»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:04 angelolicastro joined
dalek kudo/nom: 5541810 | pmichaud++ | / (5 files):
Update all &infix:<cmp> operators to return Order::Decrease/Same/Increase.
 00:04
ast: d09de35 | pmichaud++ | integration/advent2009-day20.t:
Modify spectest for new .perl output

but it seems reasonable to keep for a while longer.)		 00:06
ast: 2b52ada | pmichaud++ | S (7 files):
Merge branch 'master' of git://github.com/perl6/roast
ast: fa3ea8f | pmichaud++ | S03-operators/cmp.t:
[S03] Add Inf/Pair cmp tests.
00:14 lestrrat left, lestrrat joined 00:26 spider-mario left 00:30 replore_ joined 00:33 havenn joined 00:34 havenn left 00:47 thelazydeveloper left 00:50 Patterner left 00:56 Psyche^ joined, Psyche^ is now known as Patterner 00:59 scott_ joined, scott_ is now known as Guest92805 01:26 vlixes joined 01:36 Khisanth left 01:40 sisar joined 01:49 Khisanth joined 01:54 Enstyne joined 01:55 Enstyne left, whiteknight left, ens_ joined 01:57 ens left 02:02 pun left 02:16 tokuhirom left 02:26 harukomoto left 02:46 kst` joined, machine5 joined 02:47 pochi_ joined 02:48 ens joined 02:50 ens_ left 02:52 machine1 left, mdupont left, cotto left, kst left, arnsholt left, pochi left, yeltzooo7 left, tkr left 02:55 machine5 left, FACEFOX-DOT-COM left, FACEFOX left 02:56 orafu left 02:57 envi_ joined 02:58 orafu joined, FACEFOX-DOT-COM joined 03:00 sisar left 03:02 FACEFOX joined 03:05 machine1 joined 03:17 angelolicastro left 03:25 ponbiki joined 03:59 cognominal_ left 04:00 mdupont joined, cotto joined, tkr joined, arnsholt joined, yeltzooo7 joined 04:01 cognominal joined 04:06 cognominal left 04:08 cognominal joined 04:09 FACEFOX-DOT-COM left 04:10 FACEFOX-DOT-COM joined, FACEFOX-DOT-COM left 04:11 FACEFOX-DOT-COM joined 04:33 cognominal left 04:34 cognominal joined 04:37 libertyprime left 04:47 machine1 left 04:48 FACEFOX left 04:49 lestaway joined 04:50 mathw_ joined, moritz_ joined, [Coke]_ joined, BooK__ joined, FACEFOX joined 04:51 pjcj_ joined, domidumont1 joined 04:53 ponbiki left 04:56 ocharles- joined 04:57 lestrrat left, geistteufel left, revdiablo left, sudokode left, moritz left, [Coke] left, infosophy left, dju left, zostay left, pochi_ left, vlixes left, DarthGandalf left, nebuchadnezzar left, xenu left, daniel-s left, integral left, jlaire left, domidumont left, lestaway is now known as lestrrat, ocharles- is now known as ocharles, replore_ left 04:59 infosophy joined, infosophy is now known as Guest93445 05:00 FACEFOX-DOT-COM left 05:01 machine1 joined 05:02 FACEFOX left 05:06 FACEFOX-DOT-COM joined, replore_ joined 05:07 FACEFOX joined 05:09 birdwindupbird joined 05:12 geistteufel joined, revdiablo joined, sudokode joined, huf joined, SHODAN joined, BinGOs joined, Woodi joined, dju joined, zostay joined 05:13 daniel-s joined 05:14 WAAPLCK joined, HarryS joined, pochi_ joined, DarthGandalf joined, integral joined, jlaire joined, mikemol joined, cj joined, jrockway joined, xenu joined, HarryS left, geistteufel left 05:15 bip` joined, xenu is now known as Guest33117 05:16 machine1 left, FACEFOX left 05:18 GlitchMr joined, tadzik1 is now known as tadzik 05:20 FACEFOX joined 05:23 machine1 joined 05:38 birdwind1pbird joined
dalek rl6-bench: 96cb3a1 | (Geoffrey Broadwell)++ | bench:
s/_cmd/_run/g in preparation for handling compilers with separate compile and run passes
 05:47
rl6-bench: f8fad85 | (Geoffrey Broadwell)++ | bench:
Handle compilers that need separate compile and run passes; factor out config for name/path of various VMs; only run enabled compilers; add p5.js_v8, p6.js_v8, and p6.pl compilers; whitespace cleanups
japhb phenny, tell fglock As of this latest push, it looks like perlito5.js running on node.js is failing completely ('node.js:63 throw e;'), but perlito5.js running on d8 is working well. 05:50
phenny japhb: I'll pass that on when fglock is around.
05:52 harukomoto joined
japhb phenny, tell fglock Also, perlito6.pl is failing on rc-forest-fire ('Missing value after operator > at lib5/Perlito6/Expression.pm line 286.') but works fine on the easier stuff. perlito6.js running on d8 hangs at 100% CPU on rc-forest-fire but again, works on the easy stuff. 05:52
phenny japhb: I'll pass that on when fglock is around.
japhb Wheee, that's 9 combinations of language/compiler/VM that perl6-bench is testing. Now I just need to get pugs working locally and being tested by perl6-bench, and I think that will cover that particular line of development pretty well. Then of course there's the whole "adding more tests" dimension .... 05:55
05:55 kaleem joined 05:56 wtw joined 05:57 harukomoto left
japhb phenny, tell fglock Correction: perlito6.js running on d8 appears to hang during *compile*. Perhaps I just need to let it run a lot longer .... 05:57
phenny japhb: I'll pass that on when fglock is around.
japhb Resting brane &
05:58 HarryS joined 06:00 mucker joined 06:07 birdwind1pbird left 06:14 lestrrat left 06:15 lestrrat joined 06:17 GlitchMr left 06:25 nebuchadnezzar joined 06:29 bip` is now known as geistteufel 06:39 replore_ left 06:53 sisar joined 07:03 noam_ is now known as noam 07:17 mucker left 07:25 mucker joined 07:29 mucker left, mucker joined 07:33 jaffa4 joined
jaffa4 hi 07:33
how to set the include path for niecza?
sorear -I 07:34
jaffa4 And inside the p6?
sorear:WHy did you modidy std.p6? 07:35
sorear Because TimToady didn't have enough tuits
jaffa4 tuit? 07:36
sorear I think 'use lib' works
jaffa4 tutoriak? 07:37
sorear tuit is a well established pun 07:38
I will do this when I get around to it
I will do this when I get a round tuit
DrEeevil throws a round "tuit" at sorear 07:39
now do it! kekeke
jaffa4 manana? 07:41
sorear: have you though about using some kind of caching mechanism? 07:44
for compilation?
sorear thought about it? I did it more than a year ago 07:45
jaffa4 How to use it?
Something along the line of precompiled headers 07:46
sorear it's completely automatic.
jaffa4 precompiled modules
sorear you cannot not use it.
jaffa4 so it means the second compilation is always faster, does not it? 07:48
07:50 daniel-s left
sorear jaffa4: if you are using modules, yes 07:50
jaffa4 sorear: great, I can see it is working 07:53
07:55 fhelmberger joined
jaffa4 sorear: I cannot use your std.pm... I get an error Unable to resolve method cursor_fresh in class STD 08:01
sorear my STD.pm6 is designed to be the frontend to a compiler, nothing more 08:03
jaffa4 Meaning what? I should not be able to parse perl 6 with it? 08:04
sorear Meaning that using it outside of niecza will be *hard* 08:05
08:06 replore_ joined 08:08 daxim joined
jaffa4 sorear: because? 08:13
08:15 localhost left, huangjs joined
sorear jaffa4: you've already started to see it - you'll have to write your own cursor_fresh and a thousand other interface routines 08:22
frettled masak++ nice blog entry
sorear -> sleep 08:23
jaffa4 sorear: bye
frettled masak: both, really 08:25
masak ;) 08:28
morning, #perl6
08:28 dakkar joined
frettled indeed! 08:30
With any luck, moritz has now listened to Damian's talk at GoOpen 2012.
Guest93445 was it filmed? 08:31
08:31 Guest93445 is now known as infosophy
infosophy ...the keynote, I mean. 08:31
frettled Damian usually does not permit his talks to be recorded, IIRC. 08:32
08:32 huf left
infosophy aye - remembered him saying so at the necrolinguaphilia talk. but a keynote could be a bit different. 08:33
masak I've seen a recording of an earlier version of the talk he gave on Thursday. 08:34
frettled Ah, wait, he was supposed to start at 10:10 and finish at 10:40, I misremembered the times. 08:35
08:35 xenu joined
frettled masak: oh! 08:35
08:35 xenu is now known as Guest77831
frettled I know someone who would like to see that, hee-hee. 08:36
08:36 Guest33117 left 08:38 rafl left 08:39 huf joined
frettled Google to the rescue! 08:39
infosophy was that an introductory talk as well? 08:40
08:40 rafl joined
infosophy chould we expect an advanced presentation on temporally quaquaversal virtual nanomachine programming in multiple topologically connected quantum-relativistic spacetimes? 08:41
s/chould/should/ 08:42
masak I'd *attend* one, but I doubt it's forthcoming. 08:46
infosophy he could seek sponsorship from major pharmaceutical companies... the topic is sure to bring on major headaches 08:55
frettled It might help in making LSD legal.
08:56 cognominal left
infosophy I find myself having been being having a migraine level headache just thinking about it. 08:56
08:56 cognominal joined
masak lichtkind: er. what kind of git setup do you have that adds Emacs backup files before a commit? :) 08:59
09:04 sisar left 09:05 sisar joined
flussence whoa, stuff's been happenin' 09:15
09:24 replore_ left 09:25 xinming joined
masak yes. 09:26
there's been a hackathon.
