»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend! | Rakudo Star Released!
Set by diakopter on 6 September 2010.
00:02 tadzik joined
tadzik 'evening 00:02
TimToady like y's 00:03
00:04 patspam left 00:06 Italian_Plumber left
TimToady does rakudo HEAD use gc_massacre now? 00:10
tadzik I think so
btw, where is recent Rakudo on feather?
TimToady professes complete ignorance 00:11
00:18 eternaleye left, eternaleye joined
TimToady ooh, my hofstadter-conway $10000 sequence got further than it's ever gotten before, so it looks like the old GC limit is gone 00:20
maybe I'll be able to take the caveat out of RC
00:21 Italian_Plumber joined
lue ohai o/ 00:24
00:24 risou joined
lue How would I get a website that can run Perl 6? .oO[ Please don't say home server, please don't say home server... :) ] 00:26
TimToady define "get"
sorear google VPS 00:27
00:27 ash____ joined
TimToady for instance, try.rakudo.org/ is a website that can run Perl 6... 00:28
ash____ rakudo master has the gc_massacure in it
00:28 ash____ is now known as ash_gti
lue [rephrase] What would the best way be to have a website running Perl 6 code? [I tried coding a website again, and I can't help but notice how much better Perl 6 would be over JS :)] 00:29
ash_gti you'd need to have some way of interfacing with your http requests
like CGI
or FCGI
or a standard gateway of interacting with the request like wsgi
and a server that can hand off the request to your code 00:30
most web server's allow cgi (if not all)
you just need to have a host that allows you to compile code
tools.ietf.org/html/rfc3875 is a detailed description of the CGI interface 00:31
i don't think anyone has written a perl6 based library for it yet 00:32
well actually masak had something called Web.pm that might do it
tadzik website running Perl code, you mean a website written in Perl?
lue I want to use Perl 6 over JavaScript, that's what I'm getting at. I know full well it can't be *exactly* like JS, considering browsers don't come with P6 implementations :) 00:34
ash_gti it might be worth asking masak++ aobut web.pm
oh, you want it in the browser? 00:35
not on the server side?
tadzik please define: Perl6 over JavaScript
00:35 perlygatekeeper left
lue ash_gti: I know it won't be browser side. I just want to use it for the same stuff you tend to use Javascript for. 00:36
tadzik: I would much rather code in Perl 6 than JavaScript :) 00:37
tadzik well, you usually use JS for the browsers' side, won't you?
s/wo/do/
jnthn c7808c3...wtf. *sigh*
TimToady: Dislike. Suggestion: make the default :local. 00:38
Everytime we add something like this, we add an extra expectation on anyone implementing their own meta-objects (or at least, those that implement the HOW API). 00:40
Plus from my experience so far, I often *want* :local
(by default)
tadzik en.wikipedia.org/wiki/Diamond_problem # I see a nice room for Perl 6 00:41
lue *cough* I would like to use Perl 6 for coding various things on my site that one normally uses Javacript for (such as navigation, easy updating of stuff that appears on every page, etc.). I do not care if it's client-side or server-side [I actually expect it to be server-side]. 00:42
00:44 QinGW joined 00:49 raiph joined 00:50 eternaleye_ joined, eternaleye left
TimToady jnthn: local doesn't make much sense in the presence of mix-ins 00:55
jnthn Hmm, true. 00:58
Excluding Any/Cool/Mu just feels like a hack though.
TimToady there are good linguistic reasons 00:59
jnthn If you want it, please introduce a general "hidden from introspection" trait or something that can be looked at rather than hardcoding type names.
Fine, but the current technical solution is sub-optimal.
If I'm reading correctly, we have Mu/Any/Cool just being magically excluded. 01:00
TimToady I know, we'll name the hidden classes with a '.' on the front :)
jnthn ...I was hoping for something better. ;-)
:-)
TimToady we can certainly have a bit somewhere
jnthn I can go with the linguistic arguments, I just have a knee-jerk "oh noes" against "these types are magical". 01:01
OK, if we can make exclusion boil down to a property on the meta-object, I'm OK with it. 01:02
I just don't want all meta-objects to have to start hard-coding type names.
01:07 plobsing left 01:09 plobsing joined 01:10 jferrero left
jnthn tries to catch up on sleep a bit 01:15
01:15 patspam joined
raiph . o O ( \o/ ) 01:18
01:19 plobsing_ joined 01:20 plobsing left 01:24 Italian_Plumber left 01:28 patspam left 01:45 molaf_ joined 01:46 cdarroch left 01:47 envi^home joined 01:48 molaf_ left, LaVolta joined 01:49 molaf left 01:51 molaf joined 01:53 mikehh left 01:57 mikehh joined 02:04 nymacro joined 02:16 agentzh joined 02:20 redicaps joined 02:25 nymacro_ joined 02:27 rrgau_`` joined 02:28 nymacro left, nymacro_ is now known as nymacro 02:30 rrgau_` left 02:34 meppl left 02:37 proller_ left 02:40 proller joined 02:41 jaldhar joined, jaldhar left 02:42 jaldhar joined 02:54 nymacro left 02:58 nymacro joined 03:02 molaf left 03:09 redicaps left 03:14 rrgau_`` left
araujo tests new code for his language 03:18
03:20 kid51 left 03:32 eternaleye_ is now known as eternaleye, LionMadeOfLions joined
LaVolta wait...what?! REPL now accept directly input of unicode chars! 03:38
03:39 PZt joined
LaVolta i ❤ you guys :D 03:40
nymacro xoxo 03:41
call me
araujo fact -> { :n if <= n 1 { 1 } { * n fact - n 1 } } 03:48
:)
03:49 leprevost joined 03:55 DemoFreak left 04:01 DemoFreak joined, DemoFreak left 04:02 DemoFreak joined 04:13 ash_gti left 04:14 smosher joined 04:26 justatheory left 04:54 mfollett joined 05:10 jaldhar left 05:12 jaldhar joined 05:15 risou left, risou joined 05:20 risou left 05:30 risou joined 05:33 risou left, risou joined 05:38 risou left
araujo ideone.com/ZLLHx 05:44
06:01 mfollett left, mfollett joined 06:04 Guest23195 joined, raiph left 06:07 nymacro left 06:11 satyavvd joined 06:19 satyavvd left 06:32 molaf joined 06:48 plobsing_ left 06:49 icwiener joined
sorear I'm a little confused by how the type system interacts with binding. 06:51
I'm thinking about "sub foo(Int $x is rw) { $x = 1 }" 06:52
what can I call foo on?
my Int $k; obviously
what about my $k; ?
what about my $k = 5; ?
what about subtype Even of Int where * !%% 2; my Even $z; ? 06:53
or class SubclassOfInt is Int { }; my SubclassOfInt $z; ?
How much of this changes if 'is rw' is removed? 06:54
Does Perl 6 have anything like an InvalidArrayStoreException?
06:56 mfollett left 07:06 nymacro joined, Ross joined, Ross left, Ross joined, nymacro left 08:08 schpey joined
moritz_ good morning 08:09
08:11 nymacro joined 08:12 nymacro left 08:18 xinming left 08:19 xinming joined 08:36 Trashlord left 08:39 Trashlord joined
sorear hello moritz_ 08:45
08:45 dual left 08:54 molaf left
cosimo hi everyone 09:13
moritz_ o/ 09:15
09:16 Grrrr joined
tadzik good morning 09:30
cosimo rakudo: my $x = Mu; my $y = 42; my $z = $x // $y; say $z;
p6eval rakudo 13c8a2: OUTPUT«42␤»
09:35 icwiener left, icwiener joined 09:36 icwiener left 09:42 clintongormley joined
clintongormley Hi all - anybody have any idea what has happened to Adriano Ferreira? 09:43
I see that none of his many CPAN modules has received any loving in the last year or so
i haven't tried emailing him yet, but anybody seen him recently?
moritz_ clintongormley: do you happen to remember his nick? 09:47
clintongormley hi moritz_ - no, no idea i'm afraid
on cpan he is just 'ferreira'
sorear You might have better luck asking on #toolchain or #p5p or #perl (all MAGnet) 09:48
moritz_ clintongormley: and here too... last activity in Dec 2007: irclog.perlgeek.de/search.pl?channe...ira&q=
clintongormley hmm, wonder if he is ok 09:49
sorear: yes?
