»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend! | feather will shut down permanently on 2015-03-31
Set by jnthn on 28 February 2015.
TimToady m: say 42.fmt("%05d") 00:00
camelia rakudo-moar f5d0bc: OUTPUT«00042␤»
TimToady 'course the old $# variable is like, from Perl 3 or 4 00:01
well, actually, probably Perl 1
geekosaur whenever a2p was added, iirc
TimToady given it was put in for a2p
that would be Perl 1 then :) 00:02
jdv79 i still don't get it - i think i want elems
m: 0 .. $#{$data_tree}
camelia rakudo-moar f5d0bc: OUTPUT«5===SORRY!5=== Error while compiling /tmp/PuqTtPhl3p␤Unsupported use of $# variable; in Perl 6 please use .fmt␤at /tmp/PuqTtPhl3p:1␤------> 030 .. $#7⏏5{$data_tree}␤»
TimToady no, that's a variant of $#foo which is completely different
jdv79 what i want is $data_tree.elems, right?
TimToady I guess it doesn't pick up the {} there correctly 00:03
for that range you just wan ^$data_tree or so
*want
00:04 ssutch left
TimToady my $data_tree = <a b c d e>; say ^$data_tree 00:04
m: my $data_tree = <a b c d e>; say ^$data_tree
camelia rakudo-moar f5d0bc: OUTPUT«0..^5␤»
TimToady std: 0 .. $#{$data_tree}
camelia std 28329a7: OUTPUT«5===SORRY!5===␤Unsupported use of $# variable; in Perl 6 please use .fmt at /tmp/ScfrIKNj2L line 1:␤------> 030 .. $#7⏏5{$data_tree}␤Parse failed␤FAILED 00:00 136m␤»
jdv79 i never knew about the fmt related form - just the array usage
TimToady hmm, I guess STD gets it wrong too
$# (without brackets) was a way to set a default output format for any numeric 00:05
it's long deprecated
00:05 RabidGravy left
jdv79 ok, reading that now - removed in 5.10 it seems 00:05
anyway, thanks
TimToady so there's really no point in that error message 00:06
even if it were correct, which it isn't...
00:07 ssutch joined 00:08 Guest35307 left 00:09 ssutch left 00:11 risou is now known as risou_awy
dalek kudo/nom: bcac75c | TimToady++ | src/Perl6/Grammar.nqp:
don't bother to correct ancient $# variable
 00:12
00:13 beastd left, airdisa_ left 00:16 virtualsue_ joined 00:17 virtualsue left, virtualsue_ is now known as virtualsue 00:18 virtualsue left
dalek ast: 2389e34 | TimToady++ | S32-exceptions/misc.t:
ignore ancient $# variable
 00:18
00:22 tinyblak left 00:23 laouji joined 00:38 zakharyas joined 00:39 laouji left, laouji joined 00:44 laouji left 00:45 yqt left 00:46 bjz joined 00:47 laouji joined
[Coke] TimToady++ 00:54
00:57 zakharyas left 01:13 bjz left 01:16 chenryn joined 01:18 jack_rabbit joined 01:20 chenryn left 01:22 gfldex left 01:26 atweiden joined 01:30 kaare_ left
dalek kudo/nom: fafc5b0 | hoelzro++ | src/core/PseudoStash.pm:
Add comment to else branches in PseudoStash

The else branches correspond to the case when $!mode is STATIC_CHAIN, and having the comment makes it evident when searching the file for STATIC_CHAIN that this is when that mode takes effect		 01:32
01:42 travis-ci joined
travis-ci Rakudo build failed. Rob Hoelz 'Add comment to else branches in PseudoStash 01:42
travis-ci.org/rakudo/rakudo/builds/56043141 github.com/rakudo/rakudo/compare/b...fc5b047818
01:42 travis-ci left 01:56 bjz joined 01:57 travis-ci joined
travis-ci Rakudo build failed. Rob Hoelz 'Add comment to else branches in PseudoStash 01:57
travis-ci.org/rakudo/rakudo/builds/56043141 github.com/rakudo/rakudo/compare/b...fc5b047818
01:57 travis-ci left 02:01 Hor|zon joined 02:02 bjz left 02:07 Hor|zon left, tinyblak joined 02:11 larion left 02:12 chenryn joined 02:13 larion joined, coffee` left, tinyblak_ joined 02:16 tinyblak left 02:26 bjz joined 02:47 molaf_ left 02:51 adu joined 02:52 telex left 02:53 adu left 02:54 telex joined 03:03 molaf_ joined 03:05 noganex_ joined 03:08 noganex left
skids gist.github.com/skids/7c9089e9c941c5fc1222 # Why I've been messing around with CompUnit, by the way. 03:16
03:21 BenGoldberg left 03:24 kaleem joined 03:37 raiph left, raiph joined 03:40 lichtkind left
b2gills m: my $k = <a b c>;my $n = 2; for [X] |($k xx $n) { say "$^a $^b" } # fix for rosettacode.org/wiki/Permutations_w...ons#Perl_6 03:43
camelia rakudo-moar fafc5b: OUTPUT«a a␤a b␤a c␤b a␤b b␤b c␤c a␤c b␤c c␤»
03:51 davido_ left, Hor|zon joined 03:53 davido_ joined 03:55 laouji left 03:56 Hor|zon left, larion left
dalek kudo/tab-completion: 7aa28d2 | hoelzro++ | src/Perl6/Compiler.nqp:
Don't only complete method names
 03:59
kudo/tab-completion: bd68f7f | hoelzro++ | src/Perl6/Compiler.nqp:
Offer completions for classes from imported modules
kudo/tab-completion: 7188744 | hoelzro++ | src/Perl6/Compiler.nqp:
Add completions for imported modules
04:05 laouji joined 04:06 Sqirrel left 04:08 kaleem left 04:09 raiph left 04:12 chenryn left 04:17 simcop2387 left 04:22 Sqirrel joined 04:34 larion joined 04:56 molaf_ left 05:09 laouji_ joined 05:11 laouji left 05:35 atweiden left 05:38 chenryn joined 05:40 Hor|zon joined 05:41 skids left 05:45 Hor|zon left 05:51 yeahnoob joined 06:18 ash_gti joined 06:23 diana_olhovik_ joined 06:24 yeahnoob left 06:28 diana_olhovik_ left 06:35 ash_gti left 06:42 bjz_ joined 06:44 bjz left 06:46 simcop2387 joined 06:48 anaeem1 joined 06:58 grondilu joined 07:00 gfldex joined, SamuraiJack joined 07:05 xprime left 07:06 prime joined 07:09 diana_olhovik joined 07:20 jack_rabbit left 07:23 FROGGS joined 07:24 sirhouseboat left 07:27 fhelmberger joined 07:28 jack_rabbit joined
FROGGS o/ 07:28
07:29 Hor|zon joined 07:34 Hor|zon left 07:43 gfldex left 07:47 jack_rabbit left
TimToady b2gills: fixed, thanks 07:50
07:56 abraxxa joined 07:59 lsm01 left 08:01 lsm01 joined 08:02 Rounin joined 08:10 darutoko joined 08:15 eli-se joined 08:19 [particle]1 joined
nwc10 fwierzbicki.blogspot.com/2015/03/jy...lable.html 08:20
nwc10 is still wondering why blogspot is FAIL central without JS.
Rendering text content is hard? 08:21
08:21 [particle] left 08:28 Ven joined 08:30 chenryn left, Hor|zon joined 08:35 Hor|zon left 08:37 RabidGravy joined
jnthn morning, #perl6 08:39
08:42 [Tux] left, |Tux| left
eli-se Ven: I implemented procedures! 08:43
Ven \o/ 08:44
08:47 eli-se left
nwc10 github sometimes unhappy for everyone, or just for me? 08:47
FROGGS seems to be unhappy for me too
08:49 rindolf joined, Sir_Ragnarok left
moritz nwc10: it's being DDoSed 08:50
cdc morning #perl6
Ven doesn't really understand why one would ddos github 08:51
cdc I don't understand how | works in this code: "[Zeqv] |@s».ords", from rosettacode.org/wiki/Longest_common...fix#Perl_6
please, cuold you help me? :) 08:52
08:52 sirdancealot joined
cdc as of my understanding, it prevents @s».ords to be fully flattened 08:52
08:53 eli-se joined, Sir_Ragnarok joined 09:03 Hor|zon joined
jnthn Uh, I don't know what it's doing there either... 09:03
| means "flatten into an argument list" 09:04
09:07 larion left
cdc TimToady: could you help me to understand how | is used in "[Zeqv] |@s».ords"? 09:10
(from rosettacode.org/wiki/Longest_common...ix#Perl_6)
eli-se "[Zeqv] |@s».ords" Perl confirmed 09:11
what does the pipe mean?
jnthn "flatten into argument list", but I didn't actually know it was valid in this context...
cdc eli-se: some examples here: doc.perl6.org/type/Capture 09:12
eli-se Does it pass the elements of @s as arguments to ords? 09:13
I'm so confused. :D
cdc as of my understanding, | is used by "[Zeqv]", not by "@s».ords" 09:14
eli-se I'd avoid this code. :P 09:15
RabidGravy morning
eli-se Doesn't @ always flatten though? 09:16
Ven eli-se: no, it doesn't
eli-se Flattening rules seem arbitrary and are confusing.
