»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
raiph m: pir::foo__bar() 00:03
camelia rakudo-moar 315ec6: OUTPUT«===SORRY!===␤To compile on the MoarVM backend, QAST::VM must have an alternative 'moar' or 'moarop'␤»
raiph m: pir::foo_bar() 00:04
camelia rakudo-moar 315ec6: OUTPUT«===SORRY!===␤pir::foo_bar missing a signature␤»
raiph sjn: ^^ seems to be from trying to run on moar code with calls to pir: (parrot) ops with __ in the name ? 00:07
s/on moar/(on moar)/ 00:09
sjn raiph: yeah, that may be it 00:10
my $client = mysql_init( pir::null__P() );
is there a way around that?
psch r: $*VM.config.keys.grep("moar").say 00:11
camelia rakudo-moar 315ec6: OUTPUT«moar␤»
..rakudo-parrot 315ec6: OUTPUT«␤»
psch ISTR there was something about parrot that'd get fixed soonish, but i have no idea if your problem is related 00:12
sjn ok 00:13
psch sjn: although, in the end your problem is a pir op appearing to moar, which actually shouldn't happen 00:16
i'd still lean towards "the module doesn't seem well maintained", although i'm not completely sure what you're working with there 00:17
sjn psch looking at the examples in zavolaj 00:20
raiph sjn: was there a message of the form "Error while compiling op xxx: " just prior to the "To compile on..."? 00:21
sjn raiph: no, this was when actually trying to *run* the mysql example in zavolaj 00:22
under moar
hm, the postgresql example seems to work 00:23
good enough for me 00:24
raiph k
dalek rl6-roast-data: 8226dcd | coke++ | / (5 files):
today (automated commit)
 00:27
[Coke] 3 failures on moar-jit, 97 on moar-nojit, 7 on jvm. parrot is clean 00:28
timotimo someone get us out of this mess! 00:46
psch: yeah, unbase is a regular sub call under the hood 00:48
if we want it to curry into WhateverCode, we'd have to special-case it
psch i remember that there was a discussion about auto priming min, but i can't find it anymore 00:53
although i supposed "methods and operator-looking-sub auto prime" is a decent guideline
timotimo: i don't particularly want colon-pair unbase to be autoprimed. Whatever just looked more generally applicable in my mental model 00:54
but blocks with placeholders are the general case, and Whatever is sugar for some cases, which i'm ok with 00:55
psch .oO( although i think that's also not *quite* correct...)
m: my @a = <1 2 3>; @a[{$^a} - 1].say
camelia rakudo-moar 315ec6: OUTPUT«Cannot call 'Numeric'; none of these signatures match:␤:(Mu:U \v: *%_)␤ in sub infix:<-> at src/gen/m-CORE.setting:4476␤ in block <unit> at /tmp/vBMu5s87pl:1␤␤»
psch m: my @a = <1 2 3>; @a[* - 1].say 00:56
camelia rakudo-moar 315ec6: OUTPUT«3␤»
psch shrugs.
m: my @a = <1 2 3>; @a[+@a - 1].say
camelia rakudo-moar 315ec6: OUTPUT«3␤»
psch .oO( superstitious plus )
vendethiel whoops, time to sleep 01:00
g'night, #perl6 :-)
psch 'night vendethiel 01:01
raiph \o ven 01:05
psch sleep & 01:59
raydiak we need more 6ers in like new zealand and eastern russia to fill this daily lull in my entertainment^W^W#perl6 :) 03:23
j4janicej raydiak: Will Western Canada work?! 04:43
-!
fuck
raydiak heh why not? 04:47
j4janicej: puts you in the same time zone as me, if I'm not mistaken 04:48
j4janicej raydiak: you're mistaken but you're real close :)
i'm not a perler, but i wish i was :P 04:49
my focus has been on resolving gender identity issues
raydiak perl is a good hobby to have 04:50
j4janicej not really
moritz \o 08:00
vendethiel .u doughnut 08:42
yoleaux U+1F369 DOUGHNUT [So] (🍩)
vendethiel ...
masak vendethiel: I feel like there's something awesome I'm not quite getting about the metacircularity I just implemented. 10:46
vendethiel: it's like, ok, now I have the interpreter implemented using only a few primitives that the interpreter defines. I get that part. in a sense, it's beautiful, and... economical. 10:47
vendethiel: but I'm not getting the religious experience conveyed in weblog.raganwald.com/2006/11/signif...ar_22.html , for example.
