»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.
jnthn diakopter: same ;-) 00:00
jnthn wonders off to bed
diakopter \
supernovus night jnthn!
00:03 replore_ joined
sorear TimToady: What should be the result of my $x = 'a'; $x ~= $x for ^1024; say $x.chars? 00:04
TimToady: I see options 1. dies in loop 2. dies at 'chars' 3. Inf 4. 2**1024
00:04 ZapZ joined 00:05 tokuhirom left, wolfman2000__ joined 00:08 jferrero joined, wolfman2000_ left
supernovus Okay, I sent a "pull request" for those changes to IO::Socket::INET. Now to see why recv() isn't working as it previously did. 00:12
00:14 patspam left
TimToady sorear: well, it depends on if you have enough memory ;) 00:17
you might need to run it on the simulator that runs the multiverse
diakopter I posit option 0: the compiler's optimizer recognizes the overrun 00:18
TimToady I think option 1 is most likely even with most lazy rope implementations 00:20
diakopter: I suspect an optimizer that smart would generally use up more electrons than it was worth for typical Perl code 00:21
colomon hmmm. I guess I'd be very surprised if $x ~= $x for ^1024 were lazy. I mean, $x =. push: 00:25
argh
$x .= push($_) for ^1024 isn't excepted to be lazy, is it? (is what I meant with the second bit there.) 00:26
afk # putting H to bed
supernovus Okay, now if it wasn't weird enough, by changing recv() to get(), I can now send requests properly when telnetting to the port, but requests from Apache still aren't received until apache times out. Ah, the fun of socket stuff :-) 00:32
I'm calling it a night. Have a good * #perl6! 00:34
00:35 supernovus left 00:39 lestrrat joined
sorear TimToady: my ropes are not lazy, but they require O(log min(a,b)) time for appends 00:42
TimToady: so for a sequence of N doublings, total time is O(N^2), and so is total allocations
00:43 wolfman2000__ left
sorear option 1 would be implemented by adding extra code to the ~ operator to check if the logical string length has exceeded (u)int.MaxValue 00:43
option 2 would be implemented by adding extra code to .chars to detect overflow and throw
option 4 requires bignums in the internal nodes
#3 is probably easiest 00:44
actually the order of difficulty is probably #4 > #2 = #3 > #1
00:47 uasi joined
TimToady I suspect 2^64 positions in a string will be sufficient for some time yet 00:47
so forget 4 :)
sorear heh. ok. 00:48
TimToady well, depending on your bignum implementation...
sorear I wonder if what I'm doing technically counts as a "rope"
TimToady an implementation that steals pointer bits for ints could be very efficient :)
(memory-wise) 00:49
sorear I'm using a structure much closer to Hinze & Patterson's 2-3 finger trees, than to Boehm et al's 197x original
because it has much better bounds for many operations
like O(log n) worst case appends and O(1) amortized single-character appends
TimToady I always invent my own data structures anyway, so don't ask me :)
and I'm actually fine with any of those results... 00:51
00:52 patspam joined 00:53 patspam left, whiteknight left
TimToady as long as strings are relatively fast at representing everything from names to genomes 00:54
00:55 Moukeddar joined, Moukeddar left
cognominal welcome to Gattaca :) 00:56
TimToady TTAGGG & 00:57
01:05 patspam joined 01:06 whiteknight joined 01:14 wolfman2000 joined 01:19 Sarten-X joined 01:23 JimmyZ joined
[Coke] tadzik: the ping about the smolder server. 01:28
01:39 thou left 01:44 supernovus joined 01:45 whiteknight left
supernovus So, I think I know why my changes to IO::Socket::INET may not be working entirely right. The $.family, $.proto and $.type aren't being initialized in the new object, so when it builds its internal $!PIO, it doesn't have the same settings. I'm thinking someone more familiar with core work, NQP and the parrot socket library should probably take this one over. :/ 01:49
01:58 mkramer1 joined 02:04 supernovus left 02:05 patspam left 02:38 JimmyZ left
sorear jnthn: ping 02:43
jnthn: I am wanting to provide any necessary design assistance for the new NQP# 02:46
03:03 frhodes joined
sorear colomon: ping too 03:03
03:03 frhodes left
colomon sorear: I'm about to head to bed, but I can spare a couple of minutes. what's up? 03:03
sorear I have become aware of your mail 03:04
colomon ah
sorear 1, why does it make a difference where the complex trig logic is?
