»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or /msg camelia p6: ... | irclog: irc.perl6.org or colabti.org/irclogger/irclogger_log/perl6 | UTF-8 is our friend! 🦋
Set by Zoffix on 25 July 2018.
irced m: await IO::Socket::Async.connect("irc.freenode.net", 6667).then: -> $p { if $p.status { given $p.result { .print: "NICK IRS\nUSER z z z z\nJOIN #perl6\nPRIVMSG #perl6 :What are you waiting for? Kill that sleeping maniac!\n"; react { whenever .Supply { .say } } } } } 00:01
IRS What are you waiting for? Kill that sleeping maniac! 00:01
evalable6 (signal SIGHUP) :wolfe.freenode.net NOTICE * :*** Looking up your hostname...␦
:wolfe.fre…		 00:02
irced, Full output: gist.github.com/64716437e7ad893990...8ca2a312d7
tobs m: say 10.7 mod 10 00:25
evalable6 (exit code 1) Cannot resolve caller infix:<div>(Rat:D, Int:D); none of these signatures match:
(Int:D \a, Int:D \b --> Int:D)
(int $a, int $b --> int)
in block <unit> at /tmp/jifSNGIpz4 line 1
tobs wouldn't it be nice if that worked?
m: multi infix:<mod> (Rat $r, Int $i) { $r - $i * floor($r / $i) }; say 10.7 mod 10 00:26
evalable6 0.7
tobs m: multi infix:<mod> (Rat $r, Int $i) { $r - $i * floor($r / $i) }; say 10.7 mod 8
evalable6 2.7
tobs like that I guess 00:27
timotimo m: say 10.7 % 10
evalable6 0.7
timotimo tobs: mod is integer modulus, % is general modulus
like / and div
m: say 10 div 3; say 10 / 3 00:28
evalable6 3
3.333333
irced stops breathing in his sleep momentarily.
tobs thanks, timotimo. (I sometimes work in a language where mod and div are just synonyms for / and %) 00:29
timotimo no prob 00:30
irced wakes up. 00:49
irced whoa, I've always felt that fortunes had something meaningful to say to me. I just read this fortune: static from nylon underwear . so that's where it was coming from when I couldn't stay grounded when I was upgrading my pc. oh, where am I ? perl6. oops, never mind. 00:51
irced falls back to sleep.
irced hey timotimo, here's one you might not have seen... 01:13
m: [].^methods
evalable6
irced aww
p6: [].^methods
evalable6
irced aww
m: my @arr; arr.^methods; 01:15
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/iqcTqDE59Q
Undeclared routine:
arr used at line 1
timotimo in contrast to the REPL you're perhaps using, the eval bots won't output the value of the last statement
irced m: my @arr=[]; arr.^methods;
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/s0Db_gtj44
Undeclared routine:
arr used at line 1
irced m: my @arr=[]; @arr.^methods;
evalable6
timotimo so you'll need a "say" or similar
m: say [].^methods
evalable6 (iterator from-iterator from-list new STORE reification-target FLATTENABLE_LIST shape pop …
timotimo, Full output: gist.github.com/e24c06f468e73dc0be...2048dd3d7f
irced right right, thanks
timotimo YW
irced m: say now; 01:18
evalable6 Instant:1549070363.088548
irced m: say now.^methods; 01:19
evalable6 (SET-SELF new from-posix to-posix Bridge Num Rat Int narrow Date DateTime Instant pred cot…
irced, Full output: gist.github.com/96b7a414a0f6e495ee...328c261a9f
irced say now.DateTime;
evalable6 2019-02-02T01:19:13.856386Z
irced say now.Rat;
evalable6 1549070408.310907
irced say now.Rat.Nude;
timotimo it's lower cased
irced say now.Rat.nude;
evalable6 (1823255903179 1177)
irced k, thanks :-) 01:20
timotimo m: say now.Rat.Nude
evalable6 (exit code 1) No such method 'Nude' for invocant of type 'Rat'. Did you mean any of these?
Num
nude

in block <unit> at /tmp/78DvxXFE7V line 1
timotimo look it suggests other methods that do exist :)
irced ahh, eval redeems itself! 01:21
irced pets evalable6.
timotimo the "m:" in front was important :)
irced m: (1, 2, 3) <<+>> (1,2,0); 01:25
evalable6 Potential difficulties:
Useless use of <<+>> in sink context
at /tmp/aALwL38Dov:1
------> 03(1, 2, 3) 08⏏04<<+>> (1,2,0);
irced m: say (1, 2, 3) <<+>> (1,2,0); 01:26
evalable6 (2 4 3)
irced m: say [1, 2, 3] <<+>> [1,2,0];
evalable6 [2 4 3]
timotimo you want your mind blown a little bit more by hyper ops?
irced m: say < 1, 2, 3 > <<+>> < 1, 2, 0 >
evalable6 (exit code 1) Cannot convert string to number: trailing characters after number in '031⏏04,' (indicated by ⏏)
in block <unit> at /tmp/VDsIVpVtdM line 1
timotimo m: say { cookies => 9, chips => 4 } <<+>> { cookies => 1, chips => 99 } 01:27
evalable6 {chips => 103, cookies => 10}
irced tis a little brain but I'm down for it
timotimo m: say (1, (2, 3, 4), 2) <<+>> (5, (4, 4, 4), 9) 01:28
evalable6 (6 (6 7 8) 11)
irced picks up the fragments of bone and brain from the floor and attempts to reassemble his skull.
eww, brain matter
timotimo i'll be AFK for a while :) 01:29
irced okay, I'm still cleaning up from that first one
timotimo a long while 01:29
irced ok, I talk mostly to myself anywayz 01:30
irced has a insider conversation with himself.