"This is the other half of the power of Lisp: if you want to change deeper fundamental language features like whether you have a Lisp-1 or a Lisp-n, or whether all evaluation is lazy, or…, or…, or… you can, because Lisp interprets Lisp and Lisp compiles Lisp." 10:48
feels like the only thing I've managed so far is to make my original Perl 6 `eval` run slower, by implementing a Lisp `eval` on top of it ;) 10:49
I *suspect*, but cannot prove, that this is because I need to "close the circularity" in more places than I have so far. 10:50
masak thinks 11:03
basically, any newfound power/flexibility needs to rely on the fact that there's now an `eval` function in the language, which can do dynamic interpretation.
vendethiel masak: I'm not the one saying "wow, lisp is easy to implement in lisp" :) 11:21
(ftr, that quote put dijkstra away from lisp for a long time :P)
dalek nda/reporter: 5094916 | (Tobias Leich)++ | bin/panda:
implement panda option "smoke"
vendethiel and, I said it before I think, lisps made some sacrifices to remove some meta-circularity for performance reasons 11:22
masak: did you push? 11:23
.oO( a meta-circular dougt-nut 🍩 )
 11:36
masak vendethiel: I'm just trying to understand the details of the charm of metacircularity. 11:40
vendethiel charm's not my biggest asset :P 11:41
masak: why do you like operators being able to be written in perl6
both for the implementor and for the user?
masak I like two things about that. 11:42
(a) I can define an operator, and it's instantly available in my scope. that's me, the user, *extending the language*. 11:43
(b) I also think it's way cool that Perl 6 takes seriously the idea that operators are just funnily-spelled subroutines.
vendethiel for me, I add (c) I like the fact that the *language* uses the same constructs as the *user*
masak: maybe relating those points to metacircularity can help :) 11:44
masak why is (c) a practical benefit, as opposed to a "wow, that's neat" thing?
but yes, I see your point about all this relating to metacircularity. 11:45
interestingly, though, Perl 6 (as implemented in Rakudo) isn't metacircular.
NQP is, in the sense that C is. (but not in the sense that Lisp is, I think)
vendethiel because that means (a) it's not magical and faery thingie (b) it obeys rules I know (c) I'm familiar to and able better comprehend/relate/understand it
vendethiel nothing can be crazy, or at least not crazier than what I can do myself 11:46
to me, this kind of power means "not much language magic"
masak got it.
vendethiel I don't have any answers, btw. I only have pointers. I'm interested in answers as well 11:47
python as a language is saner than ruby. In "theory", it's a safer option with less corner cases. But interestingly, ruby's craziness (perlness?) has it with a lot of good APIs (imho) 11:49
.oO( it's easy with EVAL )
 11:50
masak *nod* 11:56
awww. I found a really cute use for an introspective macro in some piece of code I was writing. it would genuinely eliminate boilerplate from my code.
but when I try it, I run into some well-known lexical macro bugs. 11:57
masak makes a note to try this again later after fixing said bug
vendethiel ha! :) 11:58
(what was it?)
masak the bug, or the macro? 11:59
vendethiel I think I know the bug 12:00
so, the macro
masak mind if I keep it to myself until I can write it up properly? ;)
it would be so much cooler to show if it actually worked.
vendethiel alrighty.
masak even though the introspection it does is textual -- no Qtrees yet. 12:01
vendethiel m: say quasi{1} eqv quasi{1}
camelia rakudo-moar 315ec6: OUTPUT«False␤»
vendethiel masak: ^ how do I introspect?
masak m: macro foo($expr) { say $expr.Str; Nil }; foo "OH HAI" 12:02
camelia rakudo-moar 315ec6: OUTPUT« "OH HAI"␤»
masak m: macro foo($expr) { say ~$expr; Nil }; foo "OH HAI"
camelia rakudo-moar 315ec6: OUTPUT« "OH HAI"␤»
vendethiel I guess :-)
it's *not really* what I'm looking for, but close enough
masak cf "purely textual" above.
I agree it's not optimal :) 12:03
but it... opens up certain doors.
vendethiel the question being... 12:04
do you want to open these doors?
.oO( macros are more of an ethical and philophical issue, or so it seems )
 12:05
masak I want to open these doors. in the long run, I want to open them with Qtrees, not with text. 12:08
and yes, there is ethics in there. but the same could be said about giving users access to subroutines. very dangerous tech!
wollmers .tell timotimo thx for %target.push: %(<a b a b>.pairs).invert; I will use it in my port port of Algorithm::Diff. 12:17
yoleaux 1 Nov 2014 08:59Z <FROGGS_> wollmers: It looks like /:P5\X/ is post NFG... because right now "E\c[COMBINING GRAVE ACCENT]" ~~ /./ only matches 'E'
yoleaux wollmers: I'll pass your message to timotimo. 12:17
dalek rlito: af53198 | (Shlomi Fish)++ | / (3 files):
Made the error messages better.