actually, i'll just email him and see if i get a response
thanks all
moritz_ blog.rafaelferreira.net/ # last blog post from June 08
09:50 kjeldahl joined
sorear clintongormley: FERREIRA/Compress-unLZMA-0.04 uploaded 17 July 09:50
(2010)
clintongormley ah interesting! so he is alive and releasing
good to know
moritz_ www.perlmonks.org/?node_id=861938 any Perl 6 solutions? 09:56
sorear for @array.natatime(3) { .say } 09:58
moritz_ is that specced?
cosimo rakudo: my $x = False; say $x.perl; 09:59
p6eval rakudo 13c8a2: OUTPUT«Bool::False␤»
10:00 icwiener joined
sorear not explicitly, but I was under the impression all of List::'More'?Utils was going to be brought into core as methods 10:01
araujo adds support for nested lists 10:02
10:04 QinGW left
tadzik where is the newest Rakudo on feather? 10:06
moritz_ rakudo: say (0, 3 ...^ 10)
tadzik: problably nowhere
p6eval rakudo 13c8a2:
..OUTPUT«(timeout)821242730333639424548515457606366697275788184879093969910210510811111411712012312612913213513814114414715015315615916216516817117417718018318618919219519820120420721021321621922222522823123423724024324624925225525826126426727027327627928228528829129429730030330630931231531…
tadzik hm 10:07
mind if I build some?
moritz_ no
sorear What is Rakudo build memory usage like these days? 10:10
moritz_ hasn't monitored it after The Big GC Merge
sorear top(1)s feather1 looking for tadzik 10:11
tadzik there's htop, don't be so spartan 10:12
sorear hmm, tadzik not in who
tadzik :)
NYLI
10:15 somebody joined 10:19 LaVolta left
moritz_ rakudo: my @l = <a bb c d e f g h>; sub chunky(@l, $len) { (@l Z (' ' xx $len - 1, "\n") xx *).join.substr(0, -1) ~ "\n"; 10:23
p6eval rakudo 13c8a2: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 22␤»
moritz_ rakudo: my @l = <a bb c d e f g h>; sub chunky(@l, $len) { (@l Z (' ' xx $len - 1, "\n") xx *).join.substr(0, -1) ~ "\n"; }; print chunky(@l, 3)
p6eval rakudo 13c8a2: OUTPUT«a bb c␤d e f␤g h␤»
araujo ideone.com/iLZ1V 10:24
10:24 agentzh left
araujo I guess [ ] is sane enough :P 10:24
sorear niecza: say "@@L1" # Oops. 10:27
p6eval niecza eb45fca: OUTPUT«===SORRY!===␤Invalid hard reference syntax at /tmp/P10UPdLYXL line 1:␤------> say "@⏏@L1" # Oops.␤ expecting any of:␤ desigilname␤ twigil␤Parse failed␤»
sorear niecza: say '@@L1' # Oops.
p6eval niecza eb45fca: OUTPUT«Use of uninitialized value in substitution iterator at /home/p6eval/niecza/src/CodeGen.pm line 850.␤Use of uninitialized value in substitution iterator at /home/p6eval/niecza/src/CodeGen.pm line 850.␤␤»
10:29 somebody left
tadzik Parrot VM: PANIC: Out of mem! 10:30
moritz_ tadzik: please file a parrot bug
tadzik crap, I have to register? 10:32
sorear tadzik: trac sucks
tadzik :F
sorear the alternative is getting ~10 spam bugs /hr
tadzik well, on github you have to register too 10:33
10:33 plobsing joined
sorear ok, rakudo on gc_massacre is still using an egregious amount of memory 10:34
I killed src/Perl6/Grammar.pm at 350mb
(parrot-nqp even, not rakudo)
10:38 clintongormley left
sorear tadzik: sure that's a bug? 10:44
tadzik moritz_ | tadzik: please file a parrot bug
sorear also, why is Parrot allocating memory in chunks of 4096 bytes 10:45
moritz_ why shouldn't it? 10:46
sorear because quite a few mallocs like to allocate memory in round numbers of bytes 10:47
since free(p) needs to know how many bytes to free, malloc(4096) needs to allocate the nearest round number above 4100 bytes 10:48
moritz_ where does that magic number come from? 10:49
sorear which one?
4096 or 4?
moritz_ 4 10:50
sorear 4 is sizeof(int) and thus the minimum possible malloc overhead on a system that can run Parrot 10:51
*minimum sizeof(int)
in practice malloc overhead is almost always sizeof(double) due to alignment concerns; 16 bytes is the traditional amount to leave before 2^n 10:52
(or you just include your own allocator, like perl5 does) 10:53
plobsing sorear: I must have a magical malloc then, because perl6 -e '' seems to be running brk at increments of 4K 10:54
10:56 rgrau_ joined
sorear plobsing: what libc do you have? I should run a proper survey some time instead of sharing conventional wisdom 10:58
plobsing glibc 2.12.1-1 10:59
11:03 schpey left
sorear I have libc 2.11.2-2 11:14
allocating 2 bytes 1M times -> 15708kb 11:15
colomon sorear: been a long time since I thought about it, but a smart memory allocator theoretically doesn't need more than one bit overhead per allocation, does it? though that doesn't deal with alignment issues. 11:17
sorear colomon: if I could go back in time and change one thing... free(void*, int) wouldn't be #1 but it would be high up there 11:18
dalek ok: a6b8fdd | moritz++ | src/grammars.pod:
[grammars] try to simplify an explanation
 11:19
sorear testing locally allocations of 4093 through 4100 bytes use the same amount of real memory
glibc 2.11.2-2 appears to round up allocations to 8n+4
(i386)
setting the chunk size to 4096 increases brk 4k at a time (plobsing), although it's using 4104 bytes; my libc isn't magic but it's smarter than I gave it credit for 11:21
11:21 satyavvd joined
sorear actually it seems to be 8n+12; 4 and 5 bytes are same allocation 11:23
a little weird
colomon 32-bit or 64-bit? 11:24
sorear 32 11:25
colomon huh. maybe some sort of buffer at the end to provide primitive overrun protection? 11:26
sorear out 11:28
11:35 araujo left 11:38 zulon joined
colomon Did Rakudo release on Thursday? My impression from github is yes, but I haven't seen a release announcement anywhere yet... 11:46
moritz_ yes; announcement went to p6c
plobsing I'm eager to see the r* benchmark graph after this release.
tadzik R* is today? 11:49
moritz_ colomon: the new policy is to not send the Rakudo compiler announcement to too many places, to avoid confusion with the R* releases 11:50
colomon ah
colomon remembers he hasn't checked one of his e-mail addresses in a full week now... 11:55
moritz_ just explained the second solution in www.perlmonks.org/?node_id=861941 to my wife 11:58
(in detail, and with detours to named arguments)
mberends wife++ 11:59
and maybe, moritz_++ ;) 12:00
moritz_ :-)
12:00 meppl joined
plobsing karma wife 12:03
aloha wife has karma of 1.