Ven | is flatten, and I think it gets away with it because of the multi-dimension array
assign flattens, for flattens.
jnthn | is nothing to do with array flattening though 09:19
It's *only* about argument lists.
We should maybe find a different name for it.
cdc m: [Zeqv] ((1, 1), (1, 2))
camelia ( no output )
Ven it *flattens* it to an argument list?
cdc m: [Zeqv] |((1, 1), (1, 2))
camelia ( no output )
cdc m: say [Zeqv] |((1, 1), (1, 2))
camelia rakudo-moar fafc5b: OUTPUT«True False␤»
cdc m: say [Zeqv] ((1, 1), (1, 2)) 09:20
camelia rakudo-moar fafc5b: OUTPUT«True True True True␤»
jnthn Ven: Well, it usually shows up when you do foo(|@pos-args) or so
Ven indeed
09:20 kjs_ joined
jnthn I can see it has an effect above but it looks highly accidental to me. 09:21
Could do with a TimToady to ask... :)
But we'll have to wait some hours for that. 09:22
cdc ok :) Thanks !
sjn m: [R=>] 1..9 09:30
camelia ( no output )
sjn m: say [R=>] 1..9
camelia rakudo-moar fafc5b: OUTPUT«(((((((9 => 8) => 7) => 6) => 5) => 4) => 3) => 2) => 1␤»
sjn m: ([R=>] 1..9).gist
camelia ( no output )
jnthn .gist just formats it; need .say to get output 09:31
sjn is a bit surprised that a Pair can have a Pair as key
jnthn A Pair can have anything as its key and value, pretty much
Ven sjn: why so?
sjn Ven: maybe because of perl5 :) 09:32
Ven only strings can be keys in perl5, no?
sjn mm
jnthn And => is a fat comma in Perl 5 rather than constructing an actual Pair object, iirc 09:33
Ven that too :P 09:34
sjn mm
eli-se m: say ((<W I> Z <A R D>), <A B E A R D>) 09:35
camelia rakudo-moar fafc5b: OUTPUT«W A I R A B E A R D␤»
eli-se Poetry 6 09:38
jnthn: => in Perl 5 is the same as , except (assuming use strict) that a bareword is allowed at the left side 09:39
09:40 darutoko- joined
eli-se ideone.com/LM5vWj 09:40
09:42 darutoko left 09:43 laouji_ left 09:45 chenryn joined 09:46 larion joined 09:49 chenryn left
sjn looks more at Pairs 09:53
what's the difference between antipair and invert? 09:54
sjn has trouble seeing the point of antipair
jnthn invert is only on a Pair as a degenerate case. 09:56
See antiparis and invert on hashes to see what's going on, iirc. 09:57
10:00 chenryn joined
sjn ok, so if you want to invert a pair, and make sure it fails it fails if you're passed something else; use antipair 10:00
-it fails
eli-se antipairs annihilate with pairs, inverts don't :P
FROGGS m: say <1 2 3 4>.antipair 10:01
camelia rakudo-moar fafc5b: OUTPUT«No such method 'antipair' for invocant of type 'Parcel'␤ in block <unit> at /tmp/k7FgLGAzzV:1␤␤»
sjn don't know what that means
FROGGS m: say <1 2 3 4>.hash.antipair
camelia rakudo-moar fafc5b: OUTPUT«No such method 'antipair' for invocant of type 'Hash'␤ in block <unit> at /tmp/e1rwICLNog:1␤␤»
FROGGS w/e
jnthn antipairs
FROGGS m: say <1 2 3 4>.antipairs
camelia rakudo-moar fafc5b: OUTPUT«1 => 0 2 => 1 3 => 2 4 => 3␤»
FROGGS m: say <1 2 3 4>.hash
camelia rakudo-moar fafc5b: OUTPUT«1 => 2, 3 => 4␤»
FROGGS hmmm
is that suppose to happen? 10:02
sjn that looks very confusing
FROGGS I had expected 2 => 1, 4 => 3 for antipairs...
jnthn Looks right to be
FROGGS but I did not read the (design) docs about that
jnthn m: say <1 2 3 4>.pairs
camelia rakudo-moar fafc5b: OUTPUT«0 => 1 1 => 2 2 => 3 3 => 4␤»
FROGGS ahhh, I see 10:03
jnthn Note how .antipairs gets you what .pairs would but with key/value swapped
FROGGS yeah
makes sense
since it is not antihash :o)
jnthn m: my %h = a => 1, b => 2, c => 3; say %h.pairs; say %h.antipairs; say %h.invert;
camelia rakudo-moar fafc5b: OUTPUT«a => 1 b => 2 c => 3␤1 => a 2 => b 3 => c␤1 => a 2 => b 3 => c␤»
jnthn m: my %h = a => 1, b => 2, c => 1; say %h.pairs; say %h.antipairs; say %h.invert;
camelia rakudo-moar fafc5b: OUTPUT«a => 1 b => 2 c => 1␤1 => a 2 => b 1 => c␤1 => a 2 => b 1 => c␤»
jnthn Uh. Hmm. 10:04
I thought that's where there'd be a difference...
jnthn should apparently read the design docs too
10:05 Rounin left
sjn antipairs gives no hits on doc.perl6.org 10:06
ah, incomplete docs. never mind 10:07
10:07 espadrine joined 10:13 larion left 10:21 anaeem1 left 10:22 anaeem1_ joined, chenryn left 10:25 Ven left 10:26 Rounin joined, anaeem1_ left
dalek c: 8f2c2c2 | paultcochrane++ | lib/Type/Any.pod:
Fix typo
 10:31
masak good antenoon, #perl6 10:35
10:39 SamuraiJack left 10:40 laouji joined 10:41 [Tux] joined 10:48 eli-se left
dalek c: cf99a69 | paultcochrane++ | lib/Language/5to6.pod:
Add initial docs about given-when construct
 10:49
ast: 72db753 | jnthn++ | S10-packages/precompilation.t:
Test for RT #124162.
 10:50
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124162
jnthn (Todo'd test, fwiw)
dalek ecs: c1d2dea | paultcochrane++ | S04-control.pod:
Add missing "to" in sentence
 10:51
ecs: 3fe5f38 | paultcochrane++ | S (3 files):
Merge branch 'master' of github.com:perl6/specs
10:51 larion joined
RabidGravy I have a small query about encodings: in the p5 Term::Cap there is s/\\(\d\d\d)/pack('c',oct($1) & 0177)/eg; which I somewhat naïvely translated to s:g/\\(\d\d\d)/{ pack('C', :8($0)).decode }/ 10:52
10:52 sirdancealot left
RabidGravy (actually doing it now in the action class but the effect is the same) 10:52
10:53 chenryn joined
RabidGravy which is all well and good for some inputs, but in the big termcap file with lots of obscure historical devices there are quite a lot of \200 sequences 10:54
now this quite reasonably blows up with "Maiformed UTF-8"
jnthn I hope it's Malformed or we've a typo :) 10:56
RabidGravy so after my little brane kicked around for a bit I realised that the same effect could achieved with :8($0).chr which does work fine 10:57
masak no, it's literally shaped like the French month of Mai.
psch \o
masak o/ 10:58
RabidGravy in the merry merry month of Mai
psch is the french month of Mai the same as the german one?
masak looks like
RabidGravy anyway this works as I expect :8("200").chr.ord gives 128 etc
but it has "upgraded" the string and in the output there is a BOM which as I see it is going to break the intent of the tc entry 11:00
masak m: my $promise = Promise.new; say $promise.status; Promise::Vow.new(:$promise).keep(42); say $promise.status
camelia rakudo-moar fafc5b: OUTPUT«Planned␤Kept␤»
masak submits rakudobug
I should read source more often :>
seems you can create a new Promise::Vow just by referring to it, because it's an `our` class. 11:01
jnthn umm...wtf
I thought that was a lexical class
masak all accesses *should* go through the .vow method in Promise, but they don't have to
jnthn masak: Please try making it a lexical class and seeing if that helps things. 11:03
masak will do, just let me finish submitting the rakudobug first. 11:04
masak has a Perl 6 day today o/
jnthn: I dug down in git blame. you created it (as "Keeper") in 0358bae2394e14c7b4f42cd4e7f069a19c1bf060 -- it was 'our'-scoped already then.
nwc10 has a POETS day 11:06
RabidGravy that's really, really early nwc10 - I do hope you're going to call beer o'clock early 11:07
jnthn masak: OK...I suspect it's an oversight. I can't think of any reason I'd have wanted it to be non-lexical... 11:08
masak me either.
it feels like the pattern completes perfectly if it is a lexical.
jnthn Right.