Now /usr/bin/node displays the message's text and a backtrace. Thanks to Havvy from #Node.js on Freenode for some insights.		 12:23
ab5tract hmmm... i baggified (<=) before noticing that (<+) seems to be conceptually exactly the same 13:14
ab5tract as a baggified (<=), at least as far as i reasoned it 13:15
ab5tract does it make sense to have a separate baggy-specific operator here? 13:18
ab5tract it almost feels as if it is a holdout from before MMD was implemented? 13:23
timotimo unlikely, we must have had MMD for a very long time 13:29
yoleaux 12:17Z <wollmers> timotimo: thx for %target.push: %(<a b a b>.pairs).invert; I will use it in my port port of Algorithm::Diff.
ab5tract timotimo: hmmm.. do you have any thoughts on the overlap? should (<=) be left set only? what about (<)? because there is not a strict Baggy subset operator currently 13:34
timotimo hm, good question
i don't have a good intuition on that yet; have mostly been concerning myself with sets rather than bags 13:35
and then there's Mixes as well
are they called mixes still?
ab5tract haven't heard anything about mixes
timotimo basically bags with Num values
rather than Int
ab5tract that's a pretty confusing name 13:36
ab5tract just notices them for the first time on: doc.perl6.org/language/Sets_Bags_an...#class_Mix 13:37
here is the only mention in set_operators.pm: sub mix(*@args --> Mix) { Mix.new(@args) } 13:38
timotimo hmm
ab5tract and i don't see a test file for mixes in S03-operators 13:39
whereas set and bag both have files 13:40
timotimo m: say Mix
camelia rakudo-moar 315ec6: OUTPUT«(Mix)␤»
ab5tract m: my $m = mix( %( 1 => 2.3, 2 => 0.7 ) ); my $n = mix( %( 1 => 3.3, 2 => 1.7 ) ); say $m (<+) $n 13:49
camelia rakudo-moar 315ec6: OUTPUT«False␤»
ab5tract m: my $m = mix( %( 1 => 2.3, 2 => 0.7 ) ); my $n = mix( %( 1 => 3.3, 2 => 1.7 ) ); say $m (<) $n
camelia rakudo-moar 315ec6: OUTPUT«False␤»
ab5tract m: my $m = mix( %( 1 => 2.3, 2 => 0.7 ) ); my $n = mix( %( 1 => 3.3, 2 => 1.7 ) ); say $m (<=) $n
camelia rakudo-moar 315ec6: OUTPUT«False␤»
ab5tract all of which should return true 13:50
at least, in my opinion
masak agrees 13:51
timotimo yup 13:52
ab5tract i'm willing to code that in :)
however, i'd argue that (<+) doesn't add anything 13:53
if (<=) is meant to DWIM for bags and mixes
timotimo what is the supposed role of <+? 13:54
ab5tract "Returns True if $a is a Baggy subset of $b, i.e., if all the elements of $a are in $b and each argument of $a is weighed at least as heavily as the element is in $b." 13:55
timotimo huh.
ab5tract sorry, that is for (>+)
timotimo: hence my suspicion about MMD being missing... otherwise it's existence barely makes sense 13:56
maybe masak has a clue?
masak the only thing I can think of is that they maybe somehow differ in their coercion behavior. 13:59
but I struggle to see how that'd matter for comparison ops. 14:00
we have something similar going on, for example, between && and ?&
m: say 42 && "OH HAI"
camelia rakudo-moar 315ec6: OUTPUT«OH HAI␤»
masak m: say 42 ?& "OH HAI"
camelia rakudo-moar 315ec6: OUTPUT«True␤»
masak so (x ?& y) is the same as ?(x && y) 14:01
except that I think that ?& also doesn't short-circuit.
m: sub bar { say "no short-circuit!"; True }; say ?(0 && bar)
camelia rakudo-moar 315ec6: OUTPUT«False␤»
masak m: sub bar { say "no short-circuit!"; True }; say 0 ?& bar
camelia rakudo-moar 315ec6: OUTPUT«no short-circuit!␤False␤»
masak right.
ab5tract masak: as far as i can tell, there is no coercion in the comparison ops 14:02
masak I could see coercion from Bag to Set being relevant sometimes.
there are some bags one which would be smaller than the other (because different element counts) 14:03
ab5tract erm, sorry. that's not a true statement. we coerce from Any to Set
masak but projected down as sets they could be equal (because same elements)
ab5tract right
masak actually, the same thing might happen when projecting from mixes to bags
so *if* all the ops would survive, I'd say they'd have to survive based on a difference in coercion like that. 14:04
otherwise -- if it can be shown there is no such difference -- then yeah, we probably have too many right now :)
ab5tract but wouldn't it be better to lift any coercion out of the ops themselves?
so if you want to check the bag as a set with (<), you would call $bag.Set (<) $other.Set 14:05
masak there's an argument can be made for that, yes.
and maybe it should be.
ab5tract of course Any -> Set coercion would still make sense as a default
masak if that road is taken, it makes sense for the comparison ops we have left to use as much information in the operands as possible -- that is, always perform the moral equivalent of a mix comparison. 14:06
but I hasten to add that the default in Perl tends to be to place the coercion in the operator.
