723847238432423); say $f eqv FatRat(989898988898989898980909090939838302233848473393040383234234509484.3489523478234723847238432423); say $f == FatRat(989898988898989940247844191134641988224436627880004840480824357399.83933482381524942767485346932);
camelia rakudo-moar b243a9: OUTPUT«True␤False␤»
07:44 el_amigo is now known as st_iron, jack_rabbit left
masak ...no? :) 07:45
seems they should only be == or eqv if they're numerically equal
which is a pretty clear-cut notion 07:46
07:46 AlexDaniel left
atweiden masak: good morning to ya :) i was thinking about writing a new multi sub handling FatRat 07:48
masak ok :) 07:49
ufobat good morning :)
07:49 fireartist joined
atweiden should the eqv sub go in rakudo/src/core/Rat.pm ? 07:50
i haven't tested but it's possible Rat is affected by this too 07:51
07:52 st_iron left
atweiden any thoughts on rewriting the existing one in Numeric.pm 07:53
masak I'd say go for it; see whether the result is better 07:54
this ought to be measurable in absolute passing-tests numbers
the above evals are obviously suspicious and a strike against the current implementation 07:55
08:12 pierrot left 08:14 cpage left 08:21 buharin left 08:22 papayaman joined
papayaman mst: hey man, i know you said on twitter that you don't like perlmaven. why don't you submit PRs or issues to its github repo ? github.com/szabgab/perlweekly/issues 08:23
mst: it's not going to fix itself you know..
unless there's some telepathy going on between you and gabor.. 08:24
moritz papayaman: does this discussion belong into #perl6?
08:25 papayaman left
moritz wow, /quit'ing after leaving such a message isn't the best thing you can do to emphasize your point 08:26
08:28 zakharyas joined
tweakism trouble in paradise? 08:29
[Tux] test 21.717
test-t 13.739
csv-parser 49.050
08:36 RabidGravy joined
Hotkeys moritz: you mean joining specifically to say your message and then immediately leaving isn't the way to do things??? 08:37
:p
08:39 firstdayonthejob left
RabidGravy marning! 08:42
08:45 wamba left, salv0 joined
masak moritz: I agree it wasn't on-topic... but everytime such a thing happens, I understand a bit better the lightning-rod role mst has taken on in the p5 community. 08:45
it's like... people who won't ever rise above ad-hominem anyway, end up thinking that mst is the cause of all their worries. (and so other people can get on with being productive.) 08:47
08:49 abraxxa joined 08:53 Actualeyes left
stmuk_ the best solution for the large minority of people who find mst toxic is to attempt to ignore him 08:56
Hotkeys Idk who mst is 08:57
08:57 Actualeyes joined
El_Che ah perl6-js grant? \o/ 08:57
Hotkeys Yeah we'll take over the web soon 08:58
And then the world
El_Che and after some years, we'll still be big in Japan! 09:00
09:04 dakkar joined
RabidGravy sorry, I've taken over the world already, I've just been subtle about it 09:11
masak you're the fifth person this month to come in and say that 09:13
El_Che RabidGravy: if that's the case, can you "fix" a few parking tickets I have, you know as master of the world
09:16 cpage_ left
masak I still find it interesting that python3 will not only fail on removed 2.x syntax (like circumfix:<` `> and infix:<< <> >>), but do so in a way that gives no indication it ever worked 09:16
it's like the total opposite of the Perl 6 approach
jast El_Che: that wouldn't be very subtle, now would it
09:17 cpage_ joined
Hotkeys masak: its hard when people think of perl 5 -> 6 like they think of python 2 -> 3 09:17
Like perl 6 is brand new
tweakism masak: I think their goal is to make a clean-break and shed what they consider to be mis-features
Hotkeys Well 09:18
Its a long time in the making
But brand new kind of
tweakism that doesn't forego helpful errors, but it explains why they break backwards compatibility
Hotkeys In the sense that it isn't p5
tweakism masak: backwards-compat is valuable, but it can also be a great burden sometimes. 09:19
masak oh, agree in general
09:20 daxim left
masak but here it just feels like a missed opportunity to put in the minimal amounts of code to recognize-but-flag-as-gone the old syntax 09:20
they do in the case of `print "hi"` vs `print("hi")`, thankfully
tweakism heh, it would be kindof amusing if you could import from past 09:21
from past import print_statement
09:21 cpage_ left 09:27 bjz joined
masak well, you can always do like the JavaScript community, and compile your private dialect of python3 to standard python3 :) 09:28
tweakism lol. 09:30
I do that in JavaScript too, because the dialects that are actually available where I need to run are broken and braindead in various ways.
09:30 ZhangGong|2 left
RabidGravy I'll just mention that I really don't like coffeescript at all 09:38
gregf_ hi, is there a way to list all methods of an instance? 09:43
like so, class Foo { method bar() { } method baz(){} }; and then my $f = Foo.new; print $f.methods or something along those lines 09:44
jnthn m: class Foo { method bar() { } method baz(){} }; say Foo.^methods 09:46
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/NgJveNd_F9␤Strange text after block (missing semicolon or comma?)␤at /tmp/NgJveNd_F9:1␤------> 3class Foo { method bar() { }7⏏5 method baz(){} }; say Foo.^methods␤ expecting any of:␤ i…»
jnthn m: class Foo { method bar() { }; method baz(){} }; say Foo.^methods
camelia rakudo-moar b243a9: OUTPUT«(bar baz)␤»
jnthn Can do it on an instance too
gregf_ jnthn: cheers 09:47
09:54 virtualsue joined, cognominal left
RabidGravy if I have a CStruct and I apply a role to it with an attribute will that mess up the representation in native-land? 09:56
jnthn RabidGravy: Compile-time composition, or mix-in? 09:58
09:58 cognominal joined
jnthn role Foo { has int32 $.x }; class C is repr('CStruct') does Foo { } # this should be fine since the attributes are flattened into the role, so it's no different from CStruct's point of view 09:59
afaik, anyways :)
09:59 rindolf joined
RabidGravy either really, in portaudio there is a struct DeviceInfo but everywhere else in the API it is reference by an "index" 09:59
so it would be nice to store the index in the class somehow but not make that appear to the native struct 10:00
jnthn Oh...
I'd probably solve that with delegation 10:01
RabidGravy yeah, that was the next option
jnthn class WrapperThingy { has TheCStructType $.foo handles *; has $.index; }
Hotkeys What does handles do
I haven't seen that one before 10:02
Is it a native thing
RabidGravy it makes the methods of the attribute appear as methods of the object
no not native
m: class Foo { method bar() { say "bar" } }; class Baz { has Foo $.foo handles <bar> = Foo.new }; Baz.new.bar 10:03
camelia rakudo-moar b243a9: OUTPUT«bar␤»
Hotkeys Neat 10:04
RabidGravy the * in jnthn's example is "gimme everything"
Hotkeys Right
I assumed as much
I've gotten good at guessing what whatever does
RabidGravy you can also do it with a hash which is their-name => my-name
Hotkeys Oh so like 10:05
You could map bar to qux
And then call qux
RabidGravy yeah
gregf_ m: class Foo { method bar() { print "Bar"; } }; class Baz { has Foo $.f handles *; method yay(){ self.bar(); } };Baz.new.yay(); #<== something like that i would imagine
camelia rakudo-moar b243a9: OUTPUT«Bar»
RabidGravy m: class Foo { method bar() { say "bar" } }; class Baz { has Foo $.foo handles bar => 'bap' = Foo.new }; Baz.new.bap 10:06
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/d1OfTI0aH_␤Cannot modify an immutable Str␤at /tmp/d1OfTI0aH_:1␤»
Hotkeys Handles in my mind brings up memories of when I had to do VB.Net in high school
And handles did evenets
Events
RabidGravy m: class Foo { method bar() { say "bar" } }; class Baz { has Foo $.foo handles (bar => 'bap') = Foo.new }; Baz.new.bap
camelia rakudo-moar b243a9: OUTPUT«Method 'bap' not found for invocant of class 'Baz'␤ in block <unit> at /tmp/SV50oUkceS line 1␤␤»
RabidGravy or even the other way around
m: class Foo { method bar() { say "bar" } }; class Baz { has Foo $.foo handles (bap => 'bar') = Foo.new }; Baz.new.bap
camelia rakudo-moar b243a9: OUTPUT«bar␤»
gregf_ other way i would imagine, as thats how its in Moose :| 10:07
FROGGS .tell azawawi I'd like to extend your Selenium::WebDriver to support BlackBerry devices... and I think I'd need guidance on how to setup Firefox testing. Would be nice if we could chat a little
yoleaux FROGGS: I'll pass your message to azawawi.
