»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.
00:01 [ptc] joined 00:05 prammer joined 00:07 prammer left
timotimo gnite lizmat :) 00:08
00:08 [ptc] left 00:11 [ptc] joined 00:12 [ptc] left 00:14 [ptc] joined, firstdayonthejob left 00:15 virtualsue left 00:24 themonkeybob11 left 00:26 idiosyncrat_ joined 00:29 sortiz left, [ptc] left 00:30 [ptc] joined 00:32 [ptc] left 00:50 tardisx left 00:54 idyllei left 01:02 tardisx joined 01:06 skids joined 01:10 yeahnoob joined 01:22 QORRiE left, Actualeyes joined 01:40 SCHAAP137 left 01:49 rindolf left, cdg left 01:50 cdg joined 01:54 cdg left 02:02 tmch joined, themonkeybob11 joined 02:06 telex left 02:08 telex joined 02:09 zpmorgan joined 02:10 themonkeybob11 left 02:12 themonkeybob11 joined 02:17 themonkeybob11 left 02:23 tardisx left 02:24 partly_ joined 02:27 tardisx joined, kalkin- left 02:35 addison left, kid51 left 02:37 ParsonsNose joined 02:45 quester joined, ilbot3 left 02:47 ilbot3 joined 02:50 Actualeyes left 03:07 ParsonsNose left 03:28 JimmyZ_ joined 03:35 JimmyZ_ left 03:41 cbk_ left 03:52 noganex joined 03:55 noganex_ left, rburkholder left 04:04 Guest34004 is now known as pierrot, pierrot left, pierrot joined, pierrot left, pierrot joined 04:16 molaf joined 04:20 labster left 04:22 Actualeyes joined 04:23 themonkeybob11 joined 04:35 cognominal joined 04:50 molaf left
Brock At DC-Baltimore Perl Workshop (April 16 -- Baltimore, MD, USA) I'm going to give a hands-on tutorial getting people started with Perl 6! 04:51
awwaiid dcbpw.org/ for details & sign-up :)
I'm also thinking of running a Randori session there, though I might do that on the hackathon day
04:54 BenGoldberg left 04:56 labster joined 05:04 cognominal left 05:06 tardisx left 05:08 cdg joined 05:12 labster left 05:15 anaeem1_ joined 05:20 dalek left 05:21 yoleaux left, yoleaux joined, ChanServ sets mode: +v yoleaux, dalek joined, ChanServ sets mode: +v dalek, cdg left 05:27 anaeem1__ joined
TEttinger awwaiid: will there be softshell crab? that's much of what I know baltimore for. one of my best friends in high school grew up in maryland and was absolutely done with crab by the time he was a teenager. 05:30
awwaiid TEttinger: I'm sure it can be obtained during your visit :)
TEttinger haha
awwaiid some of my coworkers (in DC, but still) worked as teenagers doing various things with crabs, though they didn't report being sick of them quite 05:31
05:31 anaeem1_ left, anaeem1__ left 05:32 anaeem1_ joined, anaeem1_ left
TEttinger it is nice that baltimore has some uncommon features like that. pretty good football team I think, some years; it was interesting that one of the Ravens'... linemen I think, is also publishing math papers while he's a pro football player 05:32
being known for softshell crab is better than what many other cities are known for :) 05:33
awwaiid hehe 05:34
TEttinger I live in the LA area, and even though we definitely deserved the reputation for smog capital of america for many years, it's really gotten a fair bit better 05:35
05:35 nakiro joined
TEttinger still very bad traffic here. 05:35
but that's a recurring theme around all major cities it seems. public transit is kinda impossible here though, so sprawling 05:36
05:43 Cabanossi left
awwaiid I got this hired.com report on cost-of-living comparisons, and LA had a really good ratio compared to other cities (good purchasing power AND good absolute pay) 05:45
TEttinger huh 05:46
05:46 Cabanossi joined
TEttinger I guess "LA' could really mean a number of different areas 05:46
san bernardino county has massive amounts of cheap homes built during the housing boom, but it's quite the commute to downtown LA or most other areas 05:47
05:48 wamba left, wamba joined, DANtheBEASTman joined
TEttinger gas is pricey here in california due to more stringent environmental regulations, and also supply and demand (demand is HIGH) 05:49
05:49 idiosyncrat_ left
DANtheBEASTman i have a working perl6 install with rakudobrew, but having trouble figuring out how to make sure it's up to date and upgrading if not 05:49
05:51 khw left 05:52 rindolf joined
awwaiid DANtheBEASTman: try 'perl6 -v' and see what it says. The latest rakudo is 2016.02, though I kinda recommend 2016.01.1 with Rakudo Star if you don't have any reason to be more cutting edge 05:52
DANtheBEASTman: you can also use the (I think undocumented) "rakudobrew list" to see what you have installed 05:53
DANtheBEASTman cheers. a friend just told me to `perlbrew self-upgrade && perlbrew build moar` so i'm waiting on that now 05:54
awwaiid Lovely; that should hopefully get you 2016.02
DANtheBEASTman yeah i was on 2015.11-474-gf457007 (which doesn't seem tooo old..)
awwaiid thought MIGHT get you the 'nom' version
you definitely want 2015.12 at the LEAST, but as I said, really 2016.01.1 or 2016.02 are much better. There were a lot of fixes between your version and the Christmas Release 05:55
oh wait, you said "perlbrew". I should pay attention 05:56
"perlbrew" is a perl5 version manager, "rakudobrew" is a rakudo/perl6 version manager -- you probably meant rakudobrew 05:57
(esp since the params you mention are for rakudobrew. so probably just muscle memory -> IRC I'm guessing)
DANtheBEASTman yeah you're right. i said perlbrew but i meant rakudobrew.
awwaiid :)
DANtheBEASTman still new af to perl. coming from a mostly php/node.js background. first project is to port this horrible runlength encoder I made today into perl 05:58
05:58 skids left
DANtheBEASTman well, it's not too horrible, it works exactly as I intended, i just never figured out how to handle anything but letters.. gist.github.com/DanielFGray/cefe1b...364bac0ee9 06:00
awwaiid sounds fun 06:06
llfourn How do I rethrow an exception with the backtrace. It seems when I do $!.rethrow it only dies with the message.
awwaiid it is probably the decoding part that is restricted to letters from the \w . Note though that with this encoding you'll probably never be able to encode numbers since you won't be able to tell the difference between a number for the lenght and the number for an expansion 06:07
DANtheBEASTman awwaiid: yeah i'm unsure of how to actually take arbitrary inputs
awwaiid llfourn: doc.perl6.org/routine/rethrow says you're doing it right
Timbus www.rosettacode.org/wiki/Run-length...ing#Perl_6 <- if you ever need to be inspired... (also I think the workaround is no longer needed?) 06:08
llfourn awwaiid: hmm I thought so. Maybe it's a bug.
DANtheBEASTman oh dear. `rakudobrew build moar` failed.. 06:09
awwaiid DANtheBEASTman: you should work out an encoding format that can more exactly differentiate data from meta-data. Maybe a special character to separate the repeated-char from the repeated-char-count for example
Timbus that would be cool 06:10
DANtheBEASTman sprunge.us/iCdY
awwaiid DANtheBEASTman: to simplify, and assuming you don't have a reason otherwise, you might just "rm -rf /home/dan/.rakudobrew/" and start fresh with a new rakudobrew install 06:12
DANtheBEASTman will give that a go
awwaiid DANtheBEASTman: also, I suggest "rakudobrew build moar 2016.02" 06:13
that way it'll get the specific version, in case it doesn't defaut correctly
DANtheBEASTman is it worth building from the AUR on archlinux? i see there's rakudo-git .. 06:15
i guess i'll just tias 06:17
re RLE: i'd thought about returning an array, with data and metadata as separate items 06:21
06:22 virtualsue joined 06:23 sftp left, sftp joined 06:39 virtualsue left 06:48 bjz joined 06:50 domidumont joined 06:54 domidumont left 06:55 domidumont joined 06:59 cibs joined
Hotkeys DANtheBEASTman: I'm on arch and I just build from source, but I guess rakudo-git would work 06:59
07:03 xinming left 07:04 xinming joined 07:14 virtualsue joined 07:24 bjz_ joined 07:25 bjz left, rhr left 07:27 kjs_ joined 07:32 rhr joined 07:33 wamba left
dalek line-Perl5: 518b6d1 | (Brock Wilcox)++ | / (2 files):
Set up travis
 07:37
line-Perl5: 9c6fa98 | niner++ | / (2 files):
Merge pull request #54 from awwaiid/travis

Wow, thank you so much!
07:38 Relsak joined, firstdayonthejob joined, zpmorgan left 07:39 sno left 07:43 bjz joined, labster joined 07:44 CIAvash joined, tinita_ joined, bjz_ left, tinita left 07:48 RabidGravy joined 07:49 darutoko joined
RabidGravy MARNING 07:49
awwaiid nine_: yeah, I had that on the back-burner for a while :) 07:50
dalek osystem: 29fa5a0 | azawawi++ | META.list:
Add Memoize to ecosystem
 07:56
RabidGravy azawawi++ # harr 07:58
dalek rl6-most-wanted: 4f05d4d | azawawi++ | most-wanted/modules.md:
Added WIP for Memoize
07:58 AlexDaniel joined 08:02 virtualsue left 08:05 Actualeyes left 08:06 ely-se joined 08:23 zakharyas joined 08:32 fireartist joined, sno joined 08:39 Amnez777 left, Amnez777 joined 08:40 firstdayonthejob left 08:43 kjs_ left 08:59 dakkar joined, rurban joined, Gothmog_ joined 09:00 virtualsue joined 09:02 zakharyas left 09:06 [ptc] joined 09:23 quester left
dalek kudo/jvminterop: 7242188 | peschwa++ | src/vm/jvm/Perl6/Metamodel/JavaHOW.nqp:
Return the right thing when we don't find a method.