(previous jnthn)-- # not thinking about scope
[ptc] nwc10: just so you can impress your coworkers: a POETS day translates in German roughly to: Freitag ab eins, macht jeder seins 11:17
jnthn
.oO( How is the tag frei if I have to go to work? :P )
 11:18
11:18 Akagi201_ left, Akagi201 joined
jnthn has got RT#124162 fixed on Moar HEAD o/ 11:19
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124162
psch jnthn: there's a series of books that builds on that pun (i guess it is one?)
[ptc] jnthn: :P
psch jnthn: it's about one Herr Taschenbier, that gets a visit from the Sams, after a week of things happening that fit with the name of the day
jnthn :D 11:20
psch jnthn: e.g. on thursday (Donnerstag) there's thunder
de.wikipedia.org/wiki/Sams_(Buchreihe)
[ptc] jnthn: well, I guess in Germany all of the work is already done on Dienstag
psch no translated article apparently :/
but that's likely because the books aren't translated either, which would be hard i suppose 11:21
jnthn Yes, unfortunately ability to make basic puns in a language != having any competence in it :) 11:22
11:25 larion left 11:26 larion joined 11:31 |Tux| joined 11:33 raiph joined
masak tried to explain "MOP to break privacy in your tests is not a good idea" in all possible ways in a p6u post 11:37
11:37 Tux__ joined 11:38 |Tux| left 11:40 anaeem1_ joined
dalek ast: 08eb364 | jnthn++ | S10-packages/precompilation.t:
6model parametric interning + pre-comp test

Currently todo'd on all backends.		 11:42
masak ok, declaring 'my Vow' compiles and passes the immediate test of hiding the Vow class. 11:43
spectesting.
jnthn masak++
11:44 anaeem1_ left 11:53 vendethiel left 11:54 vendethiel joined 11:57 skids joined
[Tux] Something recent broke Text::CSV :( 11:57
dalek p: 579b06f | jnthn++ | docs/serialization_format.markdown:
Initial parametric intern serialization design.

Describes a new section to hold information we need to implement the cross-compilation-unit interning of parameterized types.		 11:58
11:59 laouji left
jnthn [Tux]: How does it fail? 12:00
[Tux] is investigating
skids jnthn++
Though now I have to go find another precomp bug to test my precomp Test:: stuff with :-) 12:01
RabidGravy I knew a shower would sort my previous uncertainty about encodings, of course I just shouldn't worry about it until such time as I am about to output it
[Tux] thinks something to do with boolean 12:02
jnthn shopping, lunch; bbiab &
[Tux] fail is caused by some exception ~~ /string/ 12:06
masak PASS
dalek kudo/nom: fb991be | (Carl Masak)++ | src/core/Promise.pm:
Lexicalize Promise::Vow

See RT #124190. The mechanism of a Vow being exclusive to a Promise hinges on Vow being defined as 'my'.
synopsebot Link: rt.perl.org/rt3//Public/Bug/Displa...?id=124190
masak gonna write a spectest for this after lunch, and then close the ticket. 12:07
12:07 skids left 12:08 eli-se joined, abraxxa left 12:10 travis-ci joined
travis-ci Rakudo build failed. Carl Masak 'Lexicalize Promise::Vow 12:10
travis-ci.org/rakudo/rakudo/builds/56084029 github.com/rakudo/rakudo/compare/f...991be191c9
12:10 travis-ci left
masak travis-ci: what are you talking about -- I built and tested locally :/ 12:11
:(
12:11 eli-se left
masak does a fresh checkout 12:11
moritz masak: not your fault 12:12
travis-ci.org/rakudo/rakudo/jobs/56084030
masak oh!
(phew)
moritz error: RPC failed; result=56, HTTP code = 200
fatal: The remote end hung up unexpectedly
probably due to github being still under heavy load
dalek specs: 695cab7 | paultcochrane++ | S04-control.pod: 12:13
specs: Wrap paragraphs consistently
specs:
specs: This contains *no* changes to content and is only a formatting change to
specs: give the text a consistent and clean look and feel.
12:13 dalek left, dalek joined, ChanServ sets mode: +v dalek
flussence I've a tiny feature request: --profile=json (for dumping straight to a .json without the fancy html wrapper) 12:14
yoleaux 26 Mar 2015 20:11Z <pippo> flussence: Thank you very much. That is interesting. Cheers.
12:16 kjs_ left
RabidGravy oh what joy "Finally, characters may be given as three octal digits after a \" - so, er, ":do=^J:am:le=^H:bs:cl=^NR:cm=^O%\068%+9%\068%+9:co#80:li#27:\" 12:16
do you think they are still accepting patches for FreeBSD termcap? 12:18
12:21 coffee` joined 12:24 Ven joined 12:27 xfix joined
hoelzro tadzik: panda doesn't do anything special when a target repo has submodules, does it? 12:27
[Tux] if I isolate the test, the test passes 12:29
12:30 raiph left
jnthn masak: I ran into the same thing earlier today...oh, but you're not on #moarvm so you didn't notice :) 12:31
[Tux] If anyone feels like digging: github.com/Tux/CSV/blob/master/t/55_combi.t#L46 <= is the line that fails
tadzik hoelzro: no, it does not 12:32
mayhaps it should clone --recursive just in case
hoelzro tadzik: I looked at p6-all-modules, and nothing uses submodules (yet) 12:33
but I was going to for linenoise
tadzik okay, we can add --recursive there, no problem 12:34
hoelzro cool!
12:34 anaeem1_ joined 12:35 telex left 12:36 telex joined 12:39 rmgk_ joined, rmgk left, rmgk_ is now known as rmgk
hoelzro is there a routine for sorting a list in lexicographical order in NQP? 12:42
I just realized the list of completions my tab completion branch offers isn't sorted =/
tadzik I don't suppose nqp has this sort of stuff 12:43
jnthn No, don't know of any 12:44
hoelzro tadzik: pun intended? =P 12:46
ok, I'll have to write my own routine or something
12:46 tinyblak_ left
tadzik hoelzro: hahaha, no, I did not :D 12:46
12:46 H2O1 joined, H2O1 left 12:48 raiph joined
eiro hello 12:50
Ven \o
12:51 Akagi201_ joined
RabidGravy ooh it gets better, anyone who wants to feel better about their own software and has access to a FreeBSD termcap look at the entry for IBM 3163 12:52
12:54 Akagi201 left, chenryn left
[Coke] .u rx 12:55
yoleaux No characters found
[Coke] .u prescription
yoleaux U+211E PRESCRIPTION TAKE [So] (℞)
12:58 kjs_ joined
[Tux] gist.github.com/Tux/7d23f5478dc962699f8f <= REPRODUCED! 12:59
please fix :)
memory leak or something?
13:00 jacksnipe left
[Tux] jnthn, ^^ 13:00
13:00 kurahaupo1 joined
jnthn Doubt it's a leak, could be corruption 13:02
Either way, a good fine
[Tux] tests++ 13:03
jnthn *find
I'm working on $other-hard-problem at the moment
But will put it on my todo list :)
masak [Tux]: will you submit a rakudobug, or shall I? 13:05
13:05 kurahaupo1 left
[Tux] please do: you have better inside in how to describe the problem 13:06
masak nice find, by the way.
13:06 kurahaupo1 joined, kurahaupo1 left
masak surely that can be golf'd a bit? 13:06
masak tries 13:07
reproduced locally.
and at exactly the same point in the loop.
seems very stable that way.