10:08 cognominal left
jnthn Yes, it points from the name you expose to the name it'll map into 10:09
RabidGravy I always have to look
I also have to not use a '*' in about half of cases as it doesn't like some kinds of members 10:10
FROGGS jnthn: can you tell me if this patch is just rubbish or sensible? github.com/rakudo/rakudo/commit/80...9e3cddd7bc
jnthn Makes no sense 10:11
$s is already a native string
So it'll code-gen a box just so it has one to throw away :P
10:13 pmurias joined
RabidGravy aga do do do 10:14
FROGGS jnthn: so my gut feeling was right... we got a Heisenbug here 10:16
maybe... if I'd add a serialize function to NativeRef, I'd dump its value and maybe get a clue what it is 10:17
FROGGS wants Data::Dumper in C 10:18
10:18 cognominal joined
ilmari gdb ;) 10:18
FROGGS ha!
ilmari p my_struct
FROGGS well, I could potentially add a breakpoint... 10:19
hmmm
yes, will try that
10:27 espadrine joined 10:30 TEttinger left 10:31 espadrine_ joined 10:32 espadrine left 10:36 pierrot joined 10:38 virtualsue left 10:39 profan left 10:41 profan joined 10:42 cpage_ joined 10:44 atweiden left, espadrine joined 10:46 daxim joined 10:48 espadrine_ left
masak m: try 1e0 / 0e0; say $!.^name 10:54
camelia rakudo-moar b243a9: OUTPUT«X::Numeric::DivideByZero␤»
colomon is looking at the FatRat / Rat eqv patch and trying to figure out what it is meant to be doing. 10:55
10:59 pmurias_ joined 11:02 pmurias left
gregf_ m: class Foo { method bar($a) { }; method baz($a, $b) { } }; print Foo.^methods.map(*.arity) 11:04
camelia rakudo-moar b243a9: OUTPUT«2 3»
gregf_ should'nt this be printing 1 2?
ah nevrmind, prolly self is included :| 11:05
11:06 mls_ joined 11:08 cognominal left, cognominal joined
jnthn gregf_: self is included; we don't consider self special. If you wanted to call the method as a code object directly you'd have to pass something to fill the invocant slot. 11:09
11:10 mls left 11:11 mls joined 11:12 mls_ left 11:13 musiKk_ joined 11:17 kid51 joined 11:23 cognominal left, cognominal joined
gregf_ oh - ok 11:23
so Foo.^methods[0].(Foo.new, 10) #or whatever 11:24
Woodi do self/this thingies are realy necesarry ? it's "how it works" currently but you do not pass car somewhere so there car can start engine... currently objects "use" methods in the classes so method need to know which object called it but can it be another way ? objects "slurps" method from class storage and call it ? 11:25
hi today :)
tweakism Woodi: a method isn't necessarily called on an instance of the class that defines it. 11:27
11:27 labster left
Woodi tweakism: but that is special case of OO system you have 11:27
tweakism other languages provide a keyword like 'this', but I like the explicit-self-parameter style personally. 11:28
Woodi: I don't think 'allows inheritence' is very special among OO implementations, is it?
Woodi tweakism: no idea what is allow inheritance system... 11:29
tweakism Woodi: Vehicle defines wheels(). Car inherits from Vehicle. somecar.wheels() calls Vehicle.wheels() on an instance of Car. (assuming Car doesn't override wheels()) 11:30
11:31 Praise left
Woodi tweakism: and what is *popular* in industry is not important at all. what's is better is. 11:31
tweakism *shrug* 11:32
gregf_ Woodi: self/this are not mandatory afaik
Woodi at the end you will hit valit range of specialisations usefull in particular cases :)
gregf_ for example in this case: self 'is' needed:
Woodi gregf_: but it's like current OO implementations work internally. i wonder what about something else :) 11:33
gregf_ class Foo { def bar;end }; class Bar < Foo; def baz; self.class.superclass.instance_method(:bar).bind(self).();end; end <== in ruby
Woodi : even in Java 'this' is not mandatory 11:34
Woodi gregf_: but it is there, i think. syntax isn't important 11:35
tweakism ultimately, they are just different syntaxes for the same thing, I think. not entirely certain, but I think.
the car object *does* get "passed" to the start_engine method in either case; making it an explicit parameter makes it easy to reason about. 11:36
Woodi gregf_: what this Ruby thing do ? you have variable bar in Foo but later use method bar...
tweakism some languages make a real mess w/ this (JavaScript).
Woodi tweakism: yes. but maybe something other have sense too ? and maybe it can be better in some cases ? 11:37
gregf_ Woodi: its a mental way by which a specific class calls a different method from the generic/parent class :| #ofcourse s/{//;
RabidGravy right off out for a bit 11:39
Woodi gregf_: it's make sense. why clone methods into *every* object, would be waste. but this system just use objects like varibles...
tweakism Woodi: so you're saying design a language that requires neither a self parameter nor a self/this keyword?
Woodi tweakism: trying things (or just thinking) is like science works :) 11:40
jnthn Woodi: You might find CLOS interesting :)
Woodi jnthn: I need to learn Lisp first probably... :) 11:41
jnsso it's already invented ??
jnthn*
using objects is indirection via pointer :) 11:42
jnthn Woodi: Well, they have a different way of looking at objects/methods is all. Objects don't have methods, and the "methods" can have multiple invocants
jast so you're thinking about a system to reference objects wihout referencing them?
jnthn Actors are also intereting to look at, and probably closer to the original thinking of OO. 11:43
Woodi jast: I think not. objects just take methods from class and use it on fields
11:43 RabidGravy left, kid51 left
jast depends. not all implementations of object orientiation even *have* classes... 11:43
jnthn The trouble with OO naming wise is it puts the focus on objects, whereas the interesting thing is the messages. 11:44
ilmari like javascript, to take random example
+a
jnthn found OO design a lot easier when he started desining message first, then grouping them by invariants to discover the objects, rather than trying to work out what objects should exist 11:45
*designing
Woodi jnthn: discover ? you mean paper-pencil-design ? 11:46
colomon m: say (989898988898989898980909090939838302233848473393040383234234509484.3489523478234723847238432423).nude
camelia rakudo-moar b243a9: OUTPUT«(9898989888989898989809090909398383022338484733930403832342345094843489523478234723847238432423 9999999999999999583119736832)␤»
colomon is trying to figure out how that makes any sense
moritz a simplistic approach to OO design that works surprisingly well is to start with a bunch of subroutines, and see what data they need in common and pass around everywhere
jast I take a data-centric view, too, usually 11:47
moritz then you can start putting the common data into one or several related objects, using Common Sense[tm]
jnthn Woodi: For many design tasks I use pen/paper or a whiteboard at first, but I quickly move to sketching in code instead.
jast what's almost never a good idea is starting with tons of base classes and building up a huge hierarchy
ilmari premature generalisation is the root of much mess 11:48
jnthn Woodi: But I mostly just mean that I found that "what classes should I have" and "what fields to they need" tend to come later than "what behaviors should the system have" for me.
ilmari you don't know what the general patterns are until you've implemented/designed several similar things
jast I quite like composition like with P6's roles
tweakism people want to inherit because you can, but has-a relations seem to usually be more appropriate, yes?
11:49 wamba joined
dalek p: d811340 | (Pawel Murias)++ | src/vm/js/HLL/Backend.nqp:
[js] Switch from uglifyjs to js-beautify.
 11:51
p: 4e1979c | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (2 files):
[js] Move up serialization format to version 18.
11:51 pmurias_ is now known as pmurias
pmurias jnthn: are there any ways to determine if a given closure needs to be serializable? 11:52
11:52 tmch joined 11:53 cognominal left 11:54 cognominal joined
timotimo generally develops dada-centric 11:59
pmurias jnthn: if I store lexicals using js lexicals it's more efficent but makes it hard to extract them for serializing closures :/
12:00 spebern joined
jnthn pmurias: I'm pondering, but...I don't really have a good answer. 12:02
pmurias: Rakudo will get much more aggressive at lexical -> local lowering
dalek p: 6f3aa41 | (Pawel Murias)++ | src/vm/js/nqp-runtime/serialization.js:
[js] Fix typo, ilmari++.
12:02 spebern left
jnthn pmurias: So it may become less of a problem than today 12:03
12:03 spebern joined
jnthn pmurias: But yeah, potentially any closure might be serialized and the demand to do so may even come from a totally separate compilation unit. 12:03
12:04 wamba left
pmurias I could generate both a serializable and unserializable version and just switch to the latter once we know that no serialization will happen, but it seems like something that would just end up adding to much complexity 12:06
jnthn Plus you can't know so easily...
pmurias yes, it would require user "pragmas" or some other things, and compiling Perl 6 is when we care about NQP speed the most 12:07
more and more I'm coming to the conclusion that I should have a seperate analysis pass before actually compiling to js 12:08
colomon m: say Rat.new(9898989888989898989809090909398383022338484733930403832342345094843489523478234723847238432423, 10000000000000000000000000000) 12:09
camelia rakudo-moar b243a9: OUTPUT«989898988898989898980909090939838302233848473393040383234234509484.348952347823472397436717074015␤»
pmurias which could determine things like "this sub is never rebound and only called with named parameters from those 2 places and doesn't care about dynamic variable" 12:10
12:10 spebern` joined
pmurias jnthn: would SPESH benefit from being preseeded with facts, like if I could statically infer what the type of a given variable is 12:12
?
jnthn pmurias: Not epsecially. The problem comes down to proof.
pmurias: That is, a bunch of what spesh does is throw out checks and virtualization 12:13
pmurias: But it has to know it's safe to do that
pmurias: We can in various cases generate simpler code from static analysis. 12:14
12:14 jolts left
pmurias jnthn: I'm thinking about stuff like "my class Foo {...}; my $foo := Foo.new; # we know what type $foo is once we close the classes" 12:17
12:17 Skarsnik joined
pmurias mere guess seem of very little value to SPESH as it runs at runtime and can just check the type 12:18
jnthn ...unless it's re-bound :)
12:20 spebern left, spebern joined
pmurias of course, which is why such type inference seems to complex to put in an adhoc manner inside the code generator 12:22
jnthn Yeah...well, I've figured we'll do better with dynamic optimization in general. 12:23
pmurias for moarvm it seems better and it seems like something that will be more predictable
jnthn Because people want their scripts to run fairly snappy too, and doing fancy analysis on something that won't do much at runtime anyway is kinda wasted
But at runtime you can discover what's hot and worth investing time optimizing
mst moritz: and in any case, as I *said* on twitter, when I submitted bug reports for perlmaven gabor yelled at me for criticising his code and didn't fix any of the bugs, so shrug, I already fscking tried :( 12:32
stmuk_: indeed; people are allowed to dislike me if they want to :) 12:33
mst apologises for the noise, was just following up the highlight
moritz 's ok 12:36
12:36 wamba joined
mst suspects that because I forgot to do the reply-chaining thing he'll've seen an RT of the first tweet but not the subsequent one, so that's arguably my fault anyway 12:38
12:42 virtualsue joined 12:45 spebern left 12:47 johndau joined
masak wonders if he would ever yell at a contributor, no matter who it was, for criticising his code 12:48
12:48 johndau left
tweakism meh, everyone has bad days / makes poor/hasty decisions. 12:48
some more often than others.