ab5tract i'm not sure i see another approach that doesn't involve creating two new ops for bags ( we are missing strict subbag and strict superbag ) and four new ops for mixes 14:06
masak *nod*
timotimo coercing "Any" as a general rule could only ever result in Bag, not Mix 14:07
if you have a Hash, though, that could result in a Mix
ab5tract timotimo: well, currently all Any's coerce to Set for the comparison (excluding for now (<+) in the discussion) 14:08
i would put it like this
(Any, Any) -> Set behavior
(Set, Any) -> Set behavior
ab5tract sorry, that should be (Set, Any) || (Any, Set) 14:09
(Bag, Any) || (Any, Bag) -> Bag behavior
etc
that way we still have sensible coercion
within the op itself, without blowing up the number of ops 14:13
btw, is it possible that parameter values can be an infinite list?
because coercion for LoLs in a set operator like (^) is also very complicated.. should it only ever coerce based on it's first argument? 14:14
[(^)] $a.Set, $b.Bag, $c.Bag # should that coerce all the bags to sets or the set to bags? 14:15
if the list is finite, you could give precedence to bags over sets
but if not...
ab5tract the whole lazy list thing is new ground for me, so i might not be thinking about it correctly 14:17
masak it doesn't make sense to me that the coercion happen based on knowledge of all the elements of the list. 14:20
it should happen pairwise, for each application of (^)
or not at all
I guess because I consider (^) to be a binary op, not a chaining op. rightly or wrongly. 14:21
ab5tract that makes sense to me 14:22
i should have chosen a different op for the example :) 14:23
masak no, it's very interesting that you choose this one. 14:27
because I'm kinda changing my mind :)
ab5tract heh
masak m: say True ^ True 14:28
camelia rakudo-moar 315ec6: OUTPUT«one(True, True)␤»
masak m: say ?(True ^ True)
camelia rakudo-moar 315ec6: OUTPUT«False␤»
masak m: say ?(True ^ True ^ True)
camelia rakudo-moar 315ec6: OUTPUT«False␤»
masak yeah.
m: say True ^^ True
camelia rakudo-moar 315ec6: OUTPUT«Nil␤»
ab5tract i poked at it before, and then got to this exact question and wasn't able to progress further
masak m: say True ^^ True ^^ True
camelia rakudo-moar 315ec6: OUTPUT«Nil␤»
masak ^ and ^^ are clearly chaining in nature.
so maybe for consistency, (^) should be, too. 14:29
ab5tract okay, so the precedence would be: mix > bag > set ? 14:32
masak not sure which way you mean that. 14:35
but it would make sense to me if it always used the simplest type it could for the result.
ab5tract in terms of precedence: Mix (<) Bag # coerce Bag to Mix 14:36
Set (<) Bag # coerce Set to Bag 14:37
but actually when i am thinking about it, using the simplest if the arguments are not homogeneous also makes sense 14:40
and perhaps more so :)
masak no, I think we mean the same thing. 14:42
I meant the simplest it *can*, without losing data.
ab5tract oh :)
masak so I agree with your two concrete examples.
ab5tract \o/ 14:44
ab5tract so then, should i take the approach of adding multi signatures for each combination, or should i take the approach of coercion within a single Any,Any sub based on inspecting the types of the operator? 14:46
considering that the logic for computing all of them will look the same, ie: so [&&] $a.keys.map({ $a{$_} > $b{$_} }) 14:47
masak (a) whatever makes the tests pass (b) that looks nice 14:48
ab5tract i was wondering if there were performance implications in either approach 14:49
masak that's not a question that we can/should try to answer without measuring anyway
the JIT is already more clever than we are
ab5tract hehe
it's a shame, junctions would really make the single sub approach a lot more comfortable 14:50
gist.github.com/ab5tract/7b8d2a82a2674f4d41be 14:54
ab5tract masak: i updated the gist with two potential approaches. one is more human intensive, the other i can only imagine is more computer intensive. but maybe there is a cheaper way to approach what i am using th junction + given/when for 15:05
masak have you run either variant against the spectests? 15:06
ab5tract i haven't met spectests yet
ab5tract goes for a look 15:07
vendethiel oh, fosdem is in bruxelles? 15:09
masak always in Bruxelles, AFAIK. 15:19
vendethiel I was worried I couldn't be back to france on that because of traveling stuff. But I can do france->bruxelles in like an hour 15:20
grondilu will there be P6-related stuff in FOSDEM? 15:22
masak yes.
vendethiel yep
moritz when implementing a game with several phases (where each phase has different rules), how do I best do that? 15:36
moritz have different rulesets for each phase, and a state machine? 15:36
vendethiel moritz: a state machine of a class Rule?
moritz vendethiel: currently I tend to think that the state (machine) lives inside the Game, and the Game object has a ruleset for each phase 15:38
and maybe a Ruleset also returns the next phase 15:39
vendethiel you have to make sure you won't lose the game *g*
yes, it makes sense to me as well
moritz in sovjet russia, the game loses me :-)
masak states as objects makes sense to me. 15:40
all states implementing some common interface.