This way we get a proper X::Method::NotFound, instead of a low-level "Can not invoke this object."		 09:31
kudo/jvminterop: 2cc0f06 | peschwa++ | t/03-jvm/01-interop.t:
Remove erroneous comment.
kudo/jvminterop: 17e2679 | peschwa++ | t/03-jvm/01-interop.t:
Add a test for throwing X::Method::NotFound.
psch the next obvious issue would be throwing the right thing when the user passes wrongly typed arguments to a Java method... 09:33
which seems like it involves quite a bit of shuffling things from BootJavaInterop to RakudoJavaInterop 09:34
the former can get away with just dying straight out of the corresponding SMO subtypes get_* method, but rakudo can't 09:35
09:35 [ptc] left
psch not sure what the good way to solve that is, though 09:35
09:36 yeahnoob left
psch what comes to mind immediately is type checking when trying to fit the arguments into the Java class 09:36
but that kind of replicates a lot of Binder code in the interop code, which i don't think is a good idea
09:37 [ptc] joined
psch so i probably should try and get Binder involved there directly, if possible... 09:38
*the
hm, actually the Binder should already be involved i suppose 09:44
09:54 abraxxa joined
AlexDaniel m: say grep:kv(/./, <a b c>) 09:57
camelia rakudo-moar 50a4df: OUTPUT«Regex object coerced to string (please use .gist or .perl to do that) in any canonicalize_pair at src/Perl6/World.nqp line 4040␤Regex object coerced to string (please use .gist or .perl to do that) in any canonicalize_pair at src/Perl6/World.nqp line 4…»
masak m: say grep:kv(rx/./, <a b c>) 10:00
camelia rakudo-moar 50a4df: OUTPUT«Regex object coerced to string (please use .gist or .perl to do that) in any canonicalize_pair at src/Perl6/World.nqp line 4040␤Regex object coerced to string (please use .gist or .perl to do that) in any canonicalize_pair at src/Perl6/World.nqp line 4…»
10:03 SCHAAP137 joined 10:09 Skarsnik joined
Skarsnik Hello 10:11
10:11 araujo left
timotimo o/ 10:12
Skarsnik Any idea on what can I do to run a part of my code that leak, I mean to freely kill it and restart it? I tried using fork but it lock the C lib I use to parse html 10:14
leak = fill 2GB in 24h x) 10:15
gregf_ hi 10:16
i've installed a module using panda. is there a way to check where it got installed (~/.rakudobrew does not have it). something like $INC{"Foo/Bar.pm"} 10:17
timotimo Skarsnik: just isolate the workload in a separate worker process (full process; a thread isn't enough) and have it exit when it's done enough work? 10:18
Skarsnik Yes, but how I do that without fork? I run a different script?
gregf_ `locate Tiny.pm | grep JSON` <- does not return any results *ive installed JSON::Tiny* 10:19
Skarsnik gregf_, try using it? it's put in PERL6 Install/site/panda/... something like that
and the name is marshamaled
so you can't find it with a locate/find 10:20
gregf_ Skarsnik: using it works.. but i just wanted to know where modules get installed by default
RabidGravy m: say $*REPO.repo-chain 10:21
camelia rakudo-moar 50a4df: OUTPUT«(inst#/home/camelia/.perl6/2016.02-72-g50a4df3 inst#/home/camelia/rakudo-m-inst-1/share/perl6/site inst#/home/camelia/rakudo-m-inst-1/share/perl6/vendor inst#/home/camelia/rakudo-m-inst-1/share/perl6 CompUnit::Repository::AbsolutePath.new(next-repo => Comp…»
partly_ OMFG I think i figured out my OpenSSL issue, which i talked about yesterday. An XMPP Server uses StartTLS to signal that it want to start a secure connection. I need to handle this in my application, before I switch to SSL
Skarsnik Is there a page on the doc about compunit/repo stuff? 10:22
10:22 partly_ is now known as kalkin, kalkin is now known as kalkin-
RabidGravy no, I was going to start writing it, got distracted and mde some software instead 10:22
10:23 araujo joined
RabidGravy gregf_, gist.github.com/jonathanstowe/be95...f713cff2b5 has the basic guts of a thing to find *all* the installed modules, this could be adapted 10:23
gregf_ RabidGravy: nice, thanks... hmmm quite a few locations :| 10:25
also, out os curiosity. would there be an AWS api using Perl6 anytime? *the Perl5 ones are pretty much incomplete :|* 10:27
s/os/of/
moritz there would be one if somebody made one
timotimo nobody knows. except the person who makes it
moritz "someone like you" 10:28
gregf_ we do use a lot of the API(S3/DynamoDb,Sns) :| *all PHP or Java*
heh
Skarsnik x)
gregf_ working on Perl6 makes me feel im using Ruby .... from a distance ;) 10:29
Skarsnik wow Stage parse : 108.744 it was around 70 for me previously 10:32
RabidGravy I've got to make some adaptions to HTTP::UserAgent for it to be able to properly work with AWS
Skarsnik I feel like h:ua crash on some malformed http occasionnaly 10:33
I had to put some code on a try {} block because of that x)
RabidGravy entirely possible, but without test cases it won't get fixed 10:34
Skarsnik I should keep the full trace of this code in a file
10:35 molaf joined
RabidGravy really the URI of a server that makes http that H::UA can't handle would be enough 10:44
Skarsnik hm running something with perl6-valgrind-m output nothing outside the start, is that normal? 10:45
RabidGravy, na it's occasionnal (I fetch regulary the same page) 10:46
10:46 cognominal joined, virtualsue left
timotimo you need to know that valgrind is EXTREMELY slow 10:47
expect your script to take at lest 20x longer to do anything at all
Skarsnik Yes, but I mean, it reached some say (yes it take a while) but it output nothing 10:49
timotimo you may have more luck building moarvm with --asan (a Configure.pl option), that's about 3x slower and catches many things valgrind catches, but not nearly everything
how can you know it reaches the say if it outputs nothing?
Skarsnik I mean, there is no valgrind output 10:51
timotimo that means everything is going well
valgrind only outputs things if things go wrong
==18801== Memcheck, a memory error detector
==18801== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al.
==18801== Using Valgrind-3.11.0 and LibVEX; rerun with -h for copyright info
^- it did output this, right?
Skarsnik I tried to track down this perl6 leak x)
Yes
10:52 zakharyas joined
timotimo well, you cannot detect a leak until the program ends 10:52
btw, if you're looking for leaks, you really do need to run your moar with --full-cleanup and also tell valgrind to --leak-check=full
Skarsnik Oh yeah, it only show error in memory allocation/free
timotimo if you do not use --full-cleanup, moar will just leave everything allocated and exit when it's finished 10:53
arnsholt Which generates a bunch of "potential leak" messages (or something along those lines. It's been a while since I last used valgrind) 10:54
10:56 brainwash left 10:57 labster left
Skarsnik it's like 100x slower x) 10:59
arnsholt Wouldn't surprise me. It's really slow
timotimo yes, valgrind is really, really slow. 11:01
11:04 espadrine_ joined, espadrine_ is now known as espadrine, rurban left
Skarsnik hm, aside from the 4MB lost at the end, it does not look that bad gist.github.com/Skarsnik/7863fd11b361df6fd931 11:06
11:06 RabidGravy left
jnthn ==17380== 4,404,470 (4,002,840 direct, 401,630 indirect) bytes in 200,142 blocks are definitely lost in loss record 68 of 68 11:07
timotimo you'ven't built moar with --debug?
jnthn ==17380== at 0x40291CC: malloc (vg_replace_malloc.c:296)
Skarsnik but it's close to what is lost on a real run of this for each iteration of the loop
jnthn ==17380== by 0x40B51AC: MVM_args_copy_callsite (in /home/skarsnik/devel/leak-rakudo/rakudo/install/lib/libmoar.so)
Wow, that's not good...
timotimo jnthn: i've seen something similar to this somewhere else, too
Skarsnik I use default option of Configure.pl
I added --full-cleanup to moar and --leak-check=full to the valgrind-m script 11:08
timotimo i always --debug=3 --optimize=3 and also set CFLAGS=-march=native
we need some way to give perl6-valgrind-m arguments that get passed to valgrind itself 11:09
Skarsnik want me to try with debug and opti option? 11:10
timotimo that'll at least give line numbers and filenames in the traceback :)
which i find is always nice
Skarsnik lookfor is something that recurse a lot (the xml tree is quite huge/deep) 11:11
optimize is a good idea? does it change much the behavior? 11:12
timotimo the --optimize flag is only for the C compiler 11:14
if lookfor is implemented in the native library (gumbo?) then it can't do anything about that whatsoever. if lookfor is implemented in perl6, it won't do anything about that, because in order to make that better it'd have to improve the bytecode, or the semantics :P
Skarsnik Na it's perl6 native (the xml moduke) 11:15
timotimo OK, then it will very unlikely change the performance much
recursion - well, invocation in general - is somewhat expensive in moarvm at the moment. it's on jnthn's to-do list for this year. 11:16
Skarsnik Well the execution time does not bother me, but the fact I lost 4Mb for each iteration of the loop it's really annoying :) 11:17
timotimo that's not something the optimizer will be able to fix 11:19
Skarsnik I mean, I though the optimize flag will change stuff in moar on how it handle thing 11:20
if it's just a compiler option, I guess it change nothing
brb lunch 11:24
11:30 kaare_ joined 11:40 johndau joined, Sherwool joined 11:42 TEttinger left 11:44 rurban joined 11:47 johndau left
dalek href="https://perl6.org:">perl6.org: dea96f3 | (Steve Mynott)++ | source/downloads/index.html:
add one line docker install docs
 11:50
11:52 kid51 joined 11:56 ely-se left 12:08 sufrostico joined 12:11 virtualsue joined
dalek kudo/faster-accessors: b66dc24 | jnthn++ | src/Perl6/World.nqp:
Generate correct signature for accessors.
 12:14
kudo/faster-accessors: c69c013 | jnthn++ | src/Perl6/World.nqp:
Share signature of accessors within a package.

Since they're identical, and there's no user opportunity to apply any traits to them. Shaves 160KB off CORE.setting size relative to after adding the accessor generation code.
p: 79e10b4 | jnthn++ | src/vm/moar/QAST/QASTCompilerMAST.nqp:
Workaround for unsigned attribute access bug.
timotimo jnthn: how much bigger does the core setting get with generated accessors? 12:16
i imagine it could be a sizable bit
12:17 wamba joined 12:18 Sherwool left
timotimo compares locally 12:19
jnthn Will find out soon
Going to merge that in a moment :) 12:20
dalek p: 902c10d | jnthn++ | tools/build/MOAR_REVISION:
Bump for performance, inlining fixes.