masak 'sooth 12:49
12:51 johndau joined 12:52 johndau left 12:56 kaare_ joined
moritz
.oO( how dare you insinuating that I make poor decisions! )
 12:57
masak .oO( how dare you joke about serious subject matters! ) 12:59
13:04 Cabanossi left 13:06 Cabanossi joined
hoelzro o/ #perl6 13:06
13:15 cdg joined 13:21 Some-body_ joined, DarthGandalf left 13:22 Some-body_ is now known as DarthGandalf, spebern` left 13:23 cognominal left 13:24 cognominal joined 13:27 zakharyas left 13:28 kanishka joined 13:29 ilbot3 left, ilbot3 joined 13:30 tmch left 13:32 aborazmeh joined, aborazmeh left, aborazmeh joined, CIAvash left 13:33 molaf joined 13:35 RabidGravy joined 13:36 camelia joined
moritz PSA: rakudo, nqp and roast commits will now be announced in #p6dev, not here anymore 13:36
13:37 ChanServ sets mode: +v camelia
RabidGravy I didn't know there was a #p6dev 13:38
perlpilot That's because it didn't exist until recently :) 13:40
ilmari recently as in "just now"?
13:40 skids joined, Cabanossi left
moritz yes 13:42
RabidGravy gets in there
13:42 Cabanossi joined
virtualsue i'm happy to hear that new channel day has arrived 13:48
timotimo hey sue :) 13:49
virtualsue did you get to the hackathon last saturday? 13:50
timotimo yes! :D
virtualsue kewl
timotimo i was just a little sad so many people that have been to the conference didn't attend the hackathon :|
but that just makes the reinforcement of "next time i'll sign up on time" stronger 13:51
virtualsue if it would have helped, you could have had my registration 13:52
timotimo :o
i hadn't considered that avenue of thought at all!
virtualsue however i enjoyed it no end
timotimo wait ... were you saying you weren't able to attend, or you would have offered to let me attend instead of you? 13:53
dalek kudo-star-daily: a3d4488 | coke++ | log/ (8 files):
today (automated commit)
 13:54
virtualsue i'd have watched stmuk do his talk through the window ;-) 13:55
timotimo no, i couldn't possibly have accepted that :)
virtualsue surprised you weren't there but glad you made it on sat 13:56
timotimo i messed up big time. i've been told way in advance about things, been reminded at least once, but still procrastinated signing up until it was quite a bit too late 13:57
but i did manage to get some code done during the hackathon, which felt good 13:58
13:58 vendethiel joined
virtualsue stmuk used your code in his talk i believe - SDL2::Raw 14:00
timotimo aye, i saw :)
i'm looking forward to the recording 14:01
virtualsue lots of good feedback
timotimo but SDL2::Raw is really just the most straight-forward binding to SDL2 that you could possibly come up with
so there's not too much credit to be given ;)
the examples/ folder, on the other hand ... that stuff is kinda neat!
virtualsue i'm using something quite similar with Go right now
masak virtualsue: something quite similar to Go? tell us more! 14:02
timotimo i heard virtualsue is actually the person who decides what moves AlphaGo makes 14:03
virtualsue HA
timotimo remember the few blunders AlphaGo made in the last(?) match? that was just the cat walking over the keyboard
masak I heard it was the Higgs boson interfering with the circuitry 14:04
14:06 rindolf left
virtualsue simple carelessness, not putting the cat out and blaming particle physics 14:07
DrForr_ When it gets confusing is where you're also puttingthe cat out. 14:08
RabidGravy I don't know, I rely on the "cat effect" for most of what I do
I did a set for a radio show a couple of years ago and the cat decided to attack my feet while I was about halfway through the "live" recotding 14:09
virtualsue i know - i've heard some of your stuff :-)
RabidGravy :) 14:10
tadzik timotimo: have you used your Cairo bindings with Gtk::Simple? 14:11
timotimo yes, i have
RabidGravy have you ever been to one of my (rare) live gigs?
14:12 rindolf joined
RabidGravy I know ilmari has 14:12
virtualsue unfortunately not
timotimo that's what powered my first shooter game (last year's GPW) and the live coding environment thingie
tadzik right, can you point me to that example? I can't find it on your github
timotimo github.com/timo/gtk_cairo_shooter
tadzik I decided to do my note editor thingy in p6 after all
or at least try
ah, there it is. Thanks! 14:13
timotimo i coulnd't get it to render with opengl back then; there ought to be some way to do it now, but it would have to be added to GTK::Simple, too
so it's doing everything in software rendering
tadzik should be good enough for me if it's on cairo side
timotimo yeah 14:14
good luck in any case :) 14:16
let me know how it goes
14:21 vendethiel left
tadzik yeah, will be a while, now that I realized that cpan's MIDI may not do real-time midi stream handling 14:26
timotimo ugh
tadzik ah, no, it seems to do
timotimo who ever needed that anyway
tadzik at least there was a conference talk about it :P 14:27
14:27 telex left
tadzik and I know the guy who did it :) 14:27
ah, no. There's MIDI::Realtime, but it only does writing to midi. Not reading. How convenient :D 14:28
14:28 telex joined
tadzik I hope p6's Bufs are up to this 14:28
timotimo if we'll get a nice midi module out of this, i'm all for it :P 14:29
Hotkeys There's a p6 live coding environment? 14:31
14:32 zakharyas joined
timotimo Hotkeys: just for cairo-using code 14:32
Hotkeys Ah
RabidGravy tadzik, timotimo I have sketched out Audio::PortMidi bindings right here, I may get round to it at some point 14:34
tadzik oo
timotimo PortMidi is the midi portion of PortAudio? 14:35
RabidGravy yeah
timotimo mhm, mhm.
RabidGravy It'll be subject to the same real-time constraints as portaudio but you have more wiggle room 14:36
tadzik RabidGravy: is it up somewhere?
14:39 musiKk_ left, YvargDibar joined
YvargDibar boo 14:39
timotimo network troubles? 14:40
virtualsue ghost in the wire 14:41
14:41 RabidGravy left 14:43 SauceRageuse joined
SauceRageuse yep, BT engineers messing with the overhead over the road 14:43
14:44 YvargDibar left, SauceRageuse is now known as RabidGravy
RabidGravy boo 14:44
I may have been rudely interrupted by the BT guys doing what BT guys do 14:45
(i.e. breaking everything) 14:47
14:52 RabidGravy left
virtualsue nobody appreciates infrastructure. if they don't break it, you wouldn't realise how much you need it 14:53
it's probably in the schedule: x% tactical breakage randomly spread around the customer base 14:55
14:58 wamba left 14:59 aborazmeh left
Skarsnik hm, is there a way to get back a Version to its p6 syntax? Like my $version = v1; say $version.Str; #output v1 and not 1 15:03
oh .gist >< 15:04
perlpilot or .perl
(.perl sounds more like what you want)
Skarsnik Yes 15:05
that what I just though
15:06 RabidGravy joined 15:07 sjoshi left, nadim left
RabidGravy gah, if they don't stop that I'm going to go across the road and take their tools off them 15:07
Skarsnik BT guys?
RabidGravy yeah 15:08
I think they've stopped now. There's a telephone pole across the road that feeds our line and a few of our neighbours and it's right in a bunch of trees 15:09
so the lines keep getting broken by the trees 15:10
we had a completely new one a couple of years ago
better than the hundred year old bit of copper that was there before
tadzik hmmm 15:15
does P6M Merging GLOBAL symbols failed: duplicate definition of symbol NativeCall ring a bell to anyone?