12:07 rgrau_` joined
moritz_ and now we're producing SVG graphs, and because she wants a happy face I need to relearn SVG paths :-) 12:09
12:09 rgrau_ left
moritz_ moritz.faui2k3.org/tmp/happy-face.svg is the result 12:11
tadzik :) 12:15
plobsing contented H2O? 12:17
12:18 Ross left
moritz_ modern "art" :-) 12:19
12:20 Ross joined 12:22 zulon left 12:23 schpey joined 12:36 nymacro joined 12:51 araujo joined 13:11 satyavvd left
jnthn good afternoon, #perl6 13:12
tadzik good afternoon jnthn
13:15 zulon joined 13:17 Patterner left 13:18 jest joined
moritz_ rakudo: my $x = 5; enum A b => $x; say b 13:19
p6eval rakudo 13c8a2: OUTPUT«Could not find sub &A␤ in main program body at line 22:/tmp/bVfK2UBm2y␤»
moritz_ rakudo: my $x = 5; enum A (b => $x); say b
p6eval rakudo 13c8a2: OUTPUT«===SORRY!===␤Symbol '$x' not predeclared in <anonymous> (/tmp/sUkVnkYaiT:22)␤»
moritz_ it's so easy to find LTA error messages while reading through Actions.pm :/
jest hi 13:27
considering bug github.com/perl6/book/issues/issue/20, here is a fix: pastie.org/1180989
but I'm not sure if it should go to the book repo, or rather to Pod::PseudoPod::LaTeX 13:28
13:29 Psyche^ joined, Psyche^ is now known as Patterner 13:31 sasha joined
moritz_ tries it in the book repo first 13:33
jest: pushed to the fix_index branch of the book... I need to start another PC to check if it works (can't build the book on my laptop, the latex there is too old) 13:34
jest moritz_: ok; I just checked the change on my comp, the only diefference in TeX input is '!' replaced to '"!', and '!' appears in the index correctly (it wasn't previously) 13:36
13:36 sasha left
moritz_ indeed, works on my second machine too 13:39
jest++
dalek ok: e76fe15 | moritz++ | lib/Perl6BookLatex.pm:
fix index; patch courtesy by jest++
 13:41
13:41 sasha joined, sasha left 13:43 sasha joined, tadzik left 13:46 sasha left
moritz_ www.perlmonks.org/?node_id=861950 13:51
jnthn moritz_++ # it's nice to be able to tick things off lists like that :-) 13:54
moritz_ if I were a user of logging frameworks, I'd now be tempted to write one
I guess I'll rather work on my Sudoku solver instead :-) 13:55
13:55 jest left 13:56 tadzik joined 13:58 rgrau_` left 14:02 miso_soup joined
tadzik what do you think about including a module manager with Star this month? 14:08
moritz_ +1, if we have one :-) 14:10
TimToady www.perlmonks.org/?node_id=861959 14:11
tadzik well, I can say we have once someone tests neutro on windows 14:12
moritz_ tadzik: even linux-only would be great start
tadzik moritz_: it works on linux since the first commit ;P) 14:13
:)
what an ugly face I made
TimToady: I'm amazed
btw, is there a way to, uhm, reset .munch? 14:14
ah, create a new array
colomon TimToady++ 14:17
14:21 flussence joined
TimToady btw, if someone wants to work on my current favorite LTA message, it's "Use of uninitialized value in numeric context" with nary a mention of location. 14:22
moritz_ rakudo: say Any + 3 14:24
p6eval rakudo 13c8a2: OUTPUT«Use of uninitialized value in numeric context␤3␤»
TimToady that gets more irritating as your program gets larger 14:28
14:29 Chillance joined
TimToady and as your data becomes less toylike and random 14:29
14:29 dual joined
moritz_ $ ./perl6 -e 'say Any +3 ' 14:29
Use of uninitialized value in numeric context at line 1372:CORE.setting
3
fail :-)
and inspecting the calling location would point to the definition of infix:<+> 14:30
14:32 mfollett joined
TimToady rakudo: say callframe(2).line 14:37
p6eval rakudo 13c8a2: OUTPUT«1␤»
moritz_ but how do we know how many callframes to walk? 14:38
TimToady it's supposed to be able to do pattern matching
14:38 wamba joined
jnthn Walk until not in CORE.setting? 14:38
moritz_ jnthn: that's a good idea, I think 14:39
rakudo: my %h = a => 1, b => 2; say "a"trans(["a", "b"] => ["A", "B"]) 14:41
p6eval rakudo 13c8a2: OUTPUT«===SORRY!===␤Confused at line 22, near "say \"a\"tra"␤»
moritz_ rakudo: my %h = a => 1, b => 2; say %a.keys[0].trans(["a", "b"] => ["A", "B"])
p6eval rakudo 13c8a2: OUTPUT«===SORRY!===␤Symbol '%a' not predeclared in <anonymous> (/tmp/fLta1M1iUA:22)␤»
moritz_ rakudo: my %h = a => 1, b => 2; say %h.keys[0].trans(["a", "b"] => ["A", "B"])
p6eval rakudo 13c8a2: OUTPUT«too few positional arguments: 2 passed, 3 (or more) expected␤ in main program body at line 22:/tmp/uBDci8MhTZ␤»
colomon not again.... 14:44
moritz_ indeed not agian 14:45
I'm adding a few prefix:<~> to EnumMap.pm
TimToady will be offline for most of today 14:46
&
Raynes You will be missed. 14:47
TimToady sorear: I think in-out parameters have to match type exactly in most language because of co-contravariant considerations; if we violate that, we have to do some kind of coercion somewhere.
really &
14:52 kid51 joined
kid51 In Perl 6, what is the equivalent to Perl 5's 'perl -c somefile'? 14:53
flussence perl6 -c somefile
it was broken and only fixed recently
try rakudo 2010.09 14:54
kid51 But I get (in August Rakudo Star release):
$ perl6 -c multi.pl
===SORRY!===
Unable to open filehandle from path '-c'
flussence yeah, it's been fixed since then
a few days ago IIRC
kid51 Well, I'm in a study group which specifically focuses on Rakudo Star releases, so I can't use that yet.
14:55 xabbu42_ left, xabbu42 joined
nymacro they have Rakudo* study groups? 14:59
pmichaud -c was only added a few days ago... it'll show up in the next Star release (on Tuesday) 15:01
15:07 zulon left
kid51 NYC Rakudo Star group: tech.groups.yahoo.com/group/perlsem...ssage/1016 15:07
kid51 has to get ready to go to that now
15:08 icwiener_ joined 15:09 icwiener left, risou joined
jnthn kid51: Looks like fun event :-) 15:09
Enjoy
flussence of course you could always use parsefile() as a workaround until then... 15:10
15:18 risou_ joined 15:20 risou left, jimk joined 15:22 kid51 left
nymacro now if only I could find some awesome groups in Brisbane, Australia ;) 15:22
15:23 icwiener_ left
tadzik you don't have a PM group there? 15:23
nymacro not to my knowledge 15:24
actually, maybe there is 15:25
wait. No. Past meetings: 2005
tadzik start one :) 15:26
dalek kudo: 2632afb | moritz++ | src/core/EnumMap.pm:
fix EnumMap.keys and .kv to return proper Perl 6 strings
jimk nymacro: Find a pub with wireless access. Convene meeting. 15:27
Simple
moritz_ one step missing 15:28
find others interesting in the same topic
tadzik wireless access? We usually seek for pubs with beer access
nymacro ^ this ;)
jimk 'beer' is the 'interesting topic'
tadzik Rakudo Star meeting shows that a pub with beer access is not so obvious
moritz_ there are pubs without beer access?
jimk We have one ozzie coming to group today 15:29
15:29 justatheory joined
nymacro crikey! 15:29
tadzik moritz_: when we came, there was a note "out of beer". We bought everything they had, and we were lucky it was enough
15:30 ash_ joined
nymacro I might look into the viability of running such a group in Brisbane; if there is enough interested parties and I can find some mutually acceptable interests/topic areas it will be on ;) 15:31
15:31 jimk is now known as kid51
pmichaud even small groups (3-5 people) can be very worthwhile :) 15:31
kid51 nymacro: 10 years ago, we started Perl Seminar NY in a very simple manner: I called the meeting. I presented the very first Perl 5 program I had ever put into production. 15:32
My ambition: To live long enough to translate *that* program into P6.
15:33 PacoLinux left
nymacro well said :) 15:33
kid51 pmichaud attends a Perl 6-focused group in DFW
moritz_ kid51: what did it do?
15:34 PacoLinux joined
kid51 It was when I was working in state psych hospital and had to refer patients to other hospitals. 15:34
It translated patient's (extensive) medication history from downloaded-from-mainframe format to one requested by MDs at other hospital. 15:35
Used Perl 5 formats!