m: say put "off" 01:34
evalable6 off
True
irced m: put "off"
evalable6 off
irced m: put 'off'
evalable6 off
irced m: put 'off\n'
evalable6 off\n
irced m: print 'off\n' 01:35
evalable6 off\n
irced m: print "off\n"
evalable6 off
irced m: say $*USER 01:47
evalable6 bisectable
irced m: say $*PROGRAM
evalable6 "/tmp/hBGeZNifdr".IO
irced m: say $*EXECUTABLE 01:48
evalable6 "/tmp/whateverable/rakudo-moar/92ebc333a61457e353151e376b8fdc080bfafdea/bin/perl6".IO
irced m: say $*CWD
evalable6 "/home/bisectable/git/whateverable".IO
irced m: say $*PERL 01:49
evalable6 Perl 6 (6.d)
irced m: say $*VM
evalable6 moar (2018.12.45.ga.9.d.02578.a)
ToddAndMargo If you will suffer a newbie question, what is the best way to test for a series of bytes in a Buf? I need to know the positino of 46 00 69 00 6C 00 65 00 56 00 65 00 72 00 73 00 69 00 6F 00 6E and read till I find a )D 02:29
0D
ToddAndMargo What do I use instead of "contains" on a type Buf? 03:13
ToddAndmargo What do I use instead of `pos` to find the position of a pattern in a Buf? 04:21
irced 286 souls . monks maybe . they have taken a vow of silence . and me, well, I'm still learning. 04:36
ToddAndmargo google is no help. Lots of hits on perl 5 04:38
irced one algorithm could be thus 04:41
given the length of the pattern and given a chosen "sub buffer length", extract a string of length sub buffer and check it for a match position
then move pattern-length -1 from the end of the extracted part and read again 04:42
rinse and repeat
i have never written such an algorithm so you may need to correct me :-)
it's costly but less expensive in terms of the logic that would have to go into a more efficient algorithm 04:43
that is, the mental logic
ToddAndmargo If I have to, I will. I was hoping our intrepit heros had a function for it. 04:44
irced they are ghosts, killed by hubris
ToddAndmargo Basically I need the pos equivalent for Buf
Or they may be doing what I should be doing: sleeping! 04:45
irced m: await IO::Socket::Async.connect("irc.freenode.net", 6667).then: -> $p { if $p.status { given $p.result { .print: "NICK hubris\nUSER z z z z\nJOIN #perl6\nPRIVMSG #perl6 :I will be the death of you ALL!\n"; react { whenever .Supply { .say } } } } } 04:47
hubris I will be the death of you ALL! 04:47
evalable6 (signal SIGHUP) :leguin.freenode.net NOTICE * :*** Looking up your hostname...␦
:leguin.f…		 04:47
irced, Full output: gist.github.com/739181632fbaf8c5b1...da18567470
irced see
irced a better algorithm would be to look for the next instance of the first character of a pattern then only extract the length of the pattern to check a match and move on if none 05:00
ToddAndmargo actaully I do beleive it is the binary equivalent of `index` I am looking for
irced naturally the counter is the position or subscript index of the bf 05:01
bf
o my gawd i forgot a letter of the alphabet, abcdefghijklmnopqrstvwxyz
crap
irced so basically you can roll your own. you might even be able to make it faster than an inhouse algorithm by parallelizing and then you would probably keep yours for future use. 05:03
then you'd write it in c/c++ and you'd be the buf'est pattern searcher 05:04
c++ probably, since c++17 has <threads> 05:05
err, that goes back to c++14 maybe 11, whatever.
but I digress.
m: 'abcdefghijklmnopqrstvwxyz'.elems 05:08
evalable6
irced say m: 'abcdefghijklmnopqrstvwxyz'.elems
m: say 'abcdefghijklmnopqrstvwxyz'.elems
evalable6 1
irced m: say 'abcdefghijklmnopqrstvwxyz'.length 05:09
evalable6 (exit code 1) No such method 'length' for invocant of type 'Str'. Did you mean any of these?
chars
codes

in block <unit> at /tmp/oTgY2echHm line 1
irced m: say 'abcdefghijklmnopqrstvwxyz'.chars
evalable6 25
irced jst as I sspected.
smallick m: say 'a'..'z' .elems
evalable6 "a"..1
irced m: say 'a'..'t'
evalable6 "a".."t"
irced m: say a..t 05:10
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/WQy4AYi6Ug
Undeclared routines:
a used at line 1
t used at line 1
irced m: .say for a..t
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/3IV_ah3k7E
Undeclared routines:
a used at line 1
t used at line 1
smallick m: say ('a'..'z').elems
evalable6 26
irced by george you got it! 05:11
irced m: say 'ferreira, you're not the one' 05:11
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/EJp_hYBc4U
Two ter…
irced, Full output: gist.github.com/1872ce421383027bf5...83c89d1a97
smallick irced: you left out 'u' 05:12
irced m: say "ferreira, you're not the one"
evalable6 ferreira, you're not the one
irced m: for a..t .say
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/G3_G6Kwfc4
Missing…
irced, Full output: gist.github.com/1cea4713209de67a5f...a72190ba5e
irced smallick, i know, i was having an identity crisis 05:13
smallick what that mean
irced i'm being serious 05:14
irced m: await IO::Socket::Async.connect("irc.freenode.net", 6667).then: -> $p { if $p.status { given $p.result { .print: "NICK SERIOUSPOLICE\nUSER z z z z\nJOIN #perl6\nPRIVMSG #perl6 :irced checks out. Moving on.\n"; react { whenever .Supply { .say } } } } } 05:15
SERIOUSPOLICE irced checks out. Moving on. 05:15
evalable6 (signal SIGHUP) :rajaniemi.freenode.net NOTICE * :*** Looking up your hostname...␦
:rajan…		 05:16
irced, Full output: gist.github.com/1ff3a653f82a7ff5fc...2607424ff8
irced smallick, this is where you oh, ha ha
this is where you say, oh, ha ha
irced pulls out his marionette strings and eyes the lifeless in the room. 05:17
m: say oh, ha ha
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/JUE7tYxNNi
Undeclared routines:
ha used at line 1
oh used at line 1
irced m: say 'oh, ha ha'
evalable6 oh, ha ha
irced m: say q{quote 'this!'} 05:26
evalable6 quote 'this!'
irced m: say q!quote 'this!'! 05:27
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/8aIb29c87D
Two ter…
irced, Full output: gist.github.com/03c48020bc55c6e0cb...d2e694a03e
irced m: put buf.^methods 05:36
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/VfTmJvDzMc
Undeclared routine:
buf used at line 1
irced m: put array.^methods
evalable6 Method object coerced to string (please use .gist or .perl to do that)
STORE BIND-POS DELE…
irced, Full output: gist.github.com/4ad761b7d00f9bb239...fcc719d7d3
irced m: my @buf = buf.new; put buf.^methods
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/2SrTpraOgJ
Undeclared routine:
buf used at line 1
irced eats a live butterfly. 05:44
irced toothpicks his teeth.
irced eyes Camelia knowingly.
