»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.
00:08 skids joined
RabidGravy eek 00:09
nighty night good people
00:14 vendethiel left 00:16 RabidGravy left 00:17 patrickz left, espadrine left, espadrine_ joined, espadrine_ is now known as espadrine 00:19 vendethiel joined 00:23 llfourn joined 00:24 lucasb left 00:27 llfourn left 00:37 pjscott joined, pjscott left 00:40 spider-mario left 00:43 pjscott joined 00:47 ugexe left, bitmap left 00:49 pjscott left 00:51 bitmap joined 00:53 ugexe joined
Rotwang could someone tell me what's going on? : 00:58
m: my %c = (a => 1); say 'a' ~~ %c;
camelia rakudo-moar b8a75b: OUTPUT«True␤»
Rotwang m: my %c = (a => 1); my %f = (b => [%c]); 'a' ~~ %f<b>[0]; 00:59
camelia rakudo-moar b8a75b: OUTPUT«Method 'a' not found for invocant of class 'Str'␤ in block <unit> at /tmp/J9m9pqA80n line 1␤␤»
01:02 yurivish_ joined
ugexe m: say $*CWD ~~ :e 01:05
camelia rakudo-moar b8a75b: OUTPUT«True␤»
01:05 kid51 joined
ugexe m: say $*CWD.e; say %(:e(1)) 01:06
camelia rakudo-moar b8a75b: OUTPUT«True␤e => 1␤»
Rotwang : my %c = (a => 1); my %f = (b => [%c,]); 'a' ~~ %f<b>[0]; 01:08
m: my %c = (a => 1); my %f = (b => [%c,]); 'a' ~~ %f<b>[0];
camelia ( no output )
Rotwang m: my %c = (a => 1); my %f = (b => [%c,]); say 'a' ~~ %f<b>[0];
camelia rakudo-moar b8a75b: OUTPUT«True␤»
Rotwang so if I put hash directly into array it gets "flattened" to the container? or it least it seems like it 01:09
01:10 lnrdo joined
Mouq m: my %c = (a => 1); my %f = (b => [%c,]); dd %f 01:10
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b($[{:a(1)},])}␤»
Mouq m: my %c = (a => 1); my %f = (b => [%c]); dd %f
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b($[:a(1)])}␤»
Mouq m: my %c = (a => 1); my %f = (b => [$%c]); dd %f
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b($[{:a(1)},])}␤»
skids Rotwang: single argument rule docs.perl6.org/language/functions#S...onventions 01:14
01:15 lnrdo left
Rotwang x__X 01:15
perl6 is the best thing that happened to mankind since sliced bread
(with a few notable exceptions of course)
however I doubt it will go MAINstream, mostly because of it's "quirkiness" 01:16
skids Certainly we've seen languages exist in niches for long periods of time. IMO its merits will make it a good teaching language and it will slowly gain market share as a result. 01:18
01:18 espadrine left
skids Back when I was in school, even though it was rarely used language, some of the CS professors preferred to teach advanced compilers/language in ada, because they could show examples of all the constructs they wanted to teach in that language. 01:19
(Not that ADA gained market share, but thats because it is no fun. :-) 01:20
01:21 user74747 left
BenGoldberg m: my %c = (a => 1); my %f = (b => [|%c]); dd %f 01:23
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b($[:a(1)])}␤»
BenGoldberg m: my %c = (a => 1); my %f = (b => |%c); dd %f
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b(slip$(:a(1),))}␤»
BenGoldberg m: my %c = (a => 1); my %f = (b => %c).flat; dd %f
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b(${:a(1)})}␤»
BenGoldberg m: my %c = (a => 1); my %f = (b => %c.flat); dd %f
01:23 troydm joined
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b((:a(1),).Seq)}␤» 01:23
01:24 llfourn joined
BenGoldberg m: my %c = a => 1; my %f = (b => %c.flat); dd %f 01:24
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b((:a(1),).Seq)}␤»
01:24 average left
BenGoldberg m: my %f = (:b(:a(1)); dd %f 01:24
camelia rakudo-moar b8a75b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/InJolmRaEs␤Cannot use variable %f in declaration to initialize itself␤at /tmp/InJolmRaEs:1␤------> 3my %f = (:b(:a(1)); dd %7⏏5f␤ expecting any of:␤ argument list␤ term␤»
BenGoldberg m: my %f = (:b(:a(1))); dd %f; 01:25
camelia rakudo-moar b8a75b: OUTPUT«Hash %f = {:b(:a(1))}␤»
BenGoldberg m: my %f = (:b(:a(1))); dd %f{b};
camelia rakudo-moar b8a75b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/AIJ10dJ8ic␤Undeclared routine:␤ b used at line 1␤␤»
BenGoldberg m: my %f = (:b(:a(1))); dd %f<b>;
camelia rakudo-moar b8a75b: OUTPUT«Pair %f = :a(1)␤»
01:27 joydon left 01:28 llfourn left, pjscott joined 01:29 wamba joined 01:31 pjscott left 01:35 mopp joined 01:39 mopp left 01:41 mopp joined
mopp Perl 6. 01:42
hahainternet indeed
is $foo[*-1] really acceptable syntax? 01:43
it seems rather odd
$foo[$foo.end] works but also a little ugly
wonder if it topicalises
m: my $foo = (2,4,8); say $foo[.end] 01:44
camelia rakudo-moar b8a75b: OUTPUT«2␤»
hahainternet this is not what i expected!
Mouq m: my $foo = (2,4,8); say $foo[*.end]
camelia rakudo-moar b8a75b: OUTPUT«2␤»
hahainternet Mouq: yeah i found that works, same as *-1
but that .end works is bizarre
Mouq m: my $foo = (2,4,8); say $foo
camelia rakudo-moar b8a75b: OUTPUT«(2 4 8)␤»
hahainternet it's because the empty string or similar has a .end method also 01:45
oh i just noticed yours doesn't work lol
Mouq Oh, oh. Right. *.end is being executed with an argument of the number 3
hahainternet it's a bit odd really 01:46
i like the ability to easily index from the end, but *-1 will have to do
Mouq m: my $foo = (2,4,8); say $foo[{$_ - 1}] # :P 01:47
camelia rakudo-moar b8a75b: OUTPUT«8␤»
hahainternet indeed!
skids RT 127024 and RT 127129 can be merged if someone who can do that action is around 01:49
Mouq skids: sure, I should be able to :) 01:50
skids Mouq++ 01:52
01:59 mopp left
Mouq Does anyone happen to know what DESCEND used to do in ~2010 Rakudo? I imagine it's something like "nextsame" 02:01
02:04 yurivish_ left
sprocket if i have a variable that i’m using in a react { whenver { … }; whenever { … } } situation, is it threadsafe by default? 02:04
02:08 pjscott joined 02:11 TimToady left 02:13 TimToady joined 02:15 wamba left 02:18 llfourn joined 02:20 pjscott left 02:21 pjscott joined 02:23 llfourn left 02:27 zwu joined
zwu how to delete a variable in a scope? 02:27
leont What do you mean? 02:29
zwu like del in python to delete attribute of object: del foo.bar 02:31
tadzik why would you ever want to do that 02:32
02:32 MadcapJake left
tadzik this is basically breaking the interface of your object 02:32
making your code unreliable and unpredictable in the general case 02:33
skids As far as variables in a scope, I do not believe there is a way to undeclare them. 02:35
02:35 MadcapJake joined
zwu in order to release memory by GC, but I think it is same to assign Nil to a variable instead of delete it. 02:37
02:37 pjscott left
skids The only reason I could see actually wanting to do so is to get syntax errors if someone uses it later on, or allow redeclaration later on. 02:37
Mouq zwu: Python can do that because variables are checked at run time. Such a thing doesn't make as much sense in Perl 6 02:38
Since variables are checked at compile time 02:39
skids I don't think the idea of going back through previous callframe lexpads to remove unused variables is especially efficient, but if it is, then it's something the compiler should be doing without the user needing to say so. 02:40
zwu thanks,
02:40 cdg left 02:41 cdg joined 02:45 cdg left 02:47 bitmap left, ugexex joined, molaf__ joined 02:49 ugexe left, pjscott joined
mort96 hey 02:51
02:51 molaf_ left
skids o/ 02:51
mort96 so, I've installed a module with panda. How do I include it with a script? 02:52
pjscott use Foo;
mort96 ah awesome, thanks
02:55 vendethiel left 02:58 cdg joined 02:59 yqt left, cdg left 03:00 xyf joined
dalek kudo-star-daily: ce6a800 | coke++ | log/ (8 files):
today (automated commit)
 03:02
03:02 sno left
xyf hello, how can I read content in a gb2312 format file? below command doesn't work: perl6 -e'.encode("gb2312").say for "gb2312format.txt".IO.lines' 03:06
Unknown string encoding: 'gb2312' 03:07
in block <unit> at -e line 1
03:07 llfourn joined 03:08 noganex joined 03:11 noganex_ left
ugexex utf8, utf16, utf32, ascii, iso-8859-1, and windows-1252 are the available encodings 03:13
xyf faint..... 03:14
means gbk, gb18030 etc is not available in perl6 yet?
skids Not yet. 03:15
leont Patches welcome! ;-) 03:18
xyf ;) 03:19
03:20 cygx joined
cygx note that utf32 isn't implemented as well 03:20
m: "hallo".encode("utf32")
camelia rakudo-moar b8a75b: OUTPUT«Unknown string encoding: 'utf32'␤ in block <unit> at /tmp/UdD7lvtYHX line 1␤␤»
leont No one uses that for interexchange anyway
03:21 xxpor_lap joined
leont In p5, Encode.pm uses abstract descriptions to cover most of the code, though varying encodings still need substantial support code. Such an approach may be sensible here too 03:21
mort96 so, I'm noticing that a script which does nothing but include HTTP::Client and print "hello world" takes about 1.3 seconds - that's kind of a long time.
Is there any way to make that faster, maybe by precompiling things to bytecode or something? 03:22
cygx mort96: also on subsequent runs, or only the first one?
mort96 also on subsequent runs
`time` claims it's using 99% CPU during that time too 03:23
cygx leont: if you need something right now and don't feel like patching moarvm, there's of course always &chrs ;)
mort96 in node.js, a=require("http"); console.log("hello world"); takes 0.06 seconds by comparison
leont Start-up performance isn't well optimized yet 03:24
cygx mort96: what's your rakudo version?
leont (well, performance in general isn't our strong side at the moment, people are working on it)
mort96 "This is perl6 version 2015.11 built on MoarVM version 2015.11"
from perl6 --version
sprocket the “NativeCall: Consider adding the api version of the library you want to use” message on STDERR is new - is there any way to quash it?
aside from adding a version to the NativeCall bindings
Hotkeys should get 2015.12 aka 6.c
leont Upgrading might help, yes
mort96 alright, will do that 03:25
sprocket which i didn’t see in the documentation
mort96 it's not a dealbreaker in any case
03:26 AlexDaniel joined
ugexex those encoding have various descriptors that can be used 03:26
AlexDaniel m: my $foo = (2,4,8); say $foo[{.pred}]
camelia rakudo-moar b8a75b: OUTPUT«8␤»
AlexDaniel m: my $foo = (2,4,8); say $foo[*.pred]
camelia rakudo-moar b8a75b: OUTPUT«8␤»
AlexDaniel hahainternet: ↑
skids sprocket: That is new and a bit annoying. To squash it I used e.g. is native('crypto', v0 but False)
That will disable all version checks, but is a it of a sleazy way of doing so. 03:27
AlexDaniel I can't see what's wrong with *-1 though :)
sprocket skids: is there any way of saying “version 1 or greater?”
or something to that effect?
skids sprocket: not that I could tell.
sprocket or is it meant to be a fixed version number ie) 1.4.3 03:28
skids I imagine a way to do it will apear eventually.
sprocket ok
AlexDaniel 1.0.0..* ?
AlexDaniel feels like he has seen that somewhere 03:29
03:30 psy_ left, orbus joined
skids AlexDaniel: When I looked at the code it compared the major version number to the filename or something. 03:31
03:32 psy_ joined
Mouq v1+ 03:33
leont cygx: actually, some way to add pure-perl encodings for these cases may actually be a good idea. Sometimes supporting something at all is more important than having a fast implementation.
cygx agrees with skids 03:34
the code expects a path, or a name + fixed version number
ugexex what makes locks non-composable (or bad form)? jnthn would mention that but i never found out why 03:35
03:36 nzkindest left, yurivis__ joined
skids Two locks opened in opposite order can deadlock, and it's easy to do that inside a highly composed codebase. 03:36
But, the recent concurrent-only Promise changes has its own deadlock hazards, if you have any feedback loops. 03:37
sprocket on the subject of locks and whatnot, in a react { whenever { … } } structure, are any variables used in there considered to be threadsafe?
ugexex skids: what do you think this could be problematic? github.com/rakudo/rakudo/blob/nom/...ally.pm#L2
03:38 pjscott left
skids Well, it's not actually used to .protect so.... ? 03:38
03:40 kid51 left
mort96 so, say I want to make an asynchronous HTTP request, letting me do a bunch of things while the request is going, but have some callback or something get called when it's done. How would I do that in perl 6? 03:41
skids ugexex: Basically if there's no elegant way to use a Channel or Promise or Supply to acheive what you am at, I would not worry abot using Lock.
yurivis__ Why does my $x = 1, 2, 3; make $x an Int while my @x = 1, 2, 3 makes $x an array, and typing 1, 2, 3 at the REPL prints something that looks like a list?
mort96 I could do it with threads presumably, but there seem to be nicer features for asynch things in perl 03:42
sprocket mort96: couldn’t you wrap that in a supply { … } block?
orbus mort96: I'm mostly just getting started in perl6 but you can string promises together so they fire in sequence - that might do what you're looking for?
mort96 I'm not sure what a supply {...} block is 03:43
ugexex you cant use data from an IO::Socket outside its originating thread
skids yurivis: Array is the default container-type of @-sigiled variables. A literal (1,2,3) is indeed a list. If you want to put the list in a @a, bind with := instead of assigning.
sprocket mort96: perl6advent.wordpress.com/2015/12/...nous-data/
mort96 I currently have HTTP::Client, which seems nice except that it's blocking
03:44 pjscott joined
ugexex thats a limitation of libuv and scheduler-per-thread model 03:44
yurivis__ skids: is := (or are lists) common in perl code? it seems like a feature you'd use in rare cases rather than regularly 03:45
orbus doc.perl6.org/type/Promise#method_then
ugexex you could save to a file, close the file, notify another thread when its done, and then use it
sprocket mort96: there’s also IO::Socket::Async, but that’s not an http client
yurivis__ skids: and I would expect my $x = 1, 2, 3 to do something other than what it does; currently it assigns 1 to $x and prints out (1 2 3) at the repl
skids yurivis: lists usually do not end up bound to @ variables in common idioms except when passed to @ routine parameters 03:46
ugexex you cant use openssl with io::socket::http yet either. i wrote a compatability layer but its too slow
i mean io::socket::async
mort96 orbus: if I have blocking code in a promise, will that at all allow me to run other things simultaneously as if it's in a different thread?
leont openssl is a hard to use mess :-( 03:47
orbus as far as I understand, yes
ugexex again, you cannot read data from an io::socket from 1 thread to another
skids The difference between the $ = and @ = comes down to the operand and MMD -- there are two conflicting groups that want different behaviors from "$a = 1,2,3" -- the way it works is more pleasant for people comig from C.
orbus promises get scheduled on different threads
mort96 interesting 03:48
cygx yurivis__: I think I mainly use := when interacting with nqp code
if possible, I rather go with sigilless variables instead
yurivis__ there are sigilless variables? like, my x = 1, 2, 3?
mort96 I'm used to javascript, where blocking code in a promise would block everything
yurivis__ m: my x = 1, 2, 3;
camelia rakudo-moar b8a75b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/0C9OWgDYiK␤Malformed my (did you mean to declare a sigilless \x or $x?)␤at /tmp/0C9OWgDYiK:1␤------> 3my x7⏏5 = 1, 2, 3;␤»
cygx m: my \x = 42; say x
camelia rakudo-moar b8a75b: OUTPUT«42␤»
yurivis__ m my \x = 1, 2, 3; dd x;
skids yurivis__: yes but they are SSA unless you tweak them with a bind ahead of time 03:49
yurivis__ m: my \x = 1, 2, 3; dd x;
camelia rakudo-moar b8a75b: OUTPUT«(1, 2, 3)␤»
skids yurivis__: see the last example in doc.perl6.org/type/Scalar 03:50
yurivis__ cygx: interesting! thanks! I'm going to go read about sigilless variables
orbus m: my $p=start {sleep 2; put 1}; put 2; $p.result 03:51
camelia rakudo-moar b8a75b: OUTPUT«2␤1␤»
orbus simple example, but execution continues in the main thread while the promise thread is sleeping 03:52
something that actually blocked would be a better example I guess 03:53
yurivis__ skids: thanks for the link; I understand a little better now. can you say more about what it is about "$a = 1, 2, 3" assigning $a to 1 that is pleasant for people coming from C?
ugexex it should be noted start { } doesnt always mean a new thread. it just schedules work
geekosaur not so much pleasant as expected behavior, I think
orbus true
ugexex so when you start trying to do http requests in a start { } and it works 20% of the time you will know why :) 03:54
cygx m: my ($a, @b) = 1,2,3; say $a; say @b
camelia rakudo-moar b8a75b: OUTPUT«1␤[2 3]␤»
yurivis__ It might be good practice to use explicit delimiters around lists; I was omitting them because it seemed like it would do the right thing most of the time
geekosaur (and people do get , wrong in C all the time)
orbus I was looking at that and I think the default thread pool in rakudo is 16 threads so if you start trying to ramp up more things than that it's not going to be able to schedule them all at once
but then you probably don't have that many cores anyway
yurivis__ oh, right, that's how you do multiple assignment. I kept trying to my $a = 1, $b = 2
orbus probably 03:55
ugexex you can have just 2 things and still get scheduled on a single thread
orbus hrm
yurivis__ (i guess it's "destructuring", rather than "multiple assignment" this way)
orbus well I'm hardly an expert so I'll defer to people who've studied it closer
you can also use raw threads of course, but that's discouraged 03:56
03:56 pjscott left 03:57 pjscott joined
skids m: my ($a, $b) = 1, 2; $a.say; $b.say; my ($c, $d); $c = 3, $d = 4; $c.say; $d.say; # The first form is special sugar for "my" <-- yurivis__ 03:57
camelia rakudo-moar b8a75b: OUTPUT«1␤2␤3␤4␤»
skids m: my $c = 3, my $d = 4; $c.say; $d.say; # also works 03:58
camelia rakudo-moar b8a75b: OUTPUT«3␤4␤»
yurivis__ right
mort96 so, I ran `perl6 bootstrap.pl` to install panda. Now, panda results in: Failed to open file /home/<myname>/.perl6/2015.12-74-gb8a75b9/precomp/495015BBA080C4425F24DE5DEF172CEEFE97A0B8.1451792201.46975/40/40DC78A765A19FD24C2EF4B275381AA8F12059CB.deps: no such file or directory
that's nice of it
orbus that's interesting 03:59
04:00 leont left
orbus presumably it didn't install all the way 04:00
did the bootstrap fail in the middle?
ugexex it just fails like that for some reason. i get it randomly as well 04:01
mort96 no, all tests passed
gfldex mort96: what OS?
mort96 Linux
Linux vor 4.2.5-1-ARCH #1 SMP PREEMPT Tue Oct 27 08:13:28 CET 2015 x86_64 GNU/Linux
04:01 pjscott left
ugexex time to nuke your install 04:02
orbus I dunno - it seems like precomp stuff should regenerate if not found?
04:02 pjscott joined
orbus says this knowing very little about the implementation 04:03
gfldex mort96: that's a very recent rakudo. I'm rebuilding right now and will get bck to you if panda works on my end.
ugexex i just submitted a PR to fix a different precomp error where it failed regenerating file structure
mort96 alright
orbus I've been using 2015.12 since it came out and I don't think I've seen that one 04:04
cygx the versioning scheme for percompiled code has changed, which requires nuking your install directory
orbus and I've bootstrapped panda
cygx *precompiled
orbus changed when?
cygx before christmas
04:04 cdg joined
orbus oh, k - all the stuff I've done with panda has been on 2015.12 and post-christmas 04:05
before that I had whatever the latest star was
mort96 btw, it is kind of annoying that everything wants to change my PATH all the time, instead of just letting me put a few binaries ind /usr/bin and have libraries and such in /usr/share or something
orbus you could force it to install in /usr/bin - it's just going to confuse your package manager 04:06
unless you feel like building your own packages
mort96 maybe, whenever package managers start providing perl 6
ugexex rakudo's `Distribution` object messes up `depends` field on install. it saves it like `[ [] ]` for empty, and seemingly randomly `[ [xxx] ]` with a depends (instead of `[ xxx ]`) 04:07
mort96 how can I force it to use /usr/bin?
orbus --prefix when you're building
probably --prefix=/usr
AlexDaniel .seen jnthn
yoleaux I saw jnthn 25 Dec 2015 23:52Z in #perl6: <jnthn> Thanks! Really going, before I fall asleep on the keyboard :-)
orbus you'd have to do moar that way first, then nqp, then rakudo 04:08
so like perl ./Configure --prefix=/usr
mort96 and panda?
orbus I don't know about panda
hrm 04:09
cygx AlexDaniel: jnthn is expected back in about a wekk or so
orbus I tried to force panda to install in a different directory
but it then it got confused about where its libraries were
ugexex installing will install to the perl6 directory (in addition to the local .perl6 directory for some reaosn)
04:09 cdg left
AlexDaniel cygx: I'm glad that he finally took a break. His dedication was unreal 04:10
orbus I don't think so - I've been doing --prefix=/home/<myname>/perl6/christmas/install
and it comes out with bin, share, lib etc. under there
so I don't see any reason you couldn't tell it /usr and get the same
but panda's kind of weird
gfldex mort96: i'm using (to keep things tidy) --prefix=/home/gfldex/local and that works
ugexex you can look at the code, CompUnit::Repo::Installation uses $.prefix
orbus I got panda to install in a custom location but then when I tried to run it it was confused by where it was 04:11
I forget what I did now
I think in the end I settled on symlinking it to the same place perl6 is installed 04:12
it wanted to live off in some share directory
ugexex because you still need a compunit::repository that reads from your custom location 04:13
orbus but there's probably a way because in star releases panda is in the same directory with perl6
ugexex loads rather
orbus the symlink covered my bases, so I'm okay with how it is now
but that's probably the *right* way 04:14
AlexDaniel interestingly, 2015.11 is now in debian unstable
ugexex you could use -I/custom/path (maybe PERL6LIB=/custom/path) to load them 04:15
TimToady 2015.11 is "perl6 unstable"
orbus and 2015.11 is the latest star - they probably used that
cygx orbus: just ask mort96 what can happen if you settle on symlinking ;)
AlexDaniel haha. Well, it does not mean that debian unstable has to provide unstable versions :) 04:16
TimToady and anyone who installs 2015.11 risks having to nuke their install dirs
mort96 lol
orbus mort96: what happens if I settle on symlinking?
mort96 well, if you use rakudobrew, and symlink it to /usr/bin ...
it used to delete all files in the directory it was in.
orbus oh, I don't know anything about rakudobrew 04:17
I just compiled from the tarballs
llfourn m: Bool.Int.^name.say # regression? 04:18
camelia rakudo-moar b8a75b: OUTPUT«Bool␤»
llfourn m: True.Int.^name.say # regression?
camelia rakudo-moar b8a75b: OUTPUT«Bool␤»
cygx m: say Bool ~~ Int 04:19
camelia rakudo-moar b8a75b: OUTPUT«True␤»
llfourn is that a new thing?
cygx feature, not bug
llfourn hmm ok :)
cygx relatively new
04:19 yurivis__ left
llfourn Just sent some old code into infinite loop :D 04:19
orbus currently I have perl6 in /home/orbus/perl6/christmas/install/bin 04:20
so I made a symlink for panda
/home/orbus/perl6/christmas/install/bin/panda -> /home/orbus/perl6/christmas/install/share/perl6/site/bin/panda
because that's where panda seemed to want to be
but it's okay with me executing it from /home/orbus/perl6/christmas/install/bin
skids Things work well enough using rakudobrew with --configure-opts='--prefix=/home/foo/.local' to install both rakudo and panda. Though I do have some local tweaks to rakudobrew. 04:21
llfourn m: True.Int.Str.say # so there is no way to have True => 1, False => 0 magically anymore? 04:22
camelia rakudo-moar b8a75b: OUTPUT«True␤»
ugexex maybe subset? 04:23
04:23 dalee joined
llfourn m: say 0 + True # works, reminds me of perl5 04:24
camelia rakudo-moar b8a75b: OUTPUT«1␤»
skids m: (+True).say
camelia rakudo-moar b8a75b: OUTPUT«1␤»
llfourn I guess that will have to do :)
cygx m: say Int.new(True) # Java version
camelia rakudo-moar b8a75b: OUTPUT«1␤»
cygx m: say True.Numeric 04:25
camelia rakudo-moar b8a75b: OUTPUT«1␤»
dalee has anyone else had a hard time installing File::Find?
llfourn cygx: all in all I like the feature :)
dalee panda claims it isn't installed, but when it tries to isntall it errors out because it is already installed
also, i'm pretty sure it was installed as part of the panda bootstrap procedure 04:26
cygx I wonder if True.Numeric really should return 1
after all, True ~~ Numeric already
TimToady: ^^
llfourn cygx: that would be nice :)
04:27 cdg joined
llfourn but if it already is it then it wouldn't make that much sense to change itself :\ 04:27
AlexDaniel cygx: we already have a couple of weird bugs associated with Enums getting into wrong places
at least last time I checked :) 04:28
04:29 pjscott left
TimToady yeah, I agree that's a little weird 04:29
cygx on the other hand, just adding a prefix + is a convenient way to construct an Int from an enum value... 04:36
04:36 xpen joined
llfourn m: enum <zero one two>; say +one # it works! 04:37
camelia rakudo-moar b8a75b: OUTPUT«1␤»
llfourn m: enum <zero one two>; say one.Int # But this doesn't do the same as True.Int 04:38
camelia rakudo-moar b8a75b: OUTPUT«1␤»
llfourn m: #{ even though it is Numeric too } enum <zero one two>; say one ~~ Numeric 04:39
camelia ( no output )
llfourn m: enum <zero one two>; say one ~~ Numeric # those comments don't work on camelia? 04:40
camelia rakudo-moar b8a75b: OUTPUT«True␤»
04:40 xpen left
llfourn m: #|{ even though it is Numeric too } enum <zero one two>; say one ~~ Numeric 04:40
camelia rakudo-moar b8a75b: OUTPUT«True␤»
AlexDaniel llfourn: #`{ }
llfourn oops :P
cygx m: #`{ sure they do }; say "ok"
camelia rakudo-moar b8a75b: OUTPUT«ok␤»
cygx m: #`{ sure they do } say "ok" 04:41
camelia rakudo-moar b8a75b: OUTPUT«ok␤»
cygx apparently, I've done enough Perl6 to automatically type }; if there's no newline
04:42 bitmap joined 04:43 TEttinger joined 04:44 ugexe joined 04:50 Psyche^ joined 04:54 zwu left, anaeem1_ joined 04:55 Psyche^_ left 05:00 anaeem1_ left
flussence dalee: panda has been flaky today for me too, no idea why or how to work around it, sorry 05:03
05:04 pjscott joined
dalee i've basically tracked it down to panda installing in this odd directory, /home/dale/panda/inst#/home/dale/.rakudobrew/moar-nom/install/share/perl6/site/panda/state 05:05
so now i have two different state files for the libraries that are installed
the other is /home/dale/.rakudobrew/moar-nom/install/share/perl6/site/panda/state
flussence inst#... shouldn't be a real directory. something's really gone wrong there. 05:06
dalee seems to be somehow related to CompUnit/RepositoryRegistry.pm 05:07
05:08 loren joined 05:09 vendethiel joined
flussence yeah, that's supposed to be an internal string used by that, it definitely should not be ending up in a path name. Not my area of expertise but it seems to find a new way to break every day... 05:09
dalee it looks like just copying the content of the bad state file into the real state file makes panda work properly 05:10
so it's at least a workaround for now 05:11
llfourn dalee: o.o panda has created '#inst' somewhere -- that's a bit scary
05:12 pjscott left
llfourn does '/home/dale/panda/inst#' actually exist 05:12
dalee i'm not the only one this is happening to, according to github
github.com/tadzik/File-Find/issues/17
flussence IMO those prefixes should've been changed to look more like "file:" and "inst:", so they're at least recognisable as being the start of a URL. 05:13
llfourn
.oO( maybe but what if you wanted inst#sftp://... )
 05:15
flussence make the # there a +, that's already convention in lots of other things 05:16
dalee i guess this is an issue with panda's bootstrap.pl
there's one of these: CompUnit::Repository::Installation.new("/home/dale/.rakudobrew/moar-nom/install/share/perl6/site") 05:17
skids I think the colon was echewed because it looks like a PATH delimiter
dalee that's what $target is, but $prefix = $target.path-spec returns "inst#/home/dale/.rakudobrew/moar-nom/install/share/perl6/site"
and right after that we do the mkdir 05:18
05:22 ugexe left, cdg left 05:23 cdg joined 05:28 cdg left 05:31 cdg joined 05:33 vendethiel left 05:35 loren left 05:37 pjscott joined 05:40 Cabanossi left, nige1 joined 05:42 pjscott left
dalee i think i actually have a fix for that panda problem that makes some sense: github.com/tadzik/panda/pull/281 05:42
05:44 Cabanossi joined
llfourn dalee++ 05:46
05:48 khw left, Tonik joined 05:50 bitmap left 05:52 perlawhirl joined 05:53 dalee left, dalee joined
perlawhirl hi perlers 05:56
skids o/
perlawhirl so if i have an IO::Path object... i can check if it's a directory with .d, exectuable with .e, etc 05:58
but where are these documented in the docs.
cant find it
there's is reference to some of them in the IO::Path docs, but not a rundown of each of them
llfourn heh I always though .e was exists 05:59
dalek osystem: e25526a | yowcow++ | META.list:
add String::CamelCase
osystem: 849d601 | (Zoffix Znet)++ | META.list:
Merge pull request #112 from yowcow/add-string-camelcase