RabidGravy and the termcap from www.catb.org/terminfo/ doesn't even have those capbilities 13:08
13:15 anaeem1_ left 13:16 anaeem1_ joined 13:17 raiph left
masak [Tux]: I managed to golf it quite a bit. 13:19
m: class C is Exception { method message { "OH NOES" } }; for ^208 { { die C.new; CATCH { default {} } }; print "." }
camelia rakudo-moar fb991b: OUTPUT«...............................................................................................................................................................................................................OH NOES␤ in block at /tmp/3uBvBOFJHV:1␤␤»
masak m: my $dots = 0; class C is Exception { method message { "That is $dots dots" } }; for ^208 { { die C.new; CATCH { default {} } }; $dots++ } 13:20
camelia rakudo-moar fb991b: OUTPUT«That is 207 dots␤ in block at /tmp/9q7YDKFEAC:1␤␤»
13:20 anaeem1_ left
masak so it *works* for 207 iterations, and then -- GC triggers? JIT? optimizer? -- it fails on the 208th. 13:21
masak submits rakudobug
13:21 anaeem1 joined
arnsholt Weird! 13:22
eiro what if, when i translate something, i disagree the author ?
arnsholt If I had to guess, I suppose I'd say that it might be a bug in the spesh/JIT stuff
eiro is it fair to just not translate it ?
arnsholt eiro: You translate
masak eiro: insert a translator's footnote? 13:23
arnsholt If it's really bad, maybe add a translator's note
eiro oohhh yes :)
thanks
masak yeah, the translator's note is kind of a last resort. 13:24
arnsholt If you want to critique a piece, you should write a separate thing about it
As a translator, your job is to communicate the author's intent, not your own thoughts on the matter
eiro this is not a critic but i disagree the conclusion 13:25
i'm translating 5to6 (yes ... i have some spare time today) and about whitespaces, i read: Many of those restrictions are unlikely to concern much real-life Perl code 13:26
psch j: my $dots = 0; class C is Exception { method message { "That is $dots dots" } }; for ^208 { { die C.new; CATCH { default {} } }; $dots++ }
camelia ( no output )
psch j: my $dots = 0; class C is Exception { method message { "That is $dots dots" } }; for ^500 { { die C.new; CATCH { default {} } }; $dots++ }
camelia ( no output )
eiro which is not true: i always use line returns when chainning a lot of methods or walking around deep structures 13:27
in perl6, you have to put a \ to do this
and i think an example could be usefull here
masak eiro: pardon me for saying so, but somehow it feels very French to want to interject something while translating. ;) 13:28
arnsholt psch: Good idea to constrast with JVM. And I think that finding goes well with my first instinct that it's a Moar/spesh problem
eiro masak, you made a point there
masak eiro: I'd advise translating that bit straight, but then blogging about it and maybe effecting a change in the original. everybody wins that way. 13:29
eiro i'll just STFU and translate :)
masak eiro++ # trascending his Frenchness :)
Tux says something recent caused the above failure. I'd really appreciate it if someone wanted bisect (maybe a month back or so). 13:30
eiro haha ... masak you know me for a while now: i owne you at least 1 blog post (about snowflakes) so i'm really nervous talking about blog posts
masak I could do it myself, but I would prefer to delegate so that I can use my `perl6` for other stuff.
[Tux] last PASS was yesterday or wednesday 13:31
masak eiro: you owe me a post about snowflakes? tell me more!
[Tux] I rebuild rakudo daily
psch arnsholt: spesh/jit are really far outside "stuff psch knows", but it has to be somewhere in moar i suppose... :)
i don't think codegen there should be sufficiently different
13:32 Morjo left
masak .oO( I build my Rakudo every day. do you even lift? ) :P 13:32
jnthn There was a Moar version bump yesterday, for the first time since the 2015.03 release.
-2015.03 13:33
+2015.03-16-ga752064
eiro masak, it was about the way i write code in perl5 (in perlude afaik). you told me not to lead lines with coma and i answer that it makes things more readable. then you explain me the concept of snowflake
RabidGravy anyone who use rakudobrew on FreeBSD?
masak oh, that's a prime candidate, then.
jnthn Curiously, the only spesh patches I see in the list change the thresholds for when dynamic opt kicks in. 13:34
13:34 FROGGS left
jnthn Meaning the problem in theory woulda been noticable *sooner* before... 13:34
masak eiro: wow. I remember the discussion about commas, but I have kind of forgotten about the snowflake connection.
eiro i wanted to explain my point of view in a blog post, tried a few time but never get the point i have something readable
masak, i remember the snowflake connection because the blog post draft is named "snowflakes looks alike" or something 13:36
masak :)
should make it easy to search the backlog for the conversation. 13:37
psch r: say "how many backends is that?"
camelia rakudo-moar fb991b: OUTPUT«how many backends is that?␤»
masak eiro: found it. seems like the "snowflake" metaphor is from someone else's blog post: irclog.perlgeek.de/perl6/2012-10-26#i_6101700 13:38
which explains why it didn't leave a strong impression in me.
Ven masak: why does it sound french to add stuff in translations? I'm interested :P
masak Ven: I dunno, it just does.
Ven masak: well, got any example for me?
eiro yep. but i i'm trying a new approach of my todo list: switching algorithm from "FILO" to the gandalf principle: "you shall not pass"
masak Ven: maybe I have the French pegged as people who like to intellectually disagree a lot. 13:39
Ven I think you could even drop "intellectually" most of the time :P
masak Ven: closest concrete example I can think of is "Lisp In Small Pieces", which is a very opinionated book :)
Ven BooK isn't *that* opinionated... *g* (sorry, book) 13:40
13:41 skids joined
masak hehe 13:41
BooK knew what he signed up for when he chose his nick.
13:42 anaeem1 left
masak eiro: dunno if I mentioned it in the last comma conversation we had, but your comma convention breaks the way commas work in *writing*, which is why it looks ugly. my convention protects against the same things, but doesn't break the way commas work in writing. 13:43
13:44 tinyblak joined
psch hrm, github still under fire :/ 13:45
clone worked, push doesn't
masak different read side and write side, perhaps? 13:46
13:46 airdisa joined
gtodd www.artsy.net/feature/the-algorithm-auction there is nothing perl related .... 13:46
RabidGravy ah I thought it was me and github today 13:48
jnthn psch: I've found retrying after a little time seems to work 13:49
RabidGravy one in three requests failing
13:50 SamuraiJack joined, SamuraiJack left, SamuraiJack joined
psch yeah, push went through 13:51
13:51 SamuraiJack left 13:52 raiph joined
psch "Reenable j: for r: and p6:, replace star-p: with star-j: in star:. " but i don't think camelia cares automatically? 13:52
RabidGravy don't think I'll play FreeBSD today - I knew I should have gone for a noise day 13:53
13:54 eli-se joined, dalek left, PerlJam left
psch .ask moritz can you check my perl6/evalbot commit and update camelia? 14:01
yoleaux psch: I'll pass your message to moritz.
14:02 kaleem joined
masak how would I formulate a spectest that checks for Promise::Vow not being an 'our' class? 14:04
is checking that `Promise.WHO<Vow> :exists` is False an OK way to do it?
m: my $p = Promise.new; my $vowname = $p.vow.^name; say Promise.WHO{$vowname} :exists 14:05
camelia rakudo-moar fb991b: OUTPUT«False␤»
masak maybe something like that? then even if we rename Vow (like it was renamed from Keeper to Vow before), the spectest remains meaningful. 14:06
psch ...and >20 load on hack again :/
jnthn Seems reasonable.
masak makes it so
psch should maybe have moved to the irc host after all
14:08 vendethiel left 14:10 vendethiel joined 14:20 sno left 14:21 lucas_ joined 14:33 Akagi201_ left 14:34 Akagi201 joined, Akagi201 left 14:35 Akagi201 joined 14:37 raiph left, airdisa left 14:38 airdisa joined
lucas_ Hi. So :U means either a type object or a failure object, right? How come some methods in Bool.pm has a signature of Bool:U and they still work on defined objects? 14:39
jnthn lucas_: Example? 14:40
lucas_ In other words, why do their signature must be Bool:U and not just Bool?
pick, roll, and pre/postfix ++, -- 14:41
14:42 larion left
jnthn m: True.pick 14:42
camelia rakudo-moar fb991b: OUTPUT«Cannot call pick(Bool: ); none of these signatures match:␤ (Bool:U: *%_)␤ (Bool:U: Any $n, *%_)␤ in block <unit> at /tmp/pPdm5uWQds:1␤␤»
jnthn Doesn't look like it's wrongly working on a defined object to me...