pmichaud mmmmm, formats. I haven't used formats since.... Perl 4. :-P 15:36
15:49 icwiener joined 15:53 kid51 left 16:05 patspam joined 16:07 masonkramer joined 16:10 patspam left 16:16 engine joined 16:18 xabbu42 left 16:19 risou joined 16:20 engine left, engine joined 16:21 risou_ left 16:22 whoever joined, whoever is now known as Guest3416 16:25 engine left
moritz_ nqp: loop (my $x := 0; $x < 10; $x++) { if $x > 1 { last }; say $x } 16:40
p6eval nqp: OUTPUT«PackFile_unpack: This Parrot cannot read bytecode files with version 8.3.␤␤PackFile header failed during unpack␤»
tadzik (: 16:46
16:50 zulon joined 16:57 sjn left 17:00 sjn joined 17:12 nymacro left 17:15 aloha left 17:19 tadzik left 17:24 aloha joined 17:26 literal joined 17:27 ruoso left 17:32 zulon left 17:37 mfollett left 17:41 ruoso joined 17:48 patspam joined
dalek odel: a301155 | mberends++ | java/runtime/ (2 files):
[java] add SignatureBinder
 17:49
odel: 84b214c | mberends++ | / (4 files):
[java] incomplete ports of PAST2JSTCompiler.pm and JST2Java.pm
17:53 uniejo joined 17:59 kjeldahl left, hillu joined 18:00 Ross left
masonkramer Instant and Duration are builtin types, but Weight, Currency, Temperature aren't, right? 18:03
I'm trying to do the last chapter of Higher Order Perl in Perl6, and I thought maybe it deserves some polish and to be made available on CPAN6 18:04
flussence rakudo: say Instant.WHAT; say Duration.WHAT; say Temperature.WHAT; 18:05
p6eval rakudo 2632af: OUTPUT«Instant()␤Duration()␤Could not find sub &Temperature␤ in main program body at line 22:/tmp/ptzgC6tXrQ␤»
masonkramer I thought we weren't going back to the implementation for a definitive answer about the language anymore :p 18:07
18:07 kjeldahl joined 18:09 M_o_C joined
flussence Google says they don't exist in the spec as builtins ;) 18:09
18:10 kid51 joined, dha joined
sorear good * #perl6 18:10
kid51 patspam: file questions about PDF at github.com/perl6/book/issues/ 18:11
patspam ta
what git tag should I checkout to build the latest Rakudo Star release?
18:11 heins57 joined 18:12 TheFox21 joined
colomon masonkramer++ 18:14
sorear kid51++ # love for people like me with ancient boxes 18:17
18:17 Front_slash joined
sorear makes the same change and tests Rakudo build again 18:18
(it did not work AT ALL previously - ludicrous memory usage during the parrot-nqp stage)
18:21 Front_slash left, aindilis left 18:22 risou left, aindilis joined
kid51 sorear: Ah, now I see what you're saying. 18:26
I'll have to try building Rakudo on this box again.
18:29 aindilis left 18:30 aindilis joined
patspam star: sub x(Bool){} 18:34
p6eval star 2010.07: ( no output )
patspam that that valid?
s/sub/multi/ 18:35
star: multi x(Bool $x){}; multi x(Bool){}; x(True)
p6eval star 2010.07: OUTPUT«Ambiguous dispatch to multi 'x'. Ambiguous candidates had signatures:␤:(Bool $x)␤:(Bool )␤␤ in main program body at line 22:/tmp/k64hJX__5u␤»
colomon star: multi x(Bool){ say "indeed" }; x(True) 18:36
p6eval star 2010.07: OUTPUT«indeed␤»
jnthn patspam: It's valid, but the dispatcher is correct to point out that they're ambiguous if you write both.
patspam hey jnthn!
jnthn o/ patspam
patspam so you'd never be able to call either?
sorear kid51: after 11 minutes cpu, setting build is now thrashing 18:37
360mb virt
'pir' stage
dha Out of curiosity, is the p6eval bot actually running rakudo* 2010.07, as it seems to be saying? 18:39
18:40 schpey left, aindilis left
patspam star: sub x(){ shift }; x() 18:41
p6eval star 2010.07: OUTPUT«Not enough positional parameters passed; got 0 but expected 1␤ in 'shift' at line 2846:CORE.setting␤ in 'x' at line 22:/tmp/AYG_6nuYF5␤ in main program body at line 22:/tmp/AYG_6nuYF5␤»
patspam sub x(){ shift }; x(1)
star: sub x(){ shift }; x(1) 18:43
p6eval star 2010.07: OUTPUT«Too many positional parameters passed; got 1 but expected 0␤ in 'x' at line 22:/tmp/jZAnUlSiVA␤ in main program body at line 22:/tmp/jZAnUlSiVA␤»
patspam damned if I do, damned if I don't
masonkramer Say I have two class files in my @*INC: Wire.pm and Node.pm. Nodes have Wire attributes, and Wires have Node attributes. How do I declare the Node type in Wire.pm and vice versa? use Node;/user Wire; is infinitely recursive 18:45
patspam shift: sub x($x){ shift }; x(1) 18:46
star: sub x($x){ shift }; x(1)
p6eval star 2010.07: OUTPUT«Not enough positional parameters passed; got 0 but expected 1␤ in 'shift' at line 2846:CORE.setting␤ in 'x' at line 22:/tmp/Ufp2hUnnE2␤ in main program body at line 22:/tmp/Ufp2hUnnE2␤»
patspam is there a way to define the sub so that I can still use shift a la Perl5? 18:47
plobsing star: sub x { .shift }; x(1)
p6eval star 2010.07: OUTPUT«Too many positional parameters passed; got 1 but expected 0␤ in 'x' at line 22:/tmp/qz_V4icVSS␤ in main program body at line 22:/tmp/qz_V4icVSS␤»
flussence sub x(*@a) maybe
patspam star: sub x(*@a) { shift }; x(1) 18:48
p6eval star 2010.07: OUTPUT«Not enough positional parameters passed; got 0 but expected 1␤ in 'shift' at line 2846:CORE.setting␤ in 'x' at line 22:/tmp/0ec9VED1xc␤ in main program body at line 22:/tmp/0ec9VED1xc␤»
plobsing start: sub x { shift @_ }; x(1)
patspam that works 18:49
so what is shift doing when you don't give it @_ as an argument?
plobsing implicit use of @_ (and to a lesser extent $_) are changed/gone
colomon star: sub x { .shift }; x(1) 18:50
plobsing patspam: shift without @_ is simply that: shift(). No arguments.
p6eval star 2010.07: OUTPUT«Too many positional parameters passed; got 1 but expected 0␤ in 'x' at line 22:/tmp/GNnnBKqT3j␤ in main program body at line 22:/tmp/GNnnBKqT3j␤»
masonkramer perhaps you could point me to a synopsis on this?
I can't find much in the way of multi-file examples
patspam shift without arguments seems to be confusing signature checking 18:51
flussence rakudo: sub x(*@_) { @_.shift }; x(1); 18:53
jnthn patspam: There isn't a shift multi-candidate that takes no arguments.
p6eval rakudo 2632af: ( no output )
flussence the error without that "@_." isn't very friendly though
rakudo: sub x(*@_) { shift }; x(1);
p6eval rakudo 2632af: OUTPUT«(timeout)Not enough positional parameters passed; got 0 but expected 1␤ in 'shift' at line 2964:CORE.setting␤ in 'x' at line 22:/tmp/0QphK_aKeB␤ in main program body at line 22:/tmp/0QphK_aKeB␤» 18:54
flussence eh
it only gives me the first error line locally
plobsing masonkramer: class A {...}; class B { has A $.a }; class A { has B $.b } 18:57
masonkramer that works because those classes are in the same lexical scope 18:58
My classes are in their own files
if I have class Wire; need Node; # Wire.pm 18:59
class Node; need Wire;# Node.pm
I get all sorts of errors
eg ./perl6 usr/lib/Wire.pm 19:00
===SORRY!===
Illegal redeclaration of symbol 'Wire'
19:00 envi^home left 19:02 mberends left 19:03 Ross joined, Ross left, Ross joined
plobsing masonkramer: A.pm - 'use B; class A { has B $.b }' 19:03
colomon masonkramer: why do you think you need to do that?
plobsing B.pm 'class A {...}; class B { has A $.a }; use A;' 19:04
patspam star: multi x(Bool $){}; multi x(Bool){}; x(True)
plobsing main.pm 'use A; use B'
p6eval star 2010.07: OUTPUT«Ambiguous dispatch to multi 'x'. Ambiguous candidates had signatures:␤:(Bool )␤:(Bool )␤␤ in main program body at line 22:/tmp/exX1ngaDvV␤»
plobsing works fine for me
masonkramer ( working through replies... ) 19:05
colomon (plobsing's approach is would be my first thought as well.) 19:06
masonkramer colomon: wtr to why: Wires have Nodes and Nodes have Wires 19:08
so class A { ... }; is serving as the declaration of A here?
colomon yes 19:09
plobsing predeclaration, yes
19:09 masak joined
masak ahoj! 19:09
colomon o/
masonkramer where is the illegal redeclaration of symbol 'Wire' coming from? 19:11
colomon when you load Wire, it loads Node, which loads Wire.