 12:21
jnthn timotimo: nom: 11,587,320 bytes. faster-accessors: 11,457,008 12:23
timotimo huh. i was expecting the other way around :D
oh well ;)
i'm liking that
jnthn m: say "so, it's {11587 - 11457}KB smaller"
camelia rakudo-moar 50a4df: OUTPUT«so, it's 130KB smaller␤»
lizmat looks forward to seeing all these opts
12:23 kid511 joined
jnthn timotimo: Well, remember that it was serializing closures before, which meant they accidentally took in a bunch of scoping 12:23
uh
timotimo oh! 12:24
jnthn A bunch of scopes
timotimo that makes sense indeed
jnthn I was expecting to come out with about a draw though
lizmat but first I'm going to be offline for a bit
&
12:24 kid51 left, lizmat left
timotimo this potentially also means we'll see shorter gen2 root lists, too? 12:24
jnthn So, coming out with a win on space *and* time is nice
timotimo: yeah
timotimo awesome :)
jnthn Though not sure how much of a dent it'll really make
Skarsnik I don't have the line in valgrind output even with --debug=3 on moar, hm 12:25
timotimo uh oh
jnthn For anyone curious, together this work makes gist.github.com/jnthn/ee65e3be80f4a5485944 go from around 3.34s down to around 0.65s
timotimo why does this make that crash ... 12:26
Cannot unbox a type object
at gen/moar/stage2/QAST.nqp:4364 (/home/timo/perl6/install/share/nqp/lib/QAST.moarvm:push_op:2)
jnthn That looks very much like something in RT 12:27
timotimo the branch merge tickles it
jnthn What code? 12:28
timotimo trying to run my white_noise.p6 from the SDL2::Raw examples/ folder
it dies while precompiling
jnthn Got any unsigned attributes?
timotimo panda install . also does that
could be, gimme a moment
yes, a whole bunch of 'em
jnthn aha 12:29
I suspect it tickles an existing issue
timotimo in repr('CPointer') classes as well as native functions, their return types and also parameterizing Bufs
jnthn Got a golf, ooc? But I can also take the one in rT
timotimo i'll try, gimme a minute
jnthn OK 12:31
lunch here :)
timotimo i've got one 12:33
jnthn OK; gist it and I'll check if the fix for that RT that I'll do after lunch also fixes it :)
timotimo class golf { has uint32 $.format is rw; }
remove the "is rw" and it's gone. remove the "u" and it's gone
unsurprisingly, "is rw" can also go on the class instead of the attribute 12:35
so ... do i have to gist that? :D
data:text/plain;charset=utf-8;base64,Y2xhc3MgZ29sZiB7IGhhcyB1aW50MzIgICAkLmZvcm1hdCBpcyBydzsgfQ== 12:36
^- here, have it as a URL :D
12:39 rurban left, rurban joined, kid511 left
Skarsnik hm, is there something in moar to free moarstruct? 12:42
timotimo "moarstruct"? 12:44
Skarsnik I mean like freeMVMCallsite, but I found it x) 12:46
timotimo i'm not certain MVMCallsite is gc-managed 12:48
psch hm, complicated /o\
Java methods are ForeignCode objects
timotimo right. it is not.
psch which means i only get :(| is raw) for them, which is probably why the Binder doesn't really care
timotimo so it's just a Plain Old Struct. if it gets malloc'd, you'll just free it later
urgh :| 12:49
psch and BootJavaInterop can't do any Exception-y stuff, 'cause it doesn't even neccessarily know about the Exception i'd like thrown
arnsholt Skarsnik: I hope you're not trying to free things from Perl 6 code by binding Moar internals with NativeCall. That's not going to work out
Skarsnik MVMCallsite is a MVMThreadContext?
psch but nqp::callmethod is defined in QAST::Compiler 12:50
Skarsnik I try to figure what cause the leak x)
psch so i'd have to lift that up into P6 land to call a RakOps method instead of an Ops method..?
err, no, callmethod calls IndyBootstrap.methcall_noa
timotimo Skarsnik: wat? "is a"? 12:51
all i can see is that you're clearly very confused ?!?
Skarsnik can be cast if you prefer, there is no cleanup function for MVMCallSite 12:52
timotimo well, yeah, you can cast a MVMCallsite into an MVMThreadContext if you like
you're not going to get data that makes much sense out of it, though
psch which means the cleanest way might actually be subclassing ForeignCode and attaching a compatible Signature
timotimo Skarsnik: did you see MVM_callsite_destroy?
psch but that'd mean ditching all the mmd work i did up to now, 'cause i'd have to have P6 level protos instead 12:53
12:53 edehont joined
timotimo psch: argh :( 12:53
psch ...which i don't even know if that really works, seeing as the marshalling still has to happen on the Java level
12:53 skids joined
psch well, or stick with the uncomfortable alternative and reimplement parts of the Binder in RakudoJavaInterop... :| 12:54
Skarsnik hm, ok this destroy function look correct
12:59 Actualeyes joined
Skarsnik github.com/MoarVM/MoarVM/blob/mast...args.c#L72 hm, should the malloc be fsize * sizeof(src_flags)? 13:00
13:02 skids left
arnsholt Probably not. If the source of your leak is callsite objects leaking, it's almost definitely not a logic bug in the callsite class itself, but rather a problem in some part of the code that *uses* callsites 13:04
13:06 wamba left, wamba joined
rudi_s I'd like to create a list of "quoted" paths. my @a = 'foo', 'bar', 'baz' should be converted to '"foo", "bar", "baz"' (it doesn't have to correctly " in the filenames). Basically I want join which also wraps the arguments. Any ideas? 13:07
DrForr rudi_s: map{ qq{"$_"} } @paths 13:08
13:09 virtualsue left
DrForr "A join which also..." that way lies madn^wPHP :) 13:09
13:09 uruwi left
rudi_s DrForr: Awesome, thank you ;-) 13:09
psch string_real_escape_and_quote_join() # perfectly sane! 13:10
13:13 tinita_ is now known as tinita
rudi_s Sounds great ;-) 13:13
dalek p: be15097 | jnthn++ | src/vm/moar/QAST/QASTCompilerMAST.nqp:
Fix crash in unsinged attributeref code-gen.
 13:15
13:15 riatre left 13:16 riatre joined
jnthn timotimo: The NQP patches should fix things up :) 13:17
DrForr I always seem to singe my attribute refs.
jnthn And also that RT :)
hah :P
m: class MV { has uint64 $.start; method s { if $!start {} } } 13:21
camelia rakudo-moar 50a4df: OUTPUT«===SORRY!===␤Cannot unbox a type object␤»
jnthn Turns out RT #127548 golfs further - doesn't need the CStruct repr
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=127548
13:22 huggable joined 13:25 uruwi joined
[Coke] finds out there's a perl workshop in april a scant 5 hour drive from his house. nifty. 13:32
13:34 fireartist left
DrForr NLPW? 13:37
[Coke] dcbpw.org/dcbpw2016/ - mentioned in backscroll 13:38
dalek kudo/nom: 62ed92a | jnthn++ | tools/build/NQP_REVISION:
Bump NQP_REVISION for code-gen fixes, Moar fixes.
kudo/nom: fdd37a9 | jnthn++ | src/ (12 files):
Start to code-gen simple accessors.

This makes them rather simpler/faster than adding them as closures. Most importantly, though, it will enable inlining of them, which is where the real speed-up will come from.
kudo/nom: ae7f822 | jnthn++ | src/Perl6/World.nqp:
Generate correct signature for accessors.
kudo/nom: 3176cb1 | jnthn++ | src/Perl6/World.nqp:
Share signature of accessors within a package.

Since they're identical, and there's no user opportunity to apply any traits to them. Shaves 160KB off CORE.setting size relative to after adding the accessor generation code.
ast: eb2b08e | jnthn++ | S12-attributes/native.t:
Tests for RT #127548.
 13:40
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=127548
13:40 RabidGravy joined
RabidGravy Boom! 13:41
yuck weather out there
jnthn Rained here and washed all the pretty snow away :( 13:45
13:48 andreoss joined, themonkeybob11 left 13:49 perlpilot left, perlpilot joined
dalek rl6-bench: ea2a4d0 | jnthn++ | microbenchmarks.pl:
Add object accessor micro-benchmark.
 13:51
arnsholt jnthn: Better than half of it melting away and the other half freezing to ice again =) 13:52
13:53 andreoss left
stmuk wonders what the weather in Nuremberg will be like 13:53
jnthn arnsholt: True :) 13:55
llfourn m: sub foo { "foo".return }; foo().say # I don't get it 13:56
camelia rakudo-moar 3176cb: OUTPUT«Attempt to return outside of any Routine␤ in block <unit> at /tmp/KTOQC8ZyDc line 1␤␤»
13:57 nakiro left
jnthn We have a .return? :S 13:58
llfourn apparently :P
it's in Mu
psch m: sub f { my $ret = Mu.^find_method('return'); $ret("foo") }; f().say 14:00
camelia rakudo-moar 3176cb: OUTPUT«foo␤»
llfourn heh that's interesting
psch i'd guess the anon Capture doesn't get the invocant right or something..? 14:01
m: class A { method f(|c) { c.perl.say } }; A.new.f("foo")
camelia rakudo-moar 3176cb: OUTPUT«\("foo")␤»
llfourn m: sub a { b() }; sub b { "foo".return }; b().say;
camelia rakudo-moar 3176cb: OUTPUT«Attempt to return outside of any Routine␤ in block <unit> at /tmp/hPrmBPJzrF line 1␤␤»
psch note that in A.f there it doesn't print the invocant 14:02
llfourn m: sub a { b() }; sub b { "foo".return }; a().say; # woops
camelia rakudo-moar 3176cb: OUTPUT«foo␤»
llfourn so it goes one too far out
14:03 liztormato joined
jnthn Yeah, maybe it's a scope off-by-one or something 14:03
liztormato The .return was meant for idioms like " .return with $foo " 14:04
Returning only if the value is defined 14:05
llfourn liztormato: That's just what I was about to use it for until I found it was broken :)
liztormato Argh
jnthn Guessing it's not very tested :)
llfourn well .return with %hash{$key} # specifically
14:06 andreoss joined, andreoss left, andreoss joined
liztormato Well. We should add some tests then first 14:06
14:06 themonkeybob11 joined
llfourn I added to TODO list 14:08
14:09 cdg joined
andreoss m: my \x = ">>>" ; multi infix:x($a,$b){ } 14:09
camelia rakudo-moar 3176cb: OUTPUT«===SORRY!===␤Cannot unbox a type object␤»
liztormato llfourn++
14:09 cdg left
andreoss m: my $x = ">>>" ; multi infix:$x($a,$b){ } 14:09
camelia ( no output )
14:09 cdg joined
jnthn Skarsnik: I've golfed what's probably the same leak you're seeing to `multi foo($ where 1) { }; loop { foo(|(1,)) }` 14:10
psch m: my $x = ">>>" ; multi infix:$x($a,$b){ say "ok" }; 1 >>> 2
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tsN2Xry5pm␤Missing << or >>␤at /tmp/tsN2Xry5pm:1␤------> 3multi infix:$x($a,$b){ say "ok" }; 1 >>>7⏏5 2␤ expecting any of:␤ infix␤ infix stopper␤»
14:11 liztormato left, themonkeybob11 left
Skarsnik jnthn, weird, there is no multi involved. I tried to golf it without success so far 14:12
14:12 rurban1 joined
jnthn Skarsnik: If you're using Perl 6 for anything non-trivial, there'll be multis involved :) 14:12
14:12 themonkeybob11 joined
jnthn Skarsnik: There's more than one way to reach the leaky code-path though, this is just the way I engineered from the valgrind output you sent. 14:13
14:13 rurban2 joined
Skarsnik if you find a short case, it's better ^^ 14:15
edehont While looking for a way to loop over a multidimensional array, I thought something like .shape would be quite handy. Turns out it already exixts... But not in the docs.