Add String::CamelCase
geekosaur "The following file test methods are provided:" in IO::Path::Class? admittedly they don't *look* like methods 06:00
skids perlawhirl: IO just got a rework, which was pending for a long time, so noone wanted to write too many docs yet.
llfourn perlawhirl: yep .e means exists :)
like test -e
perlawhirl wait, you're right... .x for executable. 06:01
llfourn yep the case for documenting it grows!
geekosaur er IO::Path Class. design.perl6.org/S16.html#IO%3A%3APath_Class 06:02
perlawhirl skids: so the '-*' things from perl5 map accross, i imagine
skids Most of them I would think.
geekosaur -T and -B went away but those were always dubious 06:03
perlawhirl geekosaur: thanks. that could practically be copy/pasted into the IO::Path docs... or are some specific to *nix
skids design.perl6.org/S32/IO.html#File_test_methods might be closer to what is implemented than perl 5 06:04
geekosaur well, the whole notion of files being executable is specific to unix
06:04 virtualsue joined
perlawhirl ahh yes 06:04
geekosaur windows does it by extensions and matching registry entries, not by mode bits
skids (and then there is selinux) 06:05
geekosaur :p :b :c also have no windows analogs
or :u :g :k for that matter 06:06
and :S 06:07
ugexex m: my $dir = "asdfasdfasfd".IO; say $dir.e; $dir.mkdir; say $dir.e; # also they are cached, not sure if p5 does that 06:10
camelia rakudo-moar b8a75b: OUTPUT«False␤False␤»
geekosaur perl5 does not cache them as such but has a hack instead 06:11
-e _ reuses the stat buffer from the previous stat()/lstat()/file test
06:12 pjscott joined
dalee for some reason, perl6 is not looking in /home/dale/.rakudobrew/moar-nom/install/share/perl6/site/lib for modules, it only looks in /home/dale/.rakudobrew/moar-nom/install/share/perl6/site 06:12
ugexex heh, in perl6 you have to use a hack to uncache
dalee but Panda is installed in /lib
06:12 cdg left
dalee i can override that with the PERL6LIB environment variable, but that seems weird to me 06:12
ugexex i dont like how it doesnt act like a regular module either 06:13
06:13 pjscott left
AlexDaniel perlawhirl: hmm what about this? doc.perl6.org/type/IO::Path#File_Test_operators 06:14
perlawhirl AlexDaniel: ... 06:15
AlexDaniel “Instead of the colonpair syntax, you can use method calls too:”
perlawhirl facepalms
i was only looking up the top in the contents of method names
06:15 pjscott joined
perlawhirl yah, i'm using method calls in a switch, ie: given $f { when .d { ... } } 06:16
AlexDaniel you can probably do when :d 06:17
geekosaur you do know that when smartmatches?
so yes, the colonpair form works
perlawhirl yes colonpair works too
i think for readability... these ops should have long aliases too... ie $f.readable 06:18
just my thought
ugexex if $f.r#`{eadable} 06:19
gfldex perlawhirl: the eco system is over there, where you could put that module :)
perlawhirl loud and clear gfldex :)
AlexDaniel gfldex: though nobody is going to add a dependency for such a tiny feature, IMO
gfldex perlawhirl: besides, if you have quite a few of those colon pairs, it may become tiresome to read 06:20
skids Still, having a bunch of those little modules around might allow someone to merge them into a DSL/preferences module later. 06:21
06:23 kaare_ joined
geekosaur .oO { use English; } 06:23
skids Heheh. Maybe more like use Flavor::Something 06:25
06:30 pjscott left 06:31 MadcapJake left 06:32 sprocket left, perlawhirl left 06:33 cygx left 06:34 MadcapJake joined 06:36 aypea[2] joined
MadcapJake Just submitted a PR to perl6/doc that makes the `-n` flag a bit more robust. Anyone able to confirm whether `type` would do the same thing as `cat` does i.e., prints to stdout when not given anywhere to write/append to. 06:40
in Windows
06:41 pjscott joined 06:43 pjscott left
ugexex MadcapJake: got something i can c&p? 06:44
06:44 pjscott joined
MadcapJake ugexex: echo hello world | type 06:45
ugexex The syntax of the command is incorrect. The process tried to write to a nonexistent pipe. 06:46
MadcapJake i'm afraid i don't know windows command line well enough, but that, in linux, echoes "hello world" into the pipe which passes to cat and cat just prints to stdout because there is nothing given for it to redirect to 06:47
maybe 'echo "hello world" | type'
06:48 bitmap joined
ugexex same thing. i think it might have to explicitly take piped input to do that 06:49
MadcapJake I think i might be better off changing from a pipe to pager to a `>&1` or something because that should work the same in windows and linux
ugexex are you doing it in perl6? like can you do my $run = shell("command xxx", :out); shell("type {$run.out.lines.join(qq|\n|)}")? 06:51
MadcapJake ugexex: github.com/perl6/doc/blob/master/b...oc#L56-L58
i added an optional flag to p6doc that prints right to stdout instead of pluging the text into a pager 06:52
however, I don't think windows has a `cat` command
ugexex fwiw powershell has cat
not cmd though
MadcapJake i'm guesing cygwin does too. but that leaves all the traditional cmd folks without this all-important option ;) 06:53
AlexDaniel hm, can I ask a question? Why deal with all this piping stuff when you can do $run = run(‘command’, ‘arg’, :out); $run2 = run(‘command2’, :in($run.out)); ? 06:54
ugexex true, although reading docs from cmd.exe would probably suck
gfldex AlexDaniel: pipes are infinite lists 06:55
MadcapJake ugexex: oh god, that's a really good point, i haven't used windows in years but iirc, cmd.exe doesn't even let you scroll back right?
ugexex because piping should handle the memory between the processes without like <fill memory with all of cmd1> -> next process
geekosaur hm? windows can keep scrollback 06:56
or did you mean more.exe?
ugexex it does let you scroll back wih the mouse at least... although im not sure how thats even done with the keyboard
MadcapJake oh, looks like back in the day at least, scrollback was something you had to enable via increasing the screen buffer size. 06:57
AlexDaniel but that's what run is supposed to do?
geekosaur that sounds like windows 3.1...
MadcapJake geekosaur: i have no idea, i've done very little with cmd.exe, most of my programming life has been on linux
geekosaur maybe 95/98, certainly not the nt line
MadcapJake last time i used cmd.exe was probably 98-era 06:58
AlexDaniel try this:
my $x = run(‘sleep’, ‘5’, :out); say ‘hello’; $x.out.slurp-rest; say ‘done’
geekosaur set more=/e and while there's no scroll 1 page back, =line takes you to that line
ugexex if a program is written to use pipes and does say `wget file1 file2 file3 | something-else` it will do `file1 | something-else` before using memory for file2
06:59 CIAvash joined
ugexex run will wget all 3 of those files and have them in memory, and then the next run has to process from that 06:59
thats if they are written to use pipes though
gfldex further those programs can generate the data in question (logfiles are most prominent here). If you program can deal with infinite lists, it can not run out of memory. 07:01
07:01 bitmap left
ugexex run is still filling memory somewhere, just because you are reading from that lazily doesnt mean that data isnt sitting somewhere already 07:01
AlexDaniel my $x = run(‘sleep’, ‘5’, :out); say $x.out.WHAT 07:02
says (Pipe)
I'm not sure if it works correctly right now, but I think that it was supposed to work just like you describe 07:03
otherwise what's the point :)
ugexex theoretically yeah, and that would be great. but at least the jvm version just buffers it 07:04
MadcapJake ugexex: try this `echo hello world | findstr "^"`
ugexex cmd.exe says "hello world" and powershell says "hello\nworld"
AlexDaniel ugexex: even your shell buffers it. Does it buffer it fully? 07:05
MadcapJake oi vey
07:05 asie left
MadcapJake ugexex: what about `echo "hello world" | findstr "^"` 07:06
ugexex AlexDaniel: i think so, but i dont know the java library it uses well enough to say i suppose. i imagine the moar version handles it better too
07:06 asie joined
MadcapJake ugexex: and one more `echo "hello world" | find/v ""` 07:06
ugexex cmd.exe is "hello world" (with the double quotes" and powershell the same without the quotes 07:07
MadcapJake holy guacamole
ugexex the last one is "hello world" with the quotes, and powershell doesnt understand find/v (is there supposed to be a space?)
MadcapJake no idea, just searching through stackexchange answers :P 07:08
ugexex space says parameter format incorrect :/
MadcapJake ok that' 07:09
's stupendous x_x
I really need to see what the output of p6doc would look like with each, not sure if it's piped bare/quoted or whatnot 07:11
i think `findstr "^"` might be the best bet, i doubt there are any quotes, maybe I could use a different pattern that wouldn't split bare words into lines in powershell
is there a way to run Windows in a VM without having a registration key :P 07:12
ugexex FCKGW 07:13
MadcapJake lol, guess not ;) 07:14
ugexex maybe for windows just `shell xxx > doc.tmp` and then read from the file
spin up an aws instance? 07:15
AlexDaniel I have just tested that on linux and it works perfectly 07:16
piping I mean
that is, it is not going to wget all things until you actually read some of the input
MadcapJake i gotta get some rest, thanks for the help ugexex ! g'night all!
AlexDaniel I could even test that on jvm… but that is going to take some time to build 07:17
07:17 pjscott left
AlexDaniel Also, this is just waiting to blow up: shell "$*EXECUTABLE-NAME -I$i --doc=SectionFilter $path $fmt-string | $pager"; 07:20
anyway, anybody using rakudo on jvm can try this: gist.github.com/AlexDaniel/1015a8072fb43442c81c 07:25
change “beep” to any other command of your liking
that is, it beeps every 100 lines (when they were produced) 07:26
when I run this on my system I get 4 beeps immediately. Then I get more beeps when I scroll down
so, dear shell enthusiasts, it seems like it is working as expected. Sure enough one should test that on jvm and on windows, but I'd say that we'd better fix “run” instead of trying to work around it (only if it does not work somewhere, of course). 07:28
and yeah, perhaps instead of thinking about whether it stores a tiny file in the memory or not, one should fix word splitting and code execution… 07:33
07:35 BenGoldberg left 07:42 molaf__ left
AlexDaniel ugexex: so, can you try that on jvm so that I can fill a bug report if it does not work correctly 07:46
07:47 AndyDee left
AlexDaniel ugexex: though I don't think that this “longoutput” script is going to work on windows, if you are using windows (I don't know!) 07:47
07:48 darutoko joined
AlexDaniel apparently rakudobrew is refusing to build it on my pc, no idea what the problem is… 07:48
ugexex yeah it doesnt. i messed around trying to just pipe in text to echo out but get `Reading from stream failed: socket is not connected` 07:51
AlexDaniel ugexex: so, is that on linux or on windows? 07:52
ugexex windows 07:54
AlexDaniel ehh! Then perhaps it's time to find some kind of a dusty windows machine in my house… that's a challenge
ugexex: ok, thanks
ugexex seems like it might be :in(xxx)
07:55 user74747 joined
ugexex yeah it acts strange with :in 07:57
07:58 molaf joined, sno joined
ugexex with `:in($out), :out`it gives the error above. without the `:out` the proc has an exitcode of -16 but doesnt throw an exception like it should 07:59
08:01 bitmap joined
AlexDaniel oh wow, I have no idea how to install perl 6 on windows 08:01
ugexex i use rakudobrew on windows 08:02
AlexDaniel oh, so it works on windows?
ugexex yeah
AlexDaniel cygwin required? 08:03
ugexex i use visual studio for nmake. no cygwin 08:04
dalek c: 715950e | lizmat++ | doc/Language/setbagmix.pod:
Mention ∅ (U+2205 EMPTY SET)
ugexex and cl compiler that is 08:06
08:07 ugexe joined
ugexex oh boy how long until it gets DoSd off again 08:09
AlexDaniel I don't think that I will ever manage to install it on windows… 08:11
“SET PATH "%USERPROFILE%\rakudobrew\bin;%PATH%"” does not work! Hmm… 08:12
ugexex i always do `perl $full-path\bin rakudobrew build moar` 08:13
AlexDaniel oh right, I don't even have perl installed
I give up.
ugexex perl you can just use the strawberry installer 08:14
AlexDaniel that's just too much for me, I really give up
08:18 user74747 left
AlexDaniel ugexex: there is a test for that: github.com/perl6/roast/blob/master...pipe.t#L46 08:20
at least a simple case should work, what exactly have you tried?
it does not seem to be fudged for JVM 08:21
ugexex cool, ill see if it works on windows too. maybe the command has to accept piped input to not error 08:24
it dies at line 38 08:27
The spawned process exited unsuccessfully (exit code: 259) in block <unit> at S32-io/pipe.t line 38 # Looks like you planned 14 tests, but ran 10 08:28
08:28 molaf left
ugexex also the broken pipe error afterwards 08:29
ChoHag Is it just me, or are the is tests in Test.pm6 not sure what "is" is? 08:30
AlexDaniel ugexex: that's very interesting! github.com/coke/perl6-roast-data/b....out#L1483
ChoHag (With a 'slightly' out of date politically aware pun thrown in for good measure) 08:31
masak morning, #perl6
AlexDaniel ugexex: since you probably understand more than me, perhaps you can report it?
ChoHag Seems like you can check that something is a type object exactly, or do a string comparison. 08:32
ugexex i'll have to do it tomorrow, its way too late here now, but sure
ChoHag Shouldn't it be a bit 'smart'er? 08:33
AlexDaniel ugexex: thank you! 08:34
08:34 skids left 08:36 SwellJoe left 08:37 molaf joined, ugexex left 08:47 virtualsue left 08:50 hankache joined
hankache hello * 08:58
09:08 wamba joined 09:10 WAAOUGK joined 09:13 molaf left 09:15 ugexe left 09:16 bitmap left
dalek kudo/nom: ef92faa | (Dale Evans)++ | lib/NativeCall.pm6:
typo