RabidGravy yes, but it went away a few weeks ago 15:16
it was a bug in that require was merging the symbols into the local stash so things couldn't see it as being loaded 15:17
or is this a recent rakudo?
tadzik it's v6.c
timotimo that's a wee bit old by now 15:22
RabidGravy the bug was only in there for a short window in late january 15:23
15:23 virtualsue left 15:26 cognominal left 15:27 cognominal joined 15:31 donaldh joined 15:41 fireartist left
stmuk_ blogs.perl.org/users/steve_mynott/2...-2016.html 15:41
timotimo stmuk_++ :) 15:46
moritz stmuk_: "use Foo::Bar:from" can be used from Per... it ate your <perl5> here 15:52
timotimo ah, yes, html tags 15:53
stmuk_ grrr
moritz stmuk_++
stmuk_ fixed 15:54
timotimo moritz: i expect we don't have anything like quota set up for hack and www; do we perhaps want quota set up so that processes that have gone berserk won't fill up the disk with garbage?
moritz timotimo: I'd like to avoid that, and rather react when things go badly
timotimo: I set up some per-user resource limits, but it seems they don't apply to processes started by cron :( 15:55
timotimo in that case, do we want to set up collectd with some notification mechanism?
i think it already supports notification, we'd just have to turn it on and configure it some way. most probably sending mail 15:56
moritz do we? or do we want to use a monitoring system?
also, so far I've avoided installing a mail server on hack
timotimo that's probably sane
moritz I'm not opposed to somebody doing that, but only if they have an eye on it 15:57
timotimo collectd has many different ways to push out notifications. we could exec (or send a network message to) an irc bot
exec, log to file, send via network, run perl code, output to syslog, send to a unix domain socket 15:58
moritz timotimo: on the one hand, it would be neat to have something like that. On the other hand, I can't help feeling it's busywork that we don't end up needing/using very much 15:59
timotimo i imagine we'd end up seeing a "hack freezes over" situation coming a bit better when we have something that pings us
moritz timotimo: on feather, the homes simply ran full occasionally; then somebody would delete compiler and cpanm caches, and all was fine again
hm 16:00
I thought you were talking about disks filling up
that doesn't seem to be related to hack freezing
we have some monitoring for some variables going out of control (number of processes, load average), but it usually comes too late 16:01
I don't know what early indicators are that we could use
timotimo collectd seems to have a flexible configuration system for "thresholds"
16:02 TimToady left 16:03 khw joined
timotimo moritz: you know better than i how to administrate dalek; i don't see #p6dev in botnix.conf; is that by accident? 16:03
16:03 nakiro left 16:04 TimToady joined, dalek left, dalek joined, ChanServ sets mode: +v dalek
moritz timotimo: yes :-) 16:04
timotimo what's with the multiple-months of packet.*.freenode,* files in dalek-poller/ ? 16:09
the queue file in there seems to still be appended to, but the beginning of it seems *really* old and it's 26 megabytes big 16:10
Skarsnik could be because of splits?
16:13 CIAvash joined 16:15 domidumont left 16:16 novice33-13 joined, DrForr_ is now known as DrForr
novice33-13 my $line='Node=1,Item=W22,Max 10'; $line ~~ /MYREGEX/; say $<Node>, $<Item>, $<Max>; #What is MYREGEX? 16:16
16:18 virtualsue joined
gregf_ prolly something like (?<Node>(?:\w+)=(\w+)) ? 16:18
perlpilot That's a very P5 looking answer for a channel about P6 :) 16:20
timotimo MYREGEX is a regex that matches any string that contains the literal "MYREGEX" in it 16:21
16:21 itaipu joined
[Coke] timotimo: wouldn't that be /'MYREGEX'/ ? 16:21
timotimo same thing
[Coke] kk
timotimo well, i should have included / in the front and end
perlpilot [Coke]: all the characters are alphanumeric so no quotes needed 16:22
novice33-13 I need such REGEX so after matching I can say $<Node>, $<Item>, $<Max> 16:25
perlpilot Why?
Skarsnik It's a grammar then? 16:26
16:26 virtualsue left
perlpilot Currently, it would have to be a grammar AFAIK (which is why I wanted to know "why?" :) 16:26
tony-o_ novice33-13: you can assign in regex to do / $<x>=\w+ / and then $/<x> 16:27
timotimo or even just $<x> 16:28
perlpilot Oh, I suppose he could do a really explicit regex like that.
But, again, why the constraint that $<Node>, et alia must be available? Why not just get a list of pairs out of it (for instance)? 16:29
tony-o_ are they ever out of order? not all available? 16:30
RabidGravy can someone come round and do this documentation for me? I'm losing the will to live
perlpilot and why a regex instead of other, simpler techniques?
RabidGravy because it's more something or other to have two problems 16:31
16:32 perl6newbee joined
perlpilot m: "Node=1,Item=W22,Max 10".split(/<[\s,=]>/).hash.perl.say; # an example 16:32
camelia rakudo-moar b243a9: OUTPUT«{:Item("W22"), :Max("10"), :Node("1")}␤»
tony-o_ nice 16:33
timotimo m: "Node=1,Item=W22,Max 10".split(["\s", "=", ","]).hash.perl.say # does split take multiple needles? 16:34
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/C0uDrt6SFa␤Unrecognized backslash sequence: '\s'␤at /tmp/C0uDrt6SFa:1␤------> 3"Node=1,Item=W22,Max 10".split(["\7⏏5s", "=", ","]).hash.perl.say # does spli␤ expecting any of:␤ double …»
timotimo m: "Node=1,Item=W22,Max 10".split([" ", "=", ","]).hash.perl.say # does split take multiple needles?
camelia rakudo-moar b243a9: OUTPUT«{:Item("W22"), :Max("10"), :Node("1")}␤»
timotimo it does! 16:35
^- this piece of code is probably 100x faster than the one using the character class (though it only accepts the one space, rather than every kind of space)
perlpilot timotimo: nice
RabidGravy groovetastic
tony-o_ awesome 16:38
CIAvash nice!
16:43 virtualsue joined
perlpilot timotimo: I challenge your "probably 100x faster" :-) 16:44
but ... 16:46
RabidGravy "significantly faster" should fly
perlpilot lunch & 16:47
16:47 wamba joined
perlpilot RabidGravy: in my little test, it wasn't faster at all. I want other people to try 16:47
perlpilot lunch for real this time &
16:48 lostinfog joined
tony-o_ for 33 iterations: 16:53
- regex: 0.171848 per iteration
+ regex: 0.188091 per iteration
16:54 cognominal left 16:55 cognominal joined, donaldh left
timotimo is 33 iterations enough? 16:59
Skarsnik hm, why you can type constraint our variable?
*can't
timotimo i guess the string isn't long enough to give you a noticable difference 17:00
17:00 zakharyas left
tony-o_ 33 isn't very many iterations but it's probably close enough, the times include perl6 start up time and bash iteration times so take them with a grain of salt 17:01
timotimo how have you been, tony-o_? it feels like i haven't talked to you in a while 17:05
17:13 FROGGS left
novice33-13 m:my $s="Node=1,MaxR 45";$s~~/[ <ident>+ ["=" | \s+] $<val>=<-[,]>+ ]+ % ","/; for flat($<ident> Z $<val>) -> $x,$y { say("our \\$x = $y")} 17:16
tony-o_ not too shabby, have been travelling a lot so i haven't had much time for perl6 :-( 17:17
novice33-13 something like that but with EVAL instead of say and inside regular expression
p6:my $s="Node=1,MaxR 45";$s~~/[ <ident>+ ["=" | \s+] $<val>=<-[,]>+ ]+ % ","/; for flat($<ident> Z $<val>) -> $x,$y { say("our \\$x = $y")} 17:18
17:20 virtualsue left 17:23 kerframil joined
MadcapJake` is there any way I could pass a Perl 6 file handle to a native call class expecting a FILE type? 17:27
nativecall sub
17:27 dakkar left
ugexe .native-descriptor? 17:28
timotimo what ugexe said
MadcapJake` what's that?
jnthn MadcapJake`: .native-descriptor on the Perl 6 handle, then that C thingy that turns an fd into a FILE*
17:29 ufobat left
MadcapJake` ok, so native descriptor is a handle method, what's this C thingy? :P 17:29
geekosaur fdopen()
Skarsnik hm, how hard it would be to catch SIGSEG and display a Backtrace?
MadcapJake` ohh so use a C function to create it
17:30 vendethiel joined
Skarsnik Well, it's the native component of the Io::Handle on the os, it's a FILE* on unix I guess 17:30
geekosaur no? native-descriptor will be a file descriptor, which is just an int on Unix 17:31
(FILE *) is libc's stdio abstraction for buffered I/O on top of file descriptors
MadcapJake` awesome! will this work with stdout?
geekosaur mm, that's actually fairly complex 17:32
MadcapJake` or perhaps more precise, is there a nativecall variable for stdout?
Skarsnik $*IN.native-descriptor? 17:33
geekosaur other stuff already has access to the existing stdout FILE*, you might be better off using nativecall to get at that rather than creating a new FILE* over fd 1
MadcapJake` haha, seems to work?
geekosaur buffering will get weird otherwise
MadcapJake` Skarsnik, $*OUT.native-descriptor gives me 11, will that work? 17:34
geekosaur ...?
11?
afraid to ask
MadcapJake` lol
geekosaur, could you elaborate on "other stuff"?
geekosaur stdout is fd1 normally. something is presumably doing something "special"
not easily, no. stdout is a global variable that pretty much any C function could decide to use 17:35
MadcapJake` hmm, I'm not sure I follow, I have a nativecall function that accepts a FILE type and it also states in comment that it can except `stdout` as well, I would like to cover both cases in my nativecall wrapper 17:36
geekosaur stdout is just a specific, globally defined, (FILE*)
Skarsnik stdout is probably a macro x) 17:37
geekosaur sometimes it is, sometimes it isn't
MadcapJake` according to gnu.org, it's a regular variable on unix and a macro on other systems xD
would be nice if I could pass $*OUT to it and nativecall would Do The Right Thing :P 17:38
geekosaur in the old days it was macroed to an indexed lookup in an array of FILE structs. what it is now will depend on what exact stdio implementation is in use
tony-o_ timotimo: how's it been going for you?
geekosaur but if $*OUT.native-descriptor != 1 then I doubt The Right Thing is at all easy
MadcapJake` bummer 17:39
timotimo tony-o_: it's been going all right. promises of performance improvements are keeping me hopeful :P
geekosaur and buffering will still potentially be a problem because a FILE* maintains a buffer and presumably the libuv thing perl 6 wraps also maintains a buffer of its own 17:40
perlpilot m: say .native-descriptor for $*IN, $*OUT, $*ERR
camelia rakudo-moar b243a9: OUTPUT«0␤1␤2␤»
perlpilot Hmm. 17:41
MadcapJake` geekosaur, any suggestion on what I should do then?