14:16 rurban left
edehont Is that intentional? 14:16
14:17 rurban1 left, rurban joined
jnthn edehont: No, just missing, I expect. Please file an issue on the perl6/doc repo, if you have a github account 14:17
14:17 rurban1 joined
llfourn ( or a pull request :D) 14:17
jnthn Skarsnik: Yeah. Now I need to understand what's happening :)
Well, or that ;)
edehont I will give it a shot! 14:18
llfourn edehont++
14:18 themonkeybob11 left
jnthn edehont: Why do you need shape to iterate over it by the way? :) 14:19
m: my @a[2;2] = (1, 2), (3, 4); for @a { .say }
camelia rakudo-moar 3176cb: OUTPUT«1␤2␤3␤4␤»
jnthn m: my @a[2;2] = (1, 2), (3, 4); for @a.kv -> $indices, $value { say "$indices => $value" }
camelia rakudo-moar 3176cb: OUTPUT«0 0 => 1␤0 1 => 2␤1 0 => 3␤1 1 => 4␤»
14:20 rurban2 left
andreoss m: my $x = "xyz";; sub infix:$x($a, $b) { $a + $b } ; 1 xyz 3; 14:20
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/s9_JjS91Md␤Two terms in a row␤at /tmp/s9_JjS91Md:1␤------> 3";; sub infix:$x($a, $b) { $a + $b } ; 17⏏5 xyz 3;␤ expecting any of:␤ infix␤ infix stopper␤ statement end…»
andreoss what is $x supposed to be? 14:21
m: sub infix:"xyz"($a, $b) { $a + $b } ; 1 xyz 3;
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/jgvxZtzsEC␤Missing block␤at /tmp/jgvxZtzsEC:1␤------> 3sub infix:7⏏5"xyz"($a, $b) { $a + $b } ; 1 xyz 3;␤»
andreoss also why this doesn't work?
psch m: sub infix:<"xyz">($a, $b) { "ok" }; say 1 xyz 2
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/YK37GD5qcN␤Two terms in a row␤at /tmp/YK37GD5qcN:1␤------> 3ub infix:<"xyz">($a, $b) { "ok" }; say 17⏏5 xyz 2␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ …»
psch hm, seems that broke..?
ah, no 14:22
14:22 rurban left
psch < > quotes 14:22
m: sub infix:<xyz>($a, $b) { "ok" }; say 1 xyz 2
camelia rakudo-moar 3176cb: OUTPUT«ok␤»
psch andreoss: you need a delimiter around your operator
andreoss "" used to work to, didn't it?
14:22 uruwi left
jnthn m: sub infix:<<"xyz">>($a, $b) { "ok" }; say 1 xyz 2 14:23
camelia rakudo-moar 3176cb: OUTPUT«ok␤»
jnthn You'd need that to interpolate 14:24
I don't think infix:"..." has ever worked, no
It's not a valid colonpair
psch m: my $x = "xyz"; sub infix:<<$x>>($a, $b) { "ok" }; say 1 xyz 2
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/fHwrD3VrUz␤Colon pair value '$x' too complex to use in name␤at /tmp/fHwrD3VrUz:1␤------> 3my $x = "xyz"; sub infix:<<$x>>7⏏5($a, $b) { "ok" }; say 1 xyz 2␤»
psch hm, that's broken though :/
not sure since when... i'm pretty sure we should have tests for it though
jnthn I don't think it needs to work
I mean
That one *can't* work
Ever.
14:25 Ven joined
jnthn Think about the timing 14:25
psch oh, right
jnthn m: constant $x = "xyz"; sub infix:<<$x>>($a, $b) { "ok" }; say 1 xyz 2
camelia rakudo-moar 3176cb: OUTPUT«ok␤»
psch constant should yeah
jnthn And that one is fine :)
andreoss m: my $x = "xyz";; my sub infix:<<$x>>($a, $b) { $a + $b } ; 1 xyz 3;
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/gMxlzfEaG4␤Colon pair value '$x' too complex to use in name␤at /tmp/gMxlzfEaG4:1␤------> 3my $x = "xyz";; my sub infix:<<$x>>7⏏5($a, $b) { $a + $b } ; 1 xyz 3;␤»
andreoss m: INIT my $x = "xyz";; my sub infix:<<$x>>($a, $b) { $a + $b } ; 1 xyz 3; 14:26
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/AGXRTxk470␤Colon pair value '$x' too complex to use in name␤at /tmp/AGXRTxk470:1␤------> 3INIT my $x = "xyz";; my sub infix:<<$x>>7⏏5($a, $b) { $a + $b } ; 1 xyz 3;␤»
jnthn andreoss: You need constant
INIT is too late
It's after compile time
m: BEGIN my $x = "xyz";; my sub infix:<<$x>>($a, $b) { $a + $b } ; 1 xyz 3;
camelia ( no output )
jnthn m: BEGIN my $x = "xyz";; my sub infix:<<$x>>($a, $b) { $a + $b } ; say 1 xyz 3;
camelia rakudo-moar 3176cb: OUTPUT«4␤»
jnthn Needs to be at BEGIN time. 14:27
But srsly, just use constant ;)
Ven o/
jnthn o/ Ven
14:28 rdleon left
edehont jhthn: Ah... so I don't need .shape after all for what I want. Thanks. 14:30
14:30 edehont left
jnthn Another situation saved from introspection land :) 14:31
llfourn wonders if we'll get that PR 14:32
jnthn Damn, yeah. I shouldn't be so helpful. :P
llfourn haha
Skarsnik m: sub foo() {}; our $foo is export = &foo; 14:34
camelia ( no output )
14:35 uruwi joined
Skarsnik m: sub foo() {}; constant $foo is export = &foo; 14:35
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/uiTcHx8CO3␤Can't use unknown trait 'is export' in a sub declaration.␤at /tmp/uiTcHx8CO3:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ pure default DEPRECATED i…»
llfourn hmmm 14:36
14:36 abaugher left
llfourn m: consant $foo is export = "foo" 14:36
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/DGWgtYLU6r␤Variable '$foo' is not declared␤at /tmp/DGWgtYLU6r:1␤------> 3consant 7⏏5$foo is export = "foo"␤»
llfourn m: constant $foo is export = "foo"
camelia ( no output )
Skarsnik m: constant $foo is export = sub {}; 14:37
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/yZ5YBVEieT␤Can't use unknown trait 'is export' in a sub declaration.␤at /tmp/yZ5YBVEieT:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ pure default DEPRECATED i…»
llfourn m: constant $foo is export = anon sub foo {};
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/vSocjMEWuC␤Can't use unknown trait 'is export' in a sub declaration.␤at /tmp/vSocjMEWuC:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ pure default DEPRECATED i…»
14:37 abaugher joined
llfourn nice bug. 14:37
m: constant $foo is export = -> { }
camelia ( no output )
llfourn m: constant $foo is export = anon method foo { } 14:38
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ksmZHV8KSr␤Can't use unknown trait 'is export' in a method declaration.␤at /tmp/ksmZHV8KSr:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ pure default DEPRECATE…»
llfourn looks to be assigning to a routine goofs it
it somehow thinks is export is export is being applied to the routine...or something 14:39
s/is export xx 2/is export/
Skarsnik m: constant $foo is export = Mu || sub {}; 14:41
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/6cBhyHrj_t␤Can't use unknown trait 'is export' in a sub declaration.␤at /tmp/6cBhyHrj_t:1␤ expecting any of:␤ rw raw hidden-from-backtrace hidden-from-USAGE␤ pure default DEPRECATED i…»
llfourn m: constant $foo is export = True || sub foo { }
camelia ( no output )
llfourn maybe there's some hackyness in the implementation of constant is export
awwaiid So I'm currently installing Task::Star into a fresh rakudo 2016.02. While it is running I try to start a new perl6 and it blocks waiting for ""/home/awwaiid/.perl6/2016.02/precomp/.lock" . If this was a production environment that would be bad I think 14:47
llfourn I've had similar weirdness with locks... like I start a websever and can't panda install anything until I stop it 14:48
dalek rakudo/repl6: f61e067 | hoelzro++ | src/ (2 files): 14:49
rakudo/repl6: REPL6: Add rudimentary Perl 6 REPL loading support
rakudo/repl6:
rakudo/repl6: This is the start of moving advanced REPL functionality into Perl 6.
rakudo/repl6: This is to lower the barrier to entry for improving the REPL, allowing
14:49 dalek left 14:50 dalek joined, ChanServ sets mode: +v dalek 14:51 virtualsue joined 14:54 skids joined 14:55 zakharyas left 14:59 sufrostico left 15:01 sufrostico joined
stmuk awwaiid: ISTR some locking behaviour being fixed in the last few days .. not sure if it's that 15:03
there is also the awful hack of -IFakeDir which I used in p6doc :/ 15:04
15:08 themonkeybob11 joined 15:09 virtualsue left 15:11 lizmat joined
lizmat waves from a ferry 15:12
Ven waves lizmat
at*
lizmat o/ Ven
15:14 uruwi left
masak o/ lizmat 15:14
15:15 themonkeybob11 left
jnthn hopes the sea isn't waving too much at lizmat :) 15:15
Skarsnik: Currently spectesting a patch to deal with the leak
lizmat naah... we won't be leaving the harbour for at least an hour
DrForr Heh. I recommended Dylan to use the ferry next time he's headed to the UK.