vendethiel y u state? :p 15:41
masak ...because a state machine has states? 15:53
if it makes you feel all warm and fuzzy inside to think of it as a graph of labeled nodes with purely functional transitions, then you have my blessing. :P 15:54
vendethiel I'm kidding, I'm kidding. 15:57
masak .oO( it was a referentially transparent joke! ) 16:08
raydiak \o 16:27
vendethiel o 16:30
raydiak omg ven where did your arm go!?
vendethiel I don't know :(. And damn, it was already painful with only one, but now... 16:31
raydiak haha...working on anything interesting? 16:32
vendethiel nope. php. 16:33
raydiak guess it can't all be fun and games :) 16:34
[Coke] tries to fix his daughter's linux laptop and tries not to just throw it in the garbage. :P 16:51
grondilu [Coke]: which distrib? (just curious) 16:58
ab5tract does 'make spectest' give benchmark data? 17:00
timotimo "time make spectest" would 17:01
ab5tract ok
timotimo it'd not be terribly precise, mind you
ab5tract i'm just curious how to satisfy my curiousity for which approach is better: gist.github.com/ab5tract/7b8d2a82a2674f4d41be 17:03
erm, yes, that's me, curious^2 17:04
timotimo i don't think the junction approach will work particularly well :(
timotimo but i'm first afk 17:05
[Coke] grondilu: ubuntu- the self update to 14.1 froze the machine, now it's unbootable. tryingt o make a usb key to boot from, but that also is unbootable. I can see my kid's home dir if I drop into recovery mode, but have no way to get the stuff off. Will bring it into work tomorrow and see if I can get someone with more experience to help me out.
ab5tract [Coke]: my suggestion would be to try a different distro on the usb key. if that boots you can mount the home directory and transfer off whatever you want 17:08
something like crunchbang
itz I can recommend "Recovery is Possible" (RIP) linux distro for repair 17:10
ab5tract timotimo: that can also be unravelled out of using a junction to ` if $a ~~ Mix or $b ~~ Mix {} `
ab5tract but yeah, it seems to me that the better way will be individual multi definitions which cover all the combinations 17:11
[Coke] itz, ab5tract, thanks. 17:12
(part of my issue is that the other machine I have to make the keys is an os x box, which makes me question some of the directions I'm trying. I -think- I'm using the mac to create a pc bootable key, but Iunno. :)
itz depending on distro dd(1) may work for USB key install 17:14
although not always :
:)
timotimo ab5tract: the optimizer can only unravel these junctions syntactically 17:26
the given/when will not allow it to do that
ab5tract timotimo: i meant i could manually unravel them :) 17:31
timotimo you can manually write "if $a | $b ~~ Bag" 17:32
that could work, but i'm not sure if the optimizer does that for ~~, because ~~ directly translates to .ACCEPTS
in fact, i don't think it does work
ab5tract right
then i will go the long-hand route
timotimo quite sadly, having elegant code in the setting is not always an option ;( 17:33
vendethiel yeah, you have to leave timo write some code as well!
the guy wants to participate
timotimo wow, sick burn 17:34
vendethiel i k r m 7
timotimo u wot m8
vendethiel hugs his timo 17:35
.oO( yer a timo, timo )
jepeway ola, #p6 17:37
m: $*TZ = 0 # worky?
camelia rakudo-moar 315ec6: OUTPUT«Cannot modify an immutable Int␤ in block <unit> at /tmp/YCPQZ_Oe9u:1␤␤»
vendethiel m: my $*TZ = 1;
camelia ( no output )
jepeway aha! tyvm, ven. 17:38
um...why worky?
vendethiel m: my $a ::= 1; $a = 2; 17:39
camelia rakudo-moar 315ec6: OUTPUT«Cannot assign to an immutable value␤ in block <unit> at /tmp/evZKDyfYyt:1␤␤»
vendethiel oh, that's implemented?
jepeway just notices that clogs auto-update, no more button-mashing 17:41
moritz jepeway: not intentionally; I think that using the space bar for scolling also triggers the button, or something 17:44
ab5tract timotimo: it was hard to imagine that having a code block with conditional logic would somehow work better than leveraging multi method dispatch
ab5tract if so, it would indicate to me that MMD would need optimizing 17:45
timotimo very much depends
ab5tract go on :)
timotimo sometimes we can do the dispatch at compile time, sometimes we can do it at spesh time, we can always have a little cache at spesh time to give a little boost if it happens to be right the next time 17:46
timotimo er, we *will* always have a little cache 17:46
ab5tract ok
ab5tract but is that VM-independent? 17:47
timotimo no
spesh is only on moarvm
jvm does what it wants :P
vendethiel jvm does stuff. Might be what it wants, might be what you want, might be what the universe wants :P 17:47
ab5tract right 17:47
jepeway moritz: could be browser-based behavior, i s'pose (safari, fwiw) 17:50
moritz jepeway: I've observed the same on firefox, so it's likely my sloppy js+DOM knowledge 17:51
ab5tract then it seems to me that i shouldn't be coding to the optimizer but rather to a baseline "less costy" approach, no?