geekosaur ok, why'd MadcapJake` get 11 then? mistake in checking it?
ugexe i get 11 on my local as well
geekosaur hrrr
MadcapJake` why would mine and ugexe be 11 and camelia be right? 17:42
geekosaur aaaaand I just got 11 as well. wtf
ugexe difference in eval server?
geekosaur and $*IN is 9
perlpilot locally, I get 9, 11, 12
RabidGravy Isn't it a libuv artefact?
or something to do with the wrapper or something 17:43
I tracked it down before but forgot what the results where
17:44 pmurias left
geekosaur has no clue 17:44
Skarsnik I get an issue with calling a NC sub that free a struct. If I use Pointer in the sub definition instead of the struct it work. Even adding is rw does not help 17:45
17:45 pmurias joined
MadcapJake` maybe it's got to do with one of these SO answers: stackoverflow.com/questions/6163504...dout-is-11 17:45
geekosaur and no idea how this will affect anything. also no idea how exactly this is being used or whether mixing I/O abstractions will be problematic or not
ohhhhhhhhhhhh
RabidGravy it has no impact on anything that might use them no, I did a test and a "syswrite" using that fd works fine
geekosaur why would it be using dup() though 17:46
RabidGravy that is the libc's write
Skarsnik damn my perl6-gdb-m does not work ><
perlpilot geekosaur: that's what I was wondering
geekosaur well, you can try it. I have no idea what you're actually trying to do with it, so can't even guess how it will turn out 17:48
17:48 vendethiel left 17:49 vendethiel joined
MadcapJake` geekosaur, I will and I'll let you know if it's successful 17:50
novice33-13 Can I EVALuate string "my \Number = 10.5" and get variable named Number? 17:51
timotimo no 17:52
eval-ing cannot create new lexical variables in an outer/calling scope 17:53
tony-o_ timotimo: what are you working on these days?
perlpilot also, even if you could make Number, it wouldn't vary. ;) 17:54
timotimo tony-o_: recently i've been spending some time on SDL2::Raw
and of course making rakudo and moarvm better in general is always something i strive for
Skarsnik IS there a way to get the addr of a NC struct (a cstruct repr class)? 17:56
17:56 vendethiel left
timotimo yeah, get its WHERE (or WHICH?) 17:57
m: use NativeCall; my $arr = CArray[int].new(1, 2, 3); say $arr.WHERE; say $arr.WHICH 17:58
camelia rakudo-moar b243a9: OUTPUT«139811157521112␤NativeCall::Types::CArray[int]|102910000␤»
timotimo m: use NativeCall; my $arr = CArray[int].new(1, 2, 3); say $arr.WHERE.fmt("%x")
camelia rakudo-moar b243a9: OUTPUT«7f55dc113130␤»
timotimo seems legit
Skarsnik Oh I found my issue >< 17:59
17:59 abraxxa left 18:03 firstdayonthejob joined
tony-o_ timotimo: then we can stream audio to one another w perl6 18:03
Skarsnik Hm, I can't find how to export cglobal 18:04
RabidGravy tony-o well ... github.com/jonathanstowe/Audio-Por...eam-source 18:05
just toshing it up, it should hit the ecosystem this evening
18:05 kerframil left
Skarsnik our $kGumboDefaultOptions is export := cglobal(LIB, "kGumboDefaultOptions", GumboOptions); I get (Any) when displaying this variable 18:06
tony-o_ RabidGravy: i'd love to hear 'm: "Node=1,Item=W22,Max 10".split(["\\s", "=", ","]);' spoken rather than written 18:08
18:11 FROGGS joined
FROGGS o/ 18:11
18:11 SCHAAP137 joined
hoelzro what *exactly* does require() evaluate to? S11 says "the value which is installed as the alias", but what is that? something representing the compunit that was loaded? the last package declaration in that compunit? 18:11
o/ FROGGS
18:12 pmurias left
FROGGS hoelzro: I understand it as: if you 'require Foo' and Foo.pm exposes a Foo symbol, then that 18:13
hoelzro ahhhh 18:14
that makes sense; maybe I'll update design/docs
perlpilot what if it doesn't expose a Foo symbol?
hoelzro I'm seeing some odd behavior; namely, if you already have a symbol named REPL in scope, and you require REPL; you get your original REPL package from require, rather than the new one 18:15
perlpilot: good question
RabidGravy tony-o_, "my $m = Proc::Async.new(:w, 'festival', '--tts'); $m.start; $m.print(""Node=1,Item=W22,Max 10".split(["\\s", "=", ","]);'")"
there you go ;-)
FROGGS perlpilot: well, it will evaluate to ::('Foo') anyway
perlpilot: if you intend to use it as a value/type, it probably exposes that symbol... 18:16
perlpilot: if you use it by accident, you get a Failure
18:18 patrickz joined
perlpilot m: my $x = require Foo; 18:19
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/meaT9Y2fsW␤Undeclared name:␤ Foo used at line 1␤Undeclared routine:␤ require used at line 1␤␤»
perlpilot What's up with that second message?
18:20 cdg left 18:21 cdg joined
FROGGS "Undeclared routine require" 18:22
m: my $x = do require Foo;
camelia rakudo-moar b243a9: OUTPUT«Could not find Foo in:␤ /home/camelia/.perl6␤ /home/camelia/rakudo-m-inst-2/share/perl6/site␤ /home/camelia/rakudo-m-inst-2/share/perl6/vendor␤ /home/camelia/rakudo-m-inst-2/share/perl6␤ CompUnit::Repository::AbsolutePath<139892912…»
FROGGS m: my $x = do require Test;
camelia ( no output )
FROGGS m: my $x = do require Test; say Test
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/VAWrPYnIir␤Undeclared name:␤ Test used at line 1␤␤»
FROGGS m: my $x = (require Test); say Test
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/BhKeID9HFG␤Undeclared name:␤ Test used at line 1␤␤»
FROGGS :o(
it is like how you put a while and for loop in statement level 18:23
18:23 sufrostico joined 18:25 nadim joined, cdg left
perlpilot Aye, but that message is LTA. It points in the wrong direction. 18:25
18:30 domidumont joined 18:33 vendethiel joined
FROGGS my $x = while 1; 18:34
m: my $x = while 1;
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/pdvd5RwKgr␤Undeclared routine:␤ while used at line 1␤␤»
FROGGS perhaps we want to specialcase these keywords in this exception? 18:35
MadcapJake` is there a way to make a named argument optional in a multi method signature? I don't want to have a separate multi for every single named argument
18:35 pmurias joined
FROGGS MadcapJake`: these are optional by default 18:35
18:36 novice33-13 left
MadcapJake` is it because it's a multi method inside a CPointer class then? 18:36
FROGGS I dunno what your problem is
MadcapJake` I have 4 multis and all of them can optionally have :$enc left off, but I'm getting an error that there is no signature that matches 18:37
sortiz m: my $pack='Test'; my \M = (require ::($pack)); dd M; # Load Test 18:38
camelia rakudo-moar b243a9: OUTPUT«Test␤»
FROGGS sortiz++
MadcapJake`: can you gist it?
m: my $x = (require Test); say $x 18:39
camelia rakudo-moar b243a9: OUTPUT«(Test)␤»
FROGGS of course
MadcapJake`: ohh wait
18:39 xinming_ joined
FROGGS MadcapJake`: these methods are not native methods, are they? 18:39
tony-o_ m: my \r = require('Test'); r.ok 1==1; 18:40
camelia rakudo-moar b243a9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/BvdCmjiu2t␤Two terms in a row␤at /tmp/BvdCmjiu2t:1␤------> 3my \r = require('Test'); r.ok7⏏5 1==1;␤ expecting any of:␤ infix␤ infix stopper␤ statement end␤ s…»
tony-o_ m: my \r = (require Test); r.ok(1==1); 18:41
camelia rakudo-moar b243a9: OUTPUT«Method 'ok' not found for invocant of class 'Test'␤ in block <unit> at /tmp/aRC9iI27hi line 1␤␤»
MadcapJake` well internally they call native methods but they are members of a class that is repr('CPointer')
RabidGravy do the signatures contain typed pointers or carrays?
MadcapJake` yep 18:42
18:43 xinming left
RabidGravy that'll be it, they don't work nicely on methods for some reason, loosen the type consrtraint 18:43
tony-o_ is Test broken currently ?
18:45 perl6newbee left
MadcapJake` RabidGravy, sweet that did it! 18:45
tony-o_ gist.github.com/tony-o/6e9634b0c43d22c44833 18:46
that output is from a nuked/freshly built moar from rakudobrew 18:49
MadcapJake` I can't seem to get DESTROY to work with a nativecall CPointer class, does it not happen automatically? 18:51
s/happen/get called/
RabidGravy destroy dinnae got called until the GC happens
18:51 musiKk_ joined
MadcapJake` dinnae? 18:51
RabidGravy "doesn't" 18:52
MadcapJake` so if I add a say to my DESTROY submethod, will I see it print?