masonkramer This is a symmetrical dependency 19:12
so having one class predeclare the other -
but second predeclare the first 19:13
how can I chose which one does it? heh
masak rakudo: class A { ... }; class B { has A $.a }; class A { has B $.b }; say "alive! \o/" 19:14
p6eval rakudo 2632af: OUTPUT«===SORRY!===␤Unrecognized backslash sequence: '\o' at line 22, near "/\""␤»
masak :P
rakudo: class A { ... }; class B { has A $.a }; class A { has B $.b }; say "alive! \\o/"
p6eval rakudo 2632af: OUTPUT«alive! \o/␤»
colomon masonkramer: you could just as easily have both classes predeclare the other... 19:15
jnthn masak: :P 19:16
plobsing rakudo: 30.can('to-radians').signature.perl.say 19:18
p6eval rakudo 2632af: OUTPUT«:(Numeric $x: Any $base;; *%_)␤»
plobsing what is Any $base?
masak enum TrigBase is export <Radians Degrees Gradians Circles>; 19:19
plobsing: it's a value of that enum type, according to S32/Numeric 19:20
I don't find it extremely intuitive that that parameter be a required positional.
plobsing oic. My thought process was: "base... hmmm... well I want my answers in base 10..."
masonkramer predeclare each class in the other is giving me an illegal redeclaration error
nopaste.snit.ch/23637
patspam is there a practical difference between multi x(Int){} and multi x(Int $){}? 19:21
colomon ah, right.
patspam maybe I'm not up to that part of the book yet
kid51 Do we have any pastebots here? 19:23
colomon masonkramer: forgot about that annoying bit.
masak patspam: no, no practical difference. 19:24
masonkramer I can't get plobsing's original suggestion to compile either 19:25
kid51 nopaste.snit.ch/23640
19:25 wallberg joined
plobsing I like it, so long as I get to be player 1. 19:26
patspam thanks masak
we're working through the Using Perl 6 book atm, hence the barrage of questions 19:27
19:27 miso_soup left
masak patspam++ 19:27
others++ 19:28
patspam kid51++ mostly
19:32 lichtkind joined
masonkramer if main.pl6 only uses Wire *or* Node, then it basically works 19:32
19:33 mberends joined
cosimo I have a class. I didn't write any new() method. what if I need to do something to initialize the class? can i write a new(), or? is there something like Moose BUILD() method? how does that work? 19:33
jnthn Yes, yes. 19:34
lichtkind moritz_: did you install any chatbots here?
phenny lichtkind: 20 Sep 12:49Z <[Coke]> tell lichtkind - please give me the exact URL including the tag for your feed.
lichtkind: 20 Sep 12:49Z <[Coke]> tell lichtkind to email me at [email@hidden.address] for that.
lichtkind ah thanks phenny
jnthn cosimo: perlgeek.de/blog-en/perl-6/object-c...ation.html 19:35
cosimo: That's a good article.
cosimo thanks jnthn 19:36
submethod BUILD. awesome. 19:37
19:37 pmurias joined
sorear kid51: rakudo build finished; real time 70 minutes, max vsize 420MB; down from a best of 7min/350 (~July); but still better than the overnight run from March 19:38
kid51 sorear: I'll try to build at home tonight: Were you using Rakudo Star or pulling from git? 19:39
masak nothingmuch++ # blog.woobling.org/2010/09/moose-has-won.html 19:43
19:44 heins57 left
sorear kid51: got 19:44
git
hmm, nothingmuch still hasn't relurked since the fake audrey incident 19:45
19:48 ruoso left
lichtkind jnthn: cheers 19:49
jnthn lichtkind: Na zdravie!
19:49 szabgab joined
lichtkind jnthn: skoll :) 19:49
jnthn: i think now i officially to work on my grant :) 19:50
jnthn \o/ 19:51
kid51 alright, let's go to the movies
jnthn lichtkind++
19:51 dha left
patspam is there an equivalent to perldoc in Perl6? 19:51
lichtkind patspam: not yet
patspam: there is a book some people here write 19:52
patspam heh yeah, that's what I'm working through :)
lichtkind patspam: and there are the tablets i write
patspam but say when I want to learn about "shift" in Perl6
19:52 kid51 left
patspam what's my spec/doc source? 19:52
lichtkind specs are synopses
www.perlfoundation.org/perl6/index.cgi?synopses 19:53
but maybe your better served with my index A
patspam url?
lichtkind www.perlfoundation.org/perl6/index....dex_tablet 19:54
patspam thanks
we're wrapping up here, till next time!
colomon lichtkind++
19:54 patspam left
lichtkind colomon: thanks for what? 19:55
colomon "i officially to work on my grant" # not thanks so much as applause and encourage
*ment 19:56
flussence lichtkind++ for that index; I was looking for something like that 19:57
lichtkind thanks its also included in grok
colomon: it was more info because i had to delay that a bit and planet6 for me is broken since i get noone to redirect to my new blog 20:00
sorear Argh 20:01
I spent over an hour building this Rakudo, and it won't even start
stefan@stefans:/tmp$ perl6
PackFile_unpack: This Parrot cannot read bytecode files with version 8.4.
both parrot & rakudo were completely clean before the build 20:02
masak moritz_: working on a Sudoku solver?
flussence does "ldd perl6" point to the right libparrot? 20:03
sorear flussence: No. Why not? 20:04
Who broke pbc2exe?
masak rakudo: my @l = <a bb c d e f g h>; while @l.munch(3) -> $_ { .say }
20:04 ruoso joined
p6eval rakudo 2632af: OUTPUT«a bb c␤d e f␤g h␤» 20:04
sorear Why is it linking stuff against Parrot 2.5?
masonkramer I want to use the default new and do some stuff afterwards...can I use .wrap for this and yet access private attributes? 20:05
flussence sorear: what commands did you use to build?
masonkramer nopaste.snit.ch/23642
sorear flussence: make 20:06
masonkramer submethod BUILD, alright 20:07
flussence on my system I'm using "rm parrot*/; make realclean; perl Configure.pl --gen-parrot; make all install"... you might need to re-run Configure.pl 20:08
sorear goddamn Makefile wants to rebuild everything now that I've rebuilt Parrot 20:09
sorear deletes all the dependencies from Makefile 20:10
20:12 M_o_C left
sorear after far too much manual work, I now have a working perl6 20:12
masak "far too much manual work" seems to be the era we're in so far. 20:13
sorear sudo rm -rf /usr/local/src/parrot /usr/local/lib/parrot /usr/local/lib/libparrot*
that, then reinstalling Parrot, then manually rebuilding Rakudo (the Makefile is useless), then installing, did it 20:14
mberends yes, over here it's mindlessly `rm -rf parrot parrot_install` whenever PARROT_REVISION shows up in a git pull. 20:17
masak I just 'make realclean' on both my Parrot and Rakudo every time I rebuild. 20:19
20:21 TheFox21 left
masonkramer my @wires = gather for 1..10 { take Wire.new }; # is there a better way? 20:22
flussence my @wires = Wire.new xx 10; ? 20:24
masak nope
that'll give you the same object ten times.
my @wires = map { Wire.new }, ^10; 20:25
masonkramer I always forget about ^ 20:26
masak my @wires = do for ^10 { Wire.new };
masonkramer thanks masak
is do for lazy?
jnthn It's lazy but list assignment is not. 20:28
masak masonkramer: 'for' is lazy. 'do' is just making an expression out of what's normally a statement. 20:29
masonkramer do for ^10 is exactly what I want, i'm not really transforming an integer into a Wire, so I don't want to say map 20:30
plobsing rakudo: (do for ^3 { .say }).munch(1)
p6eval rakudo 2632af: OUTPUT«0␤1␤2␤»
plobsing not lazy ATM
masak masonkramer: I think you read too much into the difference, but that's all right. :) use whatever you're comfortable with. 20:31
sorear masak: make realclean wasn't enough 20:32
I had to blow away the install too
masak sorear: you're probably using --gen-parrot. I'm not. 20:33
sorear I thought I was the only one who wasn't, actually
the install to blow away was in /usr/local
masak ok.