I know what you're all thinking. I'm not an Andy. 05:45
irced smallick, you Russian? 05:49
smallick no
i'm indian 05:50
irced oh, you in India? 05:50
smallick yes, why
irced I hear it's rough there. do you always have to watch your back? bar your windows? 05:50
smallick what do you mean? 05:51
irced I mean I hear it's slums everywhere.
smallick not everywhere 05:54
irced i bet. do you have a president or something? in the states our guy resides in a big White House. 05:55
smallick remember, india is 6th richest country, i'm not that rich, anyway
we have rashtrapati bhavan 05:56
irced hmm, well one might say Venezuela is rich because of all its oil exports, but none of its people profit...
rashtrapati bhavan?
smallick sigh 05:56
the house (bhavan) of president (rashtrapati) 05:57
irced well, of course the rashtrapati bhavan is not the slums!
smallick its created by british, anyway
irced those Brits, can't they conquer anything! major fail in the America too. 05:58
smallick i do not think it is a topic here, so i stop 05:59
irced and they didn't build anything here.
ok.
smallick hmm
irced checks the pulse of the other room attendees, one by one.
irced records time of death.
well, I hope you're not afraid of ghosts. 06:00
smallick no 06:00
smallick is not afraid of ghosts 06:01
irced nods in approval. 06:02
irced m: put "\c[DAGGER, BLACK HEART]"; 06:18
evalable6 †🖤
irced m: put "This is { put 'heresy' }!" 06:20
evalable6 heresy
This is True!
irced m: put "This is { 'heresy'.chars }!"
evalable6 This is 6!
irced m: put "This is { 'heresy'.chars }X3!"
evalable6 This is 6X3!
irced m: put "This is { 'heresy'.chars X3 }!"
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/GvPUC6VUEN
Bogus s…
irced, Full output: gist.github.com/3f7332a954f757e6a0...ab08ce7d02
irced m: put "This is { 'heresy'.chars X 3 }!"
evalable6 This is 6 3!
irced m: put "This is {{ 'heresy'.chars} X 3 }!" 06:21
evalable6 Block object coerced to string (please use .gist or .perl to do that)
This is 3!
in block <unit> at /tmp/GfxSywGLzl line 1
irced m: put "This is {{ 'heresy'.chars} x 3 }!"
evalable6 Block object coerced to string (please use .gist or .perl to do that)
This is !
in block at /tmp/s0u8muCZvm line 1
irced m: put "This is {~{ 'heresy'.chars} x 3 }!"
evalable6 Block object coerced to string (please use .gist or .perl to do that)
This is !
in block at /tmp/OFRvwRzlVX line 1
irced m: put "This is {{~ 'heresy'.chars} x 3 }!"
evalable6 Block object coerced to string (please use .gist or .perl to do that)
This is !
in block at /tmp/iNtjvh7M7Z line 1
irced m: put "\ttab\t!" 06:22
evalable6 ␉tab␉!
Xliff m: say "/webkitgtk.org/reference/webkit2gtk...List.html" ~~ /^ 'http' 's'? '://' .+ '/' *([A-Za-Z]+) '.html' $/ 06:48
evalable6 (exit code 1) 04===SORRY!04===
Unrecognized regex metacharacter - (must be q…
Xliff, Full output: gist.github.com/7e3f37f8483e3bcfae...ddbd3bdf45
Xliff m: say "/webkitgtk.org/reference/webkit2gtk...List.html" ~~ /^ 'http' 's'? '://' .+ '/' *([A..Za..Z]+) '.html' $/
evalable6 Nil
Xliff m: say "/webkitgtk.org/reference/webkit2gtk...List.html" ~~ /^ 'http' 's'? '://' .+ '/' ([A..Za..Z]+) '.html' $/
evalable6 Nil
Xliff m: say "webkitgtk.org/reference/webkit2gtk...List.html" ~~ /^ 'http' 's'? '://' .+ '/' ([A..Za..Z]+) '.html' $/ 06:48
evalable6 Nil
Xliff m: say "webkitgtk.org/reference/webkit2gtk...List.html" ~~ /^ 'http' 's'? '://' .+? '/' ([A..Za..Z]+) '.html' $/
evalable6 Nil
smallick m: say 'example.com/path/to/some.html' ~~ /^ 'http' 's'? '://' .+? '/' (.+) '.html' $/ 06:51
evalable6 ｢example.com/path/to/some.html｣
 0 => ｢path/to/some｣
smallick m: say 'example.com/path/to/some.html' ~~ /^ 'http' 's'? '://' .+? '/' (\w+) '.html' $/ 06:53
evalable6 ｢example.com/path/to/some.html｣
 0 => ｢some｣
smallick m: say 'example.com/path/to/some.html' ~~ /^ 'http' 's'? '://' .+? '/' ([ a..z A..Z ]+) '.html' / 06:55
evalable6 Potential difficulties:
Space is not significant here; please use quotes or :s (:sigsp…
smallick, Full output: gist.github.com/342f7915c2c6f132d3...3999b4066a
smallick m: say 'a.bc/d/ef/ghi.html' ~~ /^ 'http' 's'? '://' .+? '/' ([ 'a'..'z' 'A' .. 'Z' ]+) '.html' $/ 06:59
evalable6 Nil
smallick hmm 07:00
Xliff m: "WebKitBackForwardListItem" ~~ /(<[A..Z]>+) %% <[a..z]>+/; $/[0].map( .Str ).join('') 07:17
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/sCJ5emeeUT
Missing…
Xliff, Full output: gist.github.com/0265e86e104e0c67d6...2614a400d3
Xliff m: "WebKitBackForwardListItem" ~~ /(<[A..Z]>+) %% [ <[a..z]>+ ] /; $/[0].map( .Str ).join('') 07:18
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/jEwyULtwyt
Missing…
Xliff, Full output: gist.github.com/d7adf9ee09016ff12d...d3f852faa3
Xliff m: "WebKitBackForwardListItem" ~~ /(<[A..Z]>)+ %% [ <[a..z]>+ ] /; $/[0].map( .Str ).join('') 07:19
evalable6 (exit code 1) Use of uninitialized value of type Any in string context.