09:28 xinming_ left
frettled and people have been hacking 09:30
09:32 fglock joined
jaffa4 that is a crime 09:36
masak not in the sense used in programmer circles.
jaffa4 By the way, how would you parse perl 6 using niecza? 09:37
fglock o/
phenny fglock: 22 Apr 23:10Z <japhb> tell fglock Oooh, good news! Erm, unfortunately perlito5.js doesn't seem to be running under node.js anymore. I get "\nnode.js:63\n throw e;\n ^\n"
fglock: 05:50Z <japhb> tell fglock As of this latest push, it looks like perlito5.js running on node.js is failing completely ('node.js:63 throw e;'), but perlito5.js running on d8 is working well.
fglock: 05:52Z <japhb> tell fglock Also, perlito6.pl is failing on rc-forest-fire ('Missing value after operator > at lib5/Perlito6/Expression.pm line 286.') but works fine on the easier stuff. perlito6.js running on d8 hangs at 100% CPU on rc-forest-fire but again, works on the easy stuff.
fglock: 05:57Z <japhb> tell fglock Correction: perlito6.js running on d8 appears to hang during *compile*. Perhaps I just need to let it run a lot longer ....
fglock looking...
jaffa4 masak: any ideas? 09:38
masak I would say "hacking" is used in senses 4 and 5 from www.hackersdictionary.com/html/entry/hack.html
jaffa4: no, but sorear would know. 09:39
jaffa4 masak: I was just joking
and in rakudo? 09:40
09:41 cognominal left 09:42 cognominal joined
jaffa4 masak: what about rakudo? 09:43
masak jaffa4: well, "that is a crime :-P" would have made it clear you were joking. without the smiley, it wasn't clear.
jaffa4: you used to be able to get ASTs out in Rakudo, in alpha or ng. but they were useless, or Rakudo would crash. 09:44
I haven't tried it in nom. feel free to try.
09:47 cognominal left
jaffa4 The code of std: which is working on the channel? Can it be found somewhere? 09:48
masak. 09:50
frettled jaffa4: if it can be found somewhere, I would not be surprised if it was on github in the perl6 repo
masak it is.
github.com/perl6/std 09:51
github.com/perl6/std/blob/master/STD.pm6
09:51 am0c joined 09:52 cognominal joined
frettled If you were to look at how I sometimes turn a phrase, it would not be entirely implausible to deduce that there is a certain roundabout manner to it. 09:54
cognominal how come, when I type "panda list", I don't see zavolaj? 09:55
I did a "panda update"
masak frettled: how autopunny :) 10:00
frettled (:
tadzik o/ 10:02
frettled yay, tadzik++ 10:03
fglock phenny, tell japhb I can't reproduce the problem with perlito5.js; about perlito6, it will take some time to catch up
phenny fglock: I'll pass that on when japhb is around.
frettled Time for lunch. Less fancy and more expensive than the ones at the hackathon. :-/ 10:04
tadzik cognominal: try NativeCall
masak, masak, idea
a macro turns perl6 ast into perl6 ast 10:05
so if it takes a string, it's effectively a compiler (to perl 6)
so could you implement slangs using macros?
jaffa4 What were the prices like? 10:06
cognominal tadzik: can you be more specific?
jaffa4 What were the prices like at hackaathon? 10:07
cognominal NativeCall is part of zavolaj. How can it help me to install it?
tadzik cognominal: zavolaj is in the module ecosystem under a name "NativeCall" 10:08
10:09 Guest92805 left
cognominal ok. tadzik++ 10:09
tadzik jaffa4: quite high from my pov, but otoh I'm from Poland, so finding a place where I'll consider stuff cheap could be pretty hard :)
cognominal tadzik: it does not install on macosx lion : t/01-argless.t ......... dyld: lazy symbol binding failed: Symbol not found: _dlLoadLibrary 10:11
Referenced from: /Users/stephanepayrard/git/rakudo/install/lib/parrot/4.3.0-devel/dynext/nqp_dyncall_ops.bundle
Expected in: flat namespace
jaffa4 std.p6 does not compile in niecza 10:12
tadzik no eye deer, you'll have to ask jnthn or arnsholt
cognominal: ^
10:13 pernatiy left
cognominal I am not the first one to have the same problem : irclog.perlgeek.de/perl6/2012-02-13#i_5142113 Apparently was not solved 10:14
jaffa4 cognominal: that sounds bad, you are only the second one. 10:15
cognominal jaffa4: meaning? 10:16
jaffa4 cognominal: nothing special...
tadzik cognominal: bbkr eventually published his GeoIP module, so I guess he solved is problem somehow
cognominal "expect in: flat namespace" seems a common error message according to Google 10:17
10:34 sisar left
cognominal sounds to be that issue : github.com/jnthn/zavolaj/issues/7 10:35
nope 10:38
arnsholt cognominal: I've not had that particular problem on OS X (but a couple of others) 10:47
Which OS X Are you on?
But AFAIK bbkr_ has gotten it to work
cognominal arnsholt: 10.7.3 10:49
arnsholt But I think dlLoadLibrary is from dyncall (which is bundled with NQP)
Which NQP/Rakudo do you have?
10:50 spider-mario joined
cognominal last git version 10:50
arnsholt That's odd
10:51 Trashlord left
arnsholt My currently installed version is lagging a bit, since I'm working on some stuff in a branch, so it could be very recent breakage 10:51
But I have to admit I'd be surprised if that were the case 10:52
cognominal it seems people have that problem with old bundles lying around. I will check that 10:53
arnsholt Yeah, that could be it
Try wiping out install and recompiling 10:54
10:57 lestrrat left 10:58 lestrrat joined, snearch joined
masak today's autopun: twitter.com/sixthformpoet/status/19...7644583936 10:59
tadzik: yes. macros are "little compilers". I tend to think of them as routines you hook into the regular compiler. 11:00
slangs (in the sense of "sublanguages") are more about creating/substituting a whole parser/grammar/actions class, I think. 'course, subclassing ought to make it possible to use that procedure for minor changes as well. 11:01
like, adding a new keyword into the main Perl 6 slang, or something.
but we're well into the realm of speculation here. 11:02
tadzik yes, that's where I wanted to get :
:)
tadzik back home o/
masak \o/
colomon \o/
masak which is why we should usher in this era as soon as possible. to get through the sludgy parts.
tadzik: maybe there'll turn out to be some "functional equivalence" between deriving a grammar of the current slang, and a macro. we'll see. 11:03
11:03 jaffa4 left
tadzik interesting 11:04
11:05 pernatiy joined
frettled tadzik: I forgot to ask: where do people from Poland go for cheap shopping? Norwegians go to Sweden (and almost anywhere else), Swedes go to Denmark, Danes go to Germany, Germans go to Poland… 11:10
masak: d'oh, I forgot to hang over your shoulder and watch while you performed grammar magic. :) 11:11
s/grammar/macro/ 11:12
frettled confumbled the words.
tadzik frettled: heh, Poland? :P 11:13
I, for one, go to Lidl for cheap shopping :P
frettled tadzik: heh :) 11:14
tadzik or Tesco if it's after 21
frettled Someone once suggested Slovakia to me, but…
tadzik what kind of shopping do you look for?
frettled I'm not exactly looking, but the thing is that there is significant cross-border shopping of e.g. alcohol, foodstuffs (meat in particular), candy, soda, etc. 11:16
And it's typically so that people are willing to cross at most one border for that, since there are limits to how far you can drive on a day or weekend trip and still save money on your shopping.
I'm sure this can be turned into a coding contest challenge! 11:17
tadzik a beer in a bar is like 12 NOK here :) 11:18
frettled Okay, still more expensive than certain parts of Prague in 2004 :) 11:19
cognominal reinstalled rakudo from scratch and NativeCall installs correctly
frettled tadzik: 12-18 CZK for a beer, or (at that time), 3-5 NOK 11:20
cognominal: \o/
tadzik okay, that's cheap :)
cognominal nice, but intellectually frustrating. 11:22
11:22 pernatiy left
frettled cognominal: yes, there is some room for improvement ;) 11:23
spider-mario I’ve forgotten how to submit a rakudobug? 11:27
(that one’s minor, but still)
tadzik spider-mario: rakudo.org/how-to-help/ 11:28
spider-mario thanks
11:29 ens left
flussence NativeCall doesn't install through panda for me at all, it always fails the compiled-C tests :/ 11:35
spider-mario there 11:37
bug 112584
flussence while I'm at it, does rakudo have a -MCarp=verbose equivalent? I want to figure out what causes these - 11:38
==> Building NativeCall
use of uninitialized value of type Any in string context in method Str at src/gen/CORE.setting:670
11:38 huangjs left
tadzik flussence: update your rakudo 11:39
I fixed it like yesterday
arnsholt cognominal: Thanks for the update. Then it was probably some confusion in the installed libs section 11:40
tadzik r: my $a; say "lala$a"; CONTROL { say $_.backtrace.full }
p6eval rakudo 554181: OUTPUT« in sub warn at src/gen/CORE.setting:439␤ in method Str at src/gen/CORE.setting:670␤ in method Stringy at src/gen/CORE.setting:677␤ in block <anon> at /tmp/4EyzeyHHEu:1␤ in <anon> at /tmp/4EyzeyHHEu:1␤␤use of uninitialized value of type Any in string context …
tadzik flussence: ^
flussence er... I'm on 2012.04-50-g5541810 and git pull says I'm up to date
spider-mario oh, I forgot to include my rakudo version in my bug report.
oops.
arnsholt flussence: What kind of failure? 11:42
11:42 zhutingting joined
arnsholt If it has problems finding the compiled libs, it's fairly easy to get working 11:42
flussence t/01-argless.t ......... Cannot locate native library '01-argless.so' 11:43
tadzik flussence: github.com/rakudo/rakudo/commit/19370a16
arnsholt Yeah, that's the one 11:44
export LD_LIBRARY_PATH=. and try again
Or DYLD_LIBRARY_PATH=. is you're on OS X
(But you're not, since it's .so, not .bundle)
flussence tadzik: that's weird, this should work then... 11:46
tadzik make install? :)
masak frettled: is there something I can tell you or help you with now, regarding macros? or was it important that it happen in Oslo? 11:47
11:47 snearch left
flussence hm, maybe "make all install" doesn't do what I think it does, 'cause running make install is doing stuff now... 11:47
frettled masak: I was hoping for osmosis through direct observation :) 11:48
masak hehe.
frettled masak: but judging from your blog, I feel confident that new opportunities will arise
masak oh yes.
flussence arnsholt: thanks, worked 11:49
masak r: macro foo { say 2; quasi { say 4 } }; BEGIN { say 1 }; say 3; foo
p6eval rakudo 554181: OUTPUT«1␤2␤3␤4␤»
masak frettled: that's a good start towards understanding macros ;) 11:50
frettled masak: and possibly also quasi
11:52 cognominal_ joined
masak like this. a string is a piece of text that you can put in a variable and print whenever you want. a closure is a piece of compiled code that you can put in a variable and call whenever you want. a quasi is a piece of *un*compiled code that you can put back into your code and compile whenever you want. 11:53
just like qq strings can contain mentions of variables which are interpolated into the string as the string is evaluated, so quasis contain "unquasi slots" which are interpolated as the quasi is evaluated. 11:55
11:56 cognominal left
frettled So, essentially, what happens in the above code is: 1) say 1 is executed first, because it's in a BEGIN block. 2) say 3 is not executed (but it is parsed), 3) foo is parsed and because it's a macro, say 2 is immediately executed while quasi { say 4 } is compiled. Because foo is after say 3, say 3 happens before say 4. 11:57
Is that an adequate understanding?