I haven't tried rebuilding today, but it's always worked so far. 20:34
sorear anyways I now have a working perl6 again; I think I won't touch parrot or rakudo for a couple months
flussence rakudo: our multi sub infix:<**>(Block $thing, Int $count) { do for ^$count { $thing.() } }; say ~( { rand } ** 5 );
p6eval rakudo 2632af: OUTPUT«0.378775026739703 0.40071641471344 0.964106152156951 0.42610601255463 0.0209936362079901␤» 20:35
20:37 Grimnir_ joined
mberends forgot that nqp-rx (for 6model) is also blown away while rebuilding Parrot 20:40
lichtkind mberends: hai martin 20:44
20:45 wamba left
moritz_ nqp: loop (my $x := 0; $x < 10; $x++) { if $x > 1 { last }; say $x } 20:47
p6eval nqp: OUTPUT«Confused at line 1, near "loop (my $"␤current instr.: 'parrot;HLL;Grammar;panic' pc 654 (src/cheats/hll-compiler.pir:206)␤»
masak moritz_: [backlog] writing on a Sudoku solver? 20:53
moritz_ masak: yes 20:54
masak moritz_: have you found (1) Peter Norvig's very nice solution, (2) Donald Knuth's inspiring solution?
moritz_ yes and yes
masak good. and good.
moritz_ :-)
masak just checking :)
moritz_ maybe I should really do dancing links 20:55
20:55 wallberg left
masak I don't know. maybe fsvo 'really'. 20:55
moritz_ I've been wanting to do that for ages, but somehow the complexity deterred me
and in C++, simple deduction + backtracking was good enough
masak it doesn't seem as obvious in Perl as in, say C.
florz moritz_: in perl6? 20:56
colomon wait, there's a Knuth solution?!
Grimnir_ I have some problems with an example from an perl advent from december. either when I use the normal rakudo or the alpha branch, I get errors. the program is: pastie.org/1181554
moritz_ florz: sure
Grimnir_ the alpha branch says: No applicable candidates found to dispatch to for 'prefix:='
the normal branch starts with the error message: Missing block at line 25, near ";\n my $" 20:57
moritz_ masak: where "good enough" = "can solve more sudokus in the Rakudo startup time than I can solve in a month" :-)
masak colomon: news.ycombinator.com/item?id=1160781
moritz_: aye.
florz moritz_: I gave up when rakudo didn't want to construct cartesian products for me ;-)
sorear ... There is no line 25
masak florz: what about X? 20:58
Grimnir_ sorry, 2 secs...
masak rakudo: .say for <1 2 3> X <a b c>
colomon masak++
Grimnir_ Missing block at line 2, near ";\n my $"
p6eval rakudo 2632af: OUTPUT«1␤a␤1␤b␤1␤c␤2␤a␤2␤b␤2␤c␤3␤a␤3␤b␤3␤c␤»
moritz_ rakudo: say (<a b> X <c d>).perl
p6eval rakudo 2632af: OUTPUT«(("a", "c"), ("a", "d"), ("b", "c"), ("b", "d"))␤»
Grimnir_ I had some commented stuff in the beginning, but not in the pastie
florz masak: yeah, and then add another list ;-) 20:59
moritz_ $ /perl6 foo.pl abc 2
"abc" rotated 2 characters gives "cde".
with today's Rakudo
Grimnir_ oh... strange
masak florz: ah. aye, known deficiency.
Grimnir_ I'll try to get it again then
masak florz: I recommend you write your own custom sub for now.
florz: do you want me to try and write it for you?
florz masak: well, yeah, wasn't sufficiently important - and after all the point was to write a succint perl6 solver, not to write a sudoku solver ;-) 21:00
dalek p-rx/nom: 51b35b7 | jnthn++ | / (5 files):
Implement P6int/P6num/P6str reprs so far as they are in the .Net version of 6model and add them to the build.
moritz_ github.com/moritz/perl6-Sudokubtw 21:02
github.com/moritz/perl6-Sudoku btw
the initializer is still fixed on 9x9, but the rest is completely generic
and the approach with using contraints means it's very easy to change the rules, or add constraints (like for the diagonals) 21:03
blog.woobling.org/2010/09/moose-has-won.html might be of interest 21:04
masak pasted that a while ago :)
moritz_ hasn't backlogged yet
ye know, #perl6 is a write-only channel :-)
sorear ponders a perl 6 minimax demo
21:06 Camille_B joined
plobsing would like to see an LP and QP package in nqp - lpnqp (lp-nqp and lp n' qp) 21:06
moritz_ where LP = Long Playing phonograph record? 21:07
plobsing linear programming
moritz_ ah 21:08
21:08 bobkare joined
plobsing although having perl 6 play outdated media would be cool too 21:08
moritz_ and QP = quadratic programming? 21:09
plobsing yes. 21:10
jnthn Can't we just have a generic PP? :-)
plobsing sure, but it would ruin my pun and probably run slow as hell (IIRC, general QP is NP-hard) 21:12
masak Quarter Pounder is NP-hard?
moritz_ "Royale with Cheese"! 21:13
plobsing en.wikipedia.org/wiki/Quadratic_pro...Complexity
masak it better have quadratic complexity, or I'll be much disappointed :P
moritz_ ponders programming with fractional dimenionality 21:14
21:21 Camille_B left
lue ohai o/ 21:23
masak \o
flussence lue: [backlog] perl6 for browser scripting sounds insane. might be doable though 21:24
I mean, it's all DOM manipulation in the end, so all you'd need is something that converts the p6 code into JS DOM API calls 21:25
masak converting arbitrary p6 code into js is no easy task. 21:26
lue I tried coding in Javascript, but I think it was a combination of having to constantly look up stuff and liking P6 too much that made me not like it :)
flussence yeah, JS really could use inline documentation :| 21:27
masak lue: learning to like JavaScript is a good investment for the future.
moritz_ lue: after you learned Perl 6, you'll have to accept that almost any other language is a major hassle :-)
that shouldn't stop you from learning other languages though 21:28
lue I have used it quite enjoyably in the past. But that was before P6, and I had a place to put the results :P
flussence apparently I'm not the first person to think of this: wiki.mozilla.org/Labs/Bespin/Desig...umentation 21:30
lue Ah well, looks like I'll resign to JS. [ until the W3C says it's OK to put P6 in <script> tags :) ] 21:33
flussence (or someone implements P6 in JS...) 21:34
masak lue: doesn't matter much what the w3c says in this case, as long as there's no client-side support for p6 in browsers.
21:39 Chillance left
lue attempts to switch gears before he continues displaying his !knowledge about the finer points of the web 21:41
lue also can't wait for the IETF to finalize HTML5
moritz_ dislikes the word "finalize" 21:43
21:44 Guest23195 left
moritz_ would you like to use a finalized human language? 21:46
masak the requirements on computer standards aren't the same as on spoken languages for humans. 21:47
mberends final*ze should be a banned word in #perl6
21:47 Italian_Plumber joined
lue does that mean P6.1 would never happen? 21:47
moritz_ masak: and that's the problem 21:48
jnthn
.oO( I thought we called finalize "DESTROY" in Perl 6... )
masak moritz_: wait, what? no.
moritz_: stability has its place.
mberends lue: sure it will. But we have plenty of time before P6.0 is final*zed. 21:49
moritz_ masak: and change too
masak of course.
but I wouldn't consider the places in the computer industry where people have produced stable standards a "problem". 21:50
to some extent, I understand p5p's adherence to backwards compatibility over almost all else. 21:52
there are some definite advantages to not changing. Perl 6 just happens to be taking the other route, which has a number of advantages, too. 21:53
21:54 Italian_Plumber left
lue In my mind, P6 is not a progression of Perl, but a 'fork' of sorts. 21:56
21:56 gabiruh joined, Ross left
moritz_ right; it's way ahead in some areas, and way behind in others 21:57
masak lue: it's a "new Perl".
mberends lue: true. P6 does not threaten P5, it offers an alternative Perl.
masak meme courtesy of jnthn++
jnthn Surely I wasn't the first person to put it that way? :-) 21:58
masak think so.
lue
.oO(What would the next major step from P5 be called? P5v6?)
 22:00
masak lue: 5.14, 5.16, 5.18 ...
masonkramer when I try to run rakudo*'s perl6 interpreter using an absolute path to it, I get an unknown file error 22:01
if I cd to the directory containing perl6 and try again, it works
pastie.org/1181654
known issue?
masak masonkramer: 'make install'
flussence ...what masak said :) 22:02
masonkramer heh
flussence is slow typist
masonkramer thanks :)
masak we should probably have a better error message there.
flussence I dunno, that one's pretty descriptive compared to the generic "Segmentation fault" I got 22:03
masak :P 22:04
moritz_ uhm, didn't we have a better error message for that?