2, I'd be interested in switching to code generation for all the little operator functions
03:06 envi joined
colomon 1. It's mostly a matter of simplicity / consistency. For instance, tan($z) is most easily defined as sin($z) / cos($z). That's easy to do at the Perl 6 level (as Rakudo does it), and it's easy to do at the C# level (as I'm proposing Niecza might). But it's tricky to do in a mixed setting, as tan(Variable) cannot easily call sin(Variable) and cos(Variable). (Or maybe it can and I just don't see how?) 03:10
also, based on a quick, very naive analysis it seems like doing it entirely in C# should be faster? 03:11
2. Yeah, code generation can be very nice for that sort of repetitive code.
afk # off to bed 03:19
03:20 sftp joined
sorear phenny: tell colomon +1 to doing it all at the C# level. 03:20
phenny sorear: I'll pass that on when colomon is around.
03:44 daniel-s left
TimToady is starting to think that an object may only have one write owner at a time, and the purpose of an rw parameter is to change ownership temporarily 03:44
this would prevent both upstack and downstack write violations 03:45
this also resonates with the notion that an object may be owned only be one thread at a time 03:46
s/only be/by only/ 03:47
and it would be extra bookkeeping only for rw binding and write operations 03:49
(plus memory overhead of tracking current owner)
03:54 matt3black joined
TimToady also, it would not really need to be strict identity; N random bits is sufficient to reduce the chance of of not catching a write violations to 1 in 2^N, with less chance if the task is repeated with different random ids 03:55
03:56 satyavvd joined 04:02 birdwindupbird joined
TimToady bbl & 04:02
04:04 kaare_ joined 04:15 matt3black left 04:45 daniel-s joined 05:07 sftp left 05:19 koban` joined, koban` left 05:20 packetknife joined 05:23 Shozan joined 05:28 jevin joined 05:43 woosley joined 05:45 MayDaniel joined 05:54 packetknife left 05:58 packetknife joined 06:10 orafu joined 06:11 MayDaniel left 06:16 alvis left 06:19 cognominal_ joined, REPLeffect joined 06:30 alvis joined 06:52 hanekomu joined 07:04 frhodes joined 07:22 skangas joined 07:26 mj41 joined 07:30 wtw joined 07:37 pjcj joined 07:42 pnu joined 07:43 packetknife left 07:47 kaare_ left 07:48 replore__ joined, replore_ left
moritz good morning, #perl6 07:51
07:53 wamba joined
Shozan happy programmer's day! 07:55
moritz \o/ 07:56
snarkyboojum that's awesome - wish we had a Programmers' Day :) 07:59
informal or no :) 08:06
08:08 masak joined
masak Perl 6 Day! \o/ 08:08
moritz rakudo: say (1, 6, 36 ... * <= 256)[*-1 08:09
p6eval rakudo bb2a5a: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<[ ]>, couldn't find final ']' at line 1␤»
moritz rakudo: say (1, 6, 36 ... * <= 256)[*-1]
p6eval rakudo bb2a5a: OUTPUT«1␤»
moritz rakudo: say (1, 6, 36 ... * <= 365)[*-1]
p6eval rakudo bb2a5a: OUTPUT«1␤»
moritz rakudo: say (1, 6, 36 ... * <= 365)
p6eval rakudo bb2a5a: OUTPUT«1␤»
masak the criterion is backwards, innit? 08:10
moritz right
rakudo: say (1, 6, 36 ...^ * > 365)[*-1]
p6eval rakudo bb2a5a: OUTPUT«216␤»
moritz that should be programmer's day :-)
erm, Perl 6 programmer's day
b: say Date.new('2011-01-01') + 215 08:11
p6eval b 1b7dd1: OUTPUT«2011-08-04␤»
08:11 wamba left 08:12 ranguard joined, orafu left 08:13 orafu joined, satyavvd left 08:15 Alias joined
masak moritz: but I'm a Perl 6 programmer every day :P 08:15
moritz masak: congratulations. There are still days where I find my self doubting very much if I'm a Perl 6 programmer 08:16
because I feel I have a misunderstanding of some very fundamental language feature
masak I didn't say I was a *good* Perl 6 programmer. 08:17
In fact, I know I am not one.