11:58 skids left
masak yes. \o/ 11:59
in a very real sense, the contents of the quasi has replaced the 'foo' call in the code. 12:00
it happens on the AST level, not on the code-as-text level, but other than that, that's what happens.
what's being sent to and from a macro are AST tree fragments. 12:03
r: macro foo($x) { quasi { say "It's an $x.WHAT()" } }; foo("I'm a string") 12:04
p6eval rakudo 554181: OUTPUT«use of uninitialized value of type AST in string context in method Str at src/gen/CORE.setting:670␤␤It's an ␤»
12:04 sunfinite joined
masak huh. 12:04
r: macro foo($x) { quasi { say "It's an $x" } }; foo("I'm a string")
p6eval rakudo 554181: OUTPUT«It's an AST<1603438779>␤»
masak wonders why .WHAT doesn't work there
12:05 GlitchMr joined 12:10 mtk joined
masak I'd like to take the time to put in a good word for NPW 2012: act.yapc.eu/npw2012/ 12:10
sign up, submit talks -- and have the appropriate amount of fun!
frettled r: macro foo($x) { quasi { say "It's an ", $x.^name } }; foo("I'm a string") 12:12
p6eval rakudo 554181: OUTPUT«It's an AST␤»
frettled picked up one thing. 12:13
12:13 sunfinite left, pernatiy joined
frettled hmm 12:13
12:17 tokuhirom joined, spider-mario left 12:18 spider-mario joined
frettled And, not surprisingly: 12:18
r: macro foo(Str $x) { quasi { say "It's an ", $x.^name } }; foo("I'm a string")
p6eval rakudo 554181: OUTPUT«===SORRY!===␤Nominal type check failed for parameter '$x'; expected Str but got AST instead␤»
frettled just had to test :) 12:19
masak :)
it *looks* like a call to the macro foo with a Str.
but it isn't.
it's a call to the macro foo with an AST object containing a parsed string literal.
the Str doesn't exist because it's too early in the program lifecycle for runtime objects. 12:20
well, sorta.
as we enter the macro, we do switch gears from compile/parse mode to run mode. 12:21
and that transition seems to be why we see everything as AST objects from inside the macro. we're inside a bubble of run mode, looking out over a barren desert of parse-mode objects. 12:22
frettled with the odd tumbleweed of comments
12:24 spider-mario left
masak well, comments are just whitespace from a parser perspective ;) 12:24
12:25 spider-mario joined
frettled easily ignorable line noise 12:26
dalek blets: 74f768a | (Herbert Breunung)++ | README.md:
adding format rules
masak a boy wants to give an apple to a girl. to get to her, he has to pass through five gates, each with a guard. he bribes each guard with half of his apples, plus one. after he's given the apple to the girl, he has no apples left. how many did he have to begin with? 12:31
arnsholt A lot? =) 12:32
tadzik ALL THE APPLES
12:32 fhelmberger left
frettled arnsholt: I'd think he only had many. 12:33
masak seems the easiest way to solve the problem is to run the guy backwards in time.
we know he has 1 apple just before he gives it away to the girl.
flussence 25 12:34
ish :)
tadzik hmm
I think that can be solved using a sequnce operator
masak now run him backwards. he takes 1 apple from the guard, and then he takes as many as he has, that is, 2.
so now he has 4.
12:35 spider-mario left, spider-mario joined
masak this is possible because both of the operations performed here (halving, subtracting 1) are invertible, and their composition is invertible (adding 1, doubling). 12:35
tadzik r: (1, { ($^a + 1) * 2 } ...)[^10].perl.say
p6eval rakudo 554181: OUTPUT«===SORRY!===␤Confused␤at /tmp/lLIoOw9HGP:1␤»
tadzik r: (1, { ($^a + 1) * 2 } ... *)[^10].perl.say 12:36
p6eval rakudo 554181: OUTPUT«(1, 4, 10, 22, 46, 94, 190, 382, 766, 1534)␤»
frettled masak: whoa, whoa, whoa, there, unit test first!
masak :P
so, 94 apples?
tadzik++
tadzik provided there are no bugs
masak looks right to me.
frettled only 1 left 12:37
(oh ha ha)
tadzik r: (1,(*+1)*2...*)[6].say # golf'd!
p6eval rakudo 554181: OUTPUT«190␤»
tadzik eek, what
masak 5
tadzik r: (1,(*+1)*2...*)[5].say # golf'd!
p6eval rakudo 554181: OUTPUT«94␤»
tadzik there
mikec if i was the girl, i'd want to know why the first guard got way more apples
flussence r: say (1534, { $^a / 2 - 1 } ... *)[^5]
p6eval rakudo 554181: OUTPUT«1534 766 382 190 94␤»
masak mikec: this is how corruption works.
frettled mikec: because he was the most corrupt one, obviously 12:38
mikec hehe
masak it would be fun to have a module where the problem could be specified forwards, and it would reverse it for us.
an abstract algebra module.
that could solve linear equations.
frettled masak: I believe Damian has nailed that one.
tadzik hehehe
masak :P
tadzik . o O ( There's a pararell universe for that! ) 12:39
masak so you do 'sub guard($n) { $n / 2 - 1 }; solve (my $x), 1 == guard(guard(guard(guard(guard($x)))))'
or something of that order.
perhaps 'my Var $x' 12:40
frettled That ought to be guard^5($x)
tadzik hmm, there is this bfs variant that goes backwards as well as forwards
masak frettled: well, you could define a higher-order function (or operator) to do that. 12:41
tadzik: all we need here is to build the linear equation, and solve for $x
tadzik aye
frettled Maybe just abuse Unicode a bit: guard⁵($x)
tadzik :)
masak we could even make 'solve for $x' the default :)
or 'solve for the Var object in this expression'. 12:42
frettled So now we just need a operator that knows which Unicode chars are superposed numbers
an operator
I refactored that sentence too many times too quickly without testing.
dalek p/cstruct-work: 3bf5590 | (Arne Skjærholt)++ | src/6model/reprs/CArray.c:
Make sure out-of-bounds accesses on managed CArrays don't do stupid things.
 12:43
12:45 lichtkind joined, sisar joined
masak arnsholt++ 12:45
arnsholt Thanks. Now let's see what happens when I merge in master 12:46
12:46 bluescreen10 joined
masak arnsholt: is INTVAL signed? then you should perhaps check whether't be >= 0 too... 12:46
[Coke]_ sees pics of moritz and masak and everyone.
12:47 [Coke]_ is now known as [Coke]
arnsholt masak: Hmm. Good point 12:47
[Coke] thinks jnthn has a different style of victorian beard every time I see a picture!
12:47 machine1 left
[Coke] expected masak to look a little more... crazed? 12:47
[Coke] grins.
arnsholt In my defence, jnthn never checks for negative indices =) 12:48
But what's the correct answer for a negative index again? Invalid?
masak in... C? 12:49
dalek blets: 8fe801c | (Herbert Breunung)++ | docs/appendix- (2 files):
some rather subtle format fixes
masak I guess.
arnsholt No, in Perl 6. In C I know what happens =) 12:50
masak yes, .[-1] is always an error. 12:51
arnsholt Cool. Then I can just chuck in a check at the top of all the functions 12:53
Cheers!
12:54 machine1 joined
masak r: class Var {}; class Equation { has $.lhs; has $.rhs; method solve { $.rhs } }; multi infix:<==>(Var $lhs, Any $rhs) { Equation.new(:$.lhs, :$.rhs) }; say ((my Var $) == 42).solve 12:55
p6eval rakudo 554181: OUTPUT«===SORRY!===␤Variable $.lhs used where no 'self' is available␤at /tmp/U6qwXgK4yk:1␤»
masak r: class Var {}; class Equation { has $.lhs; has $.rhs; method solve { $.rhs } }; multi infix:<==>(Var $lhs, Any $rhs) { Equation.new(:$lhs, :$rhs) }; say ((my Var $) == 42).solve
p6eval rakudo 554181: OUTPUT«===SORRY!===␤Contextual $*GOAL not found␤»
masak r: class Var {}; class Equation { has $.lhs; has $.rhs; method solve { $.rhs } }; multi infix:<==>(Var $lhs, Any $rhs) { Equation.new(:$lhs, :$rhs) }; say ((my Var $x) == 42).solve
p6eval rakudo 554181: OUTPUT«42␤»
masak \o/
this has potential.