I mean, back in the days?
masak not that I recall.
jnthn Yes, alpha did.
moritz_ and I thought I had ported it to master at some point 22:05
jnthn The code is copy-pasteable to latest, if anyone wants to do os.
masak wuhwait. "Divide by zero"?
how does a non-installed Rakudo lead to a division by zero? o.O
that's wrong on at least one level. 22:06
plobsing well division is denoted by /. which also is used in paths. so it makes sense ;-) 22:07
lue
.oO( maybe the end of the line (...../perl6), not finding perl6 was interpreted as ....../0 ? )
 22:08
moritz_ src/Perl6/Compiler.pir lines 37-43
masonkramer: which version of rakudo are you using?
masak plobsing++ # disturbing rationalization :)
masonkramer moritz_: how can I tell? It's one of the * distributions (maybe the only one?) 22:09
./perl6 --version
This compiler is built with the Parrot Compiler Toolkit, parrot revision 47310.
moritz_ masonkramer: must be the first R* release
22:09 uniejo left
moritz_ the second reported it version probably 22:09
*its
better error message added on Wed Jun 30 12:34:38 2010 +0200 22:10
masonkramer ok, pulling the latest
moritz_ there'll be another release next Tuesday 22:11
masonkramer I'll pull then too :)~
there it is right in the readme 22:12
moritz_ you'd be the first to read it :/ 22:13
22:14 cbk joined
masonkramer is there a way to really dump an object? I want more than .perl is giving me 22:15
22:15 Nosnam joined, Nosnam left
masak masonkramer: could you be a bit more specific? 22:16
masonkramer I would try to cook up a little example, but I have to wait for perl6 to build :)
flussence rakudo: say DateTime.now.perl; # not good enough? 22:17
p6eval rakudo 2632af: OUTPUT«DateTime.new(year => 2010, month => 9, day => 25, hour => 21, minute => 54, second => 26.6693475246429, timezone => 0, formatter => { ... })␤»
masonkramer Oh, does it do that?
Then I guess the problem is that my objects have no data somehow
22:17 kjeldahl left
masak ".perl: does more than you'd think" :) 22:17
moritz_ or maybe your rakudo is too old :-)
rakudo: class A { has $.x = 5 }; say A.new(x => 17).perl 22:18
p6eval rakudo 2632af: OUTPUT«A.new(x => 17)␤»
22:18 ^Mike\b joined
^Mike\b Perl 5 docs say that -w not being mandatory is a bug. Will the equivalent be mandatory in perl 6? 22:18
22:19 mfollett joined
masak ^Mike\b: yes. it is. 22:19
^Mike\b cool - what about strict?
moritz_ not mandatory, but default
(both warnings and strict)
masak nod.
and right now, there's no way to turn them off.
^Mike\b \o/ 22:20
masonkramer class A { has @.things }; say A.new(things => <1 2 3>).perl
masak (and I suspect at least implementing turning off 'strict' will be difficult)
moritz_ rakudo: class A { has @.things }; say A.new(things => <1 2 3>).perl
p6eval rakudo 2632af: OUTPUT«A.new(things => ["1", "2", "3"])␤»
moritz_ masak: I don't think so, once we have enough compile time information
masak I hope you're right. 22:21
masonkramer is there any difference between things => <1 2 3> and :things(<1 2 3>) ?
lue any reason you *wouldn't* want warnings and strict, besides being lazy about your programming?
masak masonkramer: I wouldn't say there is.
flussence the ()s make precedence a bit more obvious
masak lue: yes, I've been unable to do some -n and -p one-liners with strict at times. 22:22
moritz_ you can even write it as :things<1 2 3>
masonkramer That's it? Both create Parcels?
or Pairs?
what's the difference?
moritz_ Pairs
22:22 ^Mike\b left
masak lue: and selectively turning off warnings is what one wants in some cases. 22:22
sorear jnthn: I don't like DESTROY, fwiw 22:29
jnthn sorear: Well, it's essentially guaranteeeless. 22:30
(as to when it'll get run)
Wo as implementors we can do it however works.
(as in, works well for the implementation)
sorear I mean, having finalizers as object methods feels very wrong
masonkramer rakudo: class F { has @.things; has $.foo; submethod BUILD { ... } }; say F.new(:things<1 2 3>, :foo<1>).perl 22:31
p6eval rakudo 2632af: OUTPUT«F.new(things => [], foo => Any)␤»
masak sorear: why?
sorear the resurrection problem is a major symptom of the underlying wrongness
jnthn sorear: What would you want them to be?
Resurrection isn't a problem as such.
moritz_ masonkramer: it's a known bug that BUILD interacts badly with the defualt constructor :( 22:32
jnthn It *is* a pain that it may happen.
sorear jnthn: The least insane finalizers I've seen were in GHC Haskell, which basically has Mu.attach-finalizer($sub)
(actually it's integrated with the weak reference system)
masak when would resurrection happen in DESTROY?
sorear in particular, $sub does *not* receive a reference to self
masonkramer moritz_: Yay! This one's not my fault 22:33
sorear which means that objects with finalizers can be collected immediately
only the data that the finalizer actually closes over needs to be retained
jnthn sorear: That is quite a nice design.
masak: When you for example make another reference to self exist. 22:34
masonkramer moritz_: so if I want to do some more initialization, I currently need to do my own new()?
masak ah. submethod DESTROY { $global = self } ?
sorear yes 22:36
plobsing S06 specifies that positional placeholder vars are arranged in "Unicode order". What does that mean? Graphemes? Codepoints?
masak sorear: I smell a spec patch coming up.
many of the ingredients are there; no-one particularly likes DESTROY/DESTROYALL, there's no implementation of them, and there are known problems with the spec as-is. 22:37
I didn't know about the resurrection problem until today. sorear++ for pointing that out. 22:38
jnthn masak: The scariest thing is code that depends on it. ;-)
masak :(
jnthn oh gah...once again I have to go chasing a Parrot change rather than doing useful stuff. 22:39
masak why would DESTROY want to be a (sub)method in the first place? because of bottom-up destruction, it would have no attribute objects that one'd want to access anyway.
jnthn: one almost wishes they'd finalize Parrot once and for all. :P 22:40
plobsing a finalized parrot would be a dead parrot. would still have lovely plumage though. 22:43
22:43 Italian_Plumber joined
lue
.oO(And all of the internet would laugh from behind the fourth wall)
 22:43
masak "it's not finalized, it's pinin'" 22:44
mberends "it's joined the Choir Eternal" # stability is overrated 22:45
lue Now I feel the need to write the Parrot/P6 version of the dead parrot sketch...
plobsing look mate, this VM wouldn't voom if you put a million volts through it! 22:46
mberends lue: you absoluetely should
lue "It's a pun" A pun? "No, one of those things that are backwards and forwards" Endianness? ... 22:47
masonkramer rakudo: class A { sub new { say "hi" } }; A.new.perl 22:48
p6eval rakudo 2632af: ( no output )
masak lue: my golly, I think your wit is getting sharper...
masonkramer class A { sub new { say "hi" } }; say A.new.perl
raaah 22:49
masak :P
masonkramer rakudo: class A { sub new { say "hi" } }; say A.new.perl
p6eval rakudo 2632af: OUTPUT«A.new()␤»
masak rakudo: say "RAAAH!"
p6eval rakudo 2632af: OUTPUT«RAAAH!␤»
lue thinks it would be quite serendipitous if #parrot went down to 0 nicks and disappeared when this topic came up 22:50
dalek p-rx/nom: c295c00 | jnthn++ | build/Makefile.in:
Need to put building the dynops/dynpmcs pretty early.
 22:51
p-rx/nom: afeebf6 | jnthn++ | src/ops/nqp.ops:
Make sure we don't double-init the meta-model / dynops.
p-rx/nom: d84b3db | jnthn++ | / (2 files):
Bump to latest Parrot to get .loadlib on PAST blocks, and chase a Parrot NCI PMC API change.
masonkramer is that supposed to happen?
22:51 Italian_Plumber left
masak masonkramer: I think you think that defining a sub 'new' overrides the method '.new' from Mu. 22:51
masonkramer if there's a method new and a sub new, how do I get the sub new? 22:52
masak rakudo: class A { method foo { bar() }; sub bar { say "OH HAI" } }; A.new.foo
p6eval rakudo 2632af: OUTPUT«OH HAI␤»
masak does that answer your question? 22:53
.foo and &bar have different names here, but it'd work if they had the same name too.
because the calling mechanisms are different.
masonkramer yes 22:54
sorry. I'm still in perl5 land
sorear interesting mini-survey of finalization: ftp://ftp.inf.puc-rio.br/pub/docs/techreports/04_30_leal.ps
masak masonkramer: you're not the first to be confused about this issue. 22:55
sorear: thanks for that reference.