Methods .^name, .p…
Xliff, Full output: gist.github.com/58f51ca778872711c5...0bdd416fe1
AlexDaniel nine: hello? :) 07:28
nine: camelia was down for a bit 07:29
it needs some help, I think :)
Geth doc: 51b7b95cdb | (JJ Merelo)++ | doc/Type/IO/CatHandle.pod6
Completed, rewritten and reflow closes #2600
 07:42
synopsebot Link: doc.perl6.org/type/IO::CatHandle
Geth doc: bac7d41032 | (JJ Merelo)++ | doc/Type/IO/ArgFiles.pod6
Improves ARGFILES with examples, refs #96
 08:34
synopsebot Link: doc.perl6.org/type/IO::ArgFiles
ufobat_ when you've forked a repo on github, is there a way to rebase it to the current version from which you had forked it? 08:48
i mean on the github website
smallick m: say 1 ~~ /1/ 09:31
evalable6 ｢1｣
DrForr Hey hey. 09:43
yoleaux 22 Nov 2018 14:45Z <tbrowder___> DrForr: Can one of your modules be used to debug Rakudo NQP Grammar.nqp?
pmurias AlexDaniel: finding stuff people are willing to mentor seems to be the problem with gsoc ideas :( 10:24
nine m: say "alive" 10:26
evalable6 alive
Xliff m: say sprintf('%1$s %1$s %3$s', 'a', 'b', 'c') 10:36
evalable6 a a c
nine AlexDaniel: thanks for the heads-up! Apparently my server needs a reboot every couple of months or the VMs lose their network connection :/ 10:46
gfldex lolibloggedalittle: gfldex.wordpress.com/2019/02/02/nq...n-threads/ 12:15
tadzik oh, interesting title :) 12:21
timotimo Xliff: i don't think you need to check for the number of general workers for Promise.start to not cause a worker to be spawned unnecessarily 12:30
the thread pool scheduler only spawns new workers if it thinks it's worth doing
tadzik gfldex: also, s/then/than/ in the title, I think :) 12:36
timotimo there's an env var that makes the TPS output what it's thinking. actually, two of them 12:46
one of them outputs measurements many many times per second, the other just outputs decisions like once per second 12:47
MasterDuke i suspect the gather/take is what makes the builtin dir slow 12:50
timotimo it's quite possible that making it a custom iterator would make it a bunch faster 12:52
tyil Andrew Shitov's talk starts in a couple moments! 12:59
timotimo oh, do you have a link to a stream?
fosdem.org/2019/schedule/streaming/ 13:00
live.fosdem.org/watch/k4201
tyil yes, that one :)
timotimo there's a timeline thingie with a knob but i can't rewind in time at all :o 13:03
tyil I cant seem to rewind the speaker irl either 13:04
timotimo damn
i wonder how much different this talk is from the previously given one 13:05
tyil I haven't seen a previous one (except the small preview at amsterdam.pm)
timotimo i don't remember where it was given previously 13:06
Xliff gfldex: Nice writeup! I need to save that one when someone tells me "nqp is not for user facing code"
timotimo "you can't rely on nqp behaving the same way in the future" - "but it's faster!" ? ;) 13:07
tyil Xliff: it's not ment for it, but you're free to use it
timotimo ("but you use nqp thoroughly in JSON::Fast" - "yes, but i'm also a core developer" ;) )
tyil meaning and usage don't have to align all the time 13:08
timotimo alex needs to remove the s from <statements>*
tyil andrew?
timotimo andrew*
thank you
tyil :p
timotimo i think i'm at the start of a wonderful cold
tyil for a moment there I was scared I've had his name wrong all along
timotimo nah, i'm just wrong more often than the average person 13:09
Xliff "yes, but i'm a core developer" doesn't mean much if someone is using JSON::Fast to learn perl 6. 13:15
You can't stop that. Why make people feel like they are making a mistake.
If you are "free to use it" then let people use it.
tyil because it may bite them majorly in the future
so people warn up front
Xliff You might get more nqp users, that way.
Warn them, yes. Belittle them? No.
tyil nobody's stopping them for using it after the warning
timotimo well, they are free to use it, but we won't necessarily have mercy upon their souls if they rely on something we're not intent on keeping
Xliff tyil: That's not been my experience. 13:16
timotimo i'm not sure what belittling would mean here
Xliff I've been here over 2 years and I still get that warning when I paste nqp code.
It gets tiring. 13:17
timotimo ah, you mean "did you forget to 'use nqp'" one?
Xliff LOL. no.
timotimo oh, you mean people like me complaining
Xliff Yes.
tyil I think people say it with their best intentions
timotimo i'm saying it mostly for bystanders, TBH
tyil because it'd be very annoying for people to have written hard(er) to read code, only to have it break in some release
without notice or anything 13:18
Xliff timotimo: Thanks. I will remember that.