12:57 PacoAir joined
masak you could generalize this to arbitrary operators, and arbitrary occurrences of the variable on both sides of the equation. 12:57
and you could still detect whether it's a linear equation or not, and if it is, you could solve it.
and then you write a nice blog post about the boy and the five guards, with tadzik++'s nice sequence operator, and then this equation package ;) 12:58
arnsholt Well volunteered? ;)
masak I can do it, and I see the way forward. 12:59
but I'm kinda hoping someone else will pick it up ;)
perhaps in vain. the world is structured such that if you want something done, most of the time you have to do it yourself ;)
I'll see what I can do today after $work. shouldn't be that difficult. 13:00
but anyone is welcome to task-steal this one.
flussence
.oO( now if only we could generalise it so we could feed it a serialised program state and some code then say "go back to here"... :)
frettled masak: would be lovely to do so, but all my tuits are dodecahedrons, apparently.
masak :P 13:01
13:02 JimmyZ joined 13:04 szbalint_ is now known as szbalint
infosophy frettled: is there a difference between an up and a down tuit? 13:13
masak .oO( first generation of tuits ) 13:14
frettled infosophy: there certainly are strange tuits
arnsholt masak: In case you wondered, the out-of-bounds check is done before the index arrives in the C-code, so it turns out I don't really have to do it =) 13:18
frettled arnsholt: perhaps you can convert that into a charm tuit. 13:20
infosophy frettled: I can see someone getting down tuit and another feeling up tuit - how do the charm and strange tuits work? 13:21
masak r: role Expr {}; class Var does Expr {}; class Op does Expr { has $.op; has $.l; has $.r }; multi infix:</>(Expr $l, $r) { Op.new(:op</>, :$l, :$r) }; multi infix:<->(Expr $l, $r) { Op.new(:op<->, :$l, :$r) }; sub guard { $^x / 2 - 1 }; say guard(Var).perl 13:22
p6eval rakudo 554181: OUTPUT«Op.new(op => "-", l => Op.new(op => "/", l => Var, r => 2), r => 1)␤»
masak arnsholt: ah, ok.
another cool thing would be to take the above AST and produce the inverse -- totally doable. 13:23
hm, but it occurs to me that this is just a linear equation solver in the small ;) 13:25
13:26 JimmyZ_ joined
dalek Heuristic branch merge: pushed 16 commits to nqp by arnsholt 13:26
13:27 mucker left
frettled infosophy: well, if you bring some charm tuit, people involved tend to feel a bit better about it. If you think there is something strange tuit, then you're probably right. 13:27
13:27 JimmyZ left, JimmyZ_ is now known as JimmyZ
masak 's pun detector twiddles 13:28
frettled There is definitively a pun tuit here.
13:31 spider-mario_ joined 13:32 spider-mario left, spider-mario_ is now known as spider-mario 13:37 skids joined
infosophy fervently agrees 13:40
I'm going to start saving for a nothing tuit right away
13:41 uniejo joined
masak the nothing tuits are massively overpriced, if you ask me. 13:41
frettled that reminds me that there are at several kinds of risk tuits. 13:43
You have these, at least: certain, massive, no
arnsholt: but be careful so you don't get too many used tuits 13:46
13:49 uniejo left
infosophy hands freddled an end tuit 13:52
frettled touché 13:57
infosophy s/dd/tt/ 13:58
sowwy
frettled Id's an undersdandable misdake. 13:59
infosophy must be the soft consonants we were talking about on Sunday
flussence Stage parse: 538.949 # observation: rakudo's string handling/parsing/regexy bits are *slow* 14:00
(and not just on core.setting) 14:01
14:02 cognominal___ joined
[Coke] I don't know. I think this whole pun thing has a certain ring tuit. 14:03
14:05 cognominal_ left
masak flussence: don't read too much into those numbers. 14:05
flussence: you'll note that the AST step two lines down takes 0.000 14:06
flussence: while the reality is that parsing and AST building are intertwined
flussence I thought those looked odd...
frettled r: say "odd" if 538949 % 2 14:08
p6eval rakudo 554181: OUTPUT«odd␤»
masak r: say "odd" if 538949 !%% 2 14:09
p6eval rakudo 554181: OUTPUT«odd␤»
[Coke] masak: I was in the middle of typing that. ;) 14:10
flussence ok, I'm going to try this the long way - `git clean -dfx`'ed rakudo, reinstalled it, rm -r ~/.perl6 ~/.panda; panda bootstrap.sh and then reinstall NativeCall from that. If that doesn't get rid of that warning I don't know what will
[Coke] # 04/22/2012 - niecza++ ; pugs (44.63%); rakudo (0.03%) 14:11
looks like rakudo has been failing for 3 days now.
masak [Coke]: moritz_ and I end up typing the same things all the time nowadays. I've had occasional resonances with pmichaud and jnthn now and then, too.
colomon [Coke]: wait, niecza's ahead again?
masak [Coke]: how do you mean, "has been failing for 3 days now"? 14:12
I'm pretty sure Rakudo has been in more heavy use lately than in quite a while.
[Coke]: most likely it's some parrot or nqp revision that you need to give a little shake.
[Coke] colomon: look at rakudo's percentage there. 14:13
colomon I see it, but I don't know what it means. (Or maybe I'm looking at the wrong thing?)
masak colomon: < [Coke]> looks like rakudo has been failing for 3 days now. 14:14
colomon is it really passing fewer tests that niecza?
I mean, pugs?
masak colomon: < [Coke]> looks like rakudo has been failing for 3 days now.
[Coke] colomon: it's passing about 6 tests in my checkout.
colomon !!!!!!!
masak question is, why? 14:15
masak builds Rakudo locally
[Coke] yes. so clearly it's just me or someone else would have noticed.
flussence now I'm curious... which 6 tests?
I'd have thought a completely broken perl6 wouldn't pass any... 14:16
colomon btw (dunno if anyone was listening when I said it) current rakudo is failing to build my modules when my previous two month old rakudo (and niecza) handled them easily
however, I did get a fairly complete and mostly passing spectest run out of my updated rakudo build.
flussence argh! panda's still giving me that "uninitialized value" warning 14:17
[Coke] Could not find Test in any of: /home/coke/.perl6/lib, /home/coke/sandbox/perl6pass/rakudo/install/lib/parrot/4.3.0-devel/languages/perl6/lib, .
so, that would do it. some tests don't use Test, but very few. 14:18
colomon ooo, I think that was one of the issues I was seeing.
[Coke] r: use Test; say "eek" 14:19
p6eval rakudo 554181: OUTPUT«eek␤»
colomon rakudo seemed to be completely ignoring the -I command line option to set "include" paths.
(That was only a minor issue, though, since I just switched to using PERL6LIB, which worked fine.)
masak flussence: apparently, those are really challenging to track down. 14:20
colomon but yeah, after celebrating getting ABC running on both Rakudo and Niecza, I upgraded my Rakudo and the module promptly stopped working.
[Coke] this is in an uninstalled rakudo; should be able find it's own Test, neh?
I can try to bisect this later. (bisecting rakudo runtime is slooow. ;) 14:21
masak [Coke]: (quoting moritz_) "Rakudo doesn't work if it's not installed. If you get it working without installing it, that's basically an accident."
arnsholt Excellent quote =D 14:22
Right. Both NQP and Zavolaj are now updated. Have the appropriate amount of fun =) 14:23
masak \o/
[Coke] masak: I shouldn't have to install rakudo to run the tests. 14:26
should I?
masak no, it sounds wrong. 14:27
[Coke] never had to before 3 days ago, anyway.
masak nevertheless, that's what moritz_ said, as late as two days ago.
[Coke] patches to gist.github.com/1476841#file_spec_r welcome. 14:28
14:28 PacoAir left
masak running spectests now. already far more than 6 tests passed ;) 14:28
[Coke] (git rb == git pull --rebase)
[Coke] will blow away the NQP/parrot directories and try one more time.
14:30 wtw left
[Coke] iwbni if there was a git mode where you could say "clone this repository, but only get the last year's worth of data. If I ask for something earlier than that, you can fetch it then." 14:32
(we'd be downloading a lot less parrot at that point.)
arnsholt Hmm. Lazy-evaluation git? ^_^ 14:33
timotimo git can actually do something like that
it should totally be possible to do with grafting and shallow clones (i believe they are called)
[Coke] removing npq/ parrot/ install/ didn't help. same failure. 14:41
masak ooc, what's the failure?
[Coke] Could not find Test in any of: /home/coke/.perl6/lib, /home/coke/sandbox/perl6pass/rakudo/install/lib/parrot/4.3.0-devel/languages/perl6/lib, .
tadzik right 14:42
lib removed from @*INC
masak oh!
so [Coke] is running Rakudo in some way that it doesn't find Test.pm 14:43
tadzik aye
[Coke] so when running tests locally, I need to manually add lib back in?
./perl6 -Ilib t/spec/S02-types/whatever.rakudo 14:46
(same error)
colomon [Coke]: see my comment earlier that -I apparently doesn't work any more
frettled To modify the include path, you can set the PERL6LIB environment variable:
From ./perl6 -h
PERL6LIB="lib" perl6 example.pl 14:47
(thanks to imarcusthis++)
colomon try PERL6LIB=lib ./perl6 t/spec/S02-types/whatever.rakudo
masak did -I ever work on Rakudo?
[Coke] masak: no clue. didn't need to until 3 days ago. ;)
masak it does on Niecza, but I've never seen it work on Rakudo.
colomon it certainly seemed too in the version I was using last week
*to
frettled I think it may have worked before nom.
But as I've left myself out of the loop for a year or so, don't trust me.
masak I don't remember it doing so. 14:48
colomon though I guess if it was always including "lib" as a path and not returning an error about an incorrect command line argument, I would never have known it didn't work
masak I've always used PERL6LIB, and not 'cus it's fun :)
colomon perl6 -Ilib does *not* return an error message, it just doesn't work 14:50
Wynne:Math-Prime colomon$ perl6 -I t/01-basic.t
No such option -I
versus
Wynne:Math-Prime colomon$ perl6 -Ilib t/01-basic.t
===SORRY!===
Could not find Math::Prime in any of: /Users/colomon/.perl6/lib, /Users/colomon/tools/rakudo/install/lib/parrot/4.3.0-devel/languages/perl6/lib, .
[Coke] ok. tests passing as expected again. spec_r updated to deal with the change. next test run should be more indicative of real stats.
14:50 fgomez left
[Coke] moritz++, masak++, colomon++, frettled++ 14:51
fglock [Coke]++
frettled (++)++
spider-mario is it possible to mark a bug report for rakudo as minor? 14:53
fglock phenny, tell phenny tell phenny tell phenny++
phenny Hey, I'm not as stupid as Monty you know!
fglock :P
spider-mario I’ve made a report earlier but it doesn’t seem really important to me
[Coke] phenny, tell aloha msg phenny: tell Coke "hah"
phenny [Coke]: I'll pass that on when aloha is around.