my code keeps getting new bugs without me changing it. 08:18
08:19 wamba joined
jnthn *yawn* 08:22
morning, #perl6
moritz good morning jnthn
jnthn finds himself only mildly pleased to be concious again "so soon" after going to bed... 08:23
masak good Mornington, #jnthn
er, jnthn* 08:24
08:24 alvis left
masak .oO( The #jnthn channel ) 08:24
jnthn oh my, what on earth would be discussed there?
er, actually, don't answer that :P
masak beer! 08:25
Slavic languages!
metaobject protocols!
moritz jnthn!
masak pun!
08:25 daxim joined
masak puns* 08:25
actually, that sounds like an awesome channel.
08:26 dakkar joined
masak all day, discussing MOP examples using Slavic beer-named variables, making subtle wordplays about it all. 08:26
moritz ... in multiple languages, of course
masak it sounds like #perl6... 08:27
masak is reminded of the scene in "Being John Malkovich" where Malkovich has gone into the portal 08:28
www.youtube.com/watch?v=xX9MtuqVrUQ 08:30
moritz nom: say Iterable.method(:local) 08:35
p6eval nom bb2a5a: OUTPUT«Method 'method' not found for invocant of class 'Iterable'␤ in <anon> at /tmp/meFzebBoRK:1␤ in <anon> at /tmp/meFzebBoRK:1␤␤»
moritz nom: say Iterable.^methods(:local)
p6eval nom bb2a5a: OUTPUT«elems infinite item fmt Int Num Numeric Str␤»
moritz oh, my local copy was out of date 08:36
masak jnthn: did you write a spectest for Iterable.fmt? 08:37
08:37 kaare_ joined
moritz nom: say (1.3, 3.5, 3).fmt('%d', '_') 08:38
p6eval nom bb2a5a: OUTPUT«Too many positional parameters passed; got 3 but expected between 1 and 2␤ in method fmt at src/gen/CORE.setting:1413␤ in <anon> at /tmp/7xv7t_jGKa:1␤ in <anon> at /tmp/7xv7t_jGKa:1␤␤»
moritz that too should work
nom: say List ~~ Iterable
p6eval nom bb2a5a: OUTPUT«Bool::True␤»
masak b: say (1.3, 3.5, 3).fmt('%d', '_')
p6eval b 1b7dd1: OUTPUT«1_3_3␤»
masak submits rakudobug 08:39
moritz (there's a failing spectest for that)
masak nom: say (1.3, 3.5, 3).WHAT
moritz nom: say (1.3, 3.5, 3).list.fmt('%d', '_')
p6eval nom bb2a5a: OUTPUT«Parcel()␤»
nom bb2a5a: OUTPUT«1_3_3␤»
masak nom: say (1.3, 3.5, 3).^mro
p6eval nom bb2a5a: OUTPUT«Parcel() Cool() Any() Mu()␤»
moritz wonders why Parcel !~~ Iterable 08:40
that causes lots of code duplication or redispatches in Parcel.pm
will ask pmichaud++ when he appears
08:41 alvis joined 08:42 hanekomu left 08:46 koban` joined, koban` left
jnthn masak: No. There's an RT ticket for it, so I figured somebody else would make sure of that before closing it. 08:48
moritz masak: fwiw fmt.t has lots of rakudo skips (most of them LHF to implement, no doubt) 08:50
masak jnthn: ah, ok.