just a typo		 09:19
kudo/nom: b5cb005 | lizmat++ | lib/NativeCall.pm6:
Merge pull request #669 from daleevans/patch-1

typo
[Tux] csv-ip5xs 50000 18.352 18.224 09:24
test 50000 24.391 24.263
test-t 50000 13.819 13.691
csv-parser 50000 52.760 52.631
09:25 domidumont joined
dalee does anyone know how this checksum-looking number in .rakudobrew/moar-nom/install/share/perl6/site/dist/81C4F7EC11D41848ECD6B188688BC57EFD15D2C8 is calculated? 09:25
is it tied to the file contents, module name, ?? 09:26
09:26 domidumont left 09:30 molaf joined
masak tadzik++ # "I AM RAKUDOBREW. ALSO POSSIBLY SPARTACUS. OUTLOOK CLOUDY, ASK AGAIN LATER." 09:30
Hotkeys how can I have leading zeros in a string? 09:35
09:36 RabidGravy joined
Hotkeys ie I want "8" to be "08" but "12" to stay "12" 09:36
s/ie/eg/ 09:37
dalee my $a = sprintf "%02d", 8; say $a;
09:37 firstdayonthejob left
cognominal A joyful year to * : pbs.twimg.com/media/CXqZEqMWYAA9Ovg.jpg 09:38
RabidGravy m: say 8.fmt("%02d"); # if you prefer
camelia rakudo-moar b5cb00: OUTPUT«08␤»
09:39 firstdayonthejob joined, domidumont joined
zhmylove RabidGravy: + 09:40
dalek osystem: 7267c01 | (Maxim Khailo)++ | META.list:
Adding PKafka

Adding reference to my Perl 6 implementation of an Apache Kafka client.		 09:44
osystem: 94ef367 | RabidGravy++ | META.list:
Merge pull request #113 from mempko/patch-1

Adding PKafka
c: 28f2111 | (Jake Russo)++ | bin/p6doc:
Allow -n flag on method/routine docs

now it properly passes the flag along to `MAIN` or `show-docs`.		 09:46
c: ad7fd7d | RabidGravy++ | bin/p6doc:
Merge pull request #311 from MadcapJake/master

Allow -n flag on method/routine docs
RabidGravy merge-tastic! 09:47
09:49 pmurias joined
masak cognominal: joie! I got that one! :D 09:51
I knew all those years of French would pay off somehow! \o/ 09:52
firstdayonthejob: hi
firstdayonthejob: that's an intriguing nick. is it a Perl 6 job, perchance? ;) 09:53
09:53 pmurias_ joined
masak pmurias, pmurias_: that makes two of you. 09:53
RabidGravy I have quite a few French speaking friends on facebook and I find it easier to understand what they are say if I turn the automatic translation off most of the time
09:56 pmurias left 09:57 spider-mario joined
masak in other happy news, I think I finally grok homology 09:59
haven't tried to grok cohomology yet, though :P
10:00 pmurias_ is now known as pmurias 10:01 abraxxa joined
nine Is it possible that panda is older than most of the IO::Path goodness? 10:01
abraxxa hi!
10:02 rindolf joined
masak abraxxa: hi! 10:03
10:03 vividsnow left
masak nine: not just possible 10:03
10:09 pierre-vigier joined 10:10 domidumont left 10:12 pmurias left 10:13 pmurias joined 10:14 FROGGS_ joined, bitmap joined, domidumont joined 10:16 Sqirrel left 10:18 FROGGS left, ugexe joined 10:19 bitmap left
llfourn can someone remind me of the equiv of use FindBin qw/$Bin/; use lib "$Bin/lib"; 10:20
10:21 bitmap joined
masak ooh, I'd like to know that too. I have that use case now and then. 10:22
RabidGravy $*PROGRAM.parent.child("lib").Str 10:23
?
llfourn RabidGravy: I think it should be $?FILE?
RabidGravy $?FILE.IO.parent.child("lib").Str; then 10:24
but that is probably not what you want in a library 10:25
llfourn RabidGravy: works! thanks.
dalee "P6M Merging GLOBAL symbols failed: duplicate definition of symbol Find"
llfourn use lib $?FILE.IO.parent.child("lib").Str;
10:26 vividsnow joined
llfourn feature request: $?DIR 10:26
this might warrant a mention in 5-to-6
dalee: I've seen at least one instance of that kind of annoying error around 10:27
I think it was in an RT
abraxxa dalee: I have the same error with NativeCall
hankache dalee i did rebuild panda using "rakudobrew build-panda" and this error was gone. I hope it solves your issue. 10:29
nine It's possible that this is caused by different versions of File::Find being loaded. 10:36
dalee it appears that the problem is caused by installing File::Find a second time
10:37 virtualsue joined
dalee the version that is installed by panda has version "*", when panda installs File::Find again it installs version 0.1 10:37
shouldn't this cause the old version to be uninstalled?
or does perl6 allow multiple versions of a module to be installed simultaneously?
it seems like this "short" directory contains a mapping of modules to different implementations: .rakudobrew/moar-nom/install/share/perl6/site/short/652B8A474FD7BF89A6C10654387F68B029BA4AFC 10:38
nine dalee: we at least try to allow multiple version to be installed. That's the source of much of the complexity in module management
dalee ok, so the issue is that File::Find doesn't want to co-exist with other versions of itself, even though the other version is actually source identical 10:39
nine dalee: I'm currently debugging exactly the same problem. The open question is why rakudo is picking different versions when using it multiple times. 10:40
dalee i wonder if it's related to the weird version "*" that panda installs
hankache dalee have you tried using --force ?
RabidGravy is there a more fine-grained way of catching warnings than a CONTROL block? it appears to be interfering with my object construction
llfourn didn't even know you could throw or catch warnings 10:41
dalee hankache: if i install with --force i still get the problems 10:42
nine RabidGravy: CONTROL { when CX::Warn { "whatever" } }
RabidGravy yeah, that's what I'm doing
nine The answer is: we're still not smart enough at outdating precomp files.
RabidGravy it seems to be totally altering the control flow 10:43
nine RabidGravy: ah, of course. Warnings are resumable control exceptions. You have to $_.resume to get back to where the warning was thrown
10:44 abraxxa left
RabidGravy duh, of course, thanks 10:44
kaare_ Now I saw the mention of File::Find, I' like to present a failed install I had just before New Year.
It was Ubuntu 14.04, no previous rakudo encounter.
10:44 nige1 left
Hotkeys Is there any chance we could get subscript notation for different based numbers? 10:45
kaare_ installing w/ rakudobrew, and it complained about not being able to go on because the first line of File::Find is use v6.
Hotkeys Like 100₃₆ for :36<100>
RabidGravy nine++ # that's the puppy 10:46
kaare_ Something like "perl 6 is required, but we only have 5.18.2"
10:46 nadim left
Hotkeys I don't think it would be ambiguous and we seem to be big on unicode support around here 10:46
kaare_ So it seems it found perl 5 when looking for perl6.
nine kaare_: that's caused by some PERL5 env var being set. PERL5LIB or something
RabidGravy kaare_, what nine said
10:46 nadim joined
RabidGravy it definitely is PERL5LIB 10:47
10:47 bitmap left
kaare_ nine: OK, I didn't realize that rakudo used PERL5LIB 10:47
nine kaare_: would be nice if you could find out, how exactly that affects rakudobrew
kaare_: rakudo doesn't, but rakudobrew is a Perl 5 script
10:47 ugexe left
kaare_ Oh, that makes sense now. 10:47
nine kaare_: and rakudo's Configure.pl is also a Perl 5 script.
kaare_ As I said, it's @work, so I can try to undef PERL5* from the environment, tomorrow 10:48
10:49 ugexe joined, azawawi joined
azawawi hi 10:49
RabidGravy: ping
RabidGravy erp
azawawi RabidGravy: ping! :)
RabidGravy POWM! 10:50
azawawi :)
RabidGravy: starting rakudobrew on my old rpi board
RabidGravy: and benchmarking
RabidGravy time for a three course meal then
;-)
azawawi RabidGravy: what's your board model and card memory speed? 10:51
nine azawawi: congrats on the Inline::Ruby progress :) 10:52
azawawi nine: that's not me btw awwaiid i think :)
nine: github.com/awwaiid/p6-Inline-Ruby 10:53
nine: but thanks :)
RabidGravy I have two, a v2 which is slow but not too bad, and a Rev B with single core 700Mhz and 435MiB
azawawi awwaiid azawawi 10:54
:)
RabidGravy I don't know what the SD card is though
azawawi SD class that is
kaare_ The v2 is 1 GHz, AFAIK
moritz \o 10:55
kaare_ And petter instruction set. Should be much faster
s/p/b/
RabidGravy to be honest it's probably a crap old SD card, I recycle the ones from audio gear when I get faster ones
azawawi RabidGravy: www.raspberrypi.org/products/raspb...2-model-b/ ? 10:56
RabidGravy I have one of those but I have a v1 one as well 10:57
azawawi good then we can test rakudo star Rpi releases on them :)
RabidGravy the lshw is less than descriptive with regard to the actual board
FROGGS_ I've got a B+ or so but it has not enough mem to compile rakudo :o( 10:58
RabidGravy the v2 is significantly faster
azawawi ARMv7
RabidGravy FROGGS_, that's weird as I've never had a problem
FROGGS_ I've only 512MB RAM 10:59
10:59 FROGGS_ is now known as FROGGS, pierre-vigier left
RabidGravy the v1 B has a "BCM2708, ARMv6-compatible processor rev 7 (v6l)" 10:59
azawawi im running on 512MB btw so will rakudobrew work or fail?
nine ugexe: left a comment on your PR
FROGGS azawawi: you need 1GB at least AFAIK 11:00
RabidGravy the v2 has "BCM2709, ARMv7 Processor rev 5 (v7l)" with four cores
nine ugexe: can you split the commit? I'd gladly merge the second part. Thanks for debugging this!
azawawi FROGGS: or a bigger swap file if you wanna wait :)
FROGGS hmmmm, maybe I try again at some point...
azawawi FROGGS: im trying it right now 11:01
"Updating submodules " step now
FROGGS I'm (literally) working on a C program right now for the pi
RabidGravy I built, a 2015.12 with the 512Mb on the B, not using Raspbian though
and I killed all the X stuff
FROGGS I'm using raspbian
azawawi RabidGravy: im already not using x stuff 11:02
RabidGravy: ssh session
RabidGravy yeah, but is it still running?
the gdm and so forth
nine .tell ugexe I left a comment on your PR. Can you split the commit? I'd gladly merge the second part. Thanks for debugging this!
yoleaux nine: I'll pass your message to ugexe.
11:02 hankache left
azawawi RabidGravy: i will check the services running ... i even have mysql-server running on it lol 11:05
11:06 lnrdo joined
masak can sprintf give thousands formatting? like rendering 1048576 as 1,048,576 ? 11:06
guess not.
azawawi RabidGravy: farabi btw (Perl 5 version) was working on a rpi without no hiccups :)
RabidGravy: s/on a/on an/
RabidGravy yeah, I had RabbitMQ, icecast and all sorts of stuff, I then bought a Gigabyte brix which wasn't much more expensive
aypea[2] masak: still? :(
azawawi RabidGravy: link it plz? 11:07
moritz lol, I blogged: perlgeek.de/blog-en/automating-dep...oject.html (slightly off-topic, because it's not about Perl 6)
RabidGravy it's just a micro-atx PC in a cube shaped box
'ang on 11:08
masak aypea[2]: to be honest, I'm not sure sprintf is the right tool for the job
FROGGS moritz: I was also thinking about blogging about my childrens-rooms-entertainment-app-for-raspi :o)
azawawi moritz: im getting on 403 (forbidden) on that link you posted
moritz azawawi: huh
FROGGS moritz: because it is also a rabbit hole like rakudo...
I get a 404 O.o 11:09
moritz azawawi: can you tell (or privmsg) me your public IP?
azawawi: and does it work with http (no https) for you? 11:10
11:10 lnrdo left, bitmap joined
azawawi moritz: fails on https, work on http 11:10
moritz azawawi: ok, thanks. And fuck.
RabidGravy azawawi, like one of these www.gigabyte.com/products/product-p...id=5569#ov - it's actually a "GB-BXBT-2807" 11:11
azawawi RabidGravy: thx
moritz azawawi: it looks like IPv4 and https don't work together on that site; work with IPv6 for me
dalee well, i can get rid of that duplicate symbol problem by just wiping all the precomp files, so that's nice 11:12
RabidGravy yeah thousands commas in numbers feels like a Locale thing to me, implement in the exosystem
ecosystem either
dalee but installing any "new" version of File::Find causes it again 11:13
FROGGS dalee: does that trigger precompilation by any chance? 11:17
11:19 krunen left
azawawi RabidGravy: confirmed, gdm, x server is off on my rpi box 11:20
nine FROGGS: seems like the cause of the duplicate symbol "Find" is that there are two different precomp files for File::Find and we're trying to load them both as dependencies of different other modules.
FROGGS uhh
11:21 aypea[2] left
FROGGS nine: I mean, this is kinda correct and incorrect that we load both 11:21
11:22 molaf_ joined
FROGGS in the end we need to make the scenario work I think: A->B->C:v1, A->D->C:v2 11:22
11:23 molaf left
FROGGS but here we have the very same C I guess, so the install with --force should replace the existing one and should trigger a recompilation of its reverse deps 11:23
like panda did 11:24
nine FROGGS: yes, that will probably only need a little more work, since jnthn++ has already done a lot for proper long name support and the repositories know about versions, too.
FROGGS: it's not the same versions. panda claims '*' while standalone File::Find is at 0.1
11:25 frankjh joined
FROGGS ahh, its projects.json is buggy then 11:25
11:26 ChoHag left
RabidGravy kill all the things 11:27
nine FROGGS: we precompile '*' during panda's bootstrap, where we load it from ext/File-Find/lib. We probably precompile the 0.1 version during precompilation of some Panda modules. What I don't get is why we even try to load the '*' version when we use a different repo chain. 11:28
FROGGS I don't know either
maybe we should add more debugging output in the codes that decide about candidates
11:30 tipdbmp left
nine Or....we make all rakudo users core developers. When you recompile rakudo multiple times a day, you're pretty safe from this bug... 11:31
FROGGS :S 11:32
11:32 virtualsue left
frankjh Hi, I am trying to make a native library: a method has a parameter unsigned long long, which I declared as longlong. I have two questions with this: Is longlong signed/unsigned? And how do I convert a perl6 'number' into longlong, just passing it gives : "This type cannot unbox to a native integer." 11:35
nine Oh! 11:36
I think, I found it.
azawawi FROGGS: rakudobrew build moar 2015.12 (on rpi 512MB) failed on CONF_SWAPSIZE=100, trying a bigger swap file now 11:37
nine CompUnit::Repository::FileSystem includes $*REPO.id in the precomp id. CompUnit::Repository::Installation OTOH only assigns an id at install time, so it is not dependent on the $*REPO.repo-chain at runtime at all.
RabidGravy frankjh, uint64 I think, and you may need to coerce to an int with Int($val) or somesuch 11:38
FROGGS azawawi: how did it fail?
azawawi FROGGS: ARM: Internal compiler error: in extract_insn 11:39
FROGGS azawawi: interesting
azawawi: I dont remember my error message though
azawawi FROGGS: running it again.... results in a couple of minutes
11:40 bitmap left
RabidGravy azawawi, I have a gig of swap on mine 11:40
llfourn nine: I was looking at that code I never understood why $*REPO.id was used rather than self.id? 11:41
(would love to understand)
RabidGravy "KiB Swap: 1048572 total"
FROGGS m: use NativeCall; sub foo (longlong) { 42 }; say foo 42 # frankjh: I can pass it just fine here
nine That's the commit: github.com/rakudo/rakudo/commit/0734842b
camelia rakudo-moar b5cb00: OUTPUT«42␤»
nine llfourn: because a repository further up in the chain may contain a newer version of a dependency.
11:42 frankjh left
llfourn nine: ok. Will think about that.. 11:42
11:42 frankjh1 joined
azawawi FROGGS: gist.github.com/azawawi/e7a1228385de73f27d4c 11:42
11:42 virtualsue joined
RabidGravy azawawi, ah that's a different matter 11:42
you will need to use a more recent gcc 11:43
azawawi RabidGravy: 4.6.3
RabidGravy gcc (GCC) 4.8.2 20131212 (Red Hat 4.8.2-8)
FROGGS azawawi: mine died when compiling the setting... in stage optimize or mast or so
RabidGravy you should be able to get a more recent version and then fiddle around the alternatives thing to make it the default
azawawi RabidGravy: that's stock raspbian settings fyi 11:44
RabidGravy yeah 4.8.2 definitely works using it on both Pis
frankjh1 FROGGS: yes this work here too, but my $crypto_box_NONCEBYTES = 24; my $mlen = $crypto_box_NONCEBYTES + $m.chars; Gives the unbox error 11:45
FROGGS RabidGravy: I also have gcc 4.6.3 and I was able to compile MoarVM (half a year ago)
RabidGravy azawawi, yeah but the moarvm compile tickles a bug or some infelicity in the older gcc
azawawi so what's the minimum gcc recommended?
RabidGravy FROGGS, yeah it stopped working a few months ago
FROGGS frankjh1: what is $m? 11:46
is it a string?
and where does it come from?
11:46 bitmap joined
RabidGravy I think someone pitched up with 4.8.1 which didn't work either, but 4.8.2 does (and is available in the apt repo) 11:46
azawawi runs "sudo apt-get install gcc-4.8" 11:47
frankjh1 FROGGS $m is a parameter og my perl6 sub "sub crypto_box(Str $m, ..."
FROGGS frankjh1: can you golf it down?
RabidGravy if I get I minute I may make a "Perl 6 on the Rapberry pi" type article (with some help)
11:47 tipdbmp joined
FROGGS RabidGravy++ 11:48
11:48 user74747 joined
RabidGravy it seems like sprocket is leading the way with actually making stuff for it 11:48
frankjh1 This compiles: crypto_box_int($data, $msg, 42, $nonce, $pk, $sk); This not: crypto_box_int($data, $msg, $mlen, $nonce, $pk, $sk); 11:49
azawawi RabidGravy: yeah i noticed, that's why im making a regular rakudo star release for it
RabidGravy I've got two modules in flight but I got pre-empted by a bunch of feature requests for Test::META
azawawi++ 11:53
11:53 ChoHag joined
RabidGravy regex gugus, is there a succinct way of saying "the string contains this substring and not this other one"? 11:56
11:57 nadim left, _nadim joined
frankjh1 RabidGravy: Int($mlen) gives same error. 11:57
RabidGravy :-O
what's in $mlen then? 11:58
azawawi RabidGravy: compiling on gcc 4.8 now 12:00
RabidGravy cool
virtualsue i'm going to do a Perl 6 talk, probably in Feb, and I fear that more than the Go workshop I'm going to do in 2 weeks
FROGGS frankjh1: is crypto_box_int the native function?
virtualsue but in a good way
frankjh1 RabidGravy: 35.
llfourn m: say so "foobaz" ~~ /<!before .*"bar">foo/ # my attempt 12:01
camelia rakudo-moar b5cb00: OUTPUT«True␤»
frankjh1 Yes crypto_box_int is native: sub crypto_box_int(CArray[int8], CArray[int8], longlong, CArray[int8], CArray[int8]) is symbol('crypto_box') is native('./lib/tweetnacl') { * }
I have a perl6 wrapper function around that.
12:02 bitmap left
azawawi RabidGravy: it seems to have worked. time for lunch :) 12:02
RabidGravy yay!
FROGGS frankjh1: can you 'say $mlen.WHAT' before the call to crypto_box_int'? 12:03
12:03 pmurias left
FROGGS or... make it: sub crypto_box_int(CArray[int8], CArray[int8], longlong(), CArray[int8], CArray[int8]) 12:03
or... do what I said first :o)
12:04 pmurias joined 12:08 whj2819 joined
frankjh1 FROGGS: The test runner 'prove' is eating my stdout. Somehow I do not see what I say. :) 12:08
12:08 virtualsue left
FROGGS frankjh1: try 'note' instead of 'say' 12:09
12:10 bitmap joined 12:13 TEttinger left
masak m: say ("0" x 3 ~ "1").substr(2) # [RT #123602] 12:13
camelia rakudo-moar b5cb00: OUTPUT«00␤»
masak still wrong 12:14
FROGGS m: say ("0" x 3 ~ "1")
camelia rakudo-moar b5cb00: OUTPUT«0001␤»
12:14 bitmap left
masak maybe I should "learn the ropes", hehe 12:14
12:15 bitmap joined 12:21 Actualeyes joined
frankjh1 FROGGS: Somehow it looks that it is only a sideffect that the compiling works, if I replace $mlen with 42. Now I have the error two lines above. (It did not reach the line with note $mlen.WHAT) Sorry I need to better understand my code... 12:24
FROGGS aha 12:25
well yeah, I suspected that something strange is going on
12:26 nowan left 12:28 nowan joined 12:29 bitmap left, Sqirrel joined 12:31 pdcawley joined
AlexDaniel m: .say for “hello\c[LINE SEPARATOR]world”.lines 12:38
camelia rakudo-moar b5cb00: OUTPUT«hello␤world␤»
AlexDaniel m: .say for “hello\x0085world”.lines 12:39
camelia rakudo-moar b5cb00: OUTPUT«hello␤world␤»
12:39 wamba left 12:47 lnrdo joined 12:52 araujo joined 12:53 araujo left 12:55 araujo joined, araujo left, araujo joined 13:01 vendethiel joined
RabidGravy jdv79, Zoffix : I think github.com/jonathanstowe/Test-META...tag/v0.0.3 covers your wishes as far as I am prepared to go right now 13:01
13:05 whj2819 left
azawawi RabidGravy: "stage start" reached.... oh yeah :) 13:06
bonsaikitten flussence: btw - second patch, doc strings don't apply cleanly 13:07
flussence: apart from that it's all good
13:12 lnrdo left, lnrdo joined
dalek c: a468121 | lizmat++ | doc/Language/setbagmix.pod:
Fix header level
 13:15
13:15 leont joined 13:16 bitmap joined 13:17 lnrdo left, regreg joined
ChoHag How can I ensure that assignment of an attribute's accessor's return value isn't lazy or optimised away? 13:20
lizmat if it's optimized away, it's a bug ?
13:21 _mg_ joined
ChoHag Even if the variable assigned to is subsequently unused by the time it goes out of scope? 13:21
llfourn that sounds like a bug to me can you golf it? 13:22
ChoHag ie. only even referenced in the assignment: { my $foo = $instance.bar; ...-which-never-mentions-$foo-again; }
There's no code. I want to ensure I write the code sanely if there's a need to.
If even the above example will never cause the assignment not to happen, I'm golden. 13:23
lizmat fwiw, as soon as something is lazy and can have side-effects, it can never be optimized away 13:24
if it is, it's probably somewhere losing the fact that it can have side-effects..
AlexDaniel .u 🌃 13:25
yoleaux U+1F303 NIGHT WITH STARS [So] (🌃)
13:25 regreg left 13:26 harrison_ joined, gavinhi joined
AlexDaniel .u ⌣ 13:27
yoleaux U+2323 SMILE [So] (⌣)
AlexDaniel Hm, I don't think that there is any symbol that is good enough for whatever star 13:28
13:33 kid51 joined
AlexDaniel ★ is a good star, but it's not “whatever”. ⊛ is an operator, so I don't think that it fits. ⍰ is good but it's a question mark… 13:33
ChoHag AlexDaniel: O 13:34
It's a blank face.
AlexDaniel ChoHag: blank face?
ChoHag Like ☺, but blank. 13:35
That or Vicky Pollard, but she's probably not in Unicode.
AlexDaniel I don't think that I'm following you :D
actually, ⍰ is not too bad 13:36
13:36 _mg_ left
AlexDaniel .u ⍰ 13:36
yoleaux U+2370 APL FUNCTIONAL SYMBOL QUAD QUESTION [So] (⍰)
llfourn ⍰ a box with a question mark is pretty whatever to me
leont Sounds like AlexDaniel doesn't have some character in his fonts that ChoHag does have
llfourn although it's hard to to see it because it's so small 13:37
AlexDaniel .u O
yoleaux U+004F LATIN CAPITAL LETTER O [Lu] (O)
ChoHag If asked a question, and you don't know the answer, your face would likely be blank.
AlexDaniel llfourn: I'm pretty sure that everything is rendered correctly here :)
llfourn AlexDaniel: it looks like it to me :) 13:38
ChoHag Oh no not Vicky Pollard. I'm thinking of Catherine Tate.
It's been a long time since I watched TV.
13:39 CIAvash left
azawawi stage parse over rpi 1 (512MB) = 2330.433 seconds :) 13:45
13:49 vike left
frankjh1 Hi, for my native crypto lib I need to pass the data as CArray[int8] in C (char*). How do I copy the indiviual bytes of a Str into a CArray? Or should I convert Str into someother type Buf, Blob sounds fitting, but I did not understand the role thing. 13:50
AlexDaniel any APL aficionados here? What does ‘⍰’ mean there?
azawawi is rakudo star still maintained btw given that the latest is 2015.11? 13:51
AlexDaniel azawawi: yes, it is just not there yet
azawawi slow upload connection? :)
frankjh1 my $msg = CArray[int8].new; my $m = 'Hello' ;loop ($i=0;$i<$m.chars;$i++) { $msg[$I] = $m.access_byte_at($i) } 13:54
What is the proper name for access_byte_at() ?
AlexDaniel frankjh1: good questions! 13:56
RabidGravy it's actually more complicated than that 13:57
AlexDaniel frankjh1: ‘test’.NFD[$i] # what would this give? I'm not sure… let's see
RabidGravy but something like what AlexDaniel said
AlexDaniel frankjh1: yeah, well, strings are on grapheme level by default. If you want bytes you have to do something 13:58
ahhh
hmm…
RabidGravy I'd probably go with "my Buf $m = "Hello".encode .... "
which will definitely get you bytes
AlexDaniel yeah, that's probably one way to do it
frankjh1 So you encode a Str into a Buf? And with Bufs i can access Bytes with [] ? 13:59
13:59 vike joined
RabidGravy yeah 14:00
frankjh1 Ok thanks I will try!
gfldex you better .comb and then encode on the one char string
RabidGravy then do "$msg[$_] = $buf[$_] for ^$buf.elems"
saving the C stule loop
14:00 bbkr joined
RabidGravy gfldex, isn't the nett effect the same? 14:01
AlexDaniel you can also do .ords but these are far from being bytes
14:02 virtualsue joined
RabidGravy (it does of course depend on the expected encoding of the CArray) 14:03
gfldex if you need the whole string as a byte string, comb wont make much sense. In most applications you wont need that or you would start with a Buf right away.
RabidGravy frankjh1, you do realise that you can pass a Str straight to a native sub and it will do the right thing? 14:04
gfldex and by using comb you allow the compiler to help you with iterator magic and CoW strings. I doubt moarvm is doing that already, but there is no reason why i couldn't in the future.
bbkr hi, how can I specify library version in NC cglobal function? I get tons of internal warnings "Consider adding the api version of the library you want to use, sub foo is native(c, v1)"
RabidGravy bbkr, yes 14:05
14:05 harrison_ left 14:06 pdcawley left
dalek c: dec7182 | lizmat++ | doc/Language/glossary.pod:
Some work on the glossary