15:15 Relsak left
Skarsnik good :), ping me if you commit or want me to test 15:16
rudi_s m: use Test; is-deeply {a => Blob}, {a => Blob};
camelia rakudo-moar 3176cb: OUTPUT«Invocant requires an instance of type Blob, but a type object was passed. Did you forget a .new?␤ in any at gen/moar/m-Metamodel.nqp line 1723␤ in sub is-deeply at /home/camelia/rakudo-m-inst-1/share/perl6/sources/C712FE6969F786C9380D643DF17E85D068…»
rudi_s Is this expected?
llfourn rudi_s: nah looks like a bug 15:17
Skarsnik is deeply do gist to compare I think
llfourn :S
15:17 edehont joined
rudi_s Interestingly it works with Int 15:17
m: use Test; is-deeply {a => Int}, {a => Int};
camelia rakudo-moar 3176cb: OUTPUT«ok 1 - ␤»
Skarsnik Blob is a role, that maybe why 15:18
jnthn I thought is-deeply used eqv?
m: say {a => Blob} eqv {a => Blob}
camelia rakudo-moar 3176cb: OUTPUT«Invocant requires an instance of type Blob, but a type object was passed. Did you forget a .new?␤ in any at gen/moar/m-Metamodel.nqp line 1723␤ in block <unit> at /tmp/2ikvkWKM_q line 1␤␤»
jnthn m: say Blob eqv Blob
camelia rakudo-moar 3176cb: OUTPUT«Invocant requires an instance of type Blob, but a type object was passed. Did you forget a .new?␤ in any at gen/moar/m-Metamodel.nqp line 1723␤ in block <unit> at /tmp/RNJf5GbSYV line 1␤␤»
llfourn jnthn++
jnthn I wonder if the eqv candidate for Blob is missing a smiley? 15:19
psch m: say Blob
camelia rakudo-moar 3176cb: OUTPUT«(Blob)␤»
llfourn .say for &infix:<eqv>.candidates
m: .say for &infix:<eqv>.candidates
camelia rakudo-moar 3176cb: OUTPUT«sub infix:<eqv> ($?) { #`(Sub|69036160) ... }␤sub infix:<eqv> ($a, $b) { #`(Sub|69038136) ... }␤sub infix:<eqv> (@a, @b) { #`(Sub|69037832) ... }␤sub infix:<eqv> (Stringy:D $a, Stringy:D $b) { #`(Sub|69037984) ... }␤sub infix:<eqv> (Numeric:D $a, N…»
jnthn m: say &infix:eqv.cando(Int,Int)
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/2b4M6T19Sa␤Undeclared routine:␤ infix:eqv used at line 1. Did you mean 'infix:<eq>', 'infix:<eqv>', 'infix:<∖>'?␤␤»
moritz src/core/Buf.pm:multi sub infix:<eqv>(Blob:D \a, Blob:D \b) {
jnthn m: say &infix:<eqv>.cando(Int,Int)
camelia rakudo-moar 3176cb: OUTPUT«Too many positionals passed; expected 2 arguments but got 3␤ in block <unit> at /tmp/06vyQdFGs8 line 1␤␤»
jnthn m: say &infix:<eqv>.cando(\(Int,Int))
camelia rakudo-moar 3176cb: OUTPUT«(sub infix:<eqv> ($a, $b) { #`(Sub|63176776) ... })␤»
jnthn m: say &infix:<eqv>.cando(\(Blob,Blob)) 15:20
camelia rakudo-moar 3176cb: OUTPUT«(sub infix:<eqv> (@a, @b) { #`(Sub|58109752) ... } sub infix:<eqv> ($a, $b) { #`(Sub|58110056) ... })␤»
jnthn m: say Blob ~~ Positional
camelia rakudo-moar 3176cb: OUTPUT«True␤»
llfourn o.o
moritz uhm, I hope @a constrains to Positional:D ?
jnthn Ah, so it's hitting an @a, @b candidate in that case
llfourn what moritz said
jnthn moritz: I'm guessing not...though maybe it should
llfourn m: sub (@a) { }.(List) 15:21
camelia ( no output )
llfourn +1 for that
jnthn Something we could fix in 6.d, since we'd be able to do it conditionally in the compiler
15:22 edehont left
lizmat jnthn: multi sub infix:<eqv>(Blob:D \a, Blob:D \b) 15:22
jnthn lizmat: Read the last few minutes of discussion :) 15:24
m: say List eqv List 15:25
camelia rakudo-moar 3176cb: OUTPUT«True␤»
jnthn Curious why that doesn't hit it though
m: say &infix:<eqv>.cando(\(List,List))
camelia rakudo-moar 3176cb: OUTPUT«(sub infix:<eqv> (@a, @b) { #`(Sub|73035592) ... } sub infix:<eqv> ($a, $b) { #`(Sub|73035896) ... })␤»
jnthn m: say List.list 15:26
camelia rakudo-moar 3176cb: OUTPUT«((List))␤»
jnthn m: say Blob.list
camelia rakudo-moar 3176cb: OUTPUT«((Blob))␤»
jnthn m: say List.elems
camelia rakudo-moar 3176cb: OUTPUT«1␤»
jnthn m: say Blob.elems
camelia rakudo-moar 3176cb: OUTPUT«Invocant requires an instance of type Blob, but a type object was passed. Did you forget a .new?␤ in any at gen/moar/m-Metamodel.nqp line 1723␤ in block <unit> at /tmp/8FKsGUk7rZ line 1␤␤»
jnthn Aha
I bet it's that
Skarsnik: If you try your code with MoarVM HEAD now, hopefully your leak is gone 15:27
Skarsnik I just need to git pull in moar or I can use the configure to do it?
jnthn Skarsnik: git pull && make install in Moar would do it 15:28
I would think so, anyways
llfourn m: my @a = List; my @b := List; say @a.perl,@b.perl; 15:29
camelia rakudo-moar 3176cb: OUTPUT«[List,]List␤»
timotimo mhm mhm
nemo wow. Zoffix is still not back 15:30
yoleaux 25 Feb 2016 13:08Z <ZoffixW> nemo: I'm not deathly sick :) While still not fully well, the reason I'm not around is because I've just moved and (a) I don't have Internet at my new place yet and won't until next week. (b) I'm too busy with unpacking and settling in :)
nemo he must have been really really sick
ah 15:31
thanks ms bot
and he left me a message ☺
15:31 uruwi joined
Skarsnik hm, moar is in src/vm/moar or gen? 15:32
timotimo i'm now only missing him, but not worried any more :)
jnthn Skarsnik: You should have a MoarVM directory somewhere, maybe under nqp/ ?
Skarsnik hm, interesting stage parse is back a 60 , it was 100 this morning x) 15:33
jnthn urgh, headache...time for a break I guess... & 15:34
Skarsnik yep, fixed, jnthn++ 15:36
15:37 lizmat left
Skarsnik Afterlookfor :71.616 kb, Afterlookfor :71.616 kb , Afterlookfor :71.616 kb 15:37
.... instead of growing by 4Mb each time
timotimo sweet! 15:40
Ven pretty good :D.
timotimo yo ven
Ven yo timotimo !
15:41 virtualsue joined 15:42 idiosyncrat_ joined 15:45 zpmorgan joined
rudi_s Regarding the Blob eqv Blob issue. Should I report a bug (and if yes where)? 15:45
moritz rudi_s: yes, and by mailto:rakudobug@perl.org 15:46
15:48 virtualsue left 15:49 Ven left, ptolemarch joined
rudi_s moritz: [x] 15:50
Can I easily fake the current time (now()) to get reproducible test cases? 15:51
ugexe create a mock object. or maybe you can use monkey-typing/augment or wrap 15:55
15:55 khw joined
timotimo m: now() 15:55
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/fwyXAZoqRK␤Undeclared routine:␤ now used at line 1. Did you mean 'not', 'HOW'?␤␤»
timotimo m: now
camelia ( no output )
timotimo you can't wrap a term, i don't think
so that's out
rudi_s I guess I could use $*test-fake-time // now in the code but that's a little ugly IMHO. 15:56
How would you write it? 15:57
ugexe create a routine that creates the delta of the real now and the now you want?
rudi_s (Is there a convention in perl6 for "private" $* variables?)
I could use a method which I overwrite in my test. Good idea. 15:58
ugexe m: sub now { say 420; }; say now
camelia rakudo-moar 3176cb: OUTPUT«420␤True␤»
ugexe yeah looks like that'll work 15:59
ugexe wonders how feasible automatically creating NC wrappers via machine learning is 16:02
16:02 pierrot left
Skarsnik machine learning? 16:03
rudi_s ugexe: Doesn't work if the sub is in another module. 16:04
I guess I'll just use $*test-fake-time // now;
Is there a convention in perl6 for "private" $* variables?
Perl6 itself uses uppercase, so those are out.
DrForr rudi_s: I'd probably use $*_xxx but that's just me. 16:05
timotimo m: my $*MyModule::Test = 5
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/8qDx4wnOQG␤Dynamic variables cannot have package-like names, like $*MyModule::Test␤at /tmp/8qDx4wnOQG:1␤------> 3my $*MyModule::Test7⏏5 = 5␤»
timotimo :|
rudi_s DrForr: Sounds good, thanks. 16:06
perlpilot would do the same as DrForr
16:07 themonkeybob11 joined
perlpilot (assuming I could agree with the idea of "private" $*vars) 16:07
ugexe i would just shell out to change the system time to whatever i need
dalek kudo-star-daily: 5a44c8a | coke++ | log/ (9 files):
today (automated commit)
kudo-star-daily: f47e270 | coke++ | log/ (9 files):
today (automated commit)
rudi_s hopes ugexe is joking.