timotimo "the optimizer" is cross-backend, but only works on the AST 17:52
ab5tract i see! that's good news. to the JIT then 17:53
timotimo mhm
timotimo well, at least having natives variables is very good for all backends 17:53
timotimo of course the jit on moarvm and jvm's code-gen both get very happy when they see native variables 17:54
timotimo well, to be fair, parrot is also happy about native vars 17:56
moritz timotimo: did you manage to fix your GC+nativecall issue? 17:58
from yesterday or Friday or so 17:59
timotimo yes, it turned out i modelled the CStructs from SDL with completely wrong sizes and SDL wrote some data over the boundaries of these structs 18:01
so it was my fault, really
moritz timotimo: still, cookies were promised. /msg me your snail mail address :-)
timotimo omygosh! cookies! 18:04
timotimo hum 18:25
generating an enum at compile time is giving me trouble of the "cannot invoke null object" kind
oh, i think i know what i did wrong
jepeway m: $*TZ.say 18:33
camelia rakudo-moar 315ec6: OUTPUT«3600␤»
jepeway m: $*TZ.say ; my $*TZ=0; $*TZ.say # hrm? 18:34
camelia rakudo-moar 315ec6: OUTPUT«(Any)␤0␤»
jepeway why that first (Any)? 18:34
FROGGS_ because you (re)declared $*TZ in that block 18:35
m: $*TZ.say; { my $*TZ=0; $*TZ.say }
camelia rakudo-moar 315ec6: OUTPUT«3600␤0␤»
jepeway so, "my" was compile-time; setting to 0 was runtime; printing before set meant (Any)? 18:36
FROGGS_ m: my $foo = 42; { say $foo; my $foo = 21 }
camelia rakudo-moar 315ec6: OUTPUT«===SORRY!=== Error while compiling /tmp/t5kh__Vxu2␤Lexical symbol '$foo' is already bound to an outer symbol;␤the implicit outer binding must be rewritten as OUTER::<$foo>␤before you can unambiguously declare a new '$foo' in this sc…»
FROGGS_ that explains it better perhaps
jepeway: but yes, your explanation is correct 18:37
jepeway and "my" affects entire block, not just subsequent statements within block? 18:38
FROGGS_ correct 18:39
jepeway ok. tad surprised, but grokking will come, I s'pose. 18:40
jepeway well, hopes, anyways 18:41
timotimo so ... how do i create an enum at compile-time without going through the MOP? 18:56
or through EXPORT
vendethiel- m: say ('Prefix' andthen "$_ * ") ~ 'topic name' 18:59
camelia rakudo-moar 315ec6: OUTPUT« * topic name␤»
vendethiel- m: say ('Prefix' andthen { "$_ * " }) ~ 'topic name' 19:00
camelia rakudo-moar 315ec6: OUTPUT«Prefix * topic name␤»
jepeway m: class I { has int $.i; method p(Int $ii) returns I { I.new(i => $.i + $ii) } } ; class J is I {}; my $j = J.new().p(2) ; say $j # what's a good way to keep $j a (J)? 19:58
camelia rakudo-moar 315ec6: OUTPUT«I.new(i => 2)␤»
timotimo you want to keep $j an undefined instance of J?
jepeway sorry, no, too loose with the notation, i guess. i meant "what do i do to make $j be an instance of class J." override method p(), of course, but anything smarter? 20:00
timotimo you can use "self.new" instead of I.new 20:02
that should give you a new J if you call J.p
moritz replace I.new by self.WHAT.new
jepeway oh, derp.