As part of moving stuff from S99 in here		 14:06
bbkr RabidGravy: I don't see such possibility. signature is "sub cglobal($libname, $symbol, $target-type) is export is rw", no lib version param 14:07
RabidGravy bbkr, yeah I was just looking
frankjh1 RabidGravy: No I didn't know. And I need to add 24 zeros to the start of the Str first. 14:08
Cool
RabidGravy ah
bbkr, if you pass a List ($libname, Version) as $libname, I think it should work 14:12
jdv79 RabidGravy: cool, thanks
RabidGravy (reading guess_library_name)
14:12 Skarsnik joined
Skarsnik Hello 14:14
RabidGravy jdv79, it occurred to me after I had pushed it that I could have gone with a :strict-version as well as the implied meta6 version to determine whether that test was done but that can wait for the next release now
bbkr RabidGravy: thanks, indeed this one works "cglobal(("c", v6), "errno", int)". 14:15
RabidGravy cool, I don't have to blame Skarsnik then ;-)
Skarsnik Ho RabidGravy, you probably want to change your is native trait that use a Callable. It's a bad thing because it get called the first time the sub is executed. You probably want to call that once x) 14:17
RabidGravy eh?
Skarsnik I should probably tag the routine as cached x) 14:18
RabidGravy oh I see, I already did do that :) it works better with the resources thing 14:19
Skarsnik it can be bad if you perform path search in system and such if you have like 20+ native sub x) 14:20
14:20 gavinhi left 14:21 pjscott joined, pjscott left 14:23 colomon left
RabidGravy yeah, I'm doing something like "my constant HELPER = %?RESOURCES<libraries/lastloghelper>.Str;" everywhere now, but that could be extended to perform something more complicated 14:25
14:28 lucasb joined 14:29 colomon joined
azawawi RabidGravy: Done, moar-2015.12 built on RPi 1 512MB, ( real = 145m10.208s, user = 123m23.310s, sys = 3m22.210s) # final results 14:30
RabidGravy haha 14:31
yeah, it would be cool if there could be a package for that really
azawawi RabidGravy: working on it :) 14:32
Skarsnik did you manage to plug 2 sata disk on a pi? 14:33
azawawi Skarsnik: it should be possible but not atm 14:34
14:35 vendethiel left 14:36 colomon left 14:37 colomon joined
masak this script gist.github.com/masak/81409152053431cac80b , which strangelyconsistent.org/blog/youre-...-all-alike says used to take 38 minutes, now takes 20. so someone++ for optimizing Rakudo by 50% :) 14:41
14:41 colomon left
RabidGravy :) 14:41
Skarsnik Does adding a type to the @ speed up a bit? 14:42
gfldex Skarsnik: it will make it slower 14:43
moritz except for native types like int
nine If debugging is like reading a book, this precompilation issue is the Malazan Chronicles, full of short lived characters and when you think, you know what's going on, a single sentence changes everything again. 14:44
14:46 telex left
tadzik masak: you make it sound like the Borg :D 14:47
RabidGravy 496 modules! c'mon another four and we can have a party :)
Skarsnik I have 2 modules to write x) 14:48
14:48 colomon joined, telex joined
masak I'm now running with a completely unrolled @cycles for loop, and next I will replace the regex by bit ops 14:49
14:50 frankjh1 left
Skarsnik how .pm file are installed btw? Like I want to add an app to ecocystem but its pm are 'private' to the app. I don't really want they became usable by other 14:50
14:50 colomon left 14:51 colomon joined
lucasb
.oO( I don't know where your .pm files are... but I will find them... and I will use them! )
 14:51
masak .oO( how is .pm file formed? ) 14:52
moritz Skarsnik: I don't think there's a mechanism for this. You can write lexical classes that can't be used from the outside, but then they must be inside the file where it's used
masak .oO( how module get pragnent? )
gfldex Skarsnik: you may have to build your own build system that does #include by hand 14:53
moritz Skarsnik: the p5 solution is simply to document a class as private
gfldex can panda be asked to execute some script before it installs .pm ?
masak .oO( they need to do way instain build your own build system that does #include by hand... who kill their module ) 14:54
nine gfldex: Build.pm 14:55
gfldex Skarsnik: your build system goes there ^^^ 14:56
lucasb masak: I guess you already saw this book: pragprog.com/book/jbmaze/mazes-for-programmers
Skarsnik x) 14:57
stmuk_ wonders if memorizing the precomp sha1 hashing would speed things up 15:02
lucasb m: say "\xfa6b" 15:05
camelia rakudo-moar b5cb00: OUTPUT«恵␤»
masak lucasb: I did not. thank you.
lucasb m: say "\xfa6c"
camelia rakudo-moar b5cb00: OUTPUT«(signal SEGV)»
masak ok, unrolling brought the time down to 17m53s
lucasb: would you mind submitting that as a bug?
shouldn't segfault :) 15:06
gfldex lucasb: Sir put that character down!
lucasb masak: I think someone already submitted...
m: say "\xfa6d"
camelia rakudo-moar b5cb00: OUTPUT«舘␤»
lucasb I think 0xfa6 is just special... idk
there are more "special" segfaulty chars less than 0xffff 15:07
15:08 CIAvash joined
gfldex m: say "\xEFA9AC" 15:08
camelia rakudo-moar b5cb00: OUTPUT«(signal SEGV)»
masak lucasb: let's find all of them :) 15:09
15:09 azawawi left
nine stux|RC: if that would speed things up, nqp's sha implementation really needs a major overhaul. SHA1 should hash a couple of 100 MB/s on a somewhat recent machine 15:11
Skarsnik wtf is this type: long long unsigned int 15:12
nine stmuk_: ^^^
RabidGravy Skarsnik I saw one of those the other day
Skarsnik it's in sqlite3
stmuk_ nine: ok
Skarsnik size is 64 bits x) 15:13
RabidGravy that's not where I saw it, but it's a uint64
the weather here is hideous
flussence bonsaikitten: ahh, bummer. I should've tested those better. Won't be needed soon(er or later) either way 15:14
masak holy moly. we have a winner.
nine It's -4.4°C here. I wonder if it's time to activate my winter jacket for cycling. 15:15
masak switching completely to native ints, the whole search takes 1m46s
\o/
RabidGravy It's not particularly cold, it's just the rain hasn't stopped all day
bonsaikitten flussence: yeah, thanks for figuring this out, and I'll see what I can use from your nqp/moarvm ebuild mods once I find some more motivation
masak I'm also not ruling out that the bit shifting could be simplified through some clever trick I haven't thought of
Gray encoding, perhaps
Skarsnik RabidGravy, hm are these standard or plateform dependant type? 15:16
RabidGravy I think a testing on an x86-64 and an ARM and they were both the same 15:17
15:17 sprocket joined
RabidGravy but that's not precluding some other architecture doing something else 15:17
Skarsnik I should probably do something special with typedef that point to this type of type x)
flussence bonsaikitten: in the meantime, I'll keep on prodding the people here until all those hacks can go away. I'm pretty good at that :) 15:18
bonsaikitten flussence: teamwork! ;)
Skarsnik RabidGravy, the l l u i is sqlite3_uint64 for example 15:19
flussence things'll be much smoother once everyone's back from christmas vacations/hangovers...
RabidGravy I'm struggling to remember where I saw that type now 15:20
Skarsnik has num64 $.estimatedCost; 15:25
has sqlite_int64 $.estimatedRows;
It's probably better like this
orbus I just tried it on gcc on x86_32 and it's 64 bits as well 15:26
15:27 sprocket left
masak whips out TAoCP 4A to help speed this up 15:27
orbus and this is in stdint.h 15:29
typedef unsigned long long int uint64_t;
Skarsnik x) 15:30
orbus well, actually that's defined if __WORDSIZE==64
is NOT rather 15:31
if it is it's 64 it's typedef unsigned long int uint64_t;
so yeah, probably better to use the header types to be on the safe side 15:32
stmuk_ it took far too long for me to spot the issue with github.com/tony-o/perl6-html-parse.../META.info 15:34
15:34 nige1 joined
stmuk_ I wonder if panda should warn if the "name" doesn't match "provides" - could also be a security issue 15:34
RabidGravy stmuk_, that would break e.g. Task::Star or something that only distributed a script 15:36
Skarsnik name : App::GPTrixie and I don't provide a file like this xD 15:37
stmuk_ it does mean I can distribute a module which appears to be "Acme::Whatever" and is actually installed as LDAP::Password and steals info
RabidGravy yes, I'm just thinking it through
stmuk_ I'm not sure warning would actually break anything 15:38
RabidGravy no, but the case you have there, if you don't crap out then it's too late 15:39
Skarsnik what about App module?
15:40 xpen joined
Skarsnik App::foo will probably not provide a App::foo.something 15:40
RabidGravy and actually even if you check that the Acme::Foo name does appear in the provides then that doesn't stop them putting else in there as described 15:41
huf a warning doesnt help anyway, nobody reads warnings.
stmuk_ Task::Star has an empty provides anyway 15:42
RabidGravy interesting catch there stmuk_
15:42 spider-mario left
RabidGravy well the empty provides case is easy, "if there is anything in the provides, one entry must match the name" 15:43
stmuk_ maybe if provides is defined and doesn't match the same top level name space as name it should warn
Skarsnik I could write a DBIish::Helper (or what look like a DBIish related module) and replace the real DBIish to stole db password probably? (or more bad: with http::ua)
RabidGravy but a) that doesn't stop me putting another entry in the provides that does something bad
and b) as Skarsnik points out, say "App::Foo" may have modules that it uses for its own purposes which have an unrelated name 15:44
15:45 raiph joined
stmuk_ I think the provides was probably ignored before precomp 15:46
RabidGravy stmuk_, that latter would mean that if there was some framework which needed a plugin in a certain namespace and I wanted to ship it with my module which was in a different namespace I would have to package it separately 15:47
nine ABC doesn't contain an ABC module but lots of ABC::
stmuk_ I think panda should just display what it actually installs 15:48
RabidGravy the provides was definitely working at some point before the recent precomp stuff but may have gone away while panda wasn't compiling
Skarsnik test::meta should provide an small app to not have to write code x)
RabidGravy nah, but I'll accept a PR against META6 for a generator script :) 15:49
it should display what it is installing yes
15:49 n0tjack joined
Skarsnik ==> Testing Test::META 15:50
===SORRY!===
Could not find META6:ver<0.0.4..*> in:
damn
I should probably update panda x)
15:50 zakharyas joined
RabidGravy yeah, you need to do a panda update and then probably force install META6 15:51
"META6:ver<0.0.4>" doesn't work in requires at the moment
I think that panda may try to install with the version from it's project.json rather than the one in the META.info 15:53
but I've done playing with that stuff for the day 15:54
15:57 sprocket joined, patrickz joined
nine RabidGravy: yes it does and it's causing problems 15:58
15:58 Actualeyes left 15:59 psy_ left
RabidGravy maybe a simple expedient of hashing the project.json and retrieving a new one if the remote one doesn't match the local one 16:00
or something like that
stmuk_ the remote one is generated by a 15m cron and might be out of date anyway 16:01
RabidGravy or whatever the cpan client does with the modules list
so yeah, panda should use the meta info from the distribution and not it's own 16:03
there's a slight catch if anyone wanted to use tagged releases and was updating the META.list but hey 16:04
16:04 xpen left 16:06 Zoffix joined
RabidGravy it's probably just do whatever it "panda install ." does after it has retrieved the distribution 16:07
stmuk_ I was always suspicious panda still said "Fetching" in that case 16:08
RabidGravy yeah
Zoffix pets the roboy 16:09
yoleaux 2 Jan 2016 19:31Z <RabidGravy> Zoffix: I updated Test::Meta with your suggestions but you will need to upgrade META6 manually too
Zoffix RabidGravy++
sprocket morning, p6!
Zoffix RabidGravy, why manually?
16:09 xyf left
RabidGravy can't specify required version in the META.info currently 16:09
Zoffix :/ 16:10
stmuk_ arggg of course displaying what panda installs just displays SHA1 hashes :)
Zoffix RabidGravy, by can't, do you mean panda doesn't support that yet? 16:11
RabidGravy so the module itself does but it won't grab that version and specifying 'requires" : [ "META6:ver<0.04>"]' no worky
sjn .u U+1F98B
yoleaux No characters found
RabidGravy so yes, panda doesn't support that yet
Zoffix k
16:12 Actualeyes joined, _mg_ joined
sjn babelstone.blogspot.de/2016/01/what...de-90.html # It has a butterfly! :D 16:12
tadzik stmuk_: it's copying it to a temporary location where it works on it later
16:13 pdcawley joined
RabidGravy all good 16:13
stmuk_ sure its the source for the moarvm
16:14 khw joined 16:16 mscha joined
arnsholt Anyone else getting "An exception occurred while evaluating a constant" in install-core-dist.p6? 16:17
mscha m: my $now = DateTime.now; say $now.Instant - $now.posix; #huh?
camelia rakudo-moar b5cb00: OUTPUT«Instant:36.912049␤»
arnsholt Running MoarVM/NQP/Rakudo HEAD
n0tjack *pops head in* Congrats to everyone on the official release!
Zoffix Thanks
mscha, leap seconds maybe? 16:18
stmuk_ with reference to the wrong provides issue I think panda should display what it actually installs rather than what it thinks it will
mscha Zoffix, 26 leap seconds sinds introduction in 1972 16:19
stmuk_ eg. ==> Installing HTML::Parser::XML
nstalling HTTP::Parser::XML in block at /home/steve/sandbox/rakudo/install/share/perl6/site/sources/A8B5E32BE26CA290354B6CBDBED92C1486082961 line 62
Zoffix pats self on back for correct guess
Zoffix unpats self on the back for incorrectly reading 36 as 26 16:20
16:21 hankache joined
mscha TAI, Temps Atomique International, is the international atomic time scale based on a continuous counting of the SI second. TAI is currently ahead of UTC by 36 seconds. TAI is always ahead of GPS by 19 seconds. 16:21
Zoffix Ah
RabidGravy Atomic Time! 16:22
moritz but why the .91?
looks too big to be a numeric error
is one of them rounded to the nearest integer, maybe?
flussence .posix is truncated to int, instant isn't
moritz m: given DatTime.now { say .Instant - .posix } 16:23
camelia rakudo-moar b5cb00: OUTPUT«5===SORRY!5=== Error while compiling /tmp/jfncwvQkgc␤Undeclared name:␤ DatTime used at line 1. Did you mean 'DateTime'?␤␤»
moritz m: given DateTime.now { say .Instant - .posix }
camelia rakudo-moar b5cb00: OUTPUT«Instant:36.985177␤»
flussence Instant also *isn't* a Numeric, which is why the .gist looks like that
mscha m: given DateTime.now { say [ .Instant, .posix, .Instant - .posix ] } 16:24
camelia rakudo-moar b5cb00: OUTPUT«[Instant:1451838293.682377 1451838257 Instant:36.682377]␤»
hankache hello #perl6
n0tjack m: say DateTime.now.Instant.gist;
camelia rakudo-moar b5cb00: OUTPUT«Instant:1451838309.442177␤»
moritz ok, .posix is the integer
n0tjack looks numeric to me
flussence m: say DateTime.now.Instant.^mro 16:25
camelia rakudo-moar b5cb00: OUTPUT«((Instant) (Cool) (Any) (Mu))␤»
n0tjack Cool is numeric-ish
moritz m: say DateTime.^roles
camelia rakudo-moar b5cb00: OUTPUT«((Dateish))␤»
flussence m: say DateTime.now.Instant.^roles 16:27
camelia rakudo-moar b5cb00: OUTPUT«((Real) (Numeric))␤»
flussence oh, there it is
16:27 Zoffix left
stmuk_ WARNING: installing HTTP::Parser::XML for HTML::Parser::XML 16:33
jdv79 when did this meta6 version thing become a thing?
flussence wishes git-push had an --allrmeotes switch
s/rme/rem
moritz flussence: there's a possiblity to configure branches to be pushed to several remotes at once; might that help you? 16:34
flussence: stackoverflow.com/a/14290145 16:35
flussence moritz: probably. 99% of the time it seems it already does what I want, I just have no idea it exists because they put it in a bizarre place :)
RabidGravy jdv79, some time in the last week.
16:35 _mg_ left
lucasb "HTTP::Parser::XML" : "lib/HTML/Parser/XML.pm6" # <-- HTML::Parser::XML's META.info 'provides' section 16:36
hankache nine++ #panda fixes
stmuk_ lucasb: I submitted a PR
16:36 _Vi left
arnsholt nine: Have you come across the "An exception occured while evaluating a constant" error during precomp? 16:36
lucasb stmuk_: ah, ok. I just saw the PR, stmuk_++ 16:37
RabidGravy stmuk_, just wondering how to test that in Test::META 16:38
stmuk_ you could try and fuzzy match the RHS and LHS of the provides I guess 16:39
I'm not even sure there is a valid case when they aren't the same TBH 16:40
RabidGravy I guess one could try and require the file specified on the RHS and see if the symbol on the LHS exists
stmuk_ yes that sounds better
16:42 pmurias left, BenGoldberg joined
RabidGravy and if there was a good reason for it to be different that we haven't thought of then some switch to say "yeah I know about that thanks" 16:42
RabidGravy adds it to the wishlist for Test::META 16:44
dalek osystem: 642c08b | (Sylvain Colinet)++ | META.list:
Add App::GPTrixie, a NC code generator from C header