16:08 eiro left
ugexe its not a problem if you run each test in its own vm 16:09
arnsholt ugexe: I'm not sure I'd use ML for generating wrappers
ML has many useful attributes, but for this it's not ideal, I think
ugexe arnsholt: im still new to ML, but why? it would seem to have some fairly obvious patterns 16:10
jnthn Skarsnik: Yay \o/
16:12 hoelzro left, hoelzro joined 16:13 themonkeybob11 left, domidumont left
arnsholt ugexe: Primarily because ML is inherently fuzzy 16:14
16:14 olinkl_ joined, kipd_ joined 16:15 flussenc1 joined, kipd left, Lucas_One__ joined, flussence left, Lucas_One_ left, BinGOs_ joined, drforr1 left, olinkl left, integral left, timotimo left, BinGOs left, integral_ joined, integral_ left, integral_ joined, kipd_ is now known as kipd, Lucas_One__ left, Lucas_One__ joined, timotimo joined, integral_ is now known as integral, drforr1 joined 16:16 olinkl_ is now known as olinkl 16:18 themonkeybob11 joined
japhb \o/ # More perl6-bench benchmarks when I do my morning 'git pull' 16:19
yoleaux 29 Feb 2016 08:33Z <sortiz> japhb: The DBIish install time issues are solved and a PR opened in MoarVM for a definite fix. Thanks for your report.
16:19 pierrot joined
japhb .tell sortiz Great, thank you, I'll rebuild-all and see if DBIish install is fixed 16:19
yoleaux japhb: I'll pass your message to sortiz.
16:21 uruwi left
japhb Does anyone have a mirror/fork of flussence's Text-Tabs-Wrap (and Test-Corpus, needed to test it) 16:22
skids I have it offline, by virtue of perl6-all-modules. Probably stale though. 16:24
ugexe thi sis from 2016-1-30: hack.p6c.org:5001/authors/id/J/JD/J...2.5.tar.gz
skids Yeah mine is older 16:25
16:28 prammer joined 16:29 themonke1bob11 joined 16:30 themonke1bob11 left 16:33 prammer left
japhb ugexe: Does JDV ~~ flussence? ISTR there are a couple different repos out there with the same name, and I'm not sure if they were forks, or different people porting the Perl 5 modules, or what 16:38
ugexe everything on that metacpan mirror is under JDV
japhb oh
ugexe its not a finalized service, its just something we have been using to test integration 16:39
japhb ugexe: Can you make it do directory indexes?
16:40 jantore left 16:41 jantore joined
ugexe japhb: what do you mean 16:43
timotimo hack.p6c.org:5001/ <- why not use the interface? 16:44
16:44 vendethiel joined 16:45 vendethiel- left
Ulti if I want to make some of my own code friendly to the new multiline REPL is there anything I can do? 16:46
hoelzro Ulti: how do you mean? like a slang, or...?
Ulti yeah a slang with a quote construct
ugexe you can also search it by enabling cpan in zef and using `zef search XXX`
Ulti it at least has a nice error where its running my actions
before it didnt get that far so figure now's the time to update my stuff 16:47
I can take a look at the grammar for "" if that's where the magic lies
hoelzro if you have a custom quoting construct, it *should* pick up on that as long as failures are triggered via FAILGOAL, thanks to sortiz++
llfourn Slangs won't work in REPL if introduced the normal way 16:48
Ulti hoelzro okedoke I will take a look, thanks :)
llfourn if you use github.com/LLFourn/p6-CompUnit-Util#mixin_lang it will
Hotkeys Morning
llfourn or just copy what is done in the source
Hotkeys Still 10 minutes til noon so I get to say morning
Ulti will be really cool if I can get my slang to work since it will make perl6 into an interactive research tool without much effort
Hotkeys Interesting 16:49
llfourn Ulti: the reason it's not working is because modifications to %*LANG don't carry over to REPL/EVAL
you have to manually pass them to %?LANG for it to work
we obviously don't have a proper implementation of slangs yet so it's all a bit rough. 16:50
kalkin- Is there a way to create a class or module programmaticaly and then generate perl 6 code from that? Something like Class.new("Foo::Bar").to-source-code()
llfourn m: class Class { }; say Class.new.perl; 16:51
camelia rakudo-moar 3176cb: OUTPUT«Class.new␤»
timotimo github.com/timo/ADT - this neat little module of mine generates classes programmatically
llfourn kalkin-: like that?
16:52 torbjorn joined
kalkin- How can i provide the class name? 16:52
ohh
timotimo m: class Class { }; ::('Class').new.perl.say
camelia rakudo-moar 3176cb: OUTPUT«Class.new␤»
kalkin- m: class "Foo" {}; Foo.new.perl.say 16:53
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/bq1P3j8BBL␤Unable to parse class definition␤at /tmp/bq1P3j8BBL:1␤------> 3class7⏏5 "Foo" {}; Foo.new.perl.say␤ expecting any of:␤ generic role␤»
kalkin- m: class Class {}; ::('Foo').new.perl.say
camelia rakudo-moar 3176cb: OUTPUT«Failure.new(exception => X::AdHoc.new(payload => "Failed"), backtrace => Backtrace.new)␤»
Ulti llfourn well the modification is there since I'm seeing my slang error coming up in the repl
16:53 CIAvash left
Ulti thats the issue its doing a panic and crashing out during the multiline 16:54
llfourn m: Metamodel::ClassHOW.new_type(:name("Foo")).say
camelia rakudo-moar 3176cb: OUTPUT«Method 'say' not found for invocant of class 'Foo'␤ in block <unit> at /tmp/dctEC7OELp line 1␤␤»
llfourn m: Metamodel::ClassHOW.new_type(:name("Foo")).^name.say
camelia rakudo-moar 3176cb: OUTPUT«Foo␤»
llfourn kalkin-: something like that
Util_: hmm ok nvm me then. I'd be interested to see what your slang is like. 16:55
16:55 CIAvash joined
Ulti though cool you've made that easier 16:56
kalkin- m: Metamodel::ClassHOW.new_type(:name("Foo")).perl.say
camelia rakudo-moar 3176cb: OUTPUT«Method 'perl' not found for invocant of class 'Foo'␤ in block <unit> at /tmp/uQQIXAloGJ line 1␤␤»
Ulti I'll port over so I dont have to keep up with all the cool kids on how to inject my slang in
llfourn: github.com/MattOates/BioInfo/blob/...ioInfo.pm6
llfourn :P it's just lonely old me afaik
kalkin-: you might want to compose the type 16:57
m: Metamodel::ClassHOW.new_type(:name("Foo")).^compose.perl.say
camelia rakudo-moar 3176cb: OUTPUT«Foo␤»
llfourn takes a look
kalkin-: essentially you create the new type do stuff to it like .^add_method and then .^compose.
Ulti its really nice too I can paste in a mixture of bio sequence types and get back an array of typed objects automatically :) so with multiline it becomes actually useful in the REPL which was what I originally imagined its use case 16:58
llfourn kalkin-: jnthn.net/papers/2015-fosdem-static-dynamic.pdf # page 64 has a cool example
llfourn actually takes a look 16:59
16:59 kjs_ joined
llfourn Util_: awesome 17:00
kalkin- llfourn: thanks will have a look at it. need to go afk. 17:01
llfourn Util_: btw there is this "find-the-other-quote-not-sure-what-its-called" grammar thing like '[' ~ '] <stuff> # finds <stuff> in between the brackets 17:02
m: say '[foo]' ~~ /'[' ~ ']' 'foo'/ #maybe 17:03
camelia rakudo-moar 3176cb: OUTPUT«｢[foo]｣␤»
llfourn Util_: you might be able to use it for wrapping your quites
quotes*
ugexe i thought that bracket matching grammar thing was nyi 17:06
17:06 zakharyas joined
llfourn well it seems to be yi 17:07
jnthn Long implemented
Goat matching syntax
...
*Goal* matching syntax
llfourn m: say '[fo[bar]o]' ~~ /'[' ~ ']' 'foo'/ #does it do balanced text?
camelia rakudo-moar 3176cb: OUTPUT«Nil␤»
jnthn No, just looks for the end 17:08
llfourn m: say '[fo[bar]o]' ~~ /'[' ~ ']' 'fo[bar]o'/ #does it do balanced text?
camelia rakudo-moar 3176cb: OUTPUT«｢[fo[bar]o]｣␤»
jnthn You'd need to recurse to do structure
llfourn makes sense
jnthn It's used extensively in the Perl 6 grammar
m: say [1,2
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5=== Error while compiling /tmp/XRAefm3rSy␤Unable to parse expression in array composer; couldn't find final ']' ␤at /tmp/XRAefm3rSy:1␤------> 3say [1,27⏏5<EOL>␤ expecting any of:␤ statement end␤ statement m…»
Ulti llfourn yeah I have to admit I just picked the first thing in the grammar that looked like it did roughyl what I wanted... I'll take a look
ugexe ah i was thinking of a more general rule for arbitrary brackets. like %% but somehow knowing what the RHS bracket is based on the LHS bracket
jnthn That "couldn't find final ']'" is part of the goal-matching syntax. 17:09
llfourn yep I did discover it from looking at Perl6::Gramamr
17:10 [ptc] left
Hotkeys Could do it with captures? 17:10
llfourn Hotkeys: do what? 17:11
Hotkeys Arbitrary brackets
llfourn but you can do arbitrary brackets goat matching
#greatname
Hotkeys Oh 17:12
I don't really know anything about the goal matching syntax
I only just heard of it when it was mentioned here a few minutes ago
:p
llfourn well it's pretty handy 17:13
17:13 domidumont joined
ugexe but you have to declare both brackets! i dont have time for that! 17:14
really though thats good to know. i can clean up some code with that 17:15
17:15 andreoss left
Hotkeys Just checked the regex doc page and goal matching isn't there :( 17:16
ugexe wonders if it can do alternate sets of brackets
Hotkeys Or if it is it isn't there by name
llfourn it probs isn't
ugexe its mention in design s05 though 17:17
llfourn Hotkeys: did you search for goat?
ugexe goat matching goes well with monkey typing
17:18 uruwi joined 17:19 [ptc] joined
llfourn m: say 🐐🐐🐐🐐🐐🐐🐐🐐🐐 ~~ /'🐐' ~ '🐐' .+/ 17:21
camelia rakudo-moar 3176cb: OUTPUT«5===SORRY!5===␤Argument to "say" seems to be malformed␤at /tmp/d1J4ggH24V:1␤------> 3say7⏏5 🐐🐐🐐🐐🐐🐐🐐🐐🐐 ~~ /'🐐' ~ '🐐' .+/␤Bogus postfix␤at /tmp/d1J4ggH24V:1␤------> 3say 7⏏5🐐🐐🐐🐐🐐🐐…»
17:21 goodname joined
llfourn m: say '🐐🐐🐐🐐🐐🐐🐐🐐🐐 '~~ /'🐐' ~ '🐐' .+/ 17:22
camelia rakudo-moar 3176cb: OUTPUT«｢🐐🐐🐐🐐🐐🐐🐐🐐🐐｣␤»
17:23 zakharyas left 17:31 itaipu joined 17:39 virtualsue joined, Actualeyes left 17:40 SCHAAP137 left 17:57 addison joined 17:59 abraxxa left, prammer joined, FROGGS joined 18:02 themonkeybob11 left, prammer left 18:04 firstdayonthejob joined
TreyHarris Just trying to understand an architectural decision: why are, for instance, the several descendents of X:IO with the single attribute :path (or even all the ones having to do with directories) not descendents of a single common exception? Is the complication of a more complex type heirarchy (especially containing classes that will never be directly instantiable) judged to be a greater issue than boilerplate 18:07
code repetition?