[Coke] aloha, what?
seen coke?
aloha coke was last seen in #parrot 2 days 19 hours ago saying "(you'd have to setup 2 logins, I suppose.)".
phenny aloha: 14:53Z <[Coke]> tell aloha msg phenny: tell Coke "hah"
[Coke] CloseButNoRecursion. 14:54
masak bots that allow for such recursion are designed wrong. 14:56
dalek ast: 7f5a8bb | coke++ | test_summary:
Note new rakudo requirement
 14:59
15:00 JimmyZ_ joined 15:01 JimmyZ_ left, JimmyZ left
dalek ast: 0d5bc79 | masak++ | S32-list/sort.t:
[S32-list/sort.t] removed test

This test has no support from the spec, nor did it ever. There's absolutely nothing that guarantees that otherwise identical objects get sorted into memory-address order, or indeed that those objects were *created* in memory-address order.		 15:05
15:08 birdwindupbird left 15:12 JimmyZ joined 15:17 NamelessTee joined 15:19 fhelmberger joined 15:21 zhutingting left 15:22 kaare_ joined
[Coke] ooh, one less test to pass. woot. 15:25
JimmyZ moritz_: where is your code that contains many '*'?
15:26 kaleem left
flussence JimmyZ: perlgeek.de/blog-en/perl-6/starry-obfu.html 15:26
JimmyZ flussence: thanks
it's not easy to find it 15:27
jnthn o/ from Stockholm :) 15:29
masak jnthn! \o/ 15:30
speaking of self-referential questionnaires: www.maa.org/mathhorizons/Puzzles/Fe...puzzle.htm
jnthn Note that Rakudo's t/harness sets $ENV{'PERL6LIB'} these days, after lib got removed from the default @*INC 15:37
Any other test-running script probably needs to do the same.
dalek blets: e36f858 | (Herbert Breunung)++ | docs/appendix- (2 files):
linking letter headings
JimmyZ r: gist.github.com/2471737 15:39
p6eval rakudo 554181: ( no output )
JimmyZ hmm, looks Starry Perl 6 obfu doesn't outputs anything on rakudo star 15:40
TimToady we broke it when we changed the meaning of *+* from {$_ + $_} to {^$a + $^b} 15:41
JimmyZ :) 15:44
masak but moritz_++ has since fixed it on his blog, no? 15:46
TimToady in old blog post or in a new one?
masak hm, maybe I dreamt that.
15:48 Psyche^ joined
JimmyZ r: constant @primes = 2, 3, -> $p { ($p+2, $p+4 ... -> $n { $n %% none @primes ... * >= sqrt $n })[*-1] } ... *; say @primes[^20]; 15:48
p6eval rakudo 554181: OUTPUT«===SORRY!===␤Variable @primes is not declared␤at /tmp/UX44hFAfpo:1␤»
dalek blets: a9543c2 | (Herbert Breunung)++ | docs/appendix- (2 files):
cleaning up conditionals
 15:49
masak JimmyZ: known bug. 15:51
15:52 Patterner left, Psyche^ is now known as Patterner
JimmyZ aye 15:53
15:54 am0c left 15:55 zhangkaizhao joined
jnthn Yeah, before I fixed it I wanted to discuss the whole "constant @a = 1,@a" thingy with Pm. :) 15:55
masak did you reach a conclusion that makes the above legal?
jnthn (Which happened at hackathon++)
masak: Which above? :)
masak: The @primes one should work. 15:56
masak \/
er, \o/
sorry, lost my head there for a bit.
jnthn And we can't think of any sane model where the other one works.
Actually, I'm not sure we came up with an insane one where it did either... :) 15:57
TimToady hopes that hamming numbers can also be made to work right 15:59
constant @hamming = 1, dedup (@hamming X* 2) M (@hamming X* 3) M (@hamming X* 5);
colomon infix:<M> ?
TimToady where M is a lazy infix merge
colomon ah
jnthn TimToady: If you can get a closure around your references to @hamming on the RHS then it stands a chance.
masak jnthn: "if you can get a closure around..."? are you asking TimToady to change his code, or are you talking about things the compiler might be able to do? 16:01
TimToady in a sense the right side of = could be considered a thunk
(and must be for 'has')
jnthn TimToady: Yes, that we execute before we bound @hamming.
e.g. to get the value to bind to it.
So that doesn't help 16:02
TimToady what if you bind it to as the 'push' instead of the whole array
arnsholt jnthn: Is there any particular reason child strings can't be stuck in the child_objs array in CStruct?
masak decommutes
jnthn masak: I'm saying that I can only see the above working if dedup were a macro, and thus managed to defer the @hamming lookup
TimToady: push onto...what? 16:03
TimToady @hamming.push: ...
er, .plan
sorry, need more coffee
jnthn ...
jnthn gives up for now 16:04
16:07 thou joined 16:10 snearch joined 16:13 mspaulding joined, JimmyZ left 16:14 cognominal_ joined 16:15 alester joined 16:18 cognominal___ left
TimToady the = in 'constant @array = 1..*' is not a binding, it's a definition thunk that we can use however we want (one of the reasons the declarator now parses the =) 16:21
in this case, it's more like 'my @array; @array.plan: lazy 1..*' 16:22
the identity of @array doesn't change in that
unlike in a binding
s/identity/referent/
that's what I'm trying feebly to say
to the extent that we are limited to binding, that turns out to be a side effect that prevents some FP-style programming 16:23
if we can't close the definitional loop on lazy lists, we've got a too-operational view of definitions somewhere or other 16:25
shower &
16:36 thelazydeveloper joined 16:39 cognominal_ left, cognominal_ joined 16:45 kaleem joined, tokuhirom left
masak home 16:53
to the extent I understand the forces at play here, I agree with TimToady ;)
16:53 dakkar left
jnthn I guess I could try and do it as TimToady++ is suggesting. But I know what comes next: an RT ticket because you get unhappy that your so called constant array is actually full of scalar containers that can be mutated. :) 16:56
masak well, that issue is deeper than just constant declarations. 16:57
jnthn Right.
constant is a bad name for it in that sense.
masak no-one has yet presented a consistent/complete model for "one level down" readonliness.
16:58 PacoAir joined
jnthn We could call it "computed_at_compile_time_and_not_rebindable" or something :) 16:58
masak doesn't even think that's cute today :P
jnthn Well, the conclusion on that tends to be "cures considered worse than disease" :)
I wonder if you can .plan on a List, mind.
lol...there's an Indisk near my hotel called "Holy Cow" :D 17:01
er, wrong window... 17:02
arnsholt jnthn: Did you see my question above about child_objs vs child_strs? 17:04
17:05 fhelmberger left
jnthn arnsholt: oh, yeah, then promptly forgot to answer 'cus I'm tied up with course prep for tomorrow 17:07
arnsholt: Short answer: strings are STRING*, not PMC*, and you need to make sure you call the correct GC mark function on each one.
arnsholt: If you can make sure of that while keeping just one store, you're fine.
arnsholt Oh, right. That makes sense
A separate list is probably easiest 17:08
masak TimToady: jnthn makes a good point. if we do like you propose, I *will* submit an RT ticket about constant arrays being mutable.
arnsholt It costs an extra INTVAL in the repr_data I suppose, but it's not immediately obvious how to unify the two lists 17:09
And a unified list would probably require more complex logic 17:10
jnthn -> dinner, bbl 17:11
17:15 cognominal_ left 17:16 cognominal joined
TimToady until we get the details of OKness smoothed out, $x ~~ s:g/// should probably return a Match that numifies to the number of times matched, since a Match is the best OK container we have at the moment 17:25
17:27 cognominal left 17:28 cognominal joined
moritz_ home 17:32
tadzik \o/
17:33 moritz_ is now known as moritz 17:35 Chillance joined 17:42 pernatiy left 17:43 mucker joined 17:50 zhangkaizhao left 17:59 kaleem left 18:00 Patterner left
dalek kudo/nom: d6ce081 | moritz++ | tools/build/NQP_REVISION:
bump NQP revision to get nativecall improvements
 18:01
18:02 Psyche^ joined, Psyche^ is now known as Patterner 18:04 araujo left 18:15 snearch left
dalek p: a9dc3fa | moritz++ | src/6model/reprs/CArray.c:
[CArray] avoid mixing declaration and code, for pedantic compilers
 18:16
kudo/nom: b64c1e9 | moritz++ | tools/build/NQP_REVISION:
bump NQP revision again, to get compilation fix
18:29 ajs_work joined
tadzik blogged I, lol 18:31
ttjjss.wordpress.com/2012/04/23/back-from-oslo/
colomon tadzik: how exactly did you solve the "lib" problem? 18:35
tadzik colomon: like this: github.com/rakudo/rakudo/commit/ac2b2ae3998c753 18:36
turns out it was there only to make 'make (spec)test' easier 18:37
18:37 vlixes joined, mikemol left
tadzik anyone with github/rakudo/rakudo admin right? 18:37
moritz I think only pmichaud has admin rights there 18:39
18:39 mucker left
tadzik ah. I've asked github support about the commit hook which doesn't work for my name, and I'm supposed to "press Test Hook" somehow 18:39
18:41 s1n left 18:46 daxim left
moritz tadzik: if you do a roat commit, I can press the button there 18:49
tadzik roat? roast?
moritz roast, yes
sorry
tadzik I think it was broken for me too, yes
okay, can you press it now? 18:50
masak tadzik++ # blog post 18:51
moritz tadzik: shouldn't you do a commit to roast first?