  github.com/Skarsnik/gptrixie		 16:45
stmuk_ installing IETF::RFC_Grammar for URI
Skarsnik Nooo
it was not my fork ><
stmuk_ that looks a valid case
Skarsnik I should display the git repo in my shell prefix x) 16:46
RabidGravy oh well :) 16:49
stmuk_ installing X::JSON::RPC for JSON::RPC as well 16:50
which makes sense
16:52 lucasb left 16:55 Actualeyes left 17:03 nzkindest joined
Skarsnik hm 17:06
17:07 SwellJoe joined
Skarsnik What should I use to have unencoded string? 17:07
17:07 user74747 left 17:08 Actualeyes joined 17:11 llfourn left
AlexDaniel Skarsnik: by unencoded you mean? 17:11
Skarsnik I should maybe explain my issue, it's probably not that 'easy' I am rewrite a module that handle IRC color, and it work by adding sequence like \002\003 before a text 17:13
I am not sure if that will not be mashed as a unicode graphem 17:14
17:14 nowan left
stmuk_ nine: github.com/rakudo/rakudo/pull/670 17:15
17:16 _Vi joined
AlexDaniel m: say “\x035” 17:16
camelia rakudo-moar b5cb00: OUTPUT«5␤»
AlexDaniel m: say “\x035”.chars
camelia rakudo-moar b5cb00: OUTPUT«1␤»
17:16 nowan joined
AlexDaniel m: say (“\x03” ~ “99,25”).chars 17:17
camelia rakudo-moar b5cb00: OUTPUT«6␤»
AlexDaniel Skarsnik: it seems like it wont? 17:18
Skarsnik: also, it seems like everything that follows is just pure text, not bytes 17:19
Skarsnik m: say "\00302,05\0021" 17:20
camelia rakudo-moar b5cb00: OUTPUT«␀0302,05␀021␤»
Skarsnik m: say "\x0302,05\x021" 17:21
camelia rakudo-moar b5cb00: OUTPUT«̂,05!␤»
Skarsnik m: say "\x00302,05\x0021"
camelia rakudo-moar b5cb00: OUTPUT«̂,05!␤»
AlexDaniel m: say “\x03” ~ ‘02,05’ ~ “\x02” ‘1’
camelia rakudo-moar b5cb00: OUTPUT«5===SORRY!5=== Error while compiling /tmp/h0FW1o9SSZ␤Confused␤at /tmp/h0FW1o9SSZ:1␤------> 3say “\x03” ~ ‘02,05’ ~ “\x02”7⏏5 ‘1’␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ sta…»
AlexDaniel m: say “\x03” ~ ‘02,05’ ~ “\x02” ~ ‘1’ 17:22
camelia rakudo-moar b5cb00: OUTPUT«02,051␤»
Skarsnik ok it work
hm it should be bold 17:23
RabidGravy :)
it'sa kind of cerise and blue thing
maybe that is "bold" here
17:23 hankache left
AlexDaniel m: say ‘x’ 17:24
camelia rakudo-moar b5cb00: OUTPUT«x␤»
17:24 hankache joined
Skarsnik m: say “\x03” ~ ‘02,05’ ~ ‘1’ 17:24
camelia rakudo-moar b5cb00: OUTPUT«02,051␤»
Skarsnik m: say “\x02" ~ ‘1’
camelia rakudo-moar b5cb00: OUTPUT«5===SORRY!5=== Error while compiling /tmp/g6qlXEnHID␤Unable to parse expression in smart double quotes; couldn't find final '”' ␤at /tmp/g6qlXEnHID:1␤------> 3say “\x02" ~ ‘1’7⏏5<EOL>␤ expecting any of:␤ argument list…»
Skarsnik m: say "\x02" ~ '1'; 17:25
camelia rakudo-moar b5cb00: OUTPUT«1␤»
Skarsnik m: say "\x02" ~ 'é';
camelia rakudo-moar b5cb00: OUTPUT«é␤»
RabidGravy are subsets "our" scoped? or more specificially if I define a subset "Block" in a class definition it won't screw up the real Block? 17:26
I'm thinking yes and no from experimentation 17:28
Skarsnik x)
17:29 lsm left
Skarsnik I think I will cp the text method of cpansearch.perl.org/src/SCOLINET/Ac...IRC/Art.pm it's a big wtf xD 17:30
17:30 deepika joined 17:32 lsm-desktop joined
stmuk_ 17:32
17:34 nige1 left
Skarsnik m: for ^2 { say $_ }; 17:38
camelia rakudo-moar b5cb00: OUTPUT«0␤1␤»
17:39 takdaram joined
RabidGravy gosh, the i2c library is made of crack 17:39
Skarsnik x) 17:41
use gptrixie!
17:41 takdaram left
RabidGravy I don't think it would help with this one, the "library" is actually a .h file with a lot of "inline" function definitions, so I have to make a wrapper in C to realise the inlines 17:43
autarch Exodist & leont: I have a working TAP12 formatter based on stream events in github.com/autarch/perl6-Test-Stream
17:44 nzkindest left
Skarsnik RabidGravy, oh yeah, very likely 17:44
17:45 pierre-vigier joined
flussence adds to crontab `find ~/code/perl6 -type d -name .precomp -exec rm -r {} +`... because it just cleaned up 27MB of space 17:46
17:46 nzkindest joined
pochi m: try { die "foo" }; say $!.can("perl"); say $!.^methods 17:48
camelia rakudo-moar b5cb00: OUTPUT«(perl)␤(message Numeric from-slurpy <anon> backtrace resume die is-compile-time gist Str reset-backtrace throw vault-backtrace rethrow resumable fail)␤»
pochi How come $! can do .perl, but .perl doesn't show up in methods?
most other objects that can do .perl also has it in the ^methods list 17:49
moritz pochi: .^methods suppresses methods from Any and Mu by default
hankache what would i need to build rakudo on Windows? 17:50
moritz pochi: see doc.perl6.org/routine/methods
pochi aha
17:50 pierre-vigier left
pochi where in the source code does the actual assignment to $! happen upon a try { die }? 17:53
17:59 Zoffix joined
Zoffix Should this cry and complain? 17:59
m: my %h = :foo<bar> :ber(<boor>); say %h
camelia rakudo-moar b5cb00: OUTPUT«foo => bar␤»
Zoffix Just had a bug due to this and it's only by chance that I noticed the missing comma
Zoffix is starting to think avoiding commas in arg calls is a bad habit to get into 18:00
18:00 nige1 joined
Zoffix m: my %h = :foo<bar> :ber<boor>, :meow<moo>; say %h 18:00
camelia rakudo-moar b5cb00: OUTPUT«foo => bar, meow => moo␤»
18:00 pierre-vigier joined 18:02 Ben_Goldberg joined
Zoffix Oh, this is fun: 18:04
m: my $e = { foo => ['bar']}; my %h = :foo($e<foo>[0]); %h<foo>.subst-mutate: /^'ba'/; say [ %h, $e ]
18:04 BenGoldberg left
hankache Zoffix i always put commas, semicolons, parentheses even if unnecessary. Code will look ugly but at least I don't spend half a day looking for that missing semicolon 18:04
Zoffix (missing second arg on .sub-mutate
camelia rakudo-moar b5cb00: OUTPUT«Memory allocation failed; could not allocate 16 bytes␤»
Zoffix )
hankache, I can counter that argument with: more typing and more reading
18:06 pierre-vigier left 18:07 llfourn joined 18:11 llfourn left
Zoffix Rakudobugged both of those: commaless-hash-keys: rt.perl.org/Ticket/Display.html?id=127134 and the infinite loop on .subst-mutate: rt.perl.org/Ticket/Display.html?id=127135 18:13
The rt.perl.org/Ticket/Display.html?id=127135 is LHF, as I see what the problem is (described in ticket). Just a check for number of args in Cool needs to be added along with the error throwage
hankache Zoffix++
18:13 pi1 left 18:14 pi1 joined, skids joined
dalek albot: 6492214 | (Zoffix Znet)++ | irc.perl.org.conf:
Add #perl6 to the list of channels on irc.perl.org
 18:15
hankache there is now a #perl6 on irc.perl.org ?? 18:16
18:16 pierre-vigier joined
Zoffix Seems to be 18:17
hankache oh!
18:17 pi1 left 18:18 Zoffix left
Skarsnik damn zoffix gone again xD 18:18
18:19 pi1 joined, pierre-vigier left 18:21 pi1 left, pi4 joined
timotimo probably watching AGDQ now? 18:22
18:22 dwarring joined
Skarsnik Monkey Ball is not really my thing, but it's still fun to watch 18:22
virtualsue hankache: that's not the official #perl6 channel in any way, shape or form. 18:23
it might become fun, however 18:24
hankache hiya virtualsue
yeah i just read that this one is the official 18:25
virtualsue i am in for the fun!
RabidGravy eugh, "i2c_smbus_block_process_call" takes an array of uint8 values and modifies it in place
18:25 _mg_ joined 18:26 wamba joined, SwellJoe left
RabidGravy are there any devices that would actually use that? 18:28
Skarsnik Don't add it and wait for someone to fill an issue about it? x)
dj_goku m: my $v = 12; { my $v = 13; { say 'parent ' ~ ::<$v> ~ ' grand parent ' ~ OUTER::OUTER::<$v>; } } 18:29
camelia rakudo-moar b5cb00: OUTPUT«parent 13 grand parent 12␤»
dj_goku :D
18:30 sjoshi joined
Rotwang is there something like 'trace' module in python? 18:31
or set -x in bash?
18:32 nowan left
Ben_Goldberg You could probably assign (or rather, bind with := ) a Proxy object to the variable you want to debug. 18:33
18:34 Actualeyes left 18:35 nowan joined
sjoshi Is perl6 out? 18:35
timotimo right. "assigning a proxy" will call the FETCH and give you what's inside the proxy instead of taking the proxy into the new variable
sjoshi: yup, grab it while it's hot!
sjoshi timotimo: cool, thanks!
18:35 mscha left
timotimo if you want more details, stick around :) 18:36
flussence m: use trace; say 1; say 2; sub foo { 3 }; say 3; # Rotwang
camelia rakudo-moar b5cb00: OUTPUT«2 (/tmp/F7TIqPhlCl line 1)␤say 1␤1␤3 (/tmp/F7TIqPhlCl line 1)␤say 2␤2␤4 (/tmp/F7TIqPhlCl line 1)␤sub foo { 3 }␤7 (/tmp/F7TIqPhlCl line 1)␤say 3␤3␤»
timotimo though i'm too distracted to give a speech
18:37 Actualeyes joined
Rotwang flussence: cool! T.Hanks 18:40
Ben_Goldberg Another way to debug your entire script, is to install this: github.com/jnthn/rakudo-debugger
timotimo i *think* the debugger currently has a problem? someone might have to look closer at that and fix it 18:42
it might be super easy, though
18:48 SwellJoe joined
sjoshi Does perl6 has something to offer as we have like ipython? 18:50
Juerd Not that I know of 18:52
Skarsnik not yet?
18:53 polyfloyd joined
Juerd polyfloyd: *wave* 18:53
polyfloyd \o Juerd
18:53 jameslenz left 18:54 nemsys joined 18:56 jameslenz joined 19:01 zakharyas left
Rotwang where are installed the perl6 core *.pm files when using rakudobrew? 19:01
19:01 ugexe left 19:02 firstdayonthejob left 19:03 pierre-vigier joined
RabidGravy Rotwang, the aren't install per-se 19:03
timotimo sjoshi: working on it
Rotwang RabidGravy: I'm getting an exception and would like to add debug print 19:04
timotimo Rotwang: before compiling they are concatenated into one big file
you will have to edit the source files and "make install"
mst Rotwang: .rakudobrew/moar-nom/install/share/perl6/dist/
19:04 Zoffix joined
Zoffix Hey. How do we do word boundaries in P6 regex? 19:04
timotimo mst: not helping :)
Zoffix m: say so 'yo bot' ~~ /\b 'bot' \b/ # doesn't work
camelia rakudo-moar b5cb00: OUTPUT«False␤»
timotimo Zoffix: >> and <<
mst timotimo: well, that tells you which file is which :) 19:05
Zoffix m: say so 'yo bot' ~~ />> 'bot' <</
camelia rakudo-moar b5cb00: OUTPUT«False␤»
timotimo mst: no, i don't think so
Zoffix: other directionk :)
mst: not with the core setting source files at least
mst timotimo: er, yeah
Zoffix m: say so 'yo bot' ~~ /<<'bot'>>/ 19:06
camelia rakudo-moar b5cb00: OUTPUT«True␤»
mst timotimo: if I crack open the file in there, and find the name
timotimo but thanks for helping anyway :)
Zoffix timotimo++ thanks
mst moar-nom/install/share/perl6/sources/ contains the source
Skarsnik Hey Zoffix, I think there is an issue with modules.p6.org, it always load a cached stuff before displaying result. it make using back on the main page lost the scrolling position
sjoshi timotimo: sounds cool :)
mst timotimo: I've found Test.pm at least :)
timotimo sjoshi: you can see the progress in github.com/timo/iperl6-kernel or something like that
arnsholt has been working on it recently, i had left it alone for a few years before that, though :(
RabidGravy yeah, Test NativeCall and lib I think 19:07
Zoffix Skarsnik, get a better browser :P
Skarsnik, what are you using?
Skarsnik chrome
it do the same thing on my desktop and my chromebook 19:08
19:08 llfourn joined
Rotwang timotimo: make install did what I wanted, thanks 19:08
Zoffix Skarsnik, so you're clicking on something on the main page, then click the back button, and your scroll position is lost?
19:08 user74747 joined
Skarsnik Yes because the page seens to always load something before displaying the right content 19:09
like a search result will display dumb content before refreshing itself to the result (when reloaded)
dalek href="https://modules.perl6.org:">modules.perl6.org: e6bf9ca | (Zoffix Znet)++ | public/js/main.js:
Fix unwanted search box focus in Chrome

The issue is it jumps back to search box when a BACK button is used after clicking on a search result		 19:12
Zoffix Skarsnik, ^ that probably fixes it
19:12 Zoffix left 19:13 llfourn left 19:15 pierre-vigier left
dalek kudo/nom: 080ac18 | (Steve Mynott)++ | src/core/CompUnit/PrecompilationStore/File.pm:
don't mkdir if it exists already - which it does sometimes
 19:16
kudo/nom: 0639e1b | (Steve Mynott)++ | src/core/CompUnit/Repository/Installation.pm:
use existing RAKUDO_LOG_PRECOMP to display more information about exactly what is installed - helps with catching cases where 'provides' means a module is installed somewhere non-obvious and sometimes wrong
kudo/nom: 4bb47d5 | lizmat++ | src/core/CompUnit/ (2 files):
Merge pull request #670 from stmuk/nom