Or did it just happen that way without any concious architectural "decisionmaking" process? 18:08
[Coke] Don't assume the exception hierarchy is well thought out.
right, that.
I'm sure a plan to clean that up would be considered for v6.d 18:09
TreyHarris Also, I've seen it both ways in older docs and in the synopses so I'm not sure if there's a consensus: do people talk of classes (and/or objects) having "attributes", or is that an internal thing that you shouldn't talk about outside of the class and you should refer to methods instead? 18:12
18:13 prammer joined
skids Classes are talked about as "having attributes", it's easier than explaining they have a private data slot and a (useually autoconstructed) accessor method. Moreover, ISTR jnthn working on making attribute accessors inlineable yesterday so they will be special in some sense, if virtually equivalent to that model. 18:15
18:15 virtualsue left
skids (Plus it makes sense to distinguish them in case someone cares what gets stored on serialization) 18:17
timotimo that code about making attribute accessors more inlinable is already in nom 18:18
18:20 prammer left 18:22 prammer joined, uruwi left 18:23 virtualsue joined, spider-mario joined
TreyHarris Liskov substitution is a problem, too; right now, if I do Klass.new(:attr('foo') :method('bar')), that doesn't work, and I can't do $obj.method = 42, but if attr is rw I can do $obj.attr = 42, so attributes are special 18:23
llfourn TreyHarris: you can make method is rw :) 18:24
18:24 dakkar left
[Coke] m: (^10e4).map({$_ => $_.uniname}).grep({.value~~/GOAT/}) # for someone in backscroll 18:27
camelia ( no output )
[Coke] m: say (^10e4).map({$_ => $_.uniname}).grep({.value~~/GOAT/}) # for someone in backscroll
camelia rakudo-moar 3176cb: OUTPUT«(65672 => LINEAR B IDEOGRAM B107F SHE-GOAT 65673 => LINEAR B IDEOGRAM B107M HE-GOAT)␤»
llfourn nice goat matching 18:29
jnthn m: say chr(65672)
camelia rakudo-moar 3176cb: OUTPUT«𐂈␤»
jnthn Aww
TreyHarris llfourn: that's what I thought, but gist.github.com/treyharris/3e813c150f86d993f07d isn't how it works... I don't see an example suggesting it would be used otherwise in the specs or docs but I may be blind (also, this isn't very useful obviously since you can't do anything but expose an lvalue by returning it) 18:30
18:30 wamba left, BinGOs_ is now known as BinGOs
timotimo you need to return-rw 18:30
TreyHarris oh, proxies... right. forgot about those.
timotimo alternatively no return at all
jnthn Hmm...it only sort of looks like a goat :)
18:32 BinGOs left, BinGOs joined
llfourn goes to bed 18:32
TreyHarris timotimo: but proxies are the only way to, say, internally store double the value assigned, right? and .new won't call methods regardless, or is there an adverb to make it do so?
18:32 wamba joined
timotimo you're right 18:34
18:35 aindilis left 18:36 aindilis joined
jnthn `has $.foo` means 3 things: 1) declare `has $!foo`, 2) set the meta-attribute's has_accessor property, and 3) at class composition time, if there isn't a method foo, generate a simple accessor method, paying attention to the `rw` trait 18:36
The things you can set in new pay attention to the property set in step 2
So you can write your own accessor method. 18:37
And it won't be a problem, 'cus we only generate such a method if you didn't write one yourself already.
timotimo a custom accessor method still won't allow every value assigned to it to land as "twice as much" in the attribute itself; except when building a Proxy for that 18:38
TreyHarris jnthn: right... so there's no way except for writing your own BUILD to have internal-state variables that are backends for attribute-like accessors including at construction time? Like, I want an Angle object to take either :degrees or :radians and respond to either .degrees or .radians in read or write?
jnthn TreyHarris: Yes, then you write your own BUILD
TreyHarris I see you can't multi on is rw, not that that would make sense unless is rw were special for multi dispatch 18:39
18:39 aindilis` joined
TreyHarris though it could have been, right? like there were an implicit is-rw boolean attached to every multi? 18:42
18:43 aindilis left
jnthn TreyHarris: For that use-case I'd probably make it something like class Angle { has $.radians; submethod BUILD(:$degrees, :$!radians = d2r($degrees)) { }; method degrees { d2r($!radians) }; sub d2r {... } } or so 18:43
TreyHarris: We don't really pass down context in Perl 6 in that way 18:44
You can differentiate on rw-ness if you're *receiving* something
TreyHarris jnthn: by any means other than a Proxy object? 18:45
jnthn TreyHarris: Well, I tend to regard using proxies a lot as a design smell.
18:45 aindilis` left
jnthn I'd somehwat hope an Angle object would be immutable, so I don't see a need for it in that case 18:46
18:46 kjs_ left
TreyHarris jnthn: agreed, but while I'm not a slave to the substitution principle, I'm a big fan of Liskov substitutability being easier than not 18:46
jnthn And I think more interesting objects are better designed in terms of methods expressing interesting behaviors.
18:47 renormalist joined, CIAvash left
jnthn TreyHarris: I don't see where you demonstrated it being hard. 18:48
18:48 virtualsue left
jnthn It's every bit as easy to factor out common code using roles, rather than turn to inheritance for that, though. 18:49
And easier the moment you want to compose two other bits of functionality.
So that gives a non-inheritance option for plenty of cases.
TreyHarris jnthn: right, but in general, changing some part of an object's internal representation from a storable one to a calculable one or vice versa should be possible without changing the code of consumers of the object. and it is, but not without proxies unless I'm missing something
unless you avoid from the start ever making the storable state available in "bare attribute" mode 18:50
renormalist If I want to re-write one of my modules into Perl6, are there conventions of how to name github repo and module name? Can it have the same name as its older Perl5 brother module? 18:51
TreyHarris renormalist: the github repo can't have the *same* name, no, not unless you define a master and a master-perl6 or somesuch, which would be weird. I've seen examples where the repo name is "Old-Name-p6" 18:52
18:52 idiosyncrat_ left
renormalist TreyHarris: sure, understood. And the module itself in Perl world? With having a later CPAN upload in mind? 18:53
TreyHarris (Git doesn't treat the name "master" specially in any way except for in git init; GitHub, OTOH, does treat "master" specially as the branch to which pull requests are applied.)
renormalist: there are modules in the ecosystem with exactly the same name, yes; i wouldn't choose a new name just to do so. I'd personally ask if you were porting someone else's CPAN module before using precisely the same name for your own Perl 6 version, but I'm not sure if others would agree. 18:54
jnthn TreyHarris: Yes, in those cases you'd probably look to Proxy. 18:55
18:55 prammer left 18:56 rurban1 left
TreyHarris jnthn: since I like using attribute access over getter/setter methods, that's where it's, if not "hard", demonstrably "harder", and does add a code smell--and I don't think having an attribute-access API is a code smell, but transitively it would seem to be equivalent to one 18:57
renormalist I want a simple own module first, and have it the prototype for myself before doing the complicated ones. So, let's say, Acme::Foo will then exist twice in CPAN, just separated by their PAUSE subdir Perl6/, ok, I will try.
TreyHarris: thanks
jnthn TreyHarris: I think attribute-access APIs have their place but are vastly overused in the real world (I'm thinking where a Foo type comes along with a FooService or FooManager) 18:58
TreyHarris jnthn: agreed, and I think immutable value types are their best use case. You can avoid a proxy in that case, as you pointed out above, since all assignment is done at object creation time. you still do have to have a BUILD submethod, which makes inheritance dicier than if there were a way to declare getter/setter methods as callable by Mu.new, but that doesn't seem like a big deal. 19:01
jnthn: consider me convinced pragmatically but I reserve the right to re-argue the point in a year or two with some real-world experience under my belt :) 19:02
jnthn :-) 19:04
Skarsnik when is the next moar push version in rakudo? so I can install the 'not leaking' moar/rakudo on my dedicated server x) 19:06
TreyHarris jnthn: at one point (in an Apocolypse?) I recall that value types were specifically supported, but in synopses it looks like they're only glancingly supported by .WHICH and their users. There's also in S09 a "compact struct" described, but the syntax isn't given. Does that actually exist yet? Either way, am I correct that that's all the language support now for value types? 19:08
19:09 spider-mario left 19:10 spider-mario joined, rurban joined, rurban left
jnthn TreyHarris: Yeah, the WHICH thing needs more attention. We didn't get the compact structs yet in general; we do have support for inlined structs to some degree for the nativecall stuff 19:10
Skarsnik: I'll probably do one later this evening 19:11
TreyHarris jnthn: Thanks. Yeah, searching perl6/docs--which I hadn't had cloned before--showed me that. Is simply defining a multi method WHICH (Klass:D:) { ObjAt.new(.gist) } sufficient for now and likely forward-compatability? (And if so, shouldn't there really be a Value subtype of Mu that just defines that since it's just an incantation that should generally work?) 19:16
19:17 uruwi joined
TreyHarris er, I guess that would be a role 19:17
jnthn TreyHarris: It'll work for now, though I think we'll extend it in the future 19:18
In terms of marking things, it'd probably want doing with a trait on the meta-object rather than something visible on the type itself.