14:42 airdisa left
jnthn The prefix/postfix cases are multi candidates and are defining the auto-viv semantics. 14:43
So a defined Bool hits an implementation of ++/-- elsewhere
Probably the one that delegates to .pred/.succ
psch psch@hack:~$ ps -ely | grep cron | wc -l 14:44
43
/o\ 14:45
lucas_ jnthn: Thanks! I just started reading Bool.pm source and I'm still a little confused. 14:46
14:49 yoleaux left, yoleaux joined, ChanServ sets mode: +v yoleaux, eli-se left
skids So, if I fork rakudo to PR my Test:: enhancements, should I select perl6 organization option versus myself? I don't know which way works out better. 14:55
jnthn skids: Probably yourself, if I understand the question correctly 14:56
psch skids: you could also branch in-repo i suppose
skids Don't think I have the bit for that. 14:57
psch oh right, commit access to the repos isn't the same as being in the org
14:58 risou_awy is now known as risou
psch skids: you're talking about Test::Utils and such? 14:58
those are in roast, and you have commits there
at least if you're skids on github as well :)
skids jnthn: Basically when I hit fork it now gives me a popup to choose myself or the perl6 org. And I don't know what that does exactly. 14:59
hoelzro skids: it's asking if you want the fork under your user (github.com/skids/rakudo-fork) or the Perl 6 org (github.com/perl6/rakudo-fork) 15:00
jnthn skids: I think if you choose yourself then you'll get a github.com/skids/rakudo
Which is probably what you want.
hoelzro if you have a commit bit, I think you should just do it in a branch on perl6/rakudo
15:01 yoleaux left
skids hoelzro: I only have a bit for docs/spec etc. 15:01
15:01 yoleaux joined, ChanServ sets mode: +v yoleaux
hoelzro ah, ok 15:01
then I would say fork to your user
skids It is decided then :-)
15:02 vendethiel left
psch so it wasn't about Test::Util apparently 15:02
skids psch: No, gist.github.com/skids/7c9089e9c941c5fc1222
psch on that note, i've occasionally wanted &is_run when writing tests
it's not designed, so i suppose that'll be ecosystem Test::More or something 15:03
skids I suppose the code could go anywhere, but it's intended to actually be used in roast eventually. 15:05
15:05 larion joined
psch skids: well, rakudo/lib/Test.pm is the Test that we shit to end-users. roast/packages/Test/Util.pm contains more compiler-specific testing routines 15:10
15:10 raiph joined
psch s/shit/ship/ 15:10
o.o
skids Yeah I figured that out. :-)
15:10 tinyblak left
grondilu nominates psch for the typo of the year :) 15:10
jnthn chokes on his tea :P 15:11
skids Well, ATM that code uses some nqp stuff, but once CompUnit is implemented to spec it will not be compiler specific.
psch skids: i'm just wondering if that's something users will want to test 15:12
skids It does raise the minimum bar a bit WRT exception handling probably, but just to run Test you already need to have EVAL and try. 15:13
psch i'm not really opposed to putting it there, i'm just remembering that it can sometimes be a bit grating to wait for someone to look over a PR... :)
psch should maybe start doing that occasionally as well
skids Well I'm very open to suggestions as to the destination, and also as to the actual speccing of exported functions to use it. 15:14
15:15 Ven left 15:16 [Sno] joined
RabidGravy speaking of nqp stuff, in Linux::Fuser I've used nqp::lstat - is that okay for the long term? or is there a plan to implement stat()/lstat() for e.g. IO::Path 15:16
[Coke] nqp ops in rakudo should be rare. If you have to use them, we probably need to implement something in rakudo. 15:17
in future, you'll need something like use NQP_STUFF; to declare it.
jnthn RabidGravy: That should be exposed eventually; I think lizmat++ already has worked on that in the newio branch
15:17 tinyblak joined
RabidGravy cool, I'll keep an eye on that then :) 15:20
skids psch: I guess the real question is, would an ecosystem module author need to ensure that the compiler could run a precomp unit that invokes his code before allowing panda to install it for production use. And I've seen some places where I would need that, though hopefully they will all go away. 15:21
15:21 vendethiel joined
psch skids: i guess then it mostly depends on whether those places are supposed to go away. if yes it should imo be a Test::Util routine 15:24
that is, i would answer "assuming a perfect implementation, would users want to test CompUnits?" with "no"
15:25 airdisa joined
skids Yes, how reliably we can prevent regressions or unnitoced stuff out of shipped products factors in. 15:25
psch (i think "perfect" there probably means "100% roast conformant")
(and roast has CompUnit tests, in that scenario)
but that's probably a bit too hypothetical to use as an actual argument 15:26
15:29 xfix left 15:30 xfix joined, Rounin left
skids Maybe the best thing is, put it in roast until enough people ask for it. Or Test::Util, if roast just syncs that (I don't know what goes on in roast/packages) 15:30
jnthn Test::Util lives in roast, iirc
RabidGravy where should $*USER be documented? I never found it before so I did "sub getuid() is NativeCall returns Int { ... }" to find that out 15:31
skids Ah, I see, it's not an ecosystem mod. 15:32
masak apparently it's Baidu's CDN that's been hijacked to DDoS Github. 15:34
15:34 Ven joined, fhelmberger left 15:35 smls joined
smls o/ 15:35
masak to be specific, the DDoS happens because Baidu *visitors* make a search and the results page fires off a few requests to Github.
15:35 fhelmberger joined
moritz ok, hack is nearly hanging again 15:36
yoleaux 14:01Z <psch> moritz: can you check my perl6/evalbot commit and update camelia?
moritz and, here's the thing
the load is >100
and rising
and yet, via the virt-manager console, it feels mostly responsive
masak moritz: is it possible to see which processes are causing the load?
moritz but killing processes doesn't seem to work
masak: I wouldn't know how
psch psch@hack:~$ ps -ely | grep cron | wc -l
78
moritz psch: yes, I've observed a high number of cron processes before 15:37
masak I wonder if it's something forehead-smackingly simple that's causing hack to behave the way it does.
smls RabidGravy: $*USER should be documented at doc.perl6.org/language/variables#Dy..._variables (but isn't yet)
masak and we'll all go "oooh" when we realize it.
psch moritz: right, i was just confirming that's it's the same. although in hindsight you probably already knew. 15:38
moritz psch: but since killing the processes doesn't work (a killall hangs), I guess there's something wrong with process handling in general
psch: I also tried to kill your java process (simply becauses it uses quite some RAM), and that hangs too
psch moritz: my java process is a perl6-j shell stuck at ^D :/ 15:39
s/shell/REPL/
RabidGravy smls, that's what I thought :)
15:39 Akagi201_ joined, Akagi201 left
[Coke] investigates hack... 15:39
psch moritz: kill -9 put that process <defunct> for me just now 15:40
15:40 fhelmberger left
smls m: say $*USER.WHAT # huh. 15:40
camelia rakudo-moar fb991b: OUTPUT«Cannot look up attributes in a type object␤ in method gist at src/gen/m-CORE.setting:26974␤ in sub say at src/gen/m-CORE.setting:18645␤ in block <unit> at /tmp/ASjWg_tQG2:1␤␤»
[Coke] irssi is responding just fine, but opening a new shell is hanging.
... but hitting control c in the new window made it come right back?
masak haha -- Github responded by XSS-ing the attacking clients, putting up an alert() in their browsers! :D 15:41
reading this is like reading a William Gibson novel
Ven haha 15:42
masak anyway, Github++ 15:43
moritz [Coke]: the ctrl+C probably killed a program called from your .bashrc
masak see it here in action: github.com/greatfire/ and github.com/cn-nytimes/
[Coke] might need iodump or iostat to get io stats. (or is there some other io tool installed?) 15:44
moritz atop is installed, but starting it hangs :(
[Coke] regular top works 15:45
trying to run vmstat and lsof...
masak moritz: sounds to me like the distribution is hosed somehow.
[Coke] how can load average be 120 if 0.2% of the cpu is in use? 15:47
15:47 vendethiel left
bartolin_ [Coke]: iowait? 15:47
retupmoca FWIW, the only time I've had processes refuse to be killed was when I was out of hard disk space 15:48
[Coke] how can I check io status?
RabidGravy smls, just added that and sent a pull request ;-)
moritz top says the CPUs are in 99.5% wait status
which is why the load is so high
bartolin_ [Coke]: if you run top there is a value for wait in the line Cpu
what moritz says 15:49
15:49 vendethiel joined
bartolin_ can not connect to hack 15:49
[Coke] so how can I see what's causing that without, e.g. iotop? 15:50
RabidGravy vmstat?
[Coke] vmstat hangs. :) 15:51
RabidGravy can never remember the precise incantation
oops
[Coke] all the crons are in D state pretty regularly. 15:52
15:52 maannj joined
maannj hey all 15:53
[Coke] ugh, I probably won't be able to sudo here. 15:54
moritz hi maannj 15:55
retupmoca wonders what 'df -h' would show
bartolin_ [Coke]: can you see a complete process list?
retupmoca otoh, I assume hack has a lot of disk space
moritz retupmoca: it hangs
lucas_ If there is a signature of (SomeType \a is rw) { a = ... }, the parameter could just as well be $a in the first place, right?