I'm more and more convinced that we should emulate GHC in this regard. 22:57
jnthn It'd have my +1 23:00
plobsing how does not giving finalizers an explicit reference to main solve the underlying issue? can't the same problems occur with cyclic datastructures? 23:01
s/main/self/
masak plobsing: at first blush, it sounds like you're conflating two different problems with that question. 23:02
cyclic datastructures are a problem in e.g. refcounting GCs.
23:02 icwiener left
masak which would preventing such a GC from ever calling DESTROY in the first place. 23:03
whereas resurrection occurs when a DESTROY submethod is called and creates a new surviving reference to self. 23:04
plobsing we prevent 'submethod DESTROY { $global = self }'. but what's to stop a user from 'submethod DESTROY { global = $^self-ref }'?
oops. submethod DESTROY { global = $!self-ref } 23:05
sorear ah. *this* is what I was looking for
research.microsoft.com/en-us/um/peo...weak.ps.gz
plobsing: $!self-ref doesn't work without self.
masak plobsing: the solution is not letting it be a (sub)method.
ah, good ol' simonpj 23:06
plobsing ok, sure it's not a submethod. but it still needs access to attributes somehow. and self can occur within the object graph of an attribute. 23:07
masak plobsing: if it needs access to attributes, then the GHC solution (as I understand it) is out of the question. 23:08
but I must confess to having very little experience with destructors.
does it actually need to access the attributes.
s/\./?/
and if it does, how can it reliably do that? haven't they already been GC'd?
plobsing how do you destroy a filehandle without having access to the underlying fd? 23:09
masak good question.
that's certainly a use case we must factor in.
masonkramer so is there no way to use the default new constructor with BUILD? 23:10
masak sure there is.
rakudo: class A { has $.x }; say A.new(x => 42).x 23:11
p6eval rakudo 2632af: OUTPUT«42␤»
masak rakudo: class A { has $.x; submethod BUILD(:$x) { $!x = 5 } }; say A.new(x => 42).x
p6eval rakudo 2632af: OUTPUT«5␤»
masak hugme: hug p6eval 23:12
hugme hugs p6eval
cbk masak, how would you do that if x is a hash?
masak hm. 23:13
cbk A.new( %x = "key", "value")
just guessen' here
I need something like this.
nothing I try works.
masak rakudo: class A { has %.h }; say A.new(:h<1 a 2 b>).h.perl 23:14
p6eval rakudo 2632af: OUTPUT«{"1" => "a", "2" => "b"}␤»
cbk now I just have a silly make() method
masak there are other ways, to be sure. but that one works.
cbk ok
jnthn rakudo: say (:h{ a => 1, b => 2}).perl 23:15
cbk I was putting ,'s in there.
p6eval rakudo 2632af: OUTPUT«"h" => {"a" => 1, "b" => 2}␤»
masak rakudo: class A { has %.h }; say A.new(:h(a => 1, b => 2).h.perl
p6eval rakudo 2632af: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 22␤»
masak rakudo: class A { has %.h }; say A.new(:h(a => 1, b => 2)).h.perl
p6eval rakudo 2632af: OUTPUT«{"a" => 1, "b" => 2}␤»
masak rakudo: class A { has %.h }; say A.new(:h{a => 1, b => 2}).h.perl 23:16
p6eval rakudo 2632af: OUTPUT«{"a" => 1, "b" => 2}␤»
masak \o/
cbk what is the : in front of the h mean?
masonkramer rakudo: class F { has @.things; has $.foo; submethod BUILD { self } }; say F.new(:things<1 2 3>).perl 23:17
p6eval rakudo 2632af: OUTPUT«F.new(things => [], foo => Any)␤»
masak cbk: 'this parameter is a named parameter'
masonkramer : is way of setting up a Pair here cbk
masak or what masonkramer said.
rakudo: say (:h).WHAT
p6eval rakudo 2632af: OUTPUT«Pair()␤»
masonkramer masak: what am I doing wrong? things should have <1 2 3>
cbk I got the array to work but was having problems with the hash. 23:18
masak masonkramer: you're misunderstanding BUILD.
masonkramer: S12 is your reading task for tomorrow :)
#perl6, I bid you good night. 23:20
o/
23:20 masak left
masonkramer nite 23:20
All I want to do is use the default new() method and then call a method on the object after it's created 23:21
*headdesk*. *headdesk*. *headdesk*.
around new { my ($orig, @args); my $obj = $orig->(@args); $obj.id(++$count) } 23:22
is there nothing like that?
jnthn rakudo: class Foo { method BUILD() { self.lol }; method lol() { say "oh lol" } }; Foo.new 23:24
p6eval rakudo 2632af: OUTPUT«oh lol␤»
jnthn Like that?
or
rakudo: class Foo { method new() { my $obj = self.bless(*, *%_); $obj.lol; $obj }; method lol() { say "oh lol" } }; Foo.new 23:25
p6eval rakudo 2632af: OUTPUT«oh lol␤»
dalek p-rx/nom: 6addc8f | jnthn++ | src/NQP/Actions.pm:
Load nqp_group and nqp_ops and initialize them at the start of an NQP program.
p-rx/nom: ae7a992 | jnthn++ | src/NQP/Actions.pm:
Don't have nqpop support yet, so use pirop for now.
p-rx/nom: 5a6ffa4 | jnthn++ | src/metamodel/knowhow_bootstrapper.c:
Oops, last attempt to modify things to work on latest Parrot missed something. This works.
p-rx/nom: 9937ece | jnthn++ | src/NQP/ (2 files):
knowhow declarations using the new meta-model now work.
masonkramer jnthn: ( working through that ... ) 23:26
First way is not what I'm looking for. I want to use Mu.new() - that is, I want attributes to be set via Pairs. Then, I have another attribute that should be set automatically by a method - an autoincrementing ID. I don't want to rewrite Mu.new()'s behavior in order to extend Mu.new() with something else 23:30
second way is also rewriting Mu.new(), I think
not really sure what's going on there 23:31
does that make sense? Is that possible? 23:34
23:34 pmurias left
jnthn masonkramer: Sounds like what you want is the use case for BUILD 23:35
BUILD is designed for when you want to do some extra initialization per new instance.
Though if you just want unique IDs...
rakudo: class Foo { my $cur_id = 0; has $.id = $cur_id++; }; say Foo.new.id for 1..5; 23:36
p6eval rakudo 2632af: OUTPUT«0␤1␤2␤3␤4␤»
masonkramer rakudo: class F { has @.things; has @.build_things; submethod BUILD { @!build_things = <A B C> } }; say F.new(:things<1 2 3>).perl 23:38
p6eval rakudo 2632af: OUTPUT«F.new(things => [], build_things => ["A", "B", "C"])␤»
masonkramer things should have [1, 2, 3], afaict
jnthn Oh!
Damm...that's a Rakudo bug. :/
masonkramer alright :) 23:39
glad I'm on the same page now
so this ability is specced but not quite there yet. A workaround would be to actually implement the behavior of the default Mu.new 23:45
jnthn masonkramer: Yes, it's not quite working as it should in Rakudo.
The spec has it as you expect, so far as I recall.
masonkramer how do I slurp up all the pairs that are passed to new? I'm having some trouble figuring that out. new (@args) { } sets up positional parameters 23:46
*%_?
jnthn They're already slurped into *%_ for you 23:47
masonkramer the interesting that is that, even while working around rakudobugs, my code is still nicer than perl 5's built in OO stuff
jnthn But you can delcare that explicitly
Or give it a nice name
*%attrs
Or whatever you like :-)
flussence ...does *%_ work even if I stick non-pairs in between the pairs? 23:48
dalek p-rx/nom: 771e2ba | jnthn++ | src/metamodel/how/NQPClassHOW.pm:
Start to sketch out an NQP ClassHOW.
jnthn flussence: Well, it slurps up all the named arguments in the capture, which are stored separately to any positionals. 23:50
23:51 Patterner left, Psyche^ joined, Psyche^ is now known as Patterner 23:54 Guest3416 left

Powered by Cro and the Raku® Programming Language.

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

Thank you!