Extra logging and check before creating an existing dir
19:16 lucasb joined 19:21 Ben_Goldberg is now known as BenGoldberg 19:26 fdsfdsfds joined 19:27 fdsfdsfds left, dsfdsfds joined
dsfdsfds In one of the files I have this: 19:28
use MyLib::Namespace1::MyClass;
In order to refer to MyClass I still have to specify its full name MyLib::Namespace1::MyClass for reason. How can get rid of it? I'd like to say only "MyClass".
**for some reason, otherwise the error is thrown 19:29
Skarsnik class MyClass is export
dsfdsfds where?
Rotwang is trick
m: use NativeCall; sub gettimeofday() is native("linux-vdso.so.1") { ... }; await (start { gettimeofday } xx 20)
Skarsnik in your class definition
camelia rakudo-moar b5cb00: OUTPUT«Memory allocation failed; could not allocate 4194304 bytes␤»
19:29 ugexe joined
Rotwang m: use NativeCall; sub gettimeofday() is native("linux-vdso.so.1") { ... }; await (start { gettimeofday } xx 20) 19:29
camelia rakudo-moar b5cb00: OUTPUT«(signal ABRT)*** Error in `/home/camelia/rakudo-m-inst-1/bin/moar': double free or corruption (fasttop): 0x00000000068a5110 ***␤======= Backtrace: =========␤/lib64/libc.so.6(+0x730bf)[0x7fe57cb9c0bf]␤/lib64/libc.so.6(+0x7892e)[0x7fe57cba192e]␤/lib6…»
dsfdsfds thanks 19:30
Skarsnik * not ...
I am not sure ... work?
19:30 dsfdsfds left
Skarsnik seems like it work 19:30
lizmat m: say "Happy {[+] ^64}"
camelia rakudo-moar b5cb00: OUTPUT«Happy 2016␤»
hankache lizmagic!!
19:31 nige1 left
lizmat
.oO( courtesy of Perl 6 FaceBook group )
 19:31
BenGoldberg m: use NativeCall; sub gettimeofday() is native("linux-vdso.so.1") { ... }; say gettimeofday; 19:32
camelia rakudo-moar 4bb47d: OUTPUT«(Mu)␤»
Rotwang BenGoldberg: it only happens when run from a promise 19:33
Skarsnik m: use NativeCall; sub gettimeofday() is native("linux-vdso.so.1") { * }; say gettimeofday;
camelia rakudo-moar 4bb47d: OUTPUT«(Mu)␤»
Skarsnik why Mu
moritz because you didn't declare a return type
Skarsnik m: use NativeCall; sub gettimeofday() is native("linux-vdso.so.1") returns num64 { * }; say gettimeofday; 19:34
camelia rakudo-moar 4bb47d: OUTPUT«0␤»
BenGoldberg m: use NativeCall; sub int gettimeofday(int, int) is native("linux-vdso.so.1") { ... }; say gettimeofday(0, 0);
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/xr2RbVaay2␤Did you mean to write "my int sub gettimeofday" or put "returns int" before the block?␤at /tmp/xr2RbVaay2:1␤------> 3use NativeCall; sub int gettimeofday7⏏5(int, int) is native("linux-vdso…»
BenGoldberg m: use NativeCall; my int sub gettimeofday(int, int) is native("linux-vdso.so.1") { ... }; say gettimeofday(0, 0);
camelia rakudo-moar 4bb47d: OUTPUT«Potential difficulties:␤ In 'gettimeofday' routine declaration - Not an accepted NativeCall type for parameter [1] : int␤ --> For Numerical type, use the appropriate int32/int64/num64...␤ at /tmp/0VgTzLOXnB:1␤ ------> 3nt) is native(…»
BenGoldberg m: use NativeCall; my int sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; say gettimeofday(0, 0);
camelia rakudo-moar 4bb47d: OUTPUT«Potential difficulties:␤ The returning type of 'gettimeofday' --> int is erroneous. You should not return a non NativeCall supported type (like Int inplace of int32), truncating errors can appear with different architectures␤ at /tmp/BVp2buGDy5:1…»
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; say gettimeofday(0, 0);
camelia rakudo-moar 4bb47d: OUTPUT«0␤»
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; await (start { gettimeofday(0, 0) } xx 20 ); 19:35
camelia rakudo-moar 4bb47d: OUTPUT«(signal SEGV)»
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; await (start { gettimeofday(0, 0) } xx 20 );
camelia rakudo-moar 4bb47d: OUTPUT«(signal SEGV)»
RabidGravy m: m: use NativeCall; sub gettimeofday(int32, int32) returms int32 is native("linux-vdso.so.1") { ... }; say gettimeofday(0, 0)
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tvlbtS02I3␤Missing block␤at /tmp/tvlbtS02I3:1␤------> 3tiveCall; sub gettimeofday(int32, int32)7⏏5 returms int32 is native("linux-vdso.so.␤ expecting any of:␤ new name to be defined␤»
moritz m: m: use NativeCall; sub gettimeofday(int32, int32) returns int32 is native(Str) { * }; say gettimeofday(0, 0);
camelia rakudo-moar 4bb47d: OUTPUT«0␤»
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; await (start { gettimeofday(0, 0) } xx 1 );
camelia ( no output )
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; say await (start { gettimeofday(0, 0) } xx 1 );
camelia rakudo-moar 4bb47d: OUTPUT«(0)␤»
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; say await (start { gettimeofday(0, 0) } xx 2 );
camelia rakudo-moar 4bb47d: OUTPUT«(0 0)␤»
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; say await (start { gettimeofday(0, 0) } xx 5 );
camelia rakudo-moar 4bb47d: OUTPUT«(0 0 0 0 0)␤»
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; say await (start { gettimeofday(0, 0) } xx 10 ); 19:36
camelia rakudo-moar 4bb47d: OUTPUT«(0 0 0 0 0 0 0 0 0 0)␤»
BenGoldberg m: use NativeCall; my int32 sub gettimeofday(int32, int32) is native("linux-vdso.so.1") { ... }; say await (start { gettimeofday(0, 0) } xx 20 );
camelia rakudo-moar 4bb47d: OUTPUT«Memory allocation failed; could not allocate 21496 bytes␤»
RabidGravy m: m: use NativeCall; sub gettimeofday(int32, int32) returns int32 is native("linux-vdso.so.1") { ... }; say gettimeofday(0, 0)
camelia rakudo-moar 4bb47d: OUTPUT«0␤»
Skarsnik maybe it's not a thread safe function?
Rotwang Skarsnik: had the same issue with getuid ;f
19:36 user74747 left
Rotwang besides getimeofaday is so basic, it has to be thread safe 19:37
moritz maybe NativeCall isn't thread safe?
19:38 uruwi joined
Skarsnik pubs.opengroup.org/onlinepubs/00969...g_02_09_01 19:38
19:38 user74747 joined
Rotwang Skarsnik: gettimeofaday is more of a syscall and not library function 19:38
19:39 user74747 left
Skarsnik I don't know if the NC devs are still around x) 19:39
19:39 user74747 joined, nige1 joined 19:40 user74747 left
BenGoldberg m: use NativeCall; class timeval is repr('CStruct') { has int32 $.tv_sec; has int32 $.tv_usec }; my int32 sub gettimeofday(timeval, int32) is native("linux-vdso.so.1") { ... }; my timeval $t .= new; say gettimeofday($t, 0); say $t; 19:40
camelia rakudo-moar 4bb47d: OUTPUT«0␤timeval.new(tv_sec => 1451850032, tv_usec => 0)␤»
19:40 user74747 joined
BenGoldberg m: use NativeCall; class timeval is repr('CStruct') { has int32 $.tv_sec; has int32 $.tv_usec }; my int32 sub gettimeofday(timeval, int32) is native("linux-vdso.so.1") { ... }; my timeval $t .= new; say gettimeofday($t, 0); say $t; say now 19:41
camelia rakudo-moar 4bb47d: OUTPUT«0␤timeval.new(tv_sec => 1451850071, tv_usec => 0)␤Instant:1451850107.633534␤»
19:41 kjk joined
BenGoldberg m: use NativeCall; class timeval is repr('CStruct') { has int32 $.tv_sec; has int32 $.tv_usec }; my int32 sub gettimeofday(timeval, int32) is native(*) { ... }; my timeval $t .= new; say gettimeofday($t, 0); say $t; say now 19:41
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/kJislfNkUo␤No appropriate parametric role variant available for 'NativeCall::Native'␤at /tmp/kJislfNkUo:1␤»
BenGoldberg m: use NativeCall; class timeval is repr('CStruct') { has int32 $.tv_sec; has int32 $.tv_usec }; my int32 sub gettimeofday(timeval, int32) is native { ... }; my timeval $t .= new; say gettimeofday($t, 0); say $t; say now 19:42
camelia rakudo-moar 4bb47d: OUTPUT«0␤timeval.new(tv_sec => 1451850123, tv_usec => 0)␤Instant:1451850159.826216␤»
BenGoldberg wonders why tv_usec isn't getting filled in.
Rotwang moritz: I don't see any info that NativeCall subroutines can't be used in a promise/supply 19:43
besides that would be a major drawback 19:44
RabidGravy yeah, they can, it really just depends on the thread safety of the library 19:45
Skarsnik NC is more in a state of "It works" that getting trohough testing I believe
19:45 mempko_ joined
mempko_ FYI, I published an alpha release of my Kafka client library github.com/mempko/PKafka 19:46
Happy to contribute a little to the perl6 community
BenGoldberg m: use NativeCall; class timeval is repr('CStruct') { has int32 $.tv_sec; has int32 $.tv_usec }; my int32 sub gettimeofday(timeval, int32) is native { ... }; say await( start { my timeval $t .= new; gettimeofday($t, 0), $t } xx 1 );
camelia rakudo-moar 4bb47d: OUTPUT«((0 timeval.new(tv_sec => 1451850382, tv_usec => 0)))␤»
Rotwang RabidGravy: but shouldn't those subroutines have their own address space?
BenGoldberg m: use NativeCall; class timeval is repr('CStruct') { has int32 $.tv_sec; has int32 $.tv_usec }; my int32 sub gettimeofday(timeval, int32) is native { ... }; say await( start { my timeval $t .= new; gettimeofday($t, 0); $t.tv_sec } xx 10 );
camelia rakudo-moar 4bb47d: OUTPUT«(1451850408 1451850408 1451850408 1451850408 1451850408 1451850408 1451850408 1451850408 1451850408 1451850408)␤»
RabidGravy mempko_, welcome aboard
mempko_ RabidGravy: Thanks! 19:47
BenGoldberg m: use NativeCall; class timeval is repr('CStruct') { has int32 $.tv_sec; has int32 $.tv_usec }; my int32 sub gettimeofday(timeval, int32) is native { ... }; say await( start { my timeval $t .= new; gettimeofday($t, 0); $t.tv_sec } xx 20 );
camelia rakudo-moar 4bb47d: OUTPUT«Memory allocation failed; could not allocate 4194304 bytes␤»
Rotwang I know that from the moarvm perspective those are threads
bbkr how can I use libc version in OS X if file is simply named "libc.dylib" without any version symlink? I want to silence "Consider adding the api version of the library..." warning
Skarsnik that pretty weird 19:48
RabidGravy Rotwang, things like say ncurses have static data in the library which is never going to be safe
19:49 pierre-vigier joined
Skarsnik bbkr, there is really no libc.x.dylib? 19:49
19:49 ChoHag left
bbkr /usr/lib/libc.dylib -> libSystem.dylib 19:49
no version alias
kjk << abc >>.WHAT.say; << abc 1 2 3 >>.WHAT.say; my $k = 'abc'; << $k >>.WHAT.say; << $k a b c >>.WHAT.say 19:50
RabidGravy for that do you even need to supply a library name?
BenGoldberg bbrk, what happens if you change "is native ('libc')" (or whatever) to "is native" ? 19:51
RabidGravy (i.e. it should be loaded)
bbkr Skarsnik: and there is alias - /usr/lib/libSystem.dylib -> libSystem.B.dylib, but "B" is not valid version identifier
RabidGravy ah but you mean cglobal
bbkr trying...
yes, I use it in cglobal
Rotwang I guess it shouldn't crash anyway, so I'll create a bug, since NativeCall is in core I should send bug report to [email@hidden.address] ?
Skarsnik Str should work too
kjk hello~ for my code above, the third 'say' statement says (Slip) rather than (Str), is this expected behaviour? 19:52
Skarsnik m: use NativeCall:TEST; say guess_library_name(Str);
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/ioZmyfsF1R␤Undeclared routine:␤ guess_library_name used at line 1␤␤»
Skarsnik m: use NativeCall :TEST; say guess_library_name(Str); 19:53
camelia rakudo-moar 4bb47d: OUTPUT«␤»
Skarsnik you can probably pass Str to cglobal too
bbkr like this: "cglobal(("c", "B"), "errno", int);" ? it croaks with Type check failed in binding $version; expected Version but got Str 19:54
Skarsnik vB
RabidGravy yeah
19:54 regreg joined
RabidGravy m: say vB 19:54
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/brcRxaKBv2␤Undeclared routine:␤ vB used at line 1␤␤»
Skarsnik err
RabidGravy m: say Version.new("B")
camelia rakudo-moar 4bb47d: OUTPUT«vB␤»
Skarsnik Oh right it's... ambiquious
bbkr cglobal(("c", Version.new("B")), "errno", int) now croaks with Cannot locate native library 'libc.B.dylib'. because in OS X it is named libSystem.B.dylib 19:56
Skarsnik tryc cglobal(Str, "errno", int)
since the libc is already loaded x)
bbkr that worked, but returns different value 2 than cglobal("c", "errno", int) which returns 10 19:58
timotimo isn't "int" unsafe here?
for 32bit/64bit? 19:59
Skarsnik It's a tricky question to not warn or not under os X for version (I think someone pointed that opengl only have a unversionned .dynlib) since you want the version on unix
flussence it's defined as "int" in the errno.h file, so that's what you gotta use...
Skarsnik no, it should be int32
timotimo bah, opengl on osx
i've heard terrible things 20:00
20:00 Zoffix joined
kjk p6: << abc >>.WHAT.say; << abc 1 2 3 >>.WHAT.say; my $k = 'abc'; << $k >>.WHAT.say; << $k a b c >>.WHAT.say 20:00
camelia rakudo-moar 4bb47d: OUTPUT«(Str)␤(List)␤(Slip)␤(List)␤»
Zoffix Curious. Using Perl 6's DBIish I get this error "Cannot locate native library 'libsqlite3.so': libsqlite3.so: cannot open shared object file:". Even though Perl 5's SQLite DBD works fine without it. Do they just bundle all the SQLite files with the distro? 20:01
flussence oh, hm. I guess a C int is 32-bit on 64-bit OSes then... I don't know the first thing about ABIs :)
RabidGravy Zoffix, yes 20:02
Skarsnik timotimo, good catch for missing check on cglobals, it should be just a matter of moving the validnctype outside the sanity_check routine and call it
Zoffix We probably should do the same :D
Skarsnik duh why libsqlite3.so 20:03
Zoffix Why? is it wrong?
Skarsnik it should be libsqlite3.so.1
or .3
flussence it's /usr/lib64/libsqlite3.so.0.8.6 on my system...
Skarsnik Well why the version number is missing
20:03 ugexe left
stmuk_ doc.perl6.org/language/faq#Nativeca...oo.so.1.2! 20:04
Zoffix Skarsnik, it probably just the error message. Works fine if I install the lib
Skarsnik Zoffix, did you set PERL6SQLITE or something?
Zoffix nope 20:05
Skarsnik Are you using an old DBIish x)
Zoffix Perhaps
How to find out what version I have?
Skarsnik constant LIB = %*ENV<DBIISH_SQLITE_LIB> || ('sqlite3', 0); 20:06
I should have bumped the version maybe
20:07 llfourn joined
Zoffix Ala Perl 5's perl -MFoo -E 'say Foo->VERSION' 20:07
20:07 _nadim left 20:08 _nadim joined
lucasb kjk: I agree that the return values from <<...>> is not very intuitive. If you are looking for a workaround, you can type << "$k" >> to get a Str 20:08
Skarsnik I am not sure how version work x) 20:09
20:09 tipdbmp left
Skarsnik I mean for panda/modules 20:10
20:10 tipdbmp joined
Skarsnik especially since DBIish has version everywhere 20:10
20:10 kjk_ joined 20:11 Tonik left, llfourn left
Zoffix notes the conversation above 20:12
Is SQLite stuff with DBIish "thread safe"? 20:13
autarch hmm, how to check if all members of a junction are defined?
m: all( Int, 42 ).defined; all( Int, 42 ) ~~ .defined;
camelia ( no output )
autarch m: say all( Int, 42 ).defined; say all( Int, 42 ) ~~ .defined;
camelia rakudo-moar 4bb47d: OUTPUT«True␤True␤»
Skarsnik stmuk, this faq entry should be fixed x)
flussence Zoffix: quoth the sqlite docs: “Threads are evil. Avoid them.␤␤SQLite is threadsafe. We make this concession since many users choose to ignore the advice given in the previous paragraph.” :) 20:14
Zoffix haha
20:15 darutoko left
Rotwang btw. does perl6 have GIL? 20:15
gfldex Rotwang: no
flussence there's also a «int sqlite3_threadsafe(void);» that can be used to check at runtime, but compiling it without safety isn't the default 20:16
20:16 sjoshi left, vendethiel joined
Skarsnik I need to use := to write ($a, $b) = foo() ? 20:19
kjk lucasb: thanks! Didn't know you can do that.
flussence Zoffix: aaaaand... there's a flag you can pass to sqlite3_open_v2 that tells it whether to use a threadsafe mode that allows sharing one DB connection between threads or not.
RabidGravy Skarsnik, or ($a, $b) = foo().list
Zoffix m: my ($x, $y ) = 42, 42; say so $x&$y .defined # autarch 20:20
camelia rakudo-moar 4bb47d: OUTPUT«True␤»
Zoffix m: my ($x, $y ) = 42; say so $x&$y .defined # autarch
camelia rakudo-moar 4bb47d: OUTPUT«False␤»
autarch Rotwang: my understanding is that threading is pretty core to how Perl 6 works (or is defined to work)
Zoffix Not sure why all() doesn't work
20:20 domidumont left
Zoffix And I may be wrong too. 20:20
m: my ($x, $y ) = 42, 0; say so $x&$y .defined
camelia rakudo-moar 4bb47d: OUTPUT«True␤»
Skarsnik my anwser for sqlite thread : Patch are welcome 20:21
20:21 kjk_ left
autarch Zoffix: I'm not sure that parses the way you think that parses ;) 20:21
Skarsnik and does @_ still exist? x) 20:28
20:28 ghost_ joined, kjk left
Zoffix m: my ($x, $y ) = 0, 42; say so $x&$y .defined 20:28
camelia rakudo-moar 4bb47d: OUTPUT«False␤»
Zoffix autarch, ah :(
Skarsnik du I can just write *@tab now
autarch Zoffix: I'm just guessing, but I wonder if that's "$x & ($y.defined)"
Zoffix yes
20:28 cdg joined
Zoffix Well, grep works 20:28
m: my ($x, $y ) = 42, 42; say so all($x, $y).grep: *.defined
camelia rakudo-moar 4bb47d: OUTPUT«True␤»
Zoffix IRC-- # waste of time
s/waste of time/distraction/;
20:28 Zoffix left
Juerd Er... 20:28
mst resists Zoffix-- # part of the problem
Juerd Please don't resist removing the k-line and apologising :)
If that part message is true, that is.
mst collapses giggling
Rotwang autarch: NativeCall not working in promises is pretty hughe drawback, so hopefully NativeCall people look at the perl rt
mst Zoffix++ # score!
Juerd Only now I realise that k-lines would result in quits, not parts :)
In "multi method elems (::?CLASS:D:)", what is "::?"?
flussence I *think* it's a weird/only way of writing sigil-less "$?CLASS"
lucasb I think ::?CLASS is the current class you are in
autarch Rotwang: I'm sure people are reading RT
lucasb ah, I didn't know $?CLASS existed
Juerd Wow, it's fugly :) 20:29
But thanks or explaining. Am I correct in thinking '::' is a sigil and '?' is a twigil
?
lucasb isn't that the "Don't repeat yourself" version of having to write (Foo:D:) in every method?
flussence :: is... not far off what it means in perl 5, ? is a twigil yes 20:30
RabidGravy Rotwang, autarch github.com/jonathanstowe/Audio-Lib...ut.pm#L630 works fine - just need a thread safe library and/pr methods
or
20:31 sprocket left
Rotwang RabidGravy: ? 20:31
20:32 CIAvash left
autarch Rotwang: well, you were trying getenv and gettimeofday which I'd expect to be thread safe 20:32
RabidGravy it's a nativecall thing using a promise, I've streamed hours of music through it
Rotwang autarch: yep
20:33 BenGoldberg left
autarch err, I responded to the wrong person 20:33
RabidGravy: rotwang seemed to find some issues just calling very simple glibc subs in promises
Rotwang RabidGravy: it doesn't pop up always, for example when I've added debug prints it dissapeared
20:33 geraud joined
Rotwang RabidGravy: rt.perl.org/Public/Bug/Display.html?id=127138 20:34
20:41 pierre-vigier left
Juerd print is still pretty slow (e.g. [*] 1..10000 takes 0.3 seconds, but printing it takes another 1.5 seconds). Has anyone profiled this yet? 20:42
timotimo have not yet
20:42 ugexe joined
timotimo could be it's just slow to stringify a big int via libtommath 20:42
Juerd m: my $x = ~ [*] 1..10000; say now - INIT now; 20:43
camelia rakudo-moar 4bb47d: OUTPUT«1.46464939␤»
Juerd Yeah, could be :)
timotimo m: my $x = ~ [*] 1..10000; say $x.chars
camelia rakudo-moar 4bb47d: OUTPUT«35660␤»
timotimo m: print "1" xx 35660;
camelia rakudo-moar 4bb47d: OUTPUT«1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 …»
timotimo m: print "1" x 35660; 20:44
camelia rakudo-moar 4bb47d: OUTPUT«1111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111111…»
timotimo ... it's kind of hard to get the timing in there, too :P
RabidGravy pubs.opengroup.org/onlinepubs/00969...g_02_09_01 - non "thead safe" libc funcs (getimeofday not one)
autarch Rotwang: looking at that list I note that getenv is _not_ thread safe 20:47
20:47 regreg left
dalee what is this colon doing in the push method of hashes? my %a; %a.push: (b => 1); 20:48
RabidGravy I'm quite willing to accept that NativeCall does expose you to threading issues but I would be surprised otherwsise 20:49
lucasb dalee: if you use the colon, you don't need to call method with parentheses 20:50
20:51 nige1 left
dalee lucasb thx 20:51
lucasb m: say ({ 42 } while $++ < 3)
camelia rakudo-moar 4bb47d: OUTPUT«(-> ;; $_? is raw { #`(Block|58687808) ... } -> ;; $_? is raw { #`(Block|58687880) ... } -> ;; $_? is raw { #`(Block|58687952) ... })␤»
lucasb FROGGS++ removed the error, but I think the result is not right 20:52
m: say ({ 42 } if True)
camelia rakudo-moar 4bb47d: OUTPUT«42␤»
dalek c: 1aacfb0 | lizmat++ | doc/Language/glossary.pod:
First part of S99 -> Glossary migration