[Coke] retupmoca: nothing, it hangs.
retupmoca oh
[Coke] bartolin_: sure. 15:56
moritz retupmoca: but yes, it has lots of free space
retupmoca well, you've reached the limits of my knowledge in the area
15:56 spider-mario joined
bartolin_ [Coke]: ooc, could you gist it somewhere maybe? 15:56
[Coke] gist.github.com/coke/10b64c3ce486dea00dab 15:57
bartolin_ thanks, [Coke]++ 15:58
[Coke] can anyone see syslog? I can't even.
15:59 Perl6_newbee joined
jnthn lucas_: I'd say so 16:00
bartolin_ looking at the process list I would suspect 14227 and 13927 to be problematic. but probably it's not possible to kill those? 16:01
lucas_ jnthn: thanks. 16:02
16:02 kaleem left
[Coke] I can't even kill -9 one of my processes. still shows up in ps with state of D 16:02
moritz it seems no processes terminate at all 16:03
well, they do; it seems no processes can be killed
[Coke] I suspect we'll have to reboot again, but this time maybe install iotop and leave that running somewhere. (and maybe also leave a sudo bash running somewhere)
bartolin_ hmm, but if you run ps, it seems to end (I see only one 'ps' in the process list, as expected) 16:04
[Coke] wait, something just died! 16:05
moritz load is now 133
screen -rd hangs 16:07
ok, I'll reboot
[Coke] I had a man CRON process that eventually did die.
nothing else did, though.
RabidGravy ah $*GROUP as well, let's do that while I'm looking 16:08
[Coke]
moritz RabidGravy: you could be able to push directly to perl6/doc, without the need for a pull request
at least if you accepted my invitation :-) 16:09
ok, 'reboot' did nothing; I had to force-off it in the hypervisor 16:10
16:10 [Coke] left, sergot left, masak left, psch left, Mouq left, Util left, avuserow left, dalek joined, ChanServ sets mode: +v dalek
moritz it's back. 16:11
RabidGravy moritz, not seen any invitation - but this could be the email shitting me around 16:12
16:14 masak joined
masak moritz++ others++ # for investigating this 16:15
eiro github.com/eiro/doc/commit/5745d80...f5c484af57 16:18
spider-mario what will parameterized type serialization bring, basically? 16:19
eiro got Homepage complete and 5to6 partially translated. but i guess there is no infrastructure to have it online
jnthn spider-mario: A bunch less pre-comp bugs 16:20
spider-mario oh, ok
16:20 psch joined
masak eiro: the comment on line 70 looks like it wants to be translated to French, too 16:21
eiro masak, i'm not sure about what you said about "the way commas works in writtings 16:22
you mean "writting code" ?
masak, i check
masak eiro: no, I mean writing prose 16:23
masak .oO( non, I mean writing prose ) :P
eiro :)) 16:24
16:24 PerlJam joined
eiro i have to admit i'm confused: so many opinions, so many good arguments 16:24
spider-mario La pratique du "code golf" n'est pas mise en avant dans Perl6 16:25
eiro well... i have to stop trying to be a developper and focus on the doc
spider-mario hm, other minor typos
mind if I submit a PR some time?
eiro spider-mario, french ? mayby you can fix
of course not :) 16:26
it was the idea of pushing things on github
16:27 maannj left
spider-mario also, apparently, « espace » is technically a feminine noun, even though it’s much more common to make it masculine 16:27
which route do you prefer to take? :p
RabidGravy in the meantime I've just made a pull request to document $*GROUP as well
eiro spider-mario, you mean "une espace" ? 16:29
spider-mario, the route you want
spider-mario, also interested by a french translation ?
masak, fixed 16:30
spider-mario yes
masak eiro++
grondilu is french an is pretty sure "espace" is masculine 16:33
*and
spider-mario not the typographical character, apparently
but some people argue that the computer character doesn’t fall into that category 16:34
TimToady maybe we should reintroduce the neuter gender into French :) 16:35
spider-mario I’m all for it
16:35 Ven_ joined
Ven_ 's also guilty of saying "un espace" when he should really say "une espace" :) 16:35
spider-mario if it replaces the other genders completely for common nouns 16:36
16:36 Ven left
Ven_ "computer character doesn’t fall into that category" seems like a typographical character to me, tho. 16:36
TimToady fondly remembers raising kids and saying things like "It wants a cookie." 16:37
jdv79 i think a lot of english speakers would find that irrationally offensive 16:38
dalek c: 9213d4b | (Jonathan Stowe)++ | lib/Language/variables.pod:
Document $*GROUP
c: 91cbe6d | moritz++ | lib/Language/variables.pod:
Merge pull request #66 from jonathanstowe/master

Document $*GROUP
PerlJam
.oO( "It rubss the lotion on its skin" )
jdv79 basically
spider-mario I’m also tired of all the (e), -e- and other politically correct nonsense 16:39
masak PerlJam: I was wondering whether I was alone in having that association...
spider-mario l’Académie française disapproves, too
masak of course it does :P
spider-mario they say that in addition to being ugly, it adds to the problem more than it solves 16:40
and that in somes cases, it is impossible altogether
(you won’t say « Le fauteuil et la chaise sont blanc(he)s »)
masak I like how gender is so fundamental to us that it gets expressed as a problem in spelling. 16:41
PerlJam I wonder when that first happened
spider-mario do German speakers do the same? 16:42
e.g. write „Student(in)en“
psch spider-mario: Student/Innen
16:42 FROGGS joined
psch the "I" might not be capitalized 16:43
spider-mario is it as common as in France?
I really can’t stand it anymore :/
psch in official documents it's very common, yeah
PerlJam (I'm also wondering if the French language is having an undue influence on Perl 6 ;) 16:44
16:44 [Coke] joined
PerlJam [Coke]: o/ 16:44
16:44 raiph left
spider-mario if it does, I don’t really see it 16:45
except for «», maybe :p
[Coke] PerlJam: o/ 16:46
PerlJam spider-mario: see?!? guillemets everywhere!
spider-mario it’s not as if variables were declared using « le »
le $x;
les @xs;
PerlJam Though that would make an interesting module ... use French; 16:47
But I'll leave the implementation of that to TheDamian
spider-mario use Perle; 16:48
jnthn Well, FROGGS is the slangs person... :D
16:48 raiph joined
spider-mario (as I suspected, “Perle” is taken by several other languages) 16:49
FROGGS :o)
[Coke] m: use Perl; 16:51
camelia rakudo-moar fb991b: OUTPUT«===SORRY!===␤Could not find Perl in any of: /home/camelia/rakudo-inst-2/share/perl6/lib, /home/camelia/rakudo-inst-2/share/perl6␤»
[Coke] m: use *;
camelia rakudo-moar fb991b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/BOOhVIMzVN␤Undeclared routine:␤ use used at line 1␤␤»
PerlJam std: use *; 16:53
camelia std 28329a7: OUTPUT«5===SORRY!5===␤Confused at /tmp/1rB1KXae1h line 1:␤------> 3use 7⏏5*;␤ expecting any of:␤ module_name␤ version␤Parse failed␤FAILED 00:00 133m␤»
16:53 kjs_ left
bartolin_ moritz, [Coke]: wrt hack.p6c.org: I looked at some logfiles (/var/log/syslog etc.) and it seems the log entries stopped between 13:48 and 13:50 (last entry I found was 13:48:xx), at 13:50:00 there should have been new entries from cron processes. 16:54
maybe the system partition became remounted read-only at that time (for some reason)? 16:55
16:55 Ven_ left
bartolin_ next time we should check that. 16:56
[Coke] I have iotop -d 5 running in a screen window now, fwiw.
hahainternet one of your guys machines died?