Although it won't prevent the snark when I respond that I have been nipping on nqp's edges for years.
timotimo i guess that's fair 13:19
Xliff And still no comments on github.com/rakudo/rakudo/issues/2589
timotimo once the ecosystem gets bigger, we may not have the time/patience to also fix user code that's breaking because nqp:: stuff changes
but fixing user code that uses perl 6 code only, i.e. the stuff tested in roast, we'll take that a lot more seriously
Xliff timotimo: No, the warning about nqp should be enough to clue people in about that. 13:20
Still, where there is no fix in pure Perl6 code (ala #2589), there is no recourse but to seek out a solution by any means necessary. 13:21
And if that means nqp, so be it. Hence more snark.
tyil working code is more important than pretty code most of the time :p
timotimo the other talks in that room also seem interesting 13:25
patrickb timotimo: Are you at FOSDEM atm? 13:31
timotimo no, just looking at the live streams 13:32
patrickb ah
timotimo more exactly, i just started looking at live streams
nine Xliff: I'm curious: in what situations do you actually have to use nqp? 13:36
timotimo to get around this bug for example :)
Xliff nine: github.com/rakudo/rakudo/issues/2589 13:41
And after re-reading it, I think this bug applies to all non int/num types. 13:42
Anything that converts to a pointer, this applies to.
Shucks. Missed the Perl6 talk in K.4.201 13:43
MasterDuke timotimo: did you see yesterday where leont++ pointed out that open was dying with an X::AdHoc instead of a typed exception? how hard is it to throw typed exceptions from moar? 13:47
Xliff sena_kun: \\o
sena_kun o/
timotimo hm, i'm not sure if we do it anywhere so far 13:48
timotimo if the exception can just be created with an nqp::create plus a few bindattr calls, that should be easy. i think throwing an object, like a hash for example, would also be possible 13:49
calling a constructor and such would be much more annoying, i expect
but i haven't tried any of that yet
MasterDuke it comes from here github.com/rakudo/rakudo/blob/mast...e.pm6#L156 13:51
lucs It appears that precompilation is not occuring when I run some code; the --stagestats option shows the same, rather long, parse time for each execution of my program. 13:56
How can I tell what's going on?
timotimo try RAKUDO_MODULE_DEBUG 13:57
is that what it's called?
lucs Is that an envvar I have to set?
timotimo aye
lucs Ok, trying...
lucs Looks like that's what it's called, a lot of reading to do now :) 13:59
timotimo i hope it helps :S
btw, it's possible for modules to export a "no precompilation" into other modules; Inline::Perl5 does - or at least did at some point do - this 14:00
lucs I happen to be using that one, I'll keep that in mind. 14:01
timotimo i wonder if the module debug outputs something about that. it would probably be nice if it would 14:05
but yeah, it's a lot of text
timotimo oh no, the audio isn't working right 14:21
timotimo someone fixed the audio! 14:24
janneke just mentioned perl 6 compiler :D
Xliff m: start { sleep 0.5; say 1 }; sleep(2) 14:29
camelia 1
Xliff m: start { sleep 0.5; say 1 }; sleep(1)
camelia 1
lucs Apart from maybe expecting a rather long next compilation time, what's the worse that could happen if I delete all precompiled files in my environment (yeah, I'm thinking of doing that). 14:33
timotimo shouldn't be a problem, unless some compilations have side-effects that are not reproducible 14:33
lucs Okay, thanks 14:34
timotimo though if "all precompiled files" includes for example the core setting or the nqp compiler, things will rather not work out very well 14:35
lucs Whoops :)
Bah, I'll just rebuild everything if I hit a snag 14:36
lucs timotimo: Nothing broke. Is there anywhere I can read up to understand what's up with the precompilation, what flags and options I can set to see what's happening, etc.? 14:48
(It appears to keep recompiling some of my modules all the time.)
timotimo sorry, that module debug env var is all i know of for now 14:49
lucs Okay, thanks anyway. 14:50
timotimo: For what it's worth, I get a "aborted precompilation without failure" message when trying to compile something that wants Inline::Perl5, so what you mentioned earlier appears to apply. 14:57
I'll try to use DBIish (had some problems with it in the past) instead of Perl5's DBI, and hope for the best.
nine Right now a module that uses Inline::Perl5 cannot be precompiled. I'll have to somehiw recreate the Perl 5 interpreter's state on deserialization for that to be fixed 15:10
I think that's quite possible, but not exactly trivial
lucs nine: I hope you find the inspiration and a couple of round tuits to get it done :) 15:16
Even for some of its shortcomings, Inline::Perl5 is a great and indispensable module! 15:17
nine thanks :) 15:19
tbrowder ref $*ARGFILES discussions, why not create a new dyn var called @*ARGFILES which would have ARGS that are valid files, not the handles but just their paths. it would be created during the @*ARGS creation. 16:17
hahainternet what's a 'valid file'? 16:21
leont Exactly
timotimo "but this file was *just* readable!" 16:22
hahainternet but this file is a symlink 16:23
AlexDaniel “yeah, sure, it is readable. Now.”
hahainternet ;)
timotimo "oh yeah COM4 is totally a file"
tbrowder “valid file”: an element of @*ARGS that is a path to a readable file 16:31
hahainternet what's a 'readable file'? does that mean you can seek? 16:33
or just read()?
timotimo does $*ARGFILES immediately open all files on the commandline? i hope not %) 16:34
hahainternet ultimately tbrowder the semantics inside linux don't really permit this sort of optimistic processing 16:35
for a start, filesystems can implement open and read ultimately however they like
leont Traditionally, it's "anything left in argv after argument parsing", which boils down to "anything that isn't an option"
hahainternet and FUSE may mean files are truly ephemeral
leont timotimo: it's opened on usage, as long as you edit @*ARGS before using $*ARGFILES, you're fine
tbrowder so you skeptics never passed a list of files as CLI? then you tested the file as readable or existing. and NO paths would be turned into handles.
hahainternet you then have to consider the difference between block and character inputs, as well as devices
tbrowder: the point is that this model is easily subverted and there will be someone using some system that does 16:36
for example, as i mentioned above, does ARGFILES guarantee you can seek?