RabidGravy it may never get called
MadcapJake` doesn't that mean I've leaked some memory? 18:53
RabidGravy if you want to guarantee something happening when the thing goes out of scope you may want to consider a phaser
well if the program ends then no 18:54
sortiz MadcapJake`, I use a form of Disposable patern in DBDish's drivers with a submethod DESTROY { self.dispose } as last resource, and works very well. 18:56
MadcapJake`, In the Pg and SQLite cases, the statement handlers are native allocated CPointers that need to be deallocated on time. 18:58
19:02 spider-mario joined
sortiz Oops! I haven't pushed that DBIish branch yet :-( 19:04
tony-o_ yay for dbiish being updated
might have to test/update github.com/tony-o/perl6-db-orm-quicky 19:05
sortiz tony-o_, I forgot the push 'cus I'm adding ODBC support. 19:07
sortiz Good excuse! ;-)
RabidGravy sortiz++ # one off my list ;-) 19:09
19:09 kaare_ left 19:10 pmurias left
tony-o_ sortiz: still an update and ++ 19:12
19:12 pmurias joined, TEttinger joined
tony-o_ is Pg more than prepared statement supported yet? 19:12
sortiz tony-o_, What you need? 19:14
tony-o_ in Pg it failed if you tried to prepare more than one statement up front and just call them as needed 19:15
that was over a year ago, i haven't played with anything DBIish since then 19:16
MadcapJake` sortiz, I'm not seeing any DESTROY in the DBIish repo
sortiz tony-o_, That is solved, yes.
Skarsnik tony-o_, I have this on some code and it work
sortiz MadcapJake`, You will see in a few minutes, sorry. 19:17
Skarsnik tony-o_, github.com/Skarsnik/fimstuff/blob/...ry.pm6#L21
tony-o_ Skarsnik: sortiz awesome guys, thank you
MadcapJake` oh no problem! I appreciate the advice!
19:20 sufrostico left 19:21 sufrostico joined, virtualsue joined 19:22 labster joined
Woodi m: gist.github.com/anonymous/e0019267b1320a45cf5c 19:25
camelia rakudo-moar b243a9: OUTPUT«｢abcabccaabbc｣␤ abc => ｢abc｣␤ bc => ｢bc｣␤ c => ｢c｣␤ abc => ｢abcc｣␤ bc => ｢bcc｣␤ c => ｢c｣␤ c => ｢c｣␤ abc => ｢a｣␤ abc => ｢abbc｣␤ bc => ｢bbc｣␤ c => ｢c｣␤»
Woodi so how to change what tokens generate ?
19:26 yqt joined 19:28 bartolin left, bartolin joined 19:30 shlomif joined
tony-o_ Woodi: you change the selectors .. 19:30
19:31 rindolf left, shlomif is now known as rindolf, pierrot left
Woodi tony-o_: you mena in grammar ? I ask about action: how to change result in actions ? 19:32
19:32 sufrostico left
perlpilot Woodi: perhaps you want what's in $r.ast ? 19:32
tony-o_ are you asking about the type that comes from actions? or ??..what would the end result look like ? 19:33
Woodi just don't get thet .ast stuff
19:33 nadim left, buharin joined
Woodi pleas look on example, last line in action: I changed c into 3 but it is ignored anyway 19:33
19:34 lichtkind__ joined
Woodi
.oO( how to do s/c/3/g with grammar ? ) :)
 19:34
dalek DBIish: 8d8f00f | (Salvador Ortiz)++ | lib/DBDish (3 files): 19:36
DBIish: $dbh and $sth objects implements 'dispose'
DBIish:
DBIish: The Connection and StatementHandle roles implements an informal
DBIish: Disposable interface.
19:36 dalek left, dalek joined, ChanServ sets mode: +v dalek
perlpilot m: gist.github.com/perlpilot/742924b6e1f6ffc6d1d6 19:37
camelia rakudo-moar b243a9: OUTPUT«[[3] [3 3] [] [3]]␤｢abcabccaabbc｣␤ abc => ｢abc｣␤ bc => ｢bc｣␤ c => ｢c｣␤ abc => ｢abcc｣␤ bc => ｢bcc｣␤ c => ｢c｣␤ c => ｢c｣␤ abc => ｢a｣␤ abc => ｢abbc｣␤ bc => ｢bbc｣␤ c => ｢c｣␤»
perlpilot Woodi: does that help?
Woodi: I'm still not clear on exactly what you want.
sortiz too much for dalek? ;-)
19:37 lichtkind_ left
arnsholt sortiz: I think there's a rate-limiting bug. Occasionally gets triggered when many commits are pushed at once 19:38
19:39 CIAvash left
sortiz arnsholt, That was the case, seven commits in a row, sorry. 19:40
RabidGravy poor dalek
perlpilot dalek is supposed to recognize many commits and aggregate the announcement. I wonder what its threshold is for that 19:41
Woodi perlpilot: it does something very unexpected :) 19:43
sortiz RabidGravy, BTW, I'm developing DBDish::ODBC as a separated module and want to stay that way to test the common services in DBDish for driver developers. 19:44
RabidGravy good plan, I was thinking of doing an Informix driver like that TBH
19:46 pmurias left 19:47 virtualsue left
sortiz The idea is that a driver only 'need DBDish' in its modules. (and NativeCall, if needed, in its DBDish::Foo::Native) 19:47
19:49 sufrostico joined 19:50 virtualsue joined 19:51 espadrine left
sortiz The drivers developer documentation will need a good expansion, btw. 19:52
arnsholt sortiz: Don't worry about it; not your fault =) 19:53
19:53 pmurias joined
arnsholt It's basically sign of having done solid stuff =) 19:54
19:54 sufrostico left
colomon m: say “989898988898989898980909090939838302233848473393040383234234509484.3489523478234723847238432423”.Rat.nude 19:58
camelia rakudo-moar b243a9: OUTPUT«(9898989888989898989809090909398383022338484733930403832342345094843489523478234723847238432423 10000000000000000000000000000)␤»
19:59 yqt left
colomon m: say (989898988898989898980909090939838302233848473393040383234234509484.3489523478234723847238432423).Rat.nude; 19:59
camelia rakudo-moar b243a9: OUTPUT«(9898989888989898989809090909398383022338484733930403832342345094843489523478234723847238432423 9999999999999999583119736832)␤»
colomon why are those two not the same?
sortiz MadcapJake`, Now pushed, you can see github.com/perl6/DBIish/blob/maste...e.pm6#L27, and if you add there 'note "Destroing";' you can see it working when running the tests. 20:00
lizmat hmmm.... so why is dalek still here and where is #p6dev ?
20:01 sufrostico joined
colomon is afraid something is broken in Rat literals in rakudo. 20:02
Skarsnik gah my blog post is totally mess up with the font change <<
sortiz #p6dev !?!?
hoelzro maybe #p6dev should be in the channel subject for the next few days or something 20:03
in case interested parties missed it earlier!
20:04 buharin left
sortiz I'm interested and missing that 20:04
FROGGS I dont get that:
m: class A { submethod BUILD ($thing = 111) { say $thing } }; class B is A { method new($thing) { self.bless: $thing } }; B.new(42)
camelia rakudo-moar b243a9: OUTPUT«111␤»
FROGGS m: class A { submethod BUILD (:$thing = 111) { say $thing } }; class B is A { method new($thing) { self.bless: :$thing } }; B.new(42)
camelia rakudo-moar b243a9: OUTPUT«42␤»
FROGGS what dont both output 42? 20:05
MadcapJake` sortiz, I've implemented mine as you have but it doesn't appear to be working, how can I force perl6 to run DESTROY?
hoelzro FROGGS: I thought new only passed named parameters to BUILD?
Woodi MadcapJake`: force GC ?
MadcapJake` yeah
hoelzro MadcapJake`: what are you doing that you need to force DESTROY? 20:06
MadcapJake` well I'm just testing, but I'm confused that when I place a say statement in there it isn't being printed to stdout
20:06 Praise joined
Woodi MadcapJake`: if you use GC then you relay on it. an it frees and sestroys when it decides too. 20:07
MadcapJake` but it should necessarily be run no matter what, right? 20:08
FROGGS hoelzro: I was thinking that I kinda call BUILD directly when calling bless, but I probably dont
hoelzro MadcapJake`: not necessarily
Woodi MadcapJake`: I think not
MadcapJake` why?
hoelzro not unless --full-cleanup is used with moar
jnthn MadcapJake`: No, we don't promise to run DESTROY blocks at exit.
nine MadcapJake`: no, there's no guarantee that a DESTROY will ever run
yoleaux 14 Mar 2016 22:35Z <jnthn> nine: OK, maybe if that's going to land soonish I'll hold on :)
MadcapJake` ohh, why is that? 20:09
hoelzro FROGGS: oh duh, I didn't read closely enough and missed your impl of new =/
20:09 pierrot joined
Woodi MadcapJake`: and it is industry standard to have such cleenup with GC :) 20:10
20:10 cognominal left
hoelzro (this question/answer for DESTROY should probably be added to doc.perl6.org/language/faq) 20:10
MadcapJake` Woodi, I don't follow what you mean by that
Skarsnik That a really bad if we can't easily ensure some cleanup easily x)
20:10 cognominal joined 20:11 cognominal left
sortiz MadcapJake`, that why I use the pattern 'with $foo = Foo.new { LEAVE { $foo.dispose } ... }' when need a native object deallocated when not needed any more. 20:11
Woodi MadcapJake`: en.wikipedia.org/wiki/Finalizer and related
20:11 cognominal joined
MadcapJake` Skarsnik, are you saying that you would prefer it to be guaranteed to run? 20:11
jnthn MadcapJake`: Because the GC's job isn't management of non-memory resources, and DESTROY is simply a callback when the GC can free an object's memory.
Skarsnik Yes x)
jnthn MadcapJake`: We don't conflate the two in Perl 6, like in many other languages. 20:12
Skarsnik Well it can have another name?
jnthn No
Too late.