moritz m: class I { has int $.i; method p(Int $ii) returns I { self.WHAT.new(i => $.i + $ii) }}; class J is I {}; my $j = J.new().p(2) ; say $j
camelia rakudo-moar 315ec6: OUTPUT«J.new(i => 2)␤»
jepeway m: class I { has int $.i; method p(Int $ii) returns I { self.new(i => $.i + $ii) } } ; class J is I {}; my $j = J.new().p(2) ; say $j # what's a good way to keep $j a (J)? 20:03
camelia rakudo-moar 315ec6: OUTPUT«J.new(i => 2)␤»
jepeway do i need the .WHAT? 20:04
and...what if I couldn't (easily) modify class I? 20:07
moritz then you override method p in the subclass 20:10
jepeway ayup. i guess i could grovel through all methods in I to find those returning an I and override them to return J's if I first implement J.new(I $i) 20:14
moritz in general, inheriting from a class only makes sense if that class was designed with inheritance in mind 20:18
if not, a has-a relation is often more useful and less surprising
jepeway what's p6 to call I.p() from within J.p()? a la Java's super.p()? 20:20
moritz nextsame/nextwith 20:21
FROGGS_ or self.I::p, right? 20:22
moritz yes
jepeway sure, prefer composition over inheritance. trying to implement a replacement class that overrides a few choice methods, so reached for inheritance first. 20:23
jepeway found perl6advent.wordpress.com/2010/12/1...s-cousins/ 20:29
dalek nda/reporter: cca058d | (Tobias Leich)++ | bin/panda:
initialize exclude param in target smoke
jepeway m: class DT is DateTime {} ; my DT $x .= new().later(seconds => 2) 20:36
camelia rakudo-moar 315ec6: OUTPUT«===SORRY!=== Error while compiling /tmp/CHKKZ0VTw1␤Two terms in a row␤at /tmp/CHKKZ0VTw1:1␤------> ss DT is DateTime {} ; my DT $x .= new()⏏.later(seconds => 2)␤ expecting any of:␤ method arguments…»
jepeway m: class DT is DateTime {} ; my DT $x = DT.new().later(seconds => 2) 20:37
camelia rakudo-moar 315ec6: OUTPUT«Must provide arguments to DateTime.new()␤ in method <anon> at src/gen/m-CORE.setting:13676␤ in any find_method_fallback at src/gen/m-Metamodel.nqp:2725␤ in any find_method at src/gen/m-Metamodel.nqp:988␤ in block <unit> at /tmp/8cmOXFOS33:1␤␤»
synopsebot Link: perlcabal.org/syn/S33.html#line_1
jepeway m: class DT is DateTime {} ; my DT $x = DT.new(now).later(seconds => 2)
camelia rakudo-moar 315ec6: OUTPUT«Type check failed in assignment to '$x'; expected 'DT' but got 'DateTime'␤ in block <unit> at /tmp/limvZMkFE2:1␤␤»
jepeway m: class DT is DateTime {} ; my DT $x = DT.new(now).later(hours => 2)
camelia ( no output )
jepeway cf src/core/Temporal.pm:340 or so. 20:38
bug? 20:40
moritz dunno if bug, but "less than awesome" for sure
I have a local patch that makes it more subclass-friendly 20:41
spectesting now
jepeway coolio. 20:44
well, need to run, but will backlog (& watch commits). ty, moritz++, timotimo++, FROGGS++ 20:49
moritz std: my $a = [1, 2]; 'abc' ~~ /$a[0]/ 20:54
camelia std 14ad63b: OUTPUT«Potential difficulties:␤ Apparent subscript will be treated as regex at /tmp/rk8o2TRNe_ line 1:␤------> my $a = [1, 2]; 'abc' ~~ /$a⏏[0]/␤ [0] appears to be an old-school character class; please use <[0]> if you␤ mean a character…» 20:55
timotimo have fun :)
dalek kudo/nom: 0a3c954 | moritz++ | src/core/Temporal.pm:
Make DateTime and Date more friendly to subclassing

  jepeway++
moritz it looks like tests 4 to 7 in S05-metasyntax/litvar.t all assume something other than STD.pm6 does 20:56
dalek href="https://cpandatesters.perl6.org:">cpandatesters.perl6.org: 12a8179 | (Tobias Leich)++ | TODO.md:
added TODO.md
 21:03
dalek rl6-roast-data: 6b2a7c3 | coke++ | / (5 files):
today (automated commit)
 21:07
timotimo scrolling starfield \o/ 21:09
FROGGS \o/ 21:11
lol: 21:13
Unsupported use of /i; in Perl 6 please use :i
at lib\Cache\Memcached.pm:479
------> $ip ~~ s:g [ \[ | \] ] = '' i⏏f $ip.defined;
timotimo hah, ouch
FROGGS did I mention already that I love my cpandatesters thingy? :D 21:14
FROGGS though, I really start to hate the peeps that are meant to supply my virtual server 21:15
thay guarantee that the server is set up within 24 hours, and, well, that would have been Thursday morning
raydiak wow 24 hours alone is bad enough...do they build each instance by hand or something? :) 21:17
timotimo FROGGS: if a shared host is enough for you, check out uberspace 21:19
they allow you to install and your own postgres, too
and run* 21:21
FROGGS timotimo: well, I also need cpu and run to some degree
which means that I ordered a six-core with 6 to 12 gig RAM or so 21:22
timotimo oh wow
so you're running your own tests
rather than just receiving panda reports?