tbrowder those are details that can be handled, can’t
hahainternet if it doesn't, then most of the operations you expect on 'files' won't work
tbrowder they?
hahainternet but if it does, then what do you do about ideas like /dev/stdin etc? 16:37
tbrowder how is the model easily subverted? 16:38
hahainternet provide a character device, not a file
it can be read, it cannot be seeked
provide a fuse filesystem backed 'file', it can be read, but perhaps only once
leont perl6 -e 'use v6.c; sub MAIN(*@*ARGS, :$foo) { say lines if $foo } ' --foo VERSION
That actually DWIMs
tbrowder isn’t that a user problem?
hahainternet it's a specification problem imho
there's no consistent definition that would make ARGFILES work for all
hahainternet short of '-e' tests 16:39
timotimo "50 common misconceptions programmers have about files", anyone? :)
hahainternet linux is old and messy enough that almost everything is like this 16:40
tbrowder well e tests can be done while creating @*ARGS can’t they?
hahainternet it gets frustrating as a devops guy, as a language developer i start to cry a bit
tbrowder: sure, but is it really 'ARGFILES' anymore, or 'ARGWHATEVEREXISTEDATBEGINTIME'? :p
timotimo hopefully not at BEGIN time :) 16:41
hahainternet i forget which phaser is which :)
INIT, not BEGIN i guess :p
timotimo right
hahainternet speaking of which, i have a semi theoretical question that has some minor relation to BEGIN/INIT, if anyone fancies a discussion on that front
timotimo CHECK is also still part of compile time
i can't guarantee to be a good discussion partner, but go ahead 16:42
hahainternet it's not directly related to 6 however
tbrowder and that’s what i get when i loop and test over @*ARGS, right?
hahainternet so I'm interested in the concept of 'Total Functional Programming', that is programming which has a limited form of recursion or iteration that guarantees the program will halt
hahainternet a common problem with this is that you can't, for example, implement a generic, recursive, graph traversal algorithm, as how could you possibly place an upper bound on the iteration count? 16:42
hahainternet but it strikes me that if your language has a JIT, you indeed can place an upper limit, because at run-time you can determine the graph structure before compiling the code to iterate it 16:43
i know perl6 embodies this idea of runtime phasers, and I have a nice article on 'my first JIT' somewhere
but i find it hard to sufficiently describe the problem domain i'm trying to attack 16:44
this channel a bit more about language design than most, so thought it worth a mention at least once
timotimo damn, i'm noticing i can't actually spend any attention here 16:47
since i'm in another discussion + watching a panel of fosdem 16:48
hahainternet no pressure on my behalf, i can barely comprehend what i'm trying to do atm 16:49
timotimo i can still be a rubber duck, ha 16:50
hahainternet well honestly at the moment, i think i lack the understanding of type theory well enough to formalise any of this 16:51
trying to find a good book on the axioms of set theory, so i can learn about the alternate type-centric formulations
but ultimately the goal is to see exactly how much of a 'typical' application can be written in a language which is guaranteed to halt at runtime 16:52
once you can eliminate the halting problem, you can start to make more serious guarantees about performance
tbrowder ladies and gentlemen: why can we not create @*ARGFILES when we already create $*ARGFILES which attempts to open file handles before testing for files? 17:07
leont Because $*ARGFILES already refers to @*ARGS? (or $*IN, if @*ARGS is empty) 17:09
b2gills If you want $*ARGFILES to skip files it can't open, why not change $*ARGFILES to skip files it can't open. 19:23
b2gills Or have it so that the `$*ARGFILES.lines()` iterator pass you a Failure when it comes across a file it can't open. 19:25
tbrowder i don’t want handles, i want a list of file paths 19:36
tbrowder so @*ARGFILES would be a subset of @*ARGS that pass some kind of file test 19:37
pmurias hahainternet: what does a JIT time upper bound give you? 20:33
hahainternet pmurias: a "total functional" algorithm 20:34
ie one that is guaranteed to halt
pmurias hahainternet: so how to you guarantee that you will always construct a valid "total functional" algorithm from the the graph you get as input? 20:38
hahainternet pmurias: the general concept i think is that the language will only support total functional programming, but will allow the deferring until JIT time of the actual data structure 20:39
so you'd define some static data that describes the data type, say a double linked list or something
and then you'd have some dynamic property, say the length of that list 20:40
as soon as that length is available to your program, you can JIT the algorithm
i'm really bad at explaining this it turns out
pmurias does it give you anything in this example? 20:43
if you have the length of the list you can use that to make sure a regular total functional algorithm terminates 20:44
hahainternet pmurias: well the question is more how do you write a language that can support this data becoming available at arbitrary times in 'runtime' 20:45
and how much can be compiled ahead of time 20:46
i've become a little obsessed of late in eliminating turing completeness
i want to be able to use a json parser that is proven to halt, that doesn't require static type definitions, that is optimised on machine via profiling / micro-optimising based on hardware 20:47
Geth doc: ugexe++ created pull request #2602:
CURFS and CURI basics		 20:53
uzl hello, #perl6!
hahainternet good evening 20:54
uzl Q: Are pinned issues in a repo something Github recommend to a user? Or something a member of the repo pins? 20:55
hahainternet: good evening!
hahainternet I'm not even sure I know what a pinned issue in github is! 20:56
uzl hahainternet: there was some little box on perl6/doc issues and I clicked the X to dismiss it thinking Github was recommending that to me. 20:57
But probably somebody pinned it there. 20:58
hahainternet uzl: intriguing, i'm not sure I know what that is! I'm sure if you broke anything someone will fix it :) 20:59
your bug looks good, and doc corrections are always A+
uzl hahainternet: oh, great! It wasn't something big but I didn't want to be missing around. 21:00
*messing 21:01
hahainternet: Thanks and have a great weekend ;-)!