[Coke] bartolin_: there's nothing to check. once it happens you can't do much. :)
bartolin_ a read-only filesystem would explain the symptoms
hahainternet bartolin_: modify your boot config to include a 'netconsole' module line
[Coke] (I couldn't run df, or dmesg..)
hahainternet being unable to run dmesg is a little odd 16:57
did it hang indefinitely [Coke]?
bartolin_ [Coke]: just run 'mount'
[Coke] bartolin_: I would expect that to cause failures, not blocks.
bartolin_: I'll try it, but I bet it also hangs. we'll see
hahainternet right, if it's hanging indefinitely that's a very common disk or controller failure sign
bartolin_ however, I would even suggest to test that hypotheses and manually remount the root-fs as read-only ...
hahainternet and the result of that will be a remount,ro
bartolin_: you must have some other machine you can send udp syslogs to 16:58
doing that with the netconsole module will get you these messages
it's how we validate these sorts of failures at work
bartolin_ hahainternet: sounds good, yes
[Coke] moritz: do you mean the reboot didn't do anything, or it rebooted back into the same state?
hahainternet otherwise, your symptoms sound exactly like a disk or controller failure
[Coke] 279 be/3 root 0.00 B/s 0.00 B/s 0.00 % 31.21 % [jbd2/vda6-8]
that seems to be a pretty consistent io user. peaked at 31% io. (now at 0ish, though. nevermind. :) 16:59
PerlJam
.oO( And now hack slows to a crawl from all of the system monitoring we're doing ;)
17:01 cschwenz left 17:05 eli-se joined 17:13 muraiki joined 17:14 sirdancealot joined 17:17 Util joined, spider-mario left
moritz [Coke]: huh? hack seems fine, load is low 17:19
[Coke] moritz: yes. 17:21
I started the monitor NOW because I couldn't when we were having the problem.
my reboot question was in re: backscroll. you had problem swith the reboot, trying to figure ou thte nature of the problem.
moritz [Coke]: reboot didn't seem to do anything 17:22
[Coke] ... "you mean it didn't reboot, or it rebooted and then it didn't seem to do anything". :) 17:25
moritz yes, didn't reboot 17:26
17:26 xinming joined
[Coke] ok. 17:26
moritz your first branch :-)
[Coke] I'm asking some friends, they were curious how scr0d things were. 17:32
17:41 rindolf left 17:42 airdisa left 17:43 airdisa joined 17:45 Akagi201_ left 17:46 Akagi201 joined 17:48 airdisa left 17:49 J-L joined
J-L m: use v6; my $n = 0; if True { $n = 7 }; say $n 17:50
camelia rakudo-moar fb991b: OUTPUT«7␤»
J-L m: use v6; my $n = 0; if True { $n = 7 } say $n
camelia rakudo-moar fb991b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/78autCj1VA␤Confused␤at /tmp/78autCj1VA:1␤------> 3use v6; my $n = 0; if True { $n = 7 }7⏏5 say $n␤»
17:50 Akagi201 left
TimToady std: use v6; my $n = 0; if True { $n = 7 } say $n 17:51
camelia std 28329a7: OUTPUT«5===SORRY!5===␤Strange text after block (missing comma, semicolon, comment marker?) at /tmp/gs8ltr6xXD line 1:␤------> 3use v6; my $n = 0; if True { $n = 7 }7⏏5 say $n␤ expecting any of:␤ horizontal whitespace␤ statement_control␤Parse fa…»
J-L Question: Why does the version without a ';' after the closing brace fail? It works just fine if the last line (the say() command) is on the next line.
TimToady because } serves as a statement terminator only if it's the last thing on a line
otherwise it's TTIAR 17:52
the rakudo error is LTA there...
J-L TTIAR? LTA? 17:53
cdc Two Terms In A Row
J-L Is that an error? Or is it done on purpose?
TimToady see S99
cdc Less Than Awesome
TimToady that is by design 17:54
TTIAR violates self-clocking
making it much more difficult to find other errors or report on them accurately
J-L As for S99, I found raw.githubusercontent.com/perl6/sp...ossary.pod , but it's raw POD. Is there another way I can read it that I don't realize? 17:55
17:56 eli-se left
cdc Hello TimToady, could you help me to understand how | is used in "[Zeqv] |@s».ords"? 17:56
J-L: design.perl6.org/S99.html
TimToady | is "insert into argument list"
cdc J-L: TTIAR + self-cloking -> strangelyconsistent.org/blog/macros...s-in-a-row
TimToady otherwise @s».ords is a single argument to the reduce 17:57
and there's not much point in reducing a single argument
17:57 J-L left, J-L joined
bartolin_ moritz: would you mind if I add a nagios check for hack.p6c.org which looks for read only filesystems? also I'd add myself to the recipients for those nagios notifications 17:57
TimToady the tricky thing here is that a reduce of a list infix can't just slurp its arguments, or it completely flattens sublists 17:58
bartolin_ moritz: maybe the problem has nothing to do with ro file system, but we don't know yet.
TimToady so instead it effectively binds like a ** rather than a * 17:59
but that means anything like a parcel is a single argument rather than a list of lists
17:59 telex left
TimToady so the | is necessary to strip one outer layer of () implied by the single argument 18:00
18:00 telex joined 18:01 espadrine left, vendethiel left
TimToady to put it another way, when you have positional arguments, you already know how parcels work 18:01
so foo (1,2,3) has a single positional argument if the signature is ($), but has 3 arguments if teh signature is (*@) 18:02
18:02 airdisa joined
TimToady likewise foo (1,2,3),(4,5,6) has 2 args if ($$), 4 args if ($ *@), and 6 if (*@) 18:03
18:03 noganex_ left
TimToady if you bind to ** insteaad of *, you get 2 args again, but variadically 18:03
cdc ooohh
TimToady: many thanks for your help!
if I understand correctly, the reduce binds to ** 18:04
TimToady we already had zip((1,2,3),(4,5,6)) working as expected when there are literal lists, but there was no way to interpolate a variable number of lists till now, which made it really hard to write a reduction over an arbitrrry number of lists without using EVAL
it's effectively **, but it is actually examining the raw arg list because it needs to optionaly decont sublists 18:05
we might be able to remove the decont after GLR, I dunno 18:06
18:06 rindolf joined
cdc and until then, we have to either use EVAL or |, right? 18:07
TimToady but the trap we were in was that every way of introducing a variadic lists of lists was itemizing each list, so all list infix reductions failed for variadic purposes 18:08
you will always need | to interpolate a variadic list there, I think
cdc have to meditate a bit more on this code
TimToady again @s».ords is a single parcel, in current parlance, and will behave as a single argument
(even after it's just a sublist) 18:09
if you strip off the implicit parens inside the [Z], you break [Z] (1,2,3),(4,5,6) instead 18:10
so we really need the | dereference to distinguish ((1,2,3),(4,5,6)) from (1,2,3),(4,5,6) 18:11
cdc so can we say that "[Zeqv] |@s».ords" is the short for "[Zeqv] (@s[0].ords, @s[1].ords)"? 18:12
TimToady leave out the parens on the right
or the space :)
afk for a bit &
18:12 airdisa left 18:13 smls left
cdc Thanks! 18:13
18:13 airdisa joined 18:15 muraiki left
cdc m: my @s = < this test >; say [Zeqv] @s[0].ords, @s[1].ords 18:15
camelia rakudo-moar fb991b: OUTPUT«True False False False␤»
18:16 airdisa_ joined 18:17 Rounin joined, eli-se joined, airdisa__ joined 18:18 airdisa left 18:21 airdisa_ left 18:23 muraiki joined
nine_ moritz: is there some collectd or similar running on hack? How was memory usage when it hang? Maybe all those processes were swapping hard. That could keep them alive for a very long time, even after kill -9 18:24
yoleaux 25 Mar 2015 22:34Z <japhb> nine_: I just rebuilt from scratch again this afternoon, and Inline::Perl5 installed perfectly. I don't know what changed, or if the problem is just plain flakey, but I'm happy to have it working now!
18:25 _edwin joined 18:26 kjs_ joined 18:28 kjs_ left
nine_ I would also suggest doing smartctl -a on all hard disks and when it happens again, run dmesg. That should give some messages in case of a hardware problem or point to the OOM killer running 18:28
18:28 vendethiel joined 18:30 rindolf left, airdisa__ left
lucas_ There is native_array.pm and native-array.pm in rakudo/src/core; the latter is empty. Is it going to be deleted? 18:31
18:31 airdisa joined
bartolin_ nine_: there is collect.p6c.org/ ... but we don't have data after 13:48 18:32
18:35 airdisa left
_edwin hi all! I have a question about moarvm: 18:39
currently the readlineintfh implementation expects readline/linenoise to return the line as a UTF-8 string
it turns out that at least on Windows linenoise does not do UTF-8 18:40
18:40 diana_olhovik left
_edwin what would be the better solution: a) try to return UTF-8 from linenoise b) covert from whatever encoding linenoise returns in readlineintfh? 18:41
[Coke] yes, collectd is running.
nine_: you could not run dmesg when it was happening.
bartolin_ I just see that collect.p6c.org is the same host as hack.p6c.org. so the missing data would fit with a ro file system. (I'm pretty sure I've seen data with nearly 100% io wait earlier -- probably data from ram which was gone with the reboot) 18:44
moritz nine_: collect.p6c.org/
nine_: and I don't think a virtual machine has smart data 18:45
18:46 Akagi201 joined
moritz one thing I considered is moving the collectd instance either to the phyiscal host or to another virtual machine 18:48
raydiak good NQnoon #perl6 18:50
yoleaux 26 Mar 2015 13:04Z <FROGGS> raydiak: can you add a test to roast for github.com/rakudo/rakudo/pull/395 ?