With thanks to all of the original authors of S99
20:53 pierre-vigier joined, TEttinger joined
dalee it seems like the colon is doing something else as well though, my %a = (b => 1); %a.push(c => 2); say %a; outputs b => 1 20:54
lucasb the 'c => 2' is a named parameter, which is silently discarded :( 20:55
masak m: repeat while my $x == 0 { $x = get().chars }
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Qxy2Ko45LE␤Preceding context expects a term, but found infix = instead␤at /tmp/Qxy2Ko45LE:1␤------> 3repeat while my $x ==7⏏5 0 { $x = get().chars }␤»
lucasb you would have to right %a.push((c => 2)) or %a.push('c' => 2)
masak TimToady: was surprised that ^^ gave that error message today 20:56
I mean, I can see how it happens, I just think it shouldn't :)
lucasb *to write
20:56 pierre-vigier left, pierre-vigier joined 20:57 deepika left
autarch masak: I thought that'd written as ... 20:58
m: while get().chars -> $x { }
camelia ( no output )
autarch or something like that
masak autarch: `repeat while` evaluates condition after the loop; `while` before 20:59
so they're different
autarch m: repeat while get().chars -> $x { }
camelia ( no output )
FROGGS masak: the fix to 'my $x == 0' might be very simple...
autarch it seems weird to put a "my" on the LHS of a comparison - it makes sense to me that that would be an error 21:00
21:00 _mg_ left
lucasb m: say my $ != 42 21:00
camelia rakudo-moar 4bb47d: OUTPUT«Use of uninitialized value of type Any in numeric context in block <unit> at /tmp/ygozVHeNzl line 1␤True␤»
lucasb m: say my $ == 42
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/KtyB5nmtEx␤Preceding context expects a term, but found infix = instead␤at /tmp/KtyB5nmtEx:1␤------> 3say my $ ==7⏏5 42␤»
masak autarch: then I guess we just have to disagree on that ;)
autarch heh
masak autarch: it makes perfect sense to me, at least if it's an "item context" `my`
dalee is there anywhere that this colon usage is documented? i don't seem to be able to find any mention of it anywhere 21:03
lucasb so... same thing happens for all operators that start with "=", like ==, ===, =:=, => 21:04
RabidGravy in the folk memory of the perl 6 people
21:04 aenaxi joined
lizmat doc.perl6.org/language/glossary#Col...Colon_List # dalee 21:05
doc.perl6.org/language/glossary#Adverb # also dalee
21:06 mempko_ left
dalee lizmat thank you 21:06
awwaiid lizmat: I think dalee is looking for the : in @foo.map: ...
lizmat ah, that one
awwaiid oh wait, maybe dalee is looking for all the colons
lucasb yeah, colon is overloaded in P6 for lots of uses :)
:foo <-- adverb notation for pairs 21:07
awwaiid the one in map:, I can't find a good spot for it. doc.perl6.org/language/syntax#Subroutine_calls should mention it at least but doesn't
lucasb obj.meth: 1,2,3 <-- precedence drop
foo(obj: 1,2,3) <-- invocant marker
awwaiid what's an invocant marker?
dalee hmm ya, i'm not actually sure any of those docs are related to what's being done with the hash push method 21:08
lucasb meth obj: 1,2,3 is the same as obj.meth(1,2,3)
awwaiid oh, is that like meth(obj: 1,2,3)
ahh
21:08 llfourn joined
vendethiel but not the same as meth (obj): 1,2,3 21:08
awwaiid that's a new one!
(for me)
lizmat dalee: that would be the invocant marker 21:09
21:09 SwellJoe left
dalee ah design.perl6.org/S03.html#Invocant_marker 21:09
lizmat yeah, that didn't make it to the glossary or other docs yet, it seems 21:10
working on that
nine ugexe: you here? 21:11
awwaiid I'm adding it to the list of invocation methods in syntax (without elaboration)
ugexe nine: yea
yoleaux 11:02Z <nine> ugexe: I left a comment on your PR. Can you split the commit? I'd gladly merge the second part. Thanks for debugging this!
21:13 llfourn left, RUR joined
lucasb m: my %h; %h.push: 'a'=>1,(b=>2),c=>3; say %h # just an example 21:14
camelia rakudo-moar 4bb47d: OUTPUT«a => 1, b => 2␤»
nine ugexe: I wonder if we should mkdir in Repository::Installation at all. The perl, vendor and site repositories are created fully by make install. No mkdir should be necessary for them. For custom repos we could offer a setup method that creates all directories. This method could also be used by make install to centralize the handling.
Ulti someone appears to be interested in writing some Matrix fu for Perl 6 but is being scuppered by: Partially dimensioned views of arrays not yet implemented. Sorry. 21:15
nine ugexe: separating these parts may help with ownership/permission problems
Ulti on a scale of 1 to 10 how hard would it be fore someone who knows a bit of Perl 6, and a bit of C to actually get that done?
dalek c: a55ebdd | (Brock Wilcox)++ | doc/Language/syntax.pod:
Add some examples of method invocation
 21:16
Ulti m: my num64 @matrix[10;10]; say @matrix[1];
camelia rakudo-moar 4bb47d: OUTPUT«Partially dimensioned views of arrays not yet implemented. Sorry. ␤ in block <unit> at /tmp/IKoE9xn69p line 1␤␤»
Ulti m: my num64 @matrix[10;10]; say @matrix[1;*];
camelia rakudo-moar 4bb47d: OUTPUT«Partially dimensioned views of arrays not yet implemented. Sorry. ␤ in block <unit> at /tmp/TC5FCGk3aL line 1␤␤»
Ulti m: my num64 @matrix[10;10]; say @matrix[1;0..9]; #etc 21:17
camelia rakudo-moar 4bb47d: OUTPUT«Partially dimensioned views of arrays not yet implemented. Sorry. ␤ in block <unit> at /tmp/Jugv1cIyfV line 1␤␤»
ugexe nine: maybe a user is not supposed to do this, but deleting `/site` to uninstall all modules (however inappropriate) is the only reason for the mkdir. if a user should not be allowed to do that and expect their perl6 to still work then it is indeed not neccesary
Ulti blogs.perl.org/users/pierre_vigier/...perl6.html
21:18 hankache left
nine ugexe: a real uninstall is not that far away. I guess it would be some 2 or 3 hours of light coding. All the information needed for that is already there. 21:18
ugexe yeah, but to uninstall say 100 modules
nine ugexe: when I really have to remove modules, I rm -rf install/share/perl6 && make install. Takes just a couple of seconds longer than removing site 21:19
lucasb awwaiid: lol, I think 'precedence drop' is just a loose term to mean that the method call will work as a list operator :) (someone needs to correct me :)
ugexe so if there is an uninstall candidate to quickly delete all modules in a CompUnit without it having to process each part individually it would solve that 21:20
compunit::repository rather
21:20 RUR left
nine ugexe: just to explain: I've become very cautious with changes to Installation because ShimmerFairy was right in one thing: the curli branch was a rush job. I really want to get it right this time as we now actually have the time. 21:20
A nuke method would be even easier than a uninstall :) Just remove all files in the repo and leave the directories. That's a couple of minutes of coding time. 21:21
It's the API design that's really taking time. Because whatever we add now, we'll have to support it indefinitely. 21:22
Skarsnik I think I will never get how to affect attribute in the constructor 21:23
nine In any case, I'm really appreciating your help wit hthis :)
jdv79 literally indefinitely?
lucasb while you are talking about curli, can I suggest again to make the dirs perl, site and vendor at the same level. I think this would make the directory structure layout much cleaner, instead of dirs site and vendor getting nested under the perl directory.
nine lucasb: I would have loved that pre-christmas. I should just have done it and with all the nuking, no one would have noticed. Now it's much harder because we do have backwards compatibility. 21:25
Skarsnik How I access attribute in new?
nine Skarsnik: you can't. You do that in BUILD.
moritz you don't, because the object hasn't been created yet
lucasb nine: aww :( ok, thanks for listening :)
moritz doc.perl6.org/language/objects#Obje...nstruction
21:25 pierre-v_ joined
jdv79 wonder how much can be deprecated in future lang versions 21:25
nine jdv79: well, yes. What we add now will be part of 6.d and will have to be supported forever. 21:26
RabidGravy unless of course you have a multi new and call another one from yours
jdv79 where forever may be just til a few versions forward? 21:27
otherwise spaghetti code:( 21:28
21:28 pierre-vigier left
mst there's ways to maintain compat without things going quite that badly wrong 21:29
ugexe oh i know, i've spent too many hours discussing the api. its why ive been so critical of it when it was merged in.
Skarsnik moritz, I want my new to take two arguments but in new I can't init stuff, and if I just define a BUILD I get new only accept named parameter
mst eh. we were always going to hate the first attempt after getting it some real world use
all of this has happened before; all of this will happen again
21:29 Zoffix joined
moritz Skarsnik: only accepting named parameters is great for composabity 21:29
Zoffix I'm having trouble with this:
m: my @z = <foo bar baz>; say "DELETE FROM factoids WHERE id IN({join ',', ('?',) x @z})" 21:30
camelia rakudo-moar 4bb47d: OUTPUT«DELETE FROM factoids WHERE id IN(???)␤»
21:30 kaare_ left
Zoffix why isn't it joining those '?' with commas? 21:30
moritz Skarsnik: though if you really want it, you can write a new method that translate a positional into a named, and BUILD that does more advanced magic
lucasb xx, not x :)
moritz Zoffix: use xx
Zoffix: x is string repetition, xx is list repetition
Zoffix: and no need for the akward trailing comma :-)
m: my @z = <foo bar baz>; say "DELETE FROM factoids WHERE id IN({join ',', '?' xx @z})" 21:31
camelia rakudo-moar 4bb47d: OUTPUT«DELETE FROM factoids WHERE id IN(?,?,?)␤»
nine jdv79: personally I would really like to aim for Linux kernel backwards compatibility rather than Python style. Because that's one thing, Perl 5 really got right and what makes a great business case. It also shows that we are caring about our users more than ourselves.
moritz m: my @z = <foo bar baz>; say "DELETE FROM factoids WHERE id IN({join ',', @z.map({'?'})})"
camelia rakudo-moar 4bb47d: OUTPUT«DELETE FROM factoids WHERE id IN(?,?,?)␤»
Zoffix Thanks. Out of the gazillion ways I've tried, I guess one escaped :)
nine ugexe: I would have loved to be part of those discussions :/
Skarsnik moritz, I need to add an ugly attribute just for that?
moritz Skarsnik: no
Skarsnik: you can have non-attribute named params in BUILD 21:32
21:32 lucasb left
moritz m: class A { has $.len; submethod BUILD(:$x, :$y) { $!len = $x + $y } }; say A.new(:x(5), :y(8)).lenz 21:32
camelia rakudo-moar 4bb47d: OUTPUT«Method 'lenz' not found for invocant of class 'A'␤ in block <unit> at /tmp/GzH5B2OAZK line 1␤␤»
moritz m: class A { has $.len; submethod BUILD(:$x, :$y) { $!len = $x + $y } }; say A.new(:x(5), :y(8)).len
camelia rakudo-moar 4bb47d: OUTPUT«13␤»
21:32 pdcawley left
dalek kudo/nom: 2aee270 | (Stefan Seifert)++ | src/core/CompUnit/PrecompilationRepository.pm:
Fix confusing debug information
 21:32
Skarsnik Oooh self.bless call build 21:34
RabidGravy yeah, I think it calls CREATE which calls BUILDALL which calls BUILD or something like that 21:35
moritz RabidGravy: bless calls CREATE (which creates a new, empty instance), and calls BUILDALL on it, which walks the chain of BUILD submethods from the root of the inheritance hierarchy to the bottom 21:36
RabidGravy :) 21:37
Skarsnik m: use Test; say Test::EXPORT::DEFAULT::.keys;
camelia rakudo-moar 4bb47d: OUTPUT«(&isnt &plan &pass &cmp-ok &flunk &does-ok &subtest &unlike &like &use-ok &todo &skip-rest &eval-dies-ok &is-deeply &throws-like &ok &is &diag &done-testing &is-approx &skip &dies-ok &lives-ok &eval-lives-ok &MONKEY-SEE-NO-EVAL &nok &is_approx &isa-ok &can…»
hoelzro o/ #perl6 21:39
Skarsnik hm should is_approx be removed? (DEPRECATED maybe?
hoelzro nine: regarding that BEGIN say 'hi' RT ticket I filed the other day, do you mind if I file a new one regarding the error message? it's a little LTA
Skarsnik Oh can-ok only take one method name? :( 21:40
hoelzro btw, is it just accessing $*OUT that causes the issue, or is it the output to standard output that corrupts compilation output?
Zoffix How can I have this named param optional?
m: sub foo (:$x where Dateish|Instant) {}; foo
camelia rakudo-moar 4bb47d: OUTPUT«Constraint type check failed for parameter '$x'␤ in sub foo at /tmp/VA5zUqXY1V line 1␤ in block <unit> at /tmp/VA5zUqXY1V line 1␤␤»
Skarsnik Oh 21:41
it get False
m: sub foo (:?$x where Dateish|Instant) {}; foo
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tq6vLJiuwr␤Missing block␤at /tmp/tq6vLJiuwr:1␤------> 3sub foo (:7⏏5?$x where Dateish|Instant) {}; foo␤»
Skarsnik m: sub foo (:$x? where Dateish|Instant) {}; foo
camelia rakudo-moar 4bb47d: OUTPUT«Constraint type check failed for parameter '$x'␤ in sub foo at /tmp/Pla4ZQOtQy line 1␤ in block <unit> at /tmp/Pla4ZQOtQy line 1␤␤»
dalek kudo/nom: 3734ea8 | (Stefan Seifert)++ | src/core/CompUnit/Repository/Installation.pm:
Fix file ids changing on reinstallation of a module

Precompiled modules are checked against their source files on loading to detect outdated precomp files. This is done by recording the source file's id. Thus this id must be stable and survive reinstallation of a module. Otherwise the precompilation file's modification time will be compared to an outdated source file.
Note that this also means that the source file's path must be stable. A patch to panda will fix those paths.		 21:42
Zoffix m: sub foo (:$x) {}; foo; # being optional is the default
camelia ( no output )
lizmat nine++
Skarsnik Zoffix, it just set to False I think 21:43
m: sub foo (:$x) {say $x.WHAT}; foo
camelia rakudo-moar 4bb47d: OUTPUT«(Any)␤»
Skarsnik or not
nine This was one tough nut to crack. Especially as the full fix is three changes to rakudo and panda.
Skarsnik Zoffix, look like a bug to me 21:44
Zoffix great 21:45
m: sub foo (:$x where Any|Dateish|Instant) {}; foo;
camelia ( no output )
RabidGravy doesn't to me
Zoffix RabidGravy, why?
lizmat nine: so, if I now pull rakudo and panda, how could I make sure I install new modules with the latest panda without having to do a bootstrap, which would lose all my installed modules / 21:46
?
nine lizmat: bootstrap doesn't loose modules
lizmat ahhhh
Skarsnik m: sub foo (:$x where Mu|Dateish|Instant) {}; foo;
camelia ( no output )
lizmat cool, nine++
nine boostrap.pl just installs panda
RabidGravy because it's always going to be an Any when it's not supplied
lizmat nine: ah, yes, I'm confusing with the old rebootstrap.pl :-) *phew* 21:47
Zoffix RabidGravy, but an Any can be supplied, which kinda puts a flaw into constraints
m: sub foo (:$x where Any|Dateish|Instant) {}; my $x; foo :x($x)
camelia ( no output )
21:47 hoelzro left
Skarsnik m: sub foo (:$x where Mu|Dateish|Instant) {say "foo"}; foo; foo(now); foo("Str"); 21:47
camelia rakudo-moar 4bb47d: OUTPUT«foo␤Too many positionals passed; expected 0 arguments but got 1␤ in sub foo at /tmp/bgulmTxHnQ line 1␤ in block <unit> at /tmp/bgulmTxHnQ line 1␤␤»
Zoffix ^ that should cry that $x is invalid
21:47 hoelzro joined
Zoffix Well, I mean in real code 21:47
Skarsnik m: sub foo (:$x where Mu|Dateish|Instant) {say "foo"}; foo; foo(:x(now)); foo(:x("Str")); 21:48
camelia rakudo-moar 4bb47d: OUTPUT«foo␤foo␤foo␤»
Skarsnik that not good x)
nine Oh it seems like the second fix to rakudo is not even needed.
RabidGravy m: subset Foo of Mu where Dateish|Instant; sub foo(Foo :$m) {}; foo;
camelia rakudo-moar 4bb47d: OUTPUT«Constraint type check failed for parameter '$m'␤ in sub foo at /tmp/VCUtN21uDE line 1␤ in block <unit> at /tmp/VCUtN21uDE line 1␤␤»
dalek kudo/nom: 7d4cdbc | lizmat++ | src/core/ (2 files):
Fix for RT #127135
RabidGravy m: subset Foo of Mu where * ~~ Dateish|Instant; sub foo(Foo :$m) {}; foo;
camelia rakudo-moar 4bb47d: OUTPUT«Constraint type check failed for parameter '$m'␤ in sub foo at /tmp/_BymDH4dOm line 1␤ in block <unit> at /tmp/_BymDH4dOm line 1␤␤»
Skarsnik m: sub foo (:$x where $x.defined and Dateish|Instant) {say "foo"}; foo; foo(:x(now)); foo(:x("Str")); 21:49
camelia rakudo-moar 4bb47d: OUTPUT«5===SORRY!5=== Error while compiling /tmp/jHz2TqlkW7␤Malformed parameter␤at /tmp/jHz2TqlkW7:1␤------> 3sub foo (:$x where $x.defined7⏏5 and Dateish|Instant) {say "foo"}; foo; ␤ expecting any of:␤ constraint␤ infix␤ …»
Zoffix needs curlies
m: sub foo (:$x where {not($_) or ($_ and Dateish|Instant)}) {}; foo :x(now)
nine hoelzro: I would not mind you reopening the ticket either. But if you do, I'd really appreciate a suggestion for a fix, because I have no idea :)
camelia rakudo-moar 4bb47d: OUTPUT«Constraint type check failed for parameter '$x'␤ in sub foo at /tmp/5EneOgz2xh line 1␤ in block <unit> at /tmp/5EneOgz2xh line 1␤␤»
21:50 firstdayonthejob joined
lizmat just finds out there are *no* spectests for Str.subst-mutate 21:50
Skarsnik Zoffix, for me it's a bug, there should not be weird workaround x)
Zoffix :o
Skarsnik, agreed
Zoffix rakudobugs it
nine hoelzro: it's really just that I write dependency ids to STDOUT and read them back in in the calling process 21:51
21:51 brrt joined
brrt hi #perl6 21:54
Zoffix \o 21:55
21:56 yurivish_ joined
Zoffix Done: rt.perl.org/Ticket/Display.html?id=127142 21:57
Skarsnik m: say " "x5; 21:58
camelia rakudo-moar 3734ea: OUTPUT«5===SORRY!5=== Error while compiling /tmp/6esDgbkTOM␤Two terms in a row␤at /tmp/6esDgbkTOM:1␤------> 3say " "7⏏5x5;␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statemen…»
Skarsnik m: say " "x 5;
camelia rakudo-moar 3734ea: OUTPUT« ␤»
Zoffix m: say (" "x5)
camelia rakudo-moar 3734ea: OUTPUT«5===SORRY!5=== Error while compiling /tmp/7_Dx18gB6P␤Two terms in a row␤at /tmp/7_Dx18gB6P:1␤------> 3say (" "7⏏5x5)␤ expecting any of:␤ infix␤ infix stopper␤ statement end␤ statement modifier␤ …»
RabidGravy m: subset Foo of Mu where { !$_.defined ||$_ ~~ Dateish|Instant }; sub foo(Foo :$m) {}; foo 21:59
camelia ( no output )
Skarsnik How the hell perl5 accept that, it's so ambiquois x)
RabidGravy perl5 is "special" 22:00
Zoffix Skarsnik, what's the ambiguity?
m: subset Foo of Mu where { !$_.defined ||$_ ~~ Dateish|Instant }; sub foo(Foo :$m) {}; foo :m(now)
Skarsnik x5 look like a sub name
camelia ( no output )
Zoffix Skarsnik, but it's in an invalid place
dalek c: 183d8fb | (Dale Evans)++ | doc/Type/Hash.pod:
clarifying push usage

added examples of the other two forms of calling, and documented some non-obvious behaviour of what perl5 programmers might expect from fat commas		 22:02
c: 25bdb05 | moritz++ | doc/Type/Hash.pod:
Merge pull request #313 from daleevans/patch-1

clarifying push usage
c: b874994 | moritz++ | doc/Type/Hash.pod:
Explain why hash keys need to be quoted
 22:03
c: eafa0de | LLFourn++ | htmlify.p6:
preparing for removal of extra ""