But, to be decided :) 19:19
TreyHarris wait, how does that syntax work? I was just cribbing from rakudo, but doesn't the signature (Klass:D:) mean that it's the meta-object and not the object that's the invocant of .gist? 19:20
oh... no, ignore me. I'm stoopid. 19:21
I was confused by my own generalization, if I'd written "(MyClass:D:)" I wouldn't have been, but I'm used to $klass holding a metaobject. 19:22
19:26 kjs_ joined
Hotkeys Why does rakudo-git require the jdk 19:28
19:29 cdg left
timotimo Hotkeys: because you can build a jvm back-end, too 19:29
19:30 nowan left 19:33 labster joined
Skarsnik jnthn, this fixed leak is crazy, my january rakudo: "After parsing :85.164 kb 19:35
", with today rakudo and your patch "After parsing :71.620 kb" nearly 20% gain of memory just on the start
Hotkeys timotimo: oh right
I forgot that the jvm backend is a thing
timotimo Hotkeys: why it's a requirement and not a suggestion ... i dunno
Skarsnik these size are from /proc/pid/statm 19:36
jnthn Skarsnik: Crazy *good* yes? :) 19:37
19:37 azawawi joined
azawawi hi 19:37
yoleaux 27 Feb 2016 18:34Z <moritz> azawawi: cannot reproduce (re 403 for irclog.perlgeek.de/perl6/search); if the problem persists, please /msg me your public IP address
azawawi moritz: it is now working as expected... thanks :) 19:38
what's the best way to pass named parameters to a trait (i.e. ... is something(:p1(1), :p2(2)) ) 19:39
19:43 itaipu left, itaipu joined 19:48 pierrot left 19:49 kjs_ left 19:54 pierrot joined 20:00 labster left 20:03 prammer joined 20:06 uruwi_ joined 20:07 uruwi left, prammer left 20:08 espadrine left
jnthn azawawi: You only get one argument to a trait really. You can write what you did, but you'll receive a list of pairs 20:08
(As the second arg)
FROGGS m: multi trait_mod:<is>(Routine $r, :$awesome!) { say $awesome }; sub foo is awesome(:yeah, :geez) { } # azawawi 20:09
camelia rakudo-moar 3176cb: OUTPUT«(yeah => True geez => True)␤»
20:10 darutoko left
FROGGS m: multi trait_mod:<is>(Routine $r, :$awesome!) { say (% = |$awesome) }; sub foo is awesome(:yeah, :geez) { } # azawawi 20:10
camelia rakudo-moar 3176cb: OUTPUT«geez => True, yeah => True␤»
azawawi jnthn: thanks 20:12
FROGGS++:) 20:13
20:13 sortiz joined
sortiz \o #perl6 20:14
yoleaux 16:19Z <japhb> sortiz: Great, thank you, I'll rebuild-all and see if DBIish install is fixed
20:15 prammer joined, sno left 20:19 prammer left, themonkeybob11 joined, prammer joined 20:20 domidumont left 20:21 QORRiE joined, prammer left 20:22 prammer joined 20:24 kjs_ joined 20:26 virtualsue joined 20:27 prammer left 20:29 prammer joined
azawawi FROGGS: ping 20:30
20:30 labster joined
azawawi FROGGS: github.com/azawawi/perl6-memoize/b...load.t#L36 # tests 20:31
FROGGS: github.com/azawawi/perl6-memoize/b...ize.pm6#L7 # trait implementation
FROGGS hmmm 20:32
is this still problematic in Perl 6?
github.com/azawawi/perl6-memoize/b...ize.pm6#L9
jnthn: ^^
vendethiel huh, that was a pretty sad blogpost to read.
FROGGS I know that this caused a lot of bug hunting work for me with Perl 5...
m: sub foo($bar) { my $baz = 42 if $bar; say $baz }; foo 0; foo 1; foo 0 20:34
camelia rakudo-moar 3176cb: OUTPUT«(Any)␤42␤(Any)␤»
FROGGS okay, seems to work better than in Perl 5
mst FROGGS: is what problematic?
vendethiel FROGGS: but hunting with Perl 5? :o
FROGGS m: sub foo($bar) { my $baz = 42 if $bar; say $baz }; foo 1; foo 0
camelia rakudo-moar 3176cb: OUTPUT«42␤(Any)␤»
FROGGS vendethiel: bug* :P 20:35
vendethiel FROGGS: bug hunting with Perl 5? :o
FROGGS mst: a statement modifying condition on a declaration
20:35 prammer left
FROGGS mst: it acts like a state variable 20:35
mst ah 20:36
jnthn FROGGS: I don't think it acts like state accidentaly in Perl 6, no :)
FROGGS which was not what I intended
jnthn At least, I sure hope not :)
FROGGS jnthn: yes, which is a relief
jnthn :) 20:37
azawawi FROGGS: anyway, that's 'is memoized' with simple LRU cache eviction :)
FROGGS just had one of these in my middleware service (Perl 5) a week ago
azawawi: it reads nicely I think :o)
20:43 Ven joined
azawawi so what is a better name than 'is memoized' other than 'is cached' in your opinion? 20:44
Skarsnik is azawawied 20:45
perlpilot "memoized" is an excellent name IMHO 20:46
Skarsnik guess_library_name in NC could be cached I think x)
azawawi Skarsnik: :) 20:47
perlpilot azawawi: it would be interesting if you could pass the thing that handles the caching as a parameter, but perhaps only as an academic exercise.
20:48 prammer joined
azawawi perlpilot: cool 20:48
20:48 Ven left
azawawi #TODO cache eviction should be done if needed in another thread every N seconds :) 20:49
is unforgettable :) 20:50
dalek p: d8dab2f | jnthn++ | tools/build/MOAR_REVISION:
Bump for MoarVM memory fix.
kudo/nom: 855de77 | jnthn++ | tools/build/NQP_REVISION:
Bump for MoarVM memory fix.
 20:52
jnthn Skarsnik: There's the bump
Skarsnik: Didn't get the other thing I was working on done :)
20:52 Ven joined
Skarsnik Well you are forgiven! It weird nobody get bothered by it before 20:53
20:54 QORRiE left 20:58 SCHAAP137 joined
jnthn :) 21:00
21:02 tmch left 21:03 sufrostico left
sortiz azawawi, IMO any trait of this kind should check that the Routine 'is pure' also, for completeness. 21:03
jnthn Well, arguably if you're sticking "is memoized" on it then you're stating that it makes sense 21:04
is pure doesn't actually check anything
It just means "if you constant-fold me at compile time, then it's fine", and other such things :)
sortiz I know that 'is pure' right now don't have much semantic, but can be an opportunity to give them some, like "I'm a candidate to memoization", or can be the other side: if "is memoized", imply that "is pure" :) 21:07
timotimo is pure on the optimizer does a tiny bit of stuff 21:08
er, i mean, the optimizer inspects subs for pureness in some cases 21:09
21:10 molaf left
sortiz And in time that can be extended, I suppose. So that can promote, one way or the other, its use. 21:10
azawawi sortiz: is memoized/cached were created to get faster execution time. That's their main objective :) 21:11
timotimo i think for optimize-time-evaluation 21:12
21:13 eiro joined
sortiz azawawi, I understand, my comment is about semantic closure, ie the 'is memoize' trait imply 'is pure', that can be asserted or auto applied, that's my point. 21:14
azawawi sortiz: PRs are more than welcome :) 21:16
21:17 FROGGS left, TEttinger joined
dalek kudo-star-daily: 750fe88 | coke++ | log/ (9 files):
today (automated commit)
 21:19
21:20 tmch joined 21:24 addison left
azawawi good night everyone and thanks :) 21:25
21:27 azawawi left
jnthn Time for me to rest also :) Enjoy the speed-ups/leak fixes :) o/ 21:29
sortiz o/ jnthn 21:30
21:34 virtualsue left, sno joined
[Coke] (rakudo-git requires jvm) only if you configure it to build the jvm. 21:36
*on the
which you might have done accidentally in the Configure step
21:43 kjs_ left 21:48 itaipu left, kaare_ left, cognominal left 21:49 rindolf left 21:51 tmch left 21:54 skids left 21:55 bjz left 21:56 prammer left 22:00 prammer joined 22:02 Skarsnik left 22:08 Ven left, itaipu joined 22:13 itaipu left 22:14 itaipu joined 22:16 ptolemarch left, bjz joined
rudi_s Hi. How can I "cast" an argument. I have for example (1,2,3) and I want to pass it to a function which takes a Int @foo. 22:21
22:22 bjz left 22:24 bjz joined
timotimo assign to the right kind of container 22:26
blah_func(my Int @ = (1, 2, 3)) 22:27
22:31 vendethiel left 22:34 bjz left 22:35 pdcawley_ left 22:39 vendethiel joined
timotimo www.youtube.com/watch?v=IvUU8joBb1Q - what a beautiful contraption 22:41
22:42 tardisx joined
rudi_s timotimo: Thank you. 22:42
22:43 pdcawley joined 22:44 AlexDaniel left 22:56 virtualsue joined 23:00 vendethiel left 23:02 vendethiel joined 23:03 wamba left 23:06 itaipu left 23:07 FreezerburnV joined 23:14 itaipu joined
lucs timotimo: Have you seen this one?: youtu.be/lHdcjW3bDTk 23:16
23:16 musca left, musca joined
timotimo yeah, that one's not real, though 23:17
23:18 spider-mario left
lucs Aw, you knew ;) 23:18
timotimo a year ago my dad sent it to me and i was legit wondering if he knew it was computer animation 23:19
lucs Yeah, even though it is, it's pretty neat.
timotimo yeah 23:20
i appreciate the actually-built one a bit more, though. it has the catchier track playing on it
lucs You probably know Stephen Malinowski's Music Animation Machine videos? 23:22
23:23 vendethiel left
timotimo not sure 23:25
lucs Example: youtu.be/yojDu3E9jls 23:26
He came up with a lot of different visualization techniques.
timotimo pretty! 23:27
23:30 geraud left 23:31 themonkeybob11 left 23:36 themonkeybob11 joined 23:41 virtualsue left 23:42 khw left, prammer left 23:46 themonkeybob11 left 23:50 vendethiel joined 23:53 themonkeybob11 joined 23:54 SCHAAP137 left, virtualsue joined 23:58 RabidGravy left 23:59 themonkeybob11 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!