cpan-p6 New module released to CPAN! CheckSocket (0.0.6) by 03JSTOWE 21:16
pmurias hahainternet: do you have a super concrete vision of what you want to do or just a vague set of things you want to work? 21:25
hahainternet pmurias: i wouldn't say super concrete, but i'm not trying to achieve specific things in terms of working 21:27
i'm more interested in the question "if we didn't have turing complete languages, what would we miss?" 21:28
after all, most software can be written in very functional forms
and maximum iteration limits as we've discussed before is a part of restricting iteration / recursion
so i want to know where exactly the disconnect is here, if i gave someone a simple imperative language that wasn't turing complete, are there any un-analysable scenarios? 21:29
pmurias hahainternet: tons of programms would be super awkward in a language that always terminate 21:31
hahainternet pmurias: can you give me an example?
pmurias an irc client for example
it doesn't know how many messages you will send 21:32
or a game loops of a computer game
hahainternet pmurias: that would be true if they were implemented in a naive loop, are you familiar with channels in Go? 21:33
it's quite common to simultaneously listen on an input channel, and on a 'cancel' channel
so being able to know "this program won't halt until cancel is sent, then it is guaranteed to halt" would be perfectly sufficient 21:34
it's the ambiguity i wish to eliminate
tobs I kinda get the point. In Perl 6 you don't even know if making a string will halt because of :x, :f or :c interpolation :D
hahainternet in almost all languages*
in python you don't even know if you can safely access an attribute
it's bad
tobs but these are a boundary in this case. Remove these adverbs and you know that you can always write down a string in a finite amount of time
pmurias hahainternet: I would guess you could do a lot with primitives that always terminate and with the wait for input forever parts put into a special construct 21:36
sena_kun anything that has any contact with real world already loses determinism. there is no guarantees that simple `my $a = get` will terminate in next 10 years if no-one will enter something. :) 21:37
hahainternet sena_kun: but that is a rule that can be applied
it's not ambiguous
"if the user enters a string and hits enter, the program is guaranteed to halt" 21:38
i'm fine with that :)
pmurias hahainternet: keep in mind that even if the program would terminate on a turing machine it doesn't mean it will finish before the heat death of the universe
hahainternet pmurias: the question is what 'a lot' encompasses.
pmurias: oh no doubt, but we don't even have a language for config files that allows any sort of complex manipulation and also is guaranteed to parse and halt
so even the most trivial tasks are indeterminate
tobs hahainternet: I don't know about your CS background, but there are non-Turing-complete formalisms. The most straightforward of these is called LOOP programs. 21:40
hahainternet tobs: very poor (CS background), the non-turing one I was looking at are "total functional languages"
i assume you're referring to en.wikipedia.org/wiki/LOOP_(progra..._language) ? 21:41
tobs and you can observe how you gain Turing-completeness by replacing counting-only loops by classical while loops, which are unconfusingly called WHILE programs
tobs hahainternet: yes, that 21:41
hahainternet tobs: indeed, there's a few other ones mentioned in en.wikipedia.org/wiki/Total_functi...rogramming which i've had a look through
tobs: the main question i had above was, if instead of using while loops, you deferred the determination of the maximum iteration count until runtime, and then JIT compiled the algorithm based on having that maximum count 21:42
that might be my worst explanation yet :)
pmurias hahainternet: what do you mean by JIT compiled? 21:43
hahainternet i mean the traditional meaning, but i'll try and give a more descriptive example 21:44
pmurias to machine code?
tobs hmm. Let me try: so, LOOP programs have all their loop counts known at compile time, WHILE programs have loops which decide when to terminate at runtime. This difference makes one Turing-complete, the other not. Where in the middle does your suggestion fall?
hahainternet pmurias: yes, or as low level as possible 21:45
pmurias hahainternet: I don't see why that would change anything in terms of language power
hahainternet pmurias: because as tobs highlights, at compile time LOOP programs have the iteration count known 21:45
and so you can prove they halt
WHILE programs cannot (presumably, traditional languages cannot) be proven to halt at compile time, as there's no available 'maximum iteration count'
if you don't know how many nodes a graph has, you can't set an upper limit 21:46
hahainternet but if you defer the compilation until /after/ you have retrieved the graph, you can indeed determine the node count, and therefore write LOOP equivalent code guaranteed to halt 21:46
tobs but I don't think the power of the language is influenced by the target architecture -- unless you say "my program is killed after at most 10³ steps on an x86"
tobs which I guarantee gives you an awkward language :) 21:47
hahainternet it's not so much the target architecture, as the target data 21:47
tobs or awkward set of computable functions, rather
pmurias hahainternet: so the generate LOOP code not x86 part is what you want
hahainternet oh i don't care about the specific architecture
LOOP just happens to be a pertinent example
pmurias yes you do 21:47
hahainternet i really dont, the JIT could be to perl6 code as far as i'm concerned 21:48
hahainternet the question is whether you can prove at 'write time' that can be compiled as the data becomes available 21:48
pmurias I'm not convinced generating code gives you more power than just executing it 21:49
hahainternet for example tobs in LOOP, you can prove the LOOP construct halts, despite not knowing what integer count is passed in a particular user's code 21:49
at least, it seems that way
tobs indeed 21:50
pmurias if you now that the graph has a size in a none primitive language you could just use that to bound the iteration instead of going through a code generation step
s/now/know/
hahainternet pmurias: i'm not sure what the difference is
you mean just write code that checks the size and adds a $max_iterations = $graph.nodes? 21:51
or the equivalent?
pmurias yes
hahainternet sure, that is ultimately equivalent to what i want, but the difference is that i want to do the first bit in a non-turing-complete language
hahainternet that way i get strong guarantees about its behaviour 21:52
i think i need to write a better description of what's in my head before i continue discussing it, as i am not making the goals clear at all
tobs: would you mind a PM? 21:53
tobs hahainternet: go ahead
pmurias hahainternet: I think that would help clarify stuff, I think I understand what your high level goals (powerful language that always terminates) but what seems vague how the JIT parts is of any benefit 21:55
s/how/is how/
hahainternet pmurias: understood
sena_kun "No lexical found with name '$_'" is pretty bad, right? I don't have it with `no precompilation` pragma, but it always occurs without it. 22:19
I wonder if it can be related to latest $_ changes.