18:51 vlixes left
tadzik I have no idea. The Github guy said "please press Test Hook and paste me the output of the following curl command" 18:51
I'm not sure what this button is supposed to do 18:52
moritz it re-sends the last two commits or so to the hook
which is why it would make sense for you do to commit first
tadzik moritz: how about perl6/ecosystem? The last commit there is mine 18:53
dalek osystem: cc1b79f | (Filip Sergot)++ | META.list:
Text::Markdown
 18:54
osystem: b1299b0 | (Geir Amdal)++ | META.list:
Test::Junkie
osystem: 854f487 | tadzik++ | META.list:
Merge pull request #8 from gam/master

Added Test::Junkie to META.list
tadzik huh
dalek osystem: cc1b79f | (Filip Sergot)++ | META.list:
Text::Markdown
osystem: b1299b0 | (Geir Amdal)++ | META.list:
Test::Junkie
osystem: 854f487 | tadzik++ | META.list:
Merge pull request #8 from gam/master

Added Test::Junkie to META.list
18:55 mikemol joined
moritz pressed it twice 18:55
tadzik seems to work this way
let me do some commit
18:55 s1n joined
moritz because the first didn't give me any feedback 18:55
18:56 ggoebel__ left
tadzik I did it to tadzik/panda, and it doesn't give me any feedback either 18:56
host04.appflux.net:5000/dalek?t=freenode,perl6 is the right url? 18:57
benabik Didn't it move to one of the feathers?
tadzik no eye deer 18:58
moritz? 18:59
19:02 ggoebel__ joined 19:03 birdwindupbird joined
dalek blets: 742545e | (Herbert Breunung)++ | docs/appendix-a-index.txt:
fine format fixes for section f
 19:06
19:07 tadzik is now known as knorra 19:08 knorra is now known as tadzik
japhb fglock, what version of node.js are you running? 19:13
phenny japhb: 10:03Z <fglock> tell japhb I can't reproduce the problem with perlito5.js; about perlito6, it will take some time to catch up
fglock japhb: v0.4.11 19:14
(updating now) 19:15
japhb fglock, it actually may be that my node.js is just way too *old*, not new. :-) 19:16
I'll try to build from source today and see if that gets things working. 19:17
19:17 machine1 left
fglock works with v0.6.15, quite slow though (maybe because perlito5 now uses many more exceptions) 19:19
japhb fglock, slower than older perlito5, or slower than older node.js? 19:20
(Wasn't quite sure which direction to parse that)
19:22 machine1 joined
fglock slower than older perlito5 19:23
japhb Gotcha.
ajs_work r: sub foo($a, $b) { $a + $b } ; foo( a => 1, b => 2); 19:24
p6eval rakudo b64c1e: OUTPUT«Not enough positional parameters passed; got 0 but expected 2␤ in sub foo at /tmp/ngWySJuGAr:1␤ in block <anon> at /tmp/ngWySJuGAr:1␤␤»
ajs_work Is that right?
Am I misreading perlcabal.org/syn/S06.html#Required_parameters ?
fglock ajs_work: why? 19:25
tadzik that may be a fossil
I think we got rid of passing positionals by name 19:26
[Coke] r: sub foo (:$a, :$b) {$a+$b}; foo( a => 1, b => 2);
p6eval rakudo b64c1e: ( no output )
[Coke] r: sub foo (:$a, :$b) {$a+$b}; foo( a => 1, b => 2).say
p6eval rakudo b64c1e: OUTPUT«3␤»
[Coke] tadzik: I believe so. (That way lies coldfusion) 19:27
ajs_work So, you have to explicitly declare a named parameter, now? Wow, when did that happen?
[Coke] sometime in the past 10 years. ;)
could probably suss it out by examing roast git logs. 19:28
masak last year, by the Neva.
[Coke] Neva?
masak St Petersburg.
[Coke] the river in Russia?
masak my memory is mostly scenic/episodic.
[Coke] ah.
masak there's a gist. hold on. 19:29
ajs_work That seems like a pretty structural change... how much should I be trusting the synopses these days?
masak gist.github.com/984783
ajs_work Thanks for the link
masak gets ahold of the relevant spec commits as well 19:30
moritz arnsholt: there's something wrong with the test plan in t/05-arrays.t
19:31 machine1 left
arnsholt moritz: I broke it? 19:31
masak github.com/perl6/specs/commit/74b2...e2d72733f6 19:32
github.com/perl6/specs/commit/e881...dfc5ebca0c # reverts the first one
github.com/perl6/specs/commit/94d0...733d0aec04 # current state
moritz arnsholt: I'm fixing it now...
ajs_work Is there an index that would, in some way, map synopses to external documents which revise them or visa versa?
arnsholt Ah, cool. What'd I break?
fglock japhb: this slowness may as well be a bug, in perlito5 or in v8, or something new in perlito5 that disabled the v8 optimizer 19:33
moritz arnsholt: the C side emitted 3 tests
arnsholt: but without the numbers
arnsholt: and the test plan expected 2
19:33 brrt joined
arnsholt Yeah, I thought I'd pushed the fix for that 19:33
Turns out I'd only made the change, not committed and pushed
Ooops =) 19:34
moritz++ # Fixing my mess
japhb fglock, nodnod
arnsholt Omitting the test numbers in the C code is intentional, FWIW
It's so that adding tests in the Perl code won't require updating the C code
moritz arnsholt: except that prove whines about that 19:36
19:36 machine1 joined
arnsholt Odd, haven't noticed anything about that here 19:37
Test.pm outputs a diag at the end complaining, but that's all
moritz might depend on the prove/TAP::Harness version 19:38
arnsholt Yeah. Or maybe it's just the harness outputting the diag messages? I know some versions do that 19:39
moritz oh, if the test plan it fix, it does indeed not complain 19:40
so I'll revert the part that adds the numbers
arnsholt Good idea to add the comment about the C call running tests, BTW
19:41 brrt left
jnthn tadzik++ # awesome blog post 19:42
moritz fwiw SQLite can now connect to the DB
tadzik awesome! 19:43
should I start writing Bailador::Plugin::Database now? :)
jnthn Yes! ;)
moritz sure :-)
19:44 envi_ left
jnthn moritz: What does it do after it's connected to the DB? ;) 19:44
arnsholt moritz: Cool!
moritz jnthn: it segfaults! 19:45
no, just kidding
tadzik lol
I just wanted to say that
moritz the parts that I've implemented so far seem to run fine
[Coke] rotfl!
arnsholt I'm looking into strings in structs now as well, in case you need that
jnthn heh, I was expecting the answer to actually be "it segfaults!" :D
moritz++, anyways :) 19:46
19:46 machine1 left
moritz hm, it hangs after test 16 19:46
sqlite has this oddity that it doesn't have a separate execute function
it just has 'step', which gets the next row, and executes if necessary 19:47
ajs_work Looking over S06 more carefully: it's definitely a bit confused. I have something else to do, but I might have to dig into this later. There's some notes about proto/only that I can't make sense of right now...
moritz so I can imagine that leading to a loop if handled wrongly from within p6 land
arnsholt jnthn: Speaking of strings. When I do something like $!memb = "foo" will bind_attribute_boxed's value parameter be a STRING pointer in disguise, or some kind of PMC wrapper around it?
jnthn arnsholt: Neither 19:48
Well, it depends
arnsholt: If it's using the _s variant of the op it actually well end up in _ref
arnsholt Aha
Good thing I asked then
jnthn Well, it's a type issue 19:49
The boxed one gets a PMC *
19:49 jaldhar left
jnthn STRING * ain't one of those 19:49
japhb fglock, It looks like node.js git (v0.7.9-pre) works fine, but node.js from Ubuntu 11.04 (v0.2.6) does not. I'm guessing v0.2.6 doesn't support the exceptions you're throwing around now (NPI). 19:51
arnsholt Right. Looking at CArray again, it looks like I'll get away with only handling strings in _boxed
19:51 machine1 joined
fglock oh, that's bad 19:53
it needs to be optimized again anyway, that should bring the number of exceptions down again 19:54
dalek ast: dbbbce7 | moritz++ | S (6 files):
rakudo unfudges
 19:56
moritz blug again: perlgeek.de/blog-en/perl-6/2012-osl...-rest.html
fglock (these are "control" exceptions - return/next/redo/last) 19:57
hmm - actually "next" is easy to optimize, I'll take a look later (but now &sleep) 19:59
[Coke] wonders if all these blugs are in the sixperl blugroll
19:59 thelazydeveloper left, thelazydeveloper joined 20:00 raiph joined
masak "In the evening we had very tasty Vietnamese food, and generally a good time." # nice zeugma you got going there :) 20:00
20:00 machine1 left
jnthn jnthn encountered a X::Pub::ClosedWhileStillWantingBeer at the end of the evening. 20:01
It was a fun time :)
sjn \o 20:02
sjn just came home from Go Open
Damian had two great talks that day
masak /o
sjn and, thanks for the hackathon, I had a great time! ^^
20:03 fglock left
sjn It was especially fun at the Norwegian Open Source Prize ceremony, where there (for some reason) were several made a point out of their background with Perl. :) 20:04
masak sjn++ # enabler
jnthn sjn: Thanks *so* much for your role in organizing it.
tadzik sjn++, it was an awesome experience. Thank You! 20:05
20:06 machine1 joined
raiph masak: your basics explanation for frettled worked for me. and left me unhappy with the name "macro". 20:07
masak: to the degree p6 is akin to lisp for the masses, it recasts some things. one is to rename things if appropriate. 20:08
masak: what about s/macro/nowandthen/ ?