Putting some redundant when statements to catch the new .contents format when that happens		 22:04
c: f6ac8ca | moritz++ | htmlify.p6:
Merge pull request #261 from LLFourn/master

preparing for removal of extra ""
Zoffix m: sub foo($x? where $_ < 42 ) {}; foo 22:05
camelia rakudo-moar 3734ea: OUTPUT«Use of uninitialized value of type Any in numeric context in sub foo at /tmp/562mGzNaCQ line 1␤»
Skarsnik is there something replace \002 from p5 that allow to write a number just after in a string? 22:07
Zoffix What's \002? 22:08
22:08 _nadim left
flussence \x[2]? 22:08
Skarsnik The ascii number 2
22:08 _nadim joined
Skarsnik or the raw 22:09
not even sure how to call that xD
22:09 frankjh joined, frankjh left
Zoffix \x02 22:09
hoelzro nine: ahhhh ok
RabidGravy moritz++
Zoffix m: say "\x[2]" == "\x02"
camelia rakudo-moar 3734ea: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '3⏏5\x[2]' (indicated by ⏏)␤ in block <unit> at /tmp/9TKJbeJ6qm line 1␤␤Actually thrown at:␤ in block <unit> at /tmp/9TKJbeJ6qm line 1␤␤»
22:09 llfourn joined
hoelzro I'll re-open the ticket, explain the situation, and think on a fix 22:09
Zoffix m: say "\x[2]" eq "\x02"
camelia rakudo-moar 3734ea: OUTPUT«True␤»
nine ++hoelzro 22:10
Skarsnik Zoffix, the issue with \x02 is for string like \0023 :) 22:11
22:11 nadim joined
lizmat good night, #perl6! 22:12
RabidGravy toodles
Zoffix night
22:13 brrt left
dalek kudo/nom: 8150686 | ugexe++ | src/core/CompUnit/Repository/Locally.pm:
Use an uncached IO::Path for $.prefix

method prefix fixes instances of $path.mkdir unless $path.e that pass around an IO::Path where !$path.e or !$path.w erroneously. This avoids adding stringification code throughout the rest of the codebase (or adding .IO all over)		 22:13
22:13 _nadim left 22:14 llfourn left
timotimo anything you people have been up to that ought to be mentioned on the weekly? 22:16
flussence hmmm... could mention there's packages for various OSes in the works (arch/gentoo/mac) 22:17
hoelzro timotimo: panda install $ModuleThatIndirectlyDependsOnFileFind now works without --force 22:18
nine++ # for adapting my fix to something long term-friendly
flussence I think the arch one's in best shape right now, it has panda too
Zoffix timotimo, I've been working on a factoid bot for this channel that'll be named 'huggable'. Would be done today if I hadn't run into a bunch of issues -_- Hopefully tomorrow
22:18 nadim left
Zoffix s/factoid bot/IRC::Client::Plugin::Factoid/; # but the purpose is the bot 22:19
nine timotimo: if you want to mention it, Inline::Perl5 is now 32 bit safe
22:19 nadim joined
autarch timotimo: File::LibMagic 22:19
22:19 rindolf left
autarch github.com/autarch/perl6-File-LibMagic 22:19
22:19 nadim left
hoelzro I also opened a ticket on panda to promote discussion of how to handle backwards compat, which is pending discussion =) 22:20
22:20 nadim joined
hoelzro nine: ticket re-opened, re-named, and suggestions offered 22:20
I can't speak to the quality of said suggestions, though =P 22:21
Zoffix Another caveat with comma-less named args:
m: sub SUB(:$foo, :$bar, :$baz) {say "[$foo] [$bar] [$baz]" }; my %args = :foo<foo>, :bar<bar>, :baz<baz>; SUB :foo<foo> :bar<bar> :baz<baz> :foo<newfoo>; SUB |%args :foo<newfoo>; SUB |%args, :foo<newfoo>;
camelia rakudo-moar 7d4cdb: OUTPUT«[newfoo] [bar] [baz]␤[foo] [bar] [baz]␤[newfoo] [bar] [baz]␤»
Zoffix SUB |%args, :foo<newfoo>; overrides existing arg, while SUB |%args :foo<newfoo> doesn't
RabidGravy timotimo, you might mention that we're > < far away from 500 modules in the ecosystem :)
timotimo cool 22:22
nine hoelzro: one word: Windows
hoelzro: everything would be dead simple if we only supported Unix. Then I'd just open a pipe and pass the file descriptor to the child process. 22:23
hoelzro that's usually the word I get for my suggestions =/
RabidGravy that's not what I heard ;-) 22:24
Skarsnik timotimo, type check on NC native routine, App::GPTrixie to generate NC code from C headers x)
hoelzro nine: and passing a filename via a command line option wouldn't work on Windows either?
Zoffix m: sub SUB(:$foo, :$bar, :$baz) {say "[$foo] [$bar] [$baz]" }; my %args = :foo<foo>, :bar<bar>, :baz<baz>; SUB :foo<foo> :bar<bar> :baz<baz> :foo<newfoo>; SUB |%args :foo<newfoo>; SUB |%args, :foo<newfoo>; # less verbouse example
camelia rakudo-moar 7d4cdb: OUTPUT«[newfoo] [bar] [baz]␤[foo] [bar] [baz]␤[newfoo] [bar] [baz]␤»
22:25 nzkindest left, adhoc left
Zoffix er fail 22:25
m: sub SUB(:$foo) {say "[$foo]" }; my %args = :foo<foo>; SUB :foo<foo> :foo<newfoo>; SUB |%args :foo<newfoo>; SUB |%args, :foo<newfoo>;
camelia rakudo-moar 7d4cdb: OUTPUT«[newfoo]␤[foo]␤[newfoo]␤»
timotimo Skarsnik: what does the name mean? o_O
22:25 nzkindest joined
Skarsnik nothing I did not have a clever idea for a name, so I pick up a MLP pony name xD 22:26
RabidGravy Skarsnik++ # top naming scheme 22:27
Zoffix heh
Skarsnik The Great and Powerful Trixie! 22:28
RabidGravy I'm going with this next time I can't think of a name for something
timotimo which one is trixie? 22:29
Skarsnik mlp.wikia.com/wiki/Trixie a 'vilain' character that appear in 2 episodes 22:30
RabidGravy Trixie Lulamoon 22:31
Zoffix Rakudobugged: rt.perl.org/Ticket/Display.html?id=127143 22:32
This makes it 5th ticket for today... Time to give up coding and shoot some dudes on Warframe
BTW, whoever has access, this can be closed: rt.perl.org/Ticket/Display.html?id=127118
night all 22:33
22:33 Zoffix left
RabidGravy Skarsnik, you should get this for a logotype vignette1.wikia.nocookie.net/mugen/...0526033312 22:33
nine hoelzro: that may work
Skarsnik I put her cutie mark
hoelzro nine: would you mind leaving your Windows feedback in the ticket?
nine hoelzro: tomorrow. I'm on my way to bed now. 22:34
Good night!
hoelzro 'night nine 22:35
timotimo gnite nine :)
dalek rakudo/nom: cffa688 | ugexe++ | src/core/CompUnit/Repository/ (2 files):
rakudo/nom: Bypass path attribute caching in CompUnit
rakudo/nom:
rakudo/nom: * Fixes issues where installation of modules fails on the
rakudo/nom: first run but works on subsequent tries.
22:35 dalek left 22:36 dalek joined, ChanServ sets mode: +v dalek 22:38 yqt joined
Juerd m: my uint8 $c = 255; $c++; say $c 22:39
camelia rakudo-moar 815068: OUTPUT«256␤»
Juerd What's wrong with my expectation of '0'? 22:40
And what's a nice way to get that behavior?
Skarsnik of even -1 x)
m: my int8 $c = 255; $c++; say $c;
camelia rakudo-moar 815068: OUTPUT«0␤»
22:41 pierre-v_ left
Juerd Skarsnik: Huh?! 22:41
m: my int8 $c = -5; say $c;
camelia rakudo-moar 815068: OUTPUT«-5␤»
Juerd m: my int8 $c = -5; $c++; say $c;
camelia rakudo-moar 815068: OUTPUT«-4␤»
Juerd m: my int8 $c = 255; say $c;
camelia rakudo-moar 815068: OUTPUT«-1␤»
Juerd Ah. 22:42
Skarsnik I think uint type are 'fake'
Juerd That hurts my brain :(
masak Juerd: I agree with you, a uint8 shouldn't be able to take on the value 256. 22:44
unless I'm missing something major.
Skarsnik m: say int8.REPR; uint8.REPR; 22:45
camelia rakudo-moar 815068: OUTPUT«P6int␤»
Skarsnik m: say int8.REPR; say uint8.REPR;
camelia rakudo-moar 815068: OUTPUT«P6int␤P6int␤»
Skarsnik m: say int8.REPR; say uint8.nativesize; 22:46
camelia rakudo-moar 815068: OUTPUT«P6int␤Method 'nativesize' not found for invocant of class 'uint8'␤ in block <unit> at /tmp/RUwhoEMr2U line 1␤␤»
Skarsnik m: say int8.REPR; say int8.nativesize;
camelia rakudo-moar 815068: OUTPUT«P6int␤Method 'nativesize' not found for invocant of class 'int8'␤ in block <unit> at /tmp/WZbifUOusl line 1␤␤»
Skarsnik should probably add tests for type bondary in roast?
*boundary 22:47
22:47 Tonik joined
Juerd m: my uint16 $foo = 2**16; say $foo; ++$foo; say $foo 22:48
camelia rakudo-moar 815068: OUTPUT«0␤1␤»
Juerd m: my uint16 $foo = 2**16-1; say $foo; ++$foo; say $foo
camelia rakudo-moar 815068: OUTPUT«65535␤65536␤»
Juerd Now my brain hurts even more.
m: my uint16 $foo = 2**16-1; say $foo; $foo = $foo + 1; say $foo 22:49
camelia rakudo-moar 815068: OUTPUT«65535␤0␤»
22:50 niceperl joined
flussence now *mine* does /o\ 22:53
masak auugh
22:57 vendethiel left 22:58 niceperl left
masak 'night, #perl6 22:59
Skarsnik good night masak
Juerd Good night :) 23:00
gfldex m: my uint16 $foo = 2**16-1; say $foo; ++$foo; say $foo.WHAT
camelia rakudo-moar e95c62: OUTPUT«65535␤(Int)␤»
gfldex m: my uint16 $foo = 2**16-1; say $foo.WHAT; ++$foo; say $foo.WHAT
camelia rakudo-moar e95c62: OUTPUT«(Int)␤(Int)␤»
timotimo seems like it autoboxes the Int for you when you .WHAT? 23:01
m: my uint16 $foo = 1; say $foo.VAR.WHAT
camelia rakudo-moar e95c62: OUTPUT«(IntLexRef)␤»
timotimo ^- that means we have a native int inside a lexical slot that we're refering to 23:02
Juerd I can't find where nqp::add_i is implemented in moar 23:06
timotimo need to search for add_I 23:07
it's inside the bigint ops file
the addition and a few others are implemented using macros 23:08
nadim hi, is there a way to remove a role from an object?
Juerd timotimo: Oh, from Int.pm I reckoned that add_i and add_I would be different.
timotimo oh!
yeah, they are different
add_i is most likely implemented in interp.c then 23:09
Juerd Yep. Thanks :)
timotimo because it's really just GET_REG(cur_op, 0).i64 = GET_REG(cur_op, 2).i64 + GET_REG(cur_op, 4).i64
Juerd That's a weird implementation for <64 ints :)
timotimo with sized ints, though, the code gen is generating truncate and extend ops for you 23:10
that's in nqp's QASTCompilerMAST likely
you should be able to find about that in the recent commits
as proper support for sized int lexicals and registers is rather young
Juerd This is way beyond what I comprehend. I should learn some internals stuff some day.
23:10 llfourn joined
timotimo hehe 23:11
Rotwang autarch: it totally is, look again
timotimo are you looking for something specific? is there a bug somewhere?
Rotwang besides both gettimeofaday and getenv doesn't mutate anything
autarch Rotwang: pubs.opengroup.org/onlinepubs/00969...g_02_09_01 says that getenv is not thread safe
Juerd timotimo: Well, that uint8 will hold 256 after a ++, see 30 minutes ago 23:12
timotimo oh, hmm.
Rotwang autarch: so? getenv is on the list?
23:12 cdg left
autarch Rotwang: yes 23:12
Juerd $foo = $foo + 1 works as expected, $foo++ does magic.
timotimo potentially because we have inc_i and dec_i?
(do we?)
Rotwang aaand it does not need to be threadsafe
gfldex nadim: you could reconstruct a copy of some object via MOP and leave out the methods of a given role. You will break bindings in the process and we will call you crazy.
timotimo Juerd: anyway, is there a rakudobug yet? 23:13
Juerd timotimo: I don't know. Because I don't really understand the language design regarding native ints, so I find it hard to determine whether it's a bug 23:14
timotimo totally a bug, IMO.
Juerd Thanks. Will send mail.
Skarsnik nadim, that a good question :)
m: say int8.WHAT
camelia rakudo-moar e95c62: OUTPUT«(int8)␤»
Skarsnik m: say uint8.WHAT
camelia rakudo-moar e95c62: OUTPUT«(uint8)␤»
Juerd What's the procedure regarding tests for bugs? Do I just add them to the repo and report a bug, or do I copy/paste my tests in the bug report? 23:15
Skarsnik m: say int8 $a; say $a.WHAT
camelia rakudo-moar e95c62: OUTPUT«5===SORRY!5=== Error while compiling /tmp/kW0cuPD4ZH␤Two terms in a row␤at /tmp/kW0cuPD4ZH:1␤------> 3say int87⏏5 $a; say $a.WHAT␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ …»
23:15 llfourn left
Skarsnik m: my int8 $a; say $a.WHAT 23:15
camelia rakudo-moar e95c62: OUTPUT«(Int)␤»
Rotwang autarch: this list implies that it is ok to use getenv in threads, at least that's my understanding
23:15 bpetering joined
autarch Rotwang: oh, you're right, I misread the text 23:16
no, wait, I didn't, it's saying that everything should be implemented as thread-safe _except_ the functions on that list
Skarsnik Juerd, don't forget to mention that it work for int8 x) 23:17
timotimo Juerd: if there's some code that's already copy-pastable into a test file, that'd be swell
Juerd timotimo: I can commit to roast directly but I don't know if people will be upset over the failing tests :)
Rotwang autarch: following functions1 need not be thread-safe.
The getenv() function need not be reentrant. A function that is not required to be reentrant is not required to be thread-safe 23:18
autarch: besides, even if a function isn't threadsafe it just means that you can be sure to its outcome 23:19
it doesn't mean that it will crash
Juerd Hmmm
Rotwang and crash is what happens with NativeCall
timotimo Juerd: yes, they would be upset; usually you'd mark them todo and it'd be fine
Juerd There are surprisingly few tests for native ints :)
timotimo yeah 23:20
23:20 Tonik left
Skarsnik NC test suite is rather small too x) 23:20
Juerd Any suggestion for a test file where to put these?
timotimo that's a thing i also always have a hard time with :)
Juerd I'll rakudobug it now, and write the tests tomorrow 23:21
timotimo thank you kindly :)
23:22 patrickz left
Rotwang s/can/can't/ 23:23
Juerd Bug submitted; rt.perl.org/Public/Bug/Display.html?id=127144 23:26
Good night!
timotimo Juerd: i may have just found the place where the fix ought to happen
Juerd That's great. I'll read the rest when I wake up :) 23:27
timotimo hm. no, maybe not.
RabidGravy Rotwang, I'm stll worried about this getenv. The glibc manpage for getenv says "he string pointed to by the return value of getenv() may be statically allocated, and can be modified by a subsequent call to getenv()"
sounds like a boom waiting to happen in a threaded environment 23:28
timotimo my expectation was that the reason we don't truncate properly is because inc_i and dec_i have only one register, a write 23:34
in other places we have a bunch of special handling code just for those two ops, too 23:35
23:35 FROGGS left 23:36 yurivish_ left 23:37 BenGoldberg joined
timotimo nope. i don't see the right way to go here 23:38
23:38 virtualsue left
timotimo wait .. huh, we're not even using that opcode for ++ 23:39
Skarsnik hm timotimo do you think is there a compile time way to check the type of attribute in a NC struct (that has is repr('CStruct')) ? 23:40
23:41 lucasb joined
timotimo hum ... i'm not sure you can do it without getting a bit of code into moar 23:42
if you had something like in jnthn's OO::Monitor, you could mixin something into all cstructs
Rotwang RabidGravy: autarch: we need to agree on something ;f can we agree that sleep() is thread safe?
timotimo but that'd require "cstruct foobar" instead of "class foobar is repr<CStruct>"
Rotwang sleep is defined in POSIX1.2008: pubs.opengroup.org/onlinepubs/96999...tag_16_560 23:43
and posix states that: All functions defined by this volume of POSIX.1-2008 shall be thread-safe [...]
except for those from the list we were talking about earlier
m: use NativeCall; sub csleep(uint32) is native is symbol('sleep') returns uint32 { * }; await start { csleep(5) } xx 15 23:44
camelia rakudo-moar e95c62: OUTPUT«Memory allocation failed; could not allocate 21232 bytes␤»
23:46 nadim left, nadim joined
lucasb m: my $x = 42; $x.subst-mutate(/o/,'x'); say $x.WHAT 23:47
camelia rakudo-moar e95c62: OUTPUT«(Str)␤»
lucasb ^^ should the type change even if the match was not successful? 23:48
m: my $x = 42; $x ~~ s/o/x/; say $x.WHAT 23:49
camelia rakudo-moar e95c62: OUTPUT«(Int)␤»
gfldex multi method subst-mutate( Cool:D $self is rw: $matcher, $replacement, *%named ) 23:50
lucasb type changing or not, I think both .subst-mutate and s/// should have the same outcome
gfldex no return type given, so it can do whatever it wants
RabidGravy I'm not sure sleep is a good example as it does something odd, threads or no threads
Rotwang RabidGravy: what? :D 23:51
RabidGravy I suspect it screws up the inbuilt event loop
Rotwang RabidGravy: if we cant use sleep/getenv and gadzillion other functions in NativeCall, we are screwed anyway [; 23:52
gfldex lucasb: Perl 6 is a dynamic language, if you don't tell it what type you want, you will have to deal with the fallout.
lucasb gfldex: I... I don't understand you :)
.subst-mutate *always* returns either a match object or Nil 23:53
gfldex lucasb: that is an implementation detail, not a properly defined interface and may change in the future. Because it's not a properly defined interface. 23:54
23:55 firstdayonthejob left
lucasb ok, we will have to disagree with this one :) I expect a pretty well defined behavior, whatever it is. If the behavior is always to coerce a non-string to string *even* when the match was not successful, then ok, I can live with it. 23:55
timotimo RabidGravy: unlikely that it breaks something; it should effectively be equivalent to a busy loop, no? 23:56
gfldex lucasb: subst-mutate is defined for Cool and overloaded by Str. If some other class overloads that method, a return value of Str can and most likely will be nonsense. Do not rely on return value type that is not there. 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!