(as in, many other languaes *also* don't conflate the two)
20:13 buharin joined
MadcapJake` jnthn, ok so you're saying that DESTROY is just a "hook" so to speak? is it because I'm using nativecall and the resources aren't managed by P6 that DESTROY isn't run? 20:13
jnthn MadcapJake`: I'm saying that it's just not meant to be a hook for freeing unmanaged resources, because you can't rely on when it will be run. 20:14
MadcapJake`: And so far not making it reliably run at exit is doing a good job of catching people mis-using it for that :)
MadcapJake` so to use it with unamanaged resources is impropr?
jnthn Yes
MadcapJake` ahhh i see
jnthn We provide LEAVE (do stuff at scope exit) and END (do stuff at program exit) for predictable we-promise-to-run-it things 20:15
The thing is that GC is generational also
20:15 darutoko left
timotimo aye, as soon as something gets considered old, it'll take quite a bit longer for it to be considered for eviction 20:16
jnthn So sure, in a small example you might find your objects get cleared up after a very short time
RabidGravy implementing phasers for variables would be helpful in this regard
jnthn But if the conditions are Just Right you could end up with things living for seconds or even minutes.
MadcapJake` jnthn, if I'm writing a module that uses nativecall, how can I add to LEAVE/END when a user uses my module?
Skarsnik What are the way to handle clean up stuff without doing it manually? Because it feel like you need to have array or something to keep track of stuff to close/unlock/send quit msg and do it at END?
jnthn Well, that we really need is something like C#'s using, and a standard role that you implement. I think somebody was exploring this in module space recently... 20:18
But for now it's like with file handles, taps on supplies, and so on; the expectation is the code using them will .close them
RabidGravy you can do a trick with an add_phaser on the calling block I guess
sortiz Yep, the the idea of my (now informal) Disposable role.
jnthn sortiz: Ah, it was you :) 20:19
timotimo Skarsnik: help, i can't install gptrixie
Skarsnik Because it's handy to have something that say "I am done with this object" so you can do clean up
timotimo "Could not find GPT::FileGenerator in: [...]"
Skarsnik timotimo, damn why?
Oh
sortiz On time, with macros I can add 'using Foo.Allocate() { ..... }' and add the needed LEAVE { } inside.
20:20 kanishka left
jnthn sortiz: Ideally we'll add something to Perl 6 for this because people keep on asking :) 20:20
Skarsnik should be fixed
I forget to add the file to META 20:21
MadcapJake` is there a way to add to LEAVE/END programmatically?
timotimo Skarsnik: also, i don't see how to get gccxml on my fedora 23
oh, "castxml" 20:22
i can install that, but your readme says that's bad
MadcapJake` timotimo, don't
check if you have gccxml.real
timotimo there's also no gccxml.real
Skarsnik there is for 22, weird
20:22 zakharyas joined
jnthn MadcapJake`: Well, ENDs are global so it doesn't matter really what does the END 20:22
timotimo well, my system is on 5.3.1 20:23
gcc, that is
jnthn MadcapJake`: For LEAVE, no, not yet.
timotimo your readme says i'd need 4.9
20:23 cdg joined
MadcapJake` jnthn, so if i add an END to my module, and someone uses it, will the END get run? 20:23
sortiz MadcapJake`, Yes, but now is complicated, you need mangling the QAST, with macros should be easy.
jnthn MadcapJake`: ye
*yes
Woodi MadcapJake`: what id process gets kill -9 signal ? 20:24
skids MadCapJake`: github.com/skids/perl6-Control-Bai...l.pm6#L107 if you care to go down to QAST.
MadcapJake` Woodi, no idea
timotimo Skarsnik: maybe you can generate the binding for github.com/grimfang4/sdl-gpu/ for me; probably just include/SDL_gpu.h
MadcapJake` jnthn, great! The C library I'm wrapping is a one go kind of thing and offers a clean function if you intend to parse something else, so I think using an END phaser makes sense 20:25
Skarsnik timotimo, maybe try with cast? It seem the last git version handle c99-c11
MadcapJake` Skarsnik, cool!
according to fedoraproject.org, gccxml is now obsolete as of f23 so it probably won't get packaged anymore
timotimo Installing : castxml-0.1-0.10.20160125gitfc71eb9.fc23.x86_64 1/1 20:26
Skarsnik Yes, the cast/gccxml business is annoying
FROGGS damnit, I have a case here where submethod BUILD is called twice... strange
Skarsnik I did not manage to compile the git version of cast
timotimo can i tell gptrixie to use "castxml"?
Skarsnik I mean I never managed to compile a project based on clang/llvm libs xD 20:27
GPT_GCCXML=castxml
timotimo ah, i see
Skarsnik I just added that today
timotimo lovely
error: unknown argument: '-fxml=plop.xml'
*sigh*
that's not how you castxml :) 20:28
20:28 domidumont left
Skarsnik need to patch gptrixie x) 20:28
20:28 domidumont joined
MadcapJake` Skarsnik++ # GPT_GCCXML 20:29
Woodi could someone add some API to grammars, pleas ?
MadcapJake` Woodi, what do you mean?
Woodi MadcapJake`: something... 20:30
MadcapJake` o_O
FROGGS Woodi: what are you talking about?
Woodi momento
lizmat clickbaits p6weekly.wordpress.com/2016/03/14/...ueling-up/ 20:31
Woodi could someone add some API to grammars, pleas ?gist.github.com/anonymous/e0befbf2f05fd84e1c5b 20:32
20:32 chienjo joined, virtualsue left
FROGGS Woodi: could you please try to explain the problem you experience? 20:33
20:33 domidumont left 20:35 kaare_ joined, virtualsue joined
Woodi FROGGS: I just want to work with parts catched by tokens but whatever i put into action methods is a) ignored or b) overrided somehow 20:36
20:36 AlexDaniel joined
Woodi I think make $<abc> overrides $<bc>... 20:36
20:36 MadcapMobile joined
Skarsnik timotimo, it use sdl or sdl2? 20:37
20:37 chienjo left
timotimo it can do both, i'd like the one for sdl2, but i don't know how to ensure that's the one it uses 20:37
20:39 kaare__ joined
Skarsnik why it install me libudev with libsdl2 x) 20:39
RabidGravy is there a special incantation to make internal links in to an "anchor" in POD or is just L<Whatever|#Whatever> ? 20:40
timotimo Skarsnik: that's just suggested dependencies, i expect
20:40 kaare_ left
RabidGravy and just hope that whatever generator DTRT 20:40
FROGGS Woodi: do you expect that the letters a and b show up in the output?
sortiz FROGGS, why you choose 'my str $s' when there are other '$s' in scope? Can that be related to your Heisenbug? 20:41
Woodi FROGGS: yes
FROGGS sortiz: I did not choose that, that was already there
Woodi: but you dont capture these matches in e.g. BuuAbc1 20:42
Skarsnik -I /usr/include/SDL2/ to make it use sdl2 x)
sortiz FROGGS, Oops, sorry, I'm reading the code and found that, seems 'unusual' at least :-) 20:43
timotimo a guess that makes sense
FROGGS Woodi: it "makes" what c had matches, then in bc it makes what c had produces via make (a list of c's), and then in abc it makes what bc had produced (a list of lists of c's)
Skarsnik I should have a cache or something -Functions: 1328 20:44
and it take a minute to parse the xml file x)
FROGGS sortiz: aye, but should not be a problem
Woodi FROGGS: hmm, eg. <abc> is [ 'a' <bc> ] so $<abc> should have it, IMO
20:44 pmurias left
FROGGS Woodi: $/ in method abc will also have the a, yes 20:44
20:46 kaare__ left, pmurias joined
FROGGS m: gist.github.com/FROGGS/8c9556cf97fb301f8f84 20:47
camelia rakudo-moar ba5270: OUTPUT«TOP => [a => [b => [c]] a => [b => [c c]] a => [b b => [c c]] a => [] a => [b => []] a => [b => [c]]]␤»
FROGGS Woodi: are you looking for that?
20:47 kaare__ joined 20:50 kaare__ left
Woodi FROGGS: actually I want to have default say $r[.ast]? "view" with 'c' changed into '3'. just want to know how to operate on parts. 20:51
20:51 sufrostico left
Skarsnik timotimo, did you get my dcc? 20:52
FROGGS Woodi: sorry, I dont understand
timotimo i just tried to accept it
then it immediately said failed
225 bytes really isn't very much 20:53
what went wrong?
Skarsnik timotimo, www.nyo.fr/~skarsnik/sdlgpu.txt
20:54 wamba left
timotimo that's significantly more than 225 bytes 20:54
20:54 cognominal left
Skarsnik hm the uint32_t are standard? 20:54
timotimo yeah, why wouldn't they be? 20:55
20:55 cognominal joined
timotimo have you considered taking comments in doxygen style and turning them into pod attached thingies? 20:55
Skarsnik I will need to use doxygen for that 20:56
20:56 virtualsue left
Skarsnik They get catched because I go backward to find like a empty line or a ; to find the function definition (I only have the end line) 20:56
timotimo oh 20:58
Woodi m: gist.github.com/anonymous/3a768052cdfd807e6d00 20:59
camelia rakudo-moar 93d597: OUTPUT«｢abcabccabbccaababc｣␤ abc => ｢abc｣␤ bc => ｢bc｣␤ c => ｢c｣␤ abc => ｢abcc｣␤ bc => ｢bcc｣␤ c => ｢c｣␤ c => ｢c｣␤ abc => ｢abbcc｣␤ bc => ｢bbcc｣␤ c => ｢c｣␤ c => ｢c｣␤ abc => ｢a｣…»
20:59 kaare__ joined
timotimo Skarsnik: what are these things in the "Extras stuff" section? 21:00
does that even compile?
Skarsnik my priority is also writing a libxml2 binding or something to not have to wait a minute at each run
Woodi FROGGS: this is just what parse do without actions. how change 'c' into '3' in that output ?