sena_kun oh, am I not supposed to mix EVAL with precompilation, right? 22:27
leont precompilation has a tendency to introduce bugs around capturing atypical things, I've seen several examples of that already 22:29
pmurias sena_kun: I don't think there is any sort of taboo with mixing EVAL and precompilation 22:30
sena_kun I'm trying to dig further now, maybe I can get more details...
pmurias but there are a bunch of bugs in precompilation mode, if you run a precompiled roast a lot of things break
sena_kun: there is a t/precompileandrun in the rakudo repo so if you have something golfed you can check if there is a difference between precompiled and not-precompiled modes 22:31
sena_kun well, the thing I am trying to build is like this: I am executing a bunch of code at BEGIN time. Sometimes I can get some of my Perl 6 code from a file and EVAL it to check if that produces some results or just Nil. If it produces something based on a condition, I'm working with result, but it not, just nop. It shouldn't be impossible, right? I mean, in theory. 22:33
pmurias sena_kun: doesn't seem impossible to me 22:35
sena_kun thanks. :)
hmm, the bad thing CATCH doesn't clarifies my error at all, just `===SORRY!===; No lexical found with name '$_'` isn't too informative about line... 22:36
I guess I'll comment it out line by line...
pmurias sena_kun: when I deal with weird rakudo problems I try to golf them down to the smallest chunks of code that has the eror
* error 22:37
leont Usually when the error comes from a low level it's uninformative like that, IME
sena_kun well, I have a golfed thing, but it gives me another error, absence of REPR, but not $_.
so I suspect it may be "a bug", not "the bug". :) 22:39
MasterDuke sena_kun: --ll-exception doesn't give any more info?
sena_kun oh, it gives a pretty stacktrace, yes 22:40
and starts with
> at EVAL_0:3 (<ephemeral file>:)
I wonder if it's my code for EVAL is broken instead.
will check it now...
ok, seems like the code itself breaks it, as replacing it with `Nil` stops it from dying. 22:42
sena_kun resolved that with `if $foo eq 'foo' else` instead of `given $foo { when 'foo' }`. 22:48
damn, it's back. :S 22:49
Xliff m: say (1/10).∈v 22:56
camelia 5===SORRY!5=== Error while compiling <tmp>
Malformed postfix call
at <tmp>:1
------> 3say (1/10).7⏏5∈v
Xliff m: say (1/10).inv
camelia No such method 'inv' for invocant of type 'Rat'
in block <unit> at <tmp> line 1
Xliff ?!?!
No inverse?
tobs m: say (1/10).flip # :) 22:58
camelia 1.0
Xliff m: 10.flip.say
camelia 01
timotimo hahaha
beautiful
Xliff m: 10.Num.flip.say
camelia 01
Xliff tobs: For ints, silly! # :P
m: 10.Int.flip.say 22:59
camelia 01
Xliff just got played. Well done. 22:59
timotimo m: say Rat.new(|.nude.reverse) given 2.5
camelia 0.4
Xliff (10/1).flip.say
evalable6 01
Xliff m: say Rat.new(|.nude.reverse) given 10 23:00
camelia No such method 'nude' for invocant of type 'Int'. Did you mean any of these?
none
note

in block <unit> at <tmp> line 1
Xliff m: say Rat.new(|.nude.reverse) given 10.0
camelia 0.1
Xliff m: say Rat.new(|.nude.reverse) given 10.Num
camelia No such method 'nude' for invocant of type 'Num'. Did you mean any of these?
none
note

in block <unit> at <tmp> line 1
Xliff m: say Rat.new(|.nude.reverse) given 10.Rat
camelia 0.1
Xliff say 10.Rat.flip
evalable6 01
Xliff say 10.Rat.nude.flip 23:01
evalable6 1 01
Xliff ~:(
Elronnd I don't understand exceptions in perl6. I wrote this sprunge.us/K2teev, but it has a strange error on line 3 of "Cannot invoke this object (REPR: Null; VMNull)". If I remove the CATCH block, that works fine (it dies with my message) 23:04
Elronnd how can I make happen what I want to happen? 23:04
timotimo m: say $*ERR 23:05
camelia IO::Handle<IO::Special.new("<STDERR>")>(opened)
timotimo if you get that error, it's usually an internals bug, but you look like you're doing something not so usual 23:06
Elronnd that's unusual?
what...what is the usual way to define a custom error type, throw it, and catch it?
timotimo having a CATCH block in the body of a class maybe 23:07
not to mention
throwing the exception before it's declaration is finished
Elronnd I don't want to put anything in this exception except to give it a unique name
timotimo could it be you really ment "class FatalCError is Exception { }"? 23:08
Elronnd oh, when I change it to class FatalCError is Exception {}, it seems to work
yeah
timotimo yeah, trying to create an instance of the exception inside of it's body is the problem i'd say 23:09
m: unit class Blorb; Blorb.new.say
camelia Blorb.new
timotimo m: unit class Blorb is Exception; Blorb.new.say
camelia Unthrown Blorb with no message
timotimo m: unit class Blorb is Exception; Blorb.new.throw
camelia Died with Blorb
in block <unit> at <tmp> line 1
timotimo oh?
m: unit class Blorb is Exception; Blorb.new.throw; CATCH { .say } 23:10
camelia Cannot invoke this object (REPR: Null; VMNull)
in any at <tmp> line 1
in block <unit> at <tmp> line 1
timotimo m: unit class Blorb is Exception; Blorb.new.throw; CATCH { say "here" }
camelia Cannot invoke this object (REPR: Null; VMNull)
in any at <tmp> line 1
in block <unit> at <tmp> line 1
timotimo so the throwing is fine, but not the catching 23:11
oh also, silly me; the throw happens at run time of course, the class is already done declaring at that point?
cpan-p6 New module released to CPAN! Audio-Liquidsoap (0.0.7) by 03JSTOWE 23:29

Powered by Cro and the Raku® Programming Language.

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

Thank you!