20:08 tokuhirom joined
masak raiph: no, it's really a macro. 20:09
that's what macros are.
raiph masak: what's a macro?
masak a code template.
a compiler hook.
a routine that runs during the parse of your program. 20:10
a way to fold complexity under the carpet of an abstraction
all of the above :)
20:10 araujo joined, araujo left, araujo joined
raiph masak: like "now" 20:10
masak I don't understand.
raiph masak: you're describing its benefits, what u can use it for 20:11
masak I'm not the person to answer that question. :/
I'm just implementing them.
raiph masak: but what is it, at a basic level? it's what u explained to frettle.
masak but I bet I would be just as stumped had someone asked me what to use `[*]` and other reduction operators for, or the sequence operator. and yet I use them all the time for useful, real things. 20:12
raiph it's a bit of code that runs immediately on being encountered, now, plus some stuff that's inserted at the point of call to be executed then.
masak raiph: what is a macro, at a basic level? a routine. the routine takes AST parameters and returns an AST.
raiph: well, so is a Lisp macro. 20:13
and a Julia macro. and a Scala macro. and an <X> macro, for lots of languages <X> 20:14
20:14 GlitchMr left
masak so I don't see why the naming bothers you, to be honest. 20:14
raiph cuz the name makes me think it's something to do with the big picture, or a zoom lens 20:15
moritz well, it kinda is
if you use a macro, the compiler zooms in on the macro definition
applies it, and then zooms out again 20:16
masak I bet that's the etymology of "macro instruction", yes.
it's a single instruction which abstracts over several more primitive ones.
doy "The term is used to make available to the programmer, a sequence of computing instructions as a single program statement, making the programming task less tedious and less error-prone.[1][2] (Thus, they are called "macros" because a big block of code can be expanded from a small sequence of characters)."
masak there you go. 20:17
raiph but that doesn't imply the now/then aspect
masak no, that's implied by macros in all languages that have them.
lichtkind hai hackorz 20:21
tadzik hai lichtkind
20:21 bluescreen10 left
raiph o/ lichtkind 20:21
masak light be on thee, child
lichtkind it is :)
masak hopes it isn't like headlights unto a deer ;) 20:22
20:22 araujo left
raiph masak: so i just absorbed the takes and returns ast args bit. 20:24
masak: so rather than now/then parts it's more like compiletime/runtime parts, right? 20:25
masak well, it's a funny little runtime island in an ocean of compile. 20:26
20:26 araujo joined
doy they are basically the next step beyond BEGIN blocks - BEGIN blocks let you run runtime code at compile time, macros let you run runtime code at compile time in order to generate code to run at runtime 20:27
masak aye. that's a nice way of putting it. 20:28
macros have BEGIN semantics in that they run ASAP.
jnthn It's not the macro itself that has BEGIN semantics, it's the usage of the macro that magically does.
masak aye. it's the macro invocation. 20:29
triggered by parsing a call to the macro.
jnthn Right.
BEGIN at a distance :)
lichtkind raiph: we slowly getting where i wanted to be
raiph r: macro foo { say "bar" } ; say "no call to foo so no bar gets printed" 20:30
p6eval rakudo b64c1e: OUTPUT«no call to foo so no bar gets printed␤»
timotimo can macros be used dynamically,too? like decide at runtimewhich of two macros to use ro something? 20:32
not really sure if what i'm asking for makes any sense
raiph r: macro foo { say "bar" quasi { say "qux" } } ; say "waldo" ; END { foo }
p6eval rakudo b64c1e: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 2␤»
raiph r: macro foo { say "bar" ; quasi { say "qux" } } ; say "waldo" ; END { foo }
p6eval rakudo b64c1e: OUTPUT«bar␤waldo␤qux␤»
raiph r: macro foo { say "bar" ; quasi { say "qux" } } ; say "waldo" ; BEGIN { foo } 20:33
p6eval rakudo b64c1e: OUTPUT«bar␤===SORRY!===␤Could not find sub &say␤»
20:33 bluescreen10 joined
timotimo oh? that's interesting 20:34
jnthn Very. 20:35
Decidedly a bug. I bet it'll be "fun" to find.
raiph r: macro foo { state $a; say ++$a ; quasi { say "qux" } } ; say "waldo" ; END { foo ; foo } 20:36
p6eval rakudo b64c1e: OUTPUT«1␤2␤waldo␤qux␤qux␤»
arnsholt jnthn: In CArray, every time a Str element is accessed, we create a PMC and stuff. Would it make sense to cache those like we do with CArray/CStruct/CPointer stuff? 20:41
jnthn arnsholt: Yes. 20:43
arnsholt: Well...
Yeah, it does :)
masak hm, macro in BEGIN... :) raiph++ 20:44
jnthn I heard you like BEGIN time...
masak timotimo: "at runtime" is too late, 'cus the macros have already run then.
arnsholt jnthn: Cool. I'll add that to my TODO list (although that means we have to create the PMC on bind as well 20:45
raiph what about unifying BEGIN blocks and macros?
masak timotimo: but as for multi-dispatch -- yes, macros can dispatch based on the arguments going into them. they're just routines, after all.
raiph: what about it? it sounds like a bad idea to me :)
BEGIN blocks and macros are too different.
jnthn If raiph means a "unify the underlying mechanism for running bits of stuff at compile time", then they already are fairly unified, fwiw. 20:46
masak right.
raiph masak: thx for your patience. bear with me. BEGIN is done as the first part of runtime, right? 20:48
er, no. back to doy 20:49
ajs_work Runtime for the parse, not your program.... 20:50
*parser
raiph right
masak raiph: a BEGIN block is run as soon as its '}' has been parsed.
i.e. ASAP
raiph what about macro { ...... } ;
same?
masak not at all. 20:51
just as a sub isn't run upon definition.
in macros (as in subs), definition is separate from call.
but yes, the *call* is made as soon as its last argument has been parsed. 20:52
20:52 dorlamm joined
raiph ok, hence unifying BEGIN and macro is nonsensical. ok. 20:52
masak unifying them beyond the degree to which they are already unified is probably nonsensical, yes. 20:53
they do share certain features, such as access to the COMPILING:: namespace.
raiph apologies if the following goes awry, but i'll share it so u see my thought process as i try to grok this 20:56
let's say i've got a some book text that i'm going to read to rehearse, then speak out loud 20:57
i read it, and encounter a footnote [1].
(struggling to understand myself, give me a moment :) 20:59
20:59 thelazydeveloper left
dalek blets: 1b3e922 | (Herbert Breunung)++ | docs/appendix-g-glossary.txt:
explaining unicode (that should prpably do someone with more clues about it)
 21:00
tadzik t/spec/S32-list/sort.rakudo, TODO passed: 33. Anyone else>?
masak raiph: no problem. it sounds interesting.
tadzik: I already removed that one. pull roast.
lichtkind raiph: your in oslo too? 21:02
tadzik oh, I've had unpushed roast changes
masak classic.
tadzik that explains why pmichaud++ has to fix integration/advent2009-day20.t 21:03
dalek blets: 4744509 | (Herbert Breunung)++ | docs/appendix- (2 files):
small fixes
 21:06
21:07 NamelessTee left, skids left
raiph lichtkind: would that i were that lucky (no). masak: i'm dealing with (pleasant) distractions here too (kid's b'day). i'll collect my thoughts then finish the line i started. 21:07
r: sub foo { say 1; BEGIN { say 2 } } ; sub bar { say 3; END { say 4 ] } ; say 5 21:11
p6eval rakudo b64c1e: OUTPUT«2␤===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 2␤»
raiph r: sub foo { say 1; BEGIN { say 2 } } ; sub bar { say 3; END { say 4 } } ; say 5
p6eval rakudo b64c1e: OUTPUT«2␤5␤4␤»
raiph r: macro duck { say "duck" ; foo } ; sub foo { say 1; BEGIN { say 2 } } ; sub bar { say 3; END { say 4 } } ; say 5; duck 21:12
p6eval rakudo b64c1e: OUTPUT«2␤duck␤1␤===SORRY!===␤Macro did not return AST at line 1, near ""␤»
21:13 brrt joined
raiph r: macro duck { say "duck" ; foo ; quasi { say "q" } } ; sub foo { say 1; BEGIN { say 2 } } ; sub bar { say 3; END { say 4 } } ; say 5; duck 21:13
p6eval rakudo b64c1e: OUTPUT«2␤duck␤1␤5␤q␤4␤»
raiph r: macro duck { say "duck" ; bar ; quasi { say "q" } } ; sub foo { say 1; BEGIN { say 2 } } ; sub bar { say 3; END { say 4 } } ; say 5; duck 21:14
p6eval rakudo b64c1e: OUTPUT«2␤duck␤3␤5␤q␤4␤»
21:16 jferrero left 21:19 birdwindupbird left 21:27 dorlamm left
[Coke] # 04/23/2012 - rakudo++ ; niecza (94.21%); pugs (42.09%) 21:28
"niecza", 20303, 69, 747, 1524, 22643, 24152
"pugs" , 9072, 17, 3333, 1599, 14021, 24013
"rakudo", 21550, 30, 638, 1972, 24190, 24162
cutting in to my pugs percentage! ;)
masak heh :) 21:29
jnthn 42 \o/
masak [Coke]++
japhb Speaking of pugs, I guess it's time to try to get pugs running locally ... any hints before I go down blind paths?
21:31 harukomoto joined
masak well, a few of us have been getting a certain error, and it's not entirely clear why we're getting it. 21:31
but the fix seems to be to pull down things using cabal in a certain way.
...I think.
japhb Are there notes somewhere on a "presumed working" build/install sequence? pugscode.org just points to hackage ... 21:33
masak github.com/perl6/Pugs.hs/blob/master/HACKING 21:34
[Coke] japhb: grab github -> perl6/Pugs.hs
21:35 vlixes joined
[Coke] if you have any troubles, ping us here. if you're starting with a fresh cabal install, you'll probably be fine. 21:36
if not, you might have to wipe it.
21:37 dorlamm joined
[Coke] jnthn: down from 42.31%! 21:37
21:38 bjarne joined
japhb [Coke], well, given that the system doesn't even have ghc on it yet, I'm thinking I'm good there. :-) 21:38
jnthn [Coke]: I'd say sorry, but I'm happy Rakudo is progressing :P 21:39
[Coke] jnthn: ok, fine. :) 21:40
21:43 kaare_ left 21:45 PacoAir left
dalek blets: bfc97cc | (Herbert Breunung)++ | docs/appendix-g-glossary.txt:
explain capture subrule and set links for the contexts
 21:49
blets: 0a0b44e | (Herbert Breunung)++ | docs/appendix- (2 files):
clear unless and until and some minor stuff
gfldex p6: say *; 21:51
p6eval niecza v16-23-gaa61ed5: OUTPUT«Whatever.new(...)␤»
..pugs: OUTPUT«Inf␤»
..rakudo b64c1e: OUTPUT«*␤»
jnthn 'night, #perl6
gfldex there seams to be disagreement
japhb o/ jnthn
masak gfldex: yes. Pugs is wrong. the other two do variations on a theme. 21:52
21:53 tokuhirom left
gfldex r: say *.perl; 21:53
p6eval rakudo b64c1e: OUTPUT«WhateverCode.new()␤»
gfldex is .perl meant to be compatible between implentations? 21:54
brrt why is that a whatevercode rather than whatever
21:55 jaffa4 joined
jaffa4 hi 21:55
gfldex std: my @a = 1,2,3; @a>>.{ $_ }; 21:56
p6eval std 3d13d53: OUTPUT«ok 00:00 43m␤»
jaffa4 I would like to capture this @<bracket1>=('(' <.ws>)* 21:57
japhb brrt, because performing an operation on * actually forms a code block. It allows us to write things like:
r: say 2, * ** 2 ... 256
p6eval rakudo b64c1e: OUTPUT«2 4 16 256␤»
21:58 donaldh joined
masak 'night, #perl6 21:58
21:58 dorlamm left
gfldex std: my @a = 1,2,3; @a>>.{ say $^a }; 21:59
p6eval std 3d13d53: OUTPUT«===SORRY!===␤Placeholder variable $^a may not be used outside of a block at /tmp/IVlJQVpV1J line 1:␤------> my @a = 1,2,3; @a>>.{ say ⏏$^a };␤Check failed␤FAILED 00:00 43m␤»
japhb o/ masak
jaffa4 that seems to be ok
gfldex if @a>>.{ ... } is not a block, what is it?
dalek ast: db06fbd | coke++ | S (5 files):
pugs fudge
 22:00
[Coke] colomon, sorear: 69 failing spectest for niecza now.