Skarsnik It's supposed to be typedef
like if you have a typedef int myintbecause; it generate a constant myintbecause = int32; 21:01
timotimo in this case i'll just have to throw that section out? 21:02
21:02 sufrostico joined
Skarsnik It does not work very well when taking a file that include a lot of other stuff 21:02
21:03 n1cky joined 21:05 skids left 21:06 kaare__ left
FROGGS Woodi: "make 3" in the c method 21:07
dalek osystem: c2e3774 | RabidGravy++ | META.list:
Add Audio::PortAudio

See github.com/jonathanstowe/Audio-PortAudio Kudos to psch++ for starting it :)
21:07 kaare__ joined
RabidGravy There 21:07
Skarsnik timotimo, you can probably delete them. I will have a look tomorrow of what I can do to improve the case of single file that need lot of other stuff
Woodi FROGGS: it don't change anytching for me... 21:08
21:08 zakharyas left
Woodi FROGGS: but Match is ro ? 21:08
FROGGS m: gist.github.com/FROGGS/8c9556cf97fb301f8f84 21:10
camelia rakudo-moar 93d597: OUTPUT«TOP => [a => [b => [3]] a => [b => [3 3]] a => [b b => [3 3]] a => [] a => [b => []] a => [b => [3]]]␤»
21:10 itaipu left
FROGGS Woodi: why dont you edit the gist I showed you? 21:11
Woodi I can ? :) 21:12
FROGGS well, you can at least improve on that code, so we are talking about the same :o)
Skarsnik timotimo, if you want to have fun www.nyo.fr/~skarsnik/sdluniverse.txt (everything the header imply) 21:13
Woodi FROGGS: ah, you made a and b into ()... 21:14
timotimo i wonder how long perl6 takes to parse that :D
21:15 kaare__ left 21:16 kaare__ joined
timotimo ------> enum ._27⏏ is export ( 21:16
starts out super well :)
Skarsnik must be an anonymous enum x) 21:17
timotimo yeah, multiple of those
FROGGS Woodi: yes, and I put these captures into what the action methods make
21:17 musiKk_ left
Skarsnik hm, I should auto exclude ctypes.h probably 21:19
pthread or other stuff are autoexcluded 21:20
timotimo Skarsnik: it doesn't do any ordering of things? 21:21
or automatic stubbing 21:22
Skarsnik it ordered on how it appear on the headers
21:22 zakharyas joined
Skarsnik but the file order is whatever gccxml does with the id. I think it 'should' avoid putting something that need something that came after 21:24
timotimo could you do a topological sort of the classes and output stubs where needed? 21:29
otherwise i'd have to go through and do it manually now :)
also, uint32_t wants to become uint32 actually
when i said "of course it's defined", i misunderstood you
21:30 bjz left 21:32 bjz joined
Skarsnik I am not sure if it can happen. do you have an example in this? 21:32
21:32 buharin left
Skarsnik Oh right sdl_surface need a sdl_rect 21:35
21:36 yqt joined, bjz left
timotimo i've had to stub two things so far 21:37
GPU_Renderer and GPU_Target
(i'm not using the universe version)
21:38 kid51 joined
timotimo SDL_BlitMap seems to be an empty struct 21:39
21:39 ely-se joined
timotimo SDL_RWops is really a tiny bit b0rked :) 21:40
we need some better "name mangling" to make all of this work
i have a thing that compiled successfully!
i have no clue if it'll work, but it does exist now :) 21:41
21:41 kanishka joined
Skarsnik I use gccxml generated name for anonymous stuff x) 21:44
dalek kudo-star-daily: a421619 | coke++ | log/ (8 files):
today (automated commit)
Skarsnik well the mangled name
anonymous enum are annoying :'
timotimo yeah, well, if they have dots in their name ... :) 21:45
Skarsnik anonymous struct have a name attribute empty, so I know they are anonymous
anonymous enum have a name x)
21:47 kaare__ left 21:51 zakharyas left, virtualsue joined 21:54 bjz joined 21:55 virtualsue left
timotimo fish: “perl6 -Ilib example.p6” terminated by signal SIGSEGV (Address boundary error) 21:55
so close now :)
just trying to print the $screen (an GPU_Target) got me into a segfault when trying to output one of its attributes 21:58
not printing it gives me a main loop that draws the black screen over and over :)
Skarsnik: GPT is well done! :)
can we has something to get #define in place?
there's a GPU_DEFAULT_INIT_FLAGS defined to be 0
that could show up in the binding, too, imo
21:59 bjz left
timotimo i'll just go ahead and push the current state up to github and see where it leads me 22:01
Skarsnik gist.github.com/Skarsnik/155f3c8062054075d4b0
22:01 pmurias left, zakharyas joined
timotimo you shouldn't be using | there 22:02
you need +|
Skarsnik This part is 'dumb' it's not a c preprocessor. I manually found #define starting_pattern* xxxx 22:03
timotimo oh, haha 22:04
well, you could have committed the code, too :)
Skarsnik I just run it with --define-enum=MYGPU:GPU_
MadcapJake` what does it mean for a C parameter to be "out"? 22:05
Skarsnik be out?
RabidGravy isn't that a C++ thing
MadcapJake` the C doc says «@param[out] status of this operation»
it's only in the docs 22:06
RabidGravy what is the actual parameter signature? I'd guess that it's something that is only used as an "rw" parameter, typically a pointer that will be set by the function 22:07
Skarsnik @param[out] look like a doxygen stuff
RabidGravy yeah
MadcapJake` github.com/lexborisov/myhtml/blob/...#L886-L895 22:08
Skarsnik it's the name of the parameter x)
It's probably to tell you the function will have this parameter becaming an output 22:09
I should translate to : myhtml_status_t $status is rw; 22:10
MadcapJake` ah so it's *where* the status will get stored?
Skarsnik Yes 22:11
RabidGravy oh, yeah it's a pointer so the function can alter the value 22:14
22:16 rindolf left
Skarsnik with this type of lib you probably want to call myhtml_parse and visit the given tree based on how the structure are defined 22:17
timotimo Skarsnik: next up we need a binding for OpenGL with gptrixie :) 22:19
Skarsnik I need a libxml2 binding to speed us the parsing x) 22:21
I tried libminixml but this lib is buggy 22:22
I did not even managed to have a working thing in C xD
arnsholt MadcapJake`: Presumably, an out parameter is a pointer that's written to by the library to pass out the result of the operation 22:29
22:29 lostinfog left 22:30 zakharyas left 22:34 Relsak joined 22:38 sufrostico left 22:39 sufrostico joined
timotimo Skarsnik: how would i handle things like "i know this pointer to uint16 is meant to be translated as "uint16 $foo is rw" rather than a Pointer[uint16]" 22:39
i know gpt takes a config file, but i'm not sure if it allows overrides like that 22:40
Skarsnik github.com/Skarsnik/gptrixie/issues/3 :)
timotimo i hope the title means more to you than it does to anybody 22:43
because sometimes you want Pointer[*] instead
Skarsnik It's only for fundamental type 22:46
timotimo yeah, but still
this library sometimes has "float *matrix"
for example
Skarsnik hm 22:47
It's not even a typedef?
22:47 FROGGS left
timotimo correct 22:47
github.com/timo/SDL2_GPU/blob/mast....pm6#L1940 22:48
Skarsnik typedef are ****
timotimo just one example
Skarsnik duh 404
timotimo ah, i renamed that
github.com/timo/SDL2_GPU/blob/mast....pm6#L1940
i find it strange that the paths i see here start in /root 22:49
Skarsnik Ok, so it's not always a good idea
timotimo correct
Skarsnik I work in root in this vm, don't ask why 22:50
timotimo OK
it seems like there are exactly two functions where the Pointer is needed to be turned into an "is rw" instead 22:51
GetVirtualCoords and GetVirtualResolution 22:52
Skarsnik I need to figure how to be able to say that in the conf file like function->GetVirtualCoords->param<foo>:isrw; 22:57
22:59 justine joined 23:09 ely-se left, labster left 23:12 hc_ joined, hc_ is now known as Guest95637 23:13 Guest95637 left, salv0 left 23:15 spider-mario left 23:17 vendethiel left
RabidGravy Skarsnik, you have the wrong path to the FileGenerator.pm6 in the provides of the META.info for GPT 23:21
Skarsnik ><
I closed the vm
RabidGravy do it in the browser just needs /GPT adding to the line ;-) 23:22
23:23 Relsak left 23:27 kid51 left
masak 'night, #perl6 23:28
RabidGravy toodles masak 23:29
Skarsnik, I fixed it locally and installed, no need to worry until next time you're in there
23:31 jack_rabbit joined
Skarsnik Oh it's nice the online edit stuff ^^ 23:32
23:34 justine left 23:36 chee is now known as CHEE, CHEE is now known as chee
RabidGravy yeah, not too shabby, I'm not sure I'd want to write a whole program with it but it can get you out of problems 23:37
Skarsnik Yeah to fix a typo it's nice 23:39
23:42 jameslenz left 23:44 RabidGravy left
Juerd m: sub infix:<except>(@list, $exception) { @list.grep: * !~~ $exception }; say ^10 except 5; say <foo bar baz quux> except 'bar'; say ^24 except /^1/; 23:46
camelia rakudo-moar 8cbb1e: OUTPUT«(0 1 2 3 4 6 7 8 9)␤(foo baz quux)␤(0 2 3 4 5 6 7 8 9 20 21 22 23)␤»
23:48 jack_rabbit left, Actualeyes left 23:56 [particle] left 23:57 SCHAAP137 left, coreyperry 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!