FROGGS raydiak: normally it is done in minutes, but their order form is broken, so I've got a Schroedinger's Account 21:23
raydiak haha 21:23
FROGGS timotimo: I know that the cpantesters servers are quite huge, and I want to be prepared :D 21:23
colomon :)
FROGGS and yes, being able to smoke all commits would be sweet! 21:24
dalek kudo/nom: 8739ee4 | usev6++ | src/Perl6/Grammar.nqp:
Add method to throw typed exception X::Syntax::Regex::SolitaryQuantifier
kudo/nom: 2bef929 | lizmat++ | src/Perl6/Grammar.nqp:
Merge pull request #328 from usev6/nom

Add method to throw typed exception X::Syntax::Regex::SolitaryQuantifier
FROGGS right now my windows 7 box is smoking all ecosystem dists, before my ubuntu laptop did that 21:25
but I'd like to receive reports from real users
moritz does it report automatically? 21:29
FROGGS one of the next steps is his:
dalek href="https://cpandatesters.perl6.org:">cpandatesters.perl6.org: fe83cfa | (Tobias Leich)++ | TODO.md:
added TODO: implement Etags to allow caching of css, js and fonts
FROGGS moritz: yes, with panda (branch reporter) and the env var PANDA_SUBMIT_TESTREPORTS 21:30
moritz puts it into the bashrc 21:33
raydiak reading "cpandatesters" makes my brain thrash around like it expects a single result but gets a junction of "C panda testers|cpan da testers|cpan datesters|cpanda testers" 21:36
moritz ==> Successfully installed NativeCall
Could not submit test report: Failed to connect: connection refused
FROGGS: first piece of feedback: failing to submit a test report should not make panda die 21:38
FROGGS moritz: true, very true
moritz I did a 'panda install NativeCall Pod::To::HTML URI' but it died after installing NativeCall
dalek href="https://cpandatesters.perl6.org:">cpandatesters.perl6.org: bc77a9a | (Tobias Leich)++ | TODO.md:
added todo
 21:40
timotimo the profiler isn't terribly useful for nativecall-heavy stuff
FROGGS err, moritz++
raydiak: my current plan is to set up cpandatesters.org on my own server, and then perhaps make it also available as testers.perl6.org 21:41
moritz and if you are looking for more ideas: integration into modules.perl6.org 21:42
FROGGS and github issues if possible
raydiak FROGGS: awesome...I think it's clever, was just being amused by how a person's mind can make up its own interpretation separate from the known facts in other parts of the same mind 21:45
raydiak but really what I opened this window for was to ask about .perl 21:45
how much to de care about how fast it is?
or as a more direct question, is there any reason we haven't made it handle non-trivial reference structures yet? 21:46
^^how much do we
FROGGS I think moritz has mused about making that work, right?
moritz FROGGS: not me. I think it was lizmat++ 21:47
FROGGS ohh
hmm, quite possible :o)
raydiak I keep running in to it, b/c I sometimes tend to use network structures other than trees 21:47
FROGGS would be awesome to see that one fixed of course 21:49
raydiak it doesn't seem hard, but if we're worried about how many .perls we can do per second, I am probably not the one to write it :) 21:50
raydiak thus the "how much do we care how slow .perl is" question 21:50
moritz raydiak: well, correct is more important first 21:51
FROGGS it has always been "correct is better that fast"
aye
than*
raydiak hm...I may take a stab at it then 21:53
FROGGS ++raydiak
lizmat my main reason for not pursuing self referential structures in .perl
was that I didn't have a solution for representing them 21:54
thoughts on that are very welcome
moritz my idea would be a dynamic var
raydiak iirc, p5 Data::Dumper literally puts them in separate vars
moritz with obj => $str
and $str is the string which can be used to access stuff 21:55
lizmat my @a; @a[0] = @a # how would you represent that ?
timotimo 89.89% time spent in postcircumfix:<( )> 21:56
moritz lizmat: I fear it must be printed out exactly like that 21:57
moritz lizmat: probably wrapped in a do { } block 21:58
though it'll probably more look like do { my \PERL1 = Array.new(); PERL1[0] = PERL1; PERL1 } 21:59
raydiak data::dumper gives "$VAR1 = [ $VAR1 ]" :P
moritz raydiak: my perl 5.20 Data::Dumper hangs 22:00
sleep&
raydiak moritz: works on 5.18 here..."my @a; $a[0] = \@a; say Dumper(\@a);"
'night moritz
lizmat good night moritz!
moritz good night everybody 22:01
colomon o/. 22:02
raydiak \o you dropped something :) 22:03
colomon ;)
raydiak but honestly, even if there are cases that turn out to be a real pain, compared to going in to an infinite loop, even some bizarre internal error is better imo 22:05
raydiak I wasted more hours than I'll admit to the first time I ran into this :) 22:06
raydiak lizmat: so what do you think of moritz's do block approach? 22:13
lizmat looks good in principle... atm I'm busy with some other stuff, so I can't help right now 22:14
raydiak np I'll leave you to it...I'll kick the idea around later when more people are around, perhaps 22:17
dalek nda: 3ed91f4 | Mouq++ | bin/panda:
Fix `panda -h`

  `#=` was connecting the comments to the wrong MAIN candidates		 22:20
nda: 45675dc | (Tobias Leich)++ | bin/panda:
Merge pull request #110 from Mouq/master

Fix `panda -h`
tadzik #| :( 22:35
FROGGS :#( 22:38
timotimo now i run into a null pointer somewhere 23:40
strange
BenGoldberg . o O (In soviet russia, null pointer runs into you!) 23:41
timotimo the null pointer dereference happens during compilation of the module 23:42
i can't reproduce it outside of panda >:( 23:50

Powered by Cro and the Raku® Programming Language.

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

Thank you!