_edwin fwiw, I'll experiment towards a) as it is more local and I have an idea how to try it (at least for the input) 18:51
18:51 Akagi201 left 18:55 gfldex joined
moritz bartolin_: re nagios check, please do 18:58
_edwin woohoo, I can type '{', etc., in the Rakudo REPL :) 19:00
19:06 lucas_ left
cdc eureka! 19:08
m: sub f (**@args) { for @args -> $arg { .say for $arg } }; f ((1, 2, 3), (4, 5, 6))
camelia rakudo-moar fb991b: OUTPUT«1 2 3 4 5 6␤»
cdc m: sub f (**@args) { for @args -> $arg { .say for $arg } }; f |((1, 2, 3), (4, 5, 6))
camelia rakudo-moar fb991b: OUTPUT«1 2 3␤4 5 6␤»
cdc \o/
TimToady++ for the explanation and everything else! 19:09
bartolin_ moritz: I added a nagios check (for the system partition / and /home) 19:10
TimToady cdc: .say for $arg is always going to loop once 19:11
so is equivalent to $arg.say
dalek ast: 9f9fd05 | raydiak++ | S32-str/pack.t:
Add a test for odd-length string input to pack 'H'
 19:12
nine_ bartolin_: a r/o file system would not cause processes to hang. 19:14
19:15 diana_olhovik joined
moritz more likely: something causes lots of IO load (possibly locking, not actual reads/writes), and that's the reason that the CPUs spend their time in "wait" state *and* the reason that collect couldn't write anything to disk 19:17
still, it's worth checking
bartolin_ well, maybe it's not related to the fs being ro. but so far I've seen no file which was written to after 13:50. the load did only rise slowly after 13:50 by what I've seen at collect.p6c.org 19:21
19:22 Perl6_newbee left 19:26 airdisa joined 19:37 kjs_ joined 19:38 cygx joined
cygx o/ 19:38
FROGGS o/
cygx is it a known bug that multi MAIN(Str $foo) won't dispatch if the string happens to be numeric? 19:39
ah... 19:40
Str() works
so not a bug?
[Coke] it's meant to process command line args, so that doesn't surprise me. 19:49
you really can't make a numeric command line arg.
cdc cygx: sub MAIN($foo) { $foo.WHAT.say }
timotimo this is one of the places where we want an IntStr and friends
yoleaux 26 Mar 2015 22:07Z <smls> timotimo: How about this as a rudimentary start? gist.github.com/smls/623a9831832a3f20f9e8
cdc cygx: (Int+{orig-string[Str]})
timotimo thank you, smls 19:50
cdc when passing "3" to the script, for instance
Ulti is it possible to make a default action for a grammar? 19:52
timotimo with a fallback method you should be able to 19:53
Ulti ahh ok so it looks for the action method and its not there so calls the fallback
I'll give that a go
TimToady cygx: command line args are supposed to be run automatically through val() to intuit numerics, but that is NYI in rakudo 20:06
20:09 grondilu left
bartolin_ another note about hack: there was a helpful log file from atop in /root. the last data it has is from 13:49:28 and at that time everything looks very good (a few gig free ram, low load, no io wait, nearly no disk io and so on. so I bet, it was a sudden problem with disk access 20:10
20:11 yqt joined
timotimo so maybe we should be streaming out log data through the network to a computer that sits on a different machine? 20:12
bartolin_ yes, I think that would be a good plan 20:13
I could help setting it up (but for today it's already late) 20:17
_edwin ok, the AltGr+* input problem in the Rakudo REPL is easy to fix, but looking more at linenoise, I think linenoise will need to be replaced to make Unicode work in the REPL under Windows 20:19
20:22 cygx left 20:25 zakharyas joined 20:34 airdisa left, darutoko- left 20:35 airdisa joined 20:39 airdisa left
timotimo god damnit linenoise :( 20:39
20:40 kjs_ left 20:49 raiph left
dalek kudo-star-daily: 388f608 | coke++ | log/ (9 files):
today (automated commit)
 20:52
21:01 raiph joined 21:12 skids left, xfix left
itz does the interp.o clang wait :/ 21:14
muraiki hi all. sorry if this is a dumb question, but I'm stuck and a noob at p6. I'm trying to use p5 modules via Inline::Perl5. I'm calling a method from a module that takes named parameters in the form foo(bar => 1, baz =>2). but for some reason it isn't seeing those params being passed
itz I assume that's the compiler's fault?
muraiki er, "I'm calling a method from an object"
in one instance I was able to wrap the params in {} and make it happy, but in this case it isn't working :(
the earlier instance was a method call as a class method, though 21:15
this is a method call on an object
Ulti muraiki you might want to put your code in a gist
nine_ was around earlier the author of Inline::Perl5 21:16
PerlJam muraiki: my guess right now is that the "list of pairs" you're creating on P6 side isn't making it across as a flattened list that P5 can understand.
muraiki: try using foo('bar', 1, 'baz', 2) and see if that works 21:17
Ulti should be easy enough since a .list should fix that
PerlJam muraiki: also keep in mind that I could be totally wrong :)
muraiki PerlJam: that fixed it 21:18
PerlJam muraiki: you might also want to ask nine about it, in case he knows how to make it Just Work without you changing your calling style. 21:22
muraiki thanks :D 21:23
japhb masak: Where were you reading all the interesting details of the GitHub DDOS?
PerlJam muraiki: looking at the source to Inline::Perl5, it looks like it might only take a new multi method p6_to_p5(Pair:D $value) that turns a Pair into a list of two values. 21:25
_edwin for the record, regarding linenoise: having looked now at two not-quite-readline implementations in C, I'm beginning to think that most of the line editing code for the REPL should actually be in Perl 6 21:34
21:35 airdisa joined
_edwin that would give a better chance for getting all the tricky Unicode stuff right, among other things 21:35
21:43 airdisa left 21:44 zakharyas left 21:53 airdisa joined
PerlJam _edwin: make a small prototype and maybe that will encourage others to hack on it :) 22:04
_edwin I might try, but there's a lot of things I need to find my way through :) 22:05
ugexe would there be any advantage to parsing out dependencies from QAST or parse tree instead of directly parsing the perl6 source? 22:06
flussence well, you wouldn't have to directly parse the perl6 source... 22:10
PerlJam except ... how did you get the QAST or parse tree in the first place?
ugexe with Perl6.Grammar.parse 22:11
does the parse tree require the dependencies to be parsed first? i figured it was just another representation of the source code 22:12
22:12 muraiki left
ugexe i wasnt sure about a QAST::CompUnit but thoughtthe parse tree would work :( 22:13
22:15 lichtkind joined 22:16 lichtkind left 22:18 BenGoldberg joined
ugexe m: use Perl6::Grammar:from<NQP>; use Perl6::Actions:from<NQP>; my $*LINEPOSCACHE; my $source-code = q/use Test; my $x; $x++; say $x;/; my NQPMatch $tree = Perl6::Grammar.parse($source-code, :actions(Perl6::Actions.new())); say $tree.dump; 22:20
camelia rakudo-moar fb991b: OUTPUT«- statementlist: use Test; my $x; $x++; say $x;␤ - statement: 4 matches␤ - statement_control: use Test␤ - doc: ␤ - sym: use␤ - module_name: Test␤ - longname: Test␤ - name: Test␤ - identifier: Te…»
ugexe although, if i change use Test; to something not installed I get an error instead. so i guess that answers my question 22:23
i suppose it could still be useful for a module installer when dealing with installing the sub modules 22:24
22:24 Akagi201 joined
ugexe the order to build the modules in 22:24
22:24 brrt joined
brrt \o #perl6 22:26
_edwin created a pull-request for fixing the input of '{' etc. in the Rakudo REPL on Windows (against MoarVM/linenoise) 22:27
22:28 Akagi201 left
_edwin -> sleep 22:30
22:30 _edwin left 22:31 Akagi201 joined 22:32 noganex joined 22:35 Akagi201 left 22:36 Vlavv_ left 22:42 brrt left 22:43 jack_rabbit joined 22:46 raiph left 22:49 Vlavv_ joined 23:07 raiph joined 23:12 khisanth_ joined 23:16 Khisanth left
raydiak ugexe: problem is, our parse time and compile time are not separate...e.g. what happens if a slang is used? can't parse the rest until the slang is compiled 23:17
thus "nothing can parse Perl except perl"...b/c you have to actually compile to be able to parse 23:18
back to afk 23:20
masak parse time is a part of compile time. 23:24
even in languages where things are more separable. 23:25
23:26 diana_olhovik left 23:28 RabidGravy left 23:35 skids joined
masak 'night, #perl6 23:36
23:36 rararara_ joined 23:56 coffee` left

Powered by Cro and the Raku® Programming Language.

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

Thank you!