22:00 harukomoto left
colomon platform? 22:00
jaffa4 masak: how to capture repetition, I have tried couple of forms, to no avail 22:01
[Coke] colomon: feather.
I'll get you the test run output.
perlcabal.org/~coke/ -> niecza 22:02
arglebargle.
feather admins, feather.perl6.nl/~coke is down.
colomon perlcabal.org seems to be down, too... 22:03
[Coke] colomon: do you have an account on feather?
colomon nope
[Coke] colomon: yah, same thing.
momentito
colomon: gist.github.com/2474191 22:04
22:06 harukomoto joined
TimToady gfldex: that's a STD buglet 22:08
22:08 cognominal left 22:10 harukomoto left, cognominal joined 22:11 mikemol left 22:12 bluescreen10 left
dalek ast: 10a70c6 | (Solomon Foster)++ | S03-operators/context.t:
Fudge new test.
 22:18
colomon r: [1,2,3].join<abc> 22:21
p6eval rakudo b64c1e: ( no output )
colomon r: say [1,2,3].join<abc>
p6eval rakudo b64c1e: OUTPUT«postcircumfix:<{ }> not defined for type Str␤ in method gist at src/gen/CORE.setting:8667␤ in sub say at src/gen/CORE.setting:6555␤ in block <anon> at /tmp/4CKLofnt8Z:1␤␤»
colomon n: say [1,2,3].join<abc>
p6eval niecza v16-23-gaa61ed5: OUTPUT«Unhandled exception: Cannot use hash access on an object of type Str␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/lib/CORE.setting line 320 (Any.at_key @ 7) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /tmp/cxorkH04CZ line 1 (mainlin…
colomon n: say [1,2,3].join<abc> ~~ Failure 22:22
p6eval niecza v16-23-gaa61ed5: OUTPUT«===SORRY!===␤␤Undeclared name:␤ 'Failure' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1366 (die @ 3) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1147 (P6.comp_unit @ 33) ␤ at /hom…
sorear good * #perl6
japhb o/ sorear 22:23
dalek d: fab64fc | larry++ | STD.pm6:
hash subscript should temp $*CURLEX
 22:24
TimToady gfldex++
dalek ast: 939d046 | (Solomon Foster)++ | S02-literals/listquote.t:
Fudge new test for niecza.
22:24 bluescreen10 joined
sorear whee, an incomprehensible niecza bug report. 22:25
japhb [Coke], masak: It appears to be important to 'unset GHC' before running 'sh bootstrap.sh' for cabal-install. Perhaps the GHC variable at the top of the installing-onto-Debian instructions should be 'GHC_SRC' or somesuch that does not collide with bootstrap.sh's internals?
22:25 tyatpi joined
TimToady evalbot rebuild std 22:25
p6eval OK (started asynchronously)
TimToady does that work again yet?
[Coke] I never installed on debian. 22:26
gfldex after looking at rosettacode.org/wiki/Search_a_list#Perl_6 and S32 I found it rather strange that there is no .index method
would that break lazy lists?
japhb ... whee, and now failed further into the cabal-install bootstrap at the time of 'Linking Setup ...', complaining of missing dependencies. 22:27
sorear niecza: #<<< foo
p6eval niecza v16-23-gaa61ed5: ( no output )
gfldex r: say List.^methods.grep: /pos/;
p6eval rakudo b64c1e: OUTPUT«at_pos␤»
22:28 vlixes left
gfldex r: say List.^methods.grep: /index/; 22:28
p6eval rakudo b64c1e: OUTPUT«␤»
sorear n: say +""
p6eval niecza v16-23-gaa61ed5: OUTPUT«0␤»
dalek ast: 03211aa | (Solomon Foster)++ | S02-types/catch_type_cast_mismatch.t:
Fudge new test for niecza.
 22:29
japhb Anyone know what to do about this: gist.github.com/2474289 22:30
(cabal-install failure)
dalek blets: 6e47b90 | (Herbert Breunung)++ | docs/appendix-g-glossary.txt:
adding logical ops
blets: afe0ec8 | (Herbert Breunung)++ | docs/appendix-a-index.txt:
overhaul multi and only + minor stuff
sorear would I be a bad person to close github.com/sorear/niecza/issues/120 with no further action?
22:31 cognominal left, cognominal joined
colomon Is the issue just that the error message he's getting doesn't have a line number? 22:32
TimToady looks like a leftover from the days in which #<<< was a valid comment beginner
colomon n: say +"**"
p6eval niecza v16-23-gaa61ed5: OUTPUT«Unhandled exception: Cannot parse number: **␤ at /home/p6eval/niecza/lib/CORE.setting line 1364 (die @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3403 (ANON @ 10) ␤ at /home/p6eval/niecza/lib/CORE.setting line 3405 (NumSyntax.str2num @ 4) ␤ at /…
TimToady and we temporarily had a #<<< detector, which we subsequently removed (I think)
std: #<<< foo
p6eval std fab64fc: OUTPUT«ok 00:00 39m␤»
colomon TimToady: I think he was just using #<<< to show where his problem is. 22:33
sorear colomon: I think the main issue is that English is not his first language, which makes me really apprehensive to do anything with it
22:33 sivoais left
TimToady when I run it, I do get: at /home/larry/perl6/niecza/foo line 0 (mainline @ 1) 22:35
colomon Yeah. But I'm pretty sure he was trying to complain that when he runs that program, the error message he gets does have any line number info better than
at /Users/colomon/tools/niecza/frip.pl line 0 (mainline @ 1)
TimToady the line 0 is wrongish
colomon at least, that's what I get when I try it.
22:35 jaffa4 left
TimToady jinx 22:35
22:35 raiph left 22:36 donaldh left
sorear It's been the same behavior for months and it's impossible to fix without incurring a runtime penalty 22:36
22:36 sivoais joined
TimToady the lack of a line number seems to be precisely the complaint, according to the title 22:37
and I'll note that p5 does take a run-time hit in order to set the current line number (in an instruction that also clears the tmp stack, so not an extra dispatch) 22:38
colomon sorear: might it make sense to make that an option? ie --debug (or something) turns on the line number storage even though that slows things down. 22:39
TimToady a real compiler could, presumably, reverse engineer the line number from the return address 22:40
lichtkind TimToady: im bit confused over temp, on one side its the new local on other side just temporarely overwerites values 22:42
dalek blets: 41208f3 | (Herbert Breunung)++ | docs/appendix-a-index.txt:
fix $*PERL and $*PID lines + new formating rule for P5 comparison
 22:45
22:45 tyatpi left
dalek blets: ef0b626 | (Herbert Breunung)++ | docs/appendix-a-index.txt:
linkfix for :Perl5 and :p adverbs
 22:49
22:52 skids joined 22:53 mikemol joined
dalek blets: b18b418 | (Herbert Breunung)++ | docs/appendix-a-index.txt:
clear map example
 22:58
22:59 brrt left 23:00 tyatpi joined 23:06 raiph joined
dalek blets: 55de143 | (Herbert Breunung)++ | docs/appendix-a-index.txt:
reformat :p and :path
 23:14
23:14 cognominal left 23:15 cognominal joined
sorear so.... what was the verdict? what should I do with 120? 23:16
dalek blets: 0e4194f | (Herbert Breunung)++ | docs/appendix-a-index.txt:
repair $*UID entry
 23:17
colomon "colomon: sorear: might it make sense to make that an option? ie --debug (or something) turns on the line number storage even though that slows things down."
sorear colomon: I was looking for a "yes" or a "no"
lichtkind good night
colomon if that's impractical, then I guess 120 should just get a nice explanation of why it's not desirable to store the line numbers. 23:18
sorear of course it makes sense to have an option, but my current tuit supply allows onl "close" and "don't close"
colomon it's really a less-than-awesome, not a bug. want me to write a short explanation and close it?
sorear I'm chronically conflicted over whether LTAs belong in the bug queue 23:19
23:20 fgomez joined, lichtkind left
sorear If you have a better explanataion, by all means post it 23:20
23:21 cognominal left 23:22 ranguard joined 23:23 cognominal joined
colomon "The error is what we would expect to get, but the error message would ideally have the line number of the problem. Unfortunately, tracking line numbers in this case would cause a hit on Niecza's performance. While we hope to improve this situation in the future, this is not a bug." 23:23
I've got that typed into the box there, will hit enter if you approve. :)
sorear +1 23:24
colomon commented and closed
23:26 Chillance left 23:40 pochi_ left 23:45 flussence_ joined, alester_ joined 23:48 flussence left
japhb phenny, tell fglock See gist.github.com/2474657 -- even if it will be some time before perlito6.js can compile rc-forest-fire, it would be nice to stop it from endlessly looping the same error messages. :-) 23:48
phenny japhb: I'll pass that on when fglock is around.
23:50 whiteknight joined, alester left, alester_ is now known as alester 23:52 FACEFOX left 23:53 machine1 left 23:58 FACEFOX joined 23:59 machine1 joined, bluescreen10 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!