moritz you *can* turn them all into tickets, but I'd doubt the utility of that approach
masak me too.
my criterion for submitting RT tickets is something like "someone wants it". skipped spectests are slightly below that level. 08:52
dalek kudo/nom: ecce797 | moritz++ | src/core/Parcel.pm:
Parcel.fmt
 08:53
ast: faa58d4 | moritz++ | S02-names_and_variables/fmt.t:
test Range.fmt, unfudge Parcel.fmt tests
 08:54
jnthn sorear: (NQP# ;)) Great! :-) There's a bunch of things I'd like to ask at some point soonish (CPS vs stack reconstruction, efficient invocation, etc.) 08:59
09:06 alvis left 09:10 wamba left 09:13 daniel-s left
pnu goo.gl/8uKyk - (apparent) location of current #perl6 channel users 09:14
moritz pnu++ 09:15
pnu moritz: unfotrunately it's p5, not six.. 09:16
moritz I'm not really surprised that it's lots of US and EU folks, just that there are so many in Europe
moritz finds it amusing that szabgab is on the equator in the Atlantic Ocean :-) 09:17
ah, might be (0, 0) for "no data obtained" or so 09:18
pnu yes, well that's what lookup.geo.opera.com gives for his ip..
jnthn Nice :)
pnu 26 users didn't report ip address.. those are not shown but present in the kml (visibility hidden). 09:19
moritz it might also make sense to filter out those with the 'voice' flag 09:20
(bots)
pnu i added the (p5) code and source data i used to the same gist.. gist.github.com/1213459 09:23
mortiz: true..
09:23 alvis joined
masak wow, pnu++! 09:23
pnu moritz: *
TiMBuS haha wow. i live in victoria. 09:24
but victoria, australia. not victoria, canada
close though!
moritz lol
b: say (a => 1).fmt('%s:%s') 09:27
p6eval b 1b7dd1: OUTPUT«a:1␤»
moritz b: say (a => 1).fmt('%s', '%s')
p6eval b 1b7dd1: OUTPUT«Too many positional parameters passed; got 3 but expected between 1 and 2␤ in 'Enum::fmt' at line 5610:src/gen/core.pm␤ in main program body at line 22:/tmp/MPbUoIwfXU␤»
TiMBuS also laughed at where szabgab is located 09:28
maybe we should send a rescue boat...
masak no, he's managed until now.
he'll let us know if he needs rescuing.
bbkr good localime() perl6 09:29
masak bbkr++!
moritz can't be so bad as long as his IRC is still working :-)
masak right, exactly.
09:30 dakkar left, replore__ left 09:37 daemon left, shachaf left 09:38 takesako joined, Util joined, szbalint joined, ingy joined, odoacre joined, dukeleto joined, itz joined, s1n1 joined, senobmeht joined, mattp_ joined, daemon joined, shachaf joined 09:39 dakkar joined
dalek kudo/nom: d97997e | moritz++ | src/core/Enum (2 files):
implement .fmt in Enum and EnumMap
 09:39
ast: c70a045 | moritz++ | S02-names_and_variables/fmt.t:
rakudo unfudges (fmt)
 09:40
09:51 satyavvd joined, satyavvd left 09:52 satyavvd joined 09:54 Alias_ joined 09:57 Alias left 10:06 daniel-s joined, mj41 left 10:07 frhodes left 10:16 static_perl joined 10:18 replore joined
masak grrr. I still hate how array parameter binding works. 10:26
jnthn How so? 10:27
masak you take a named parameter :@a. you pass it on as a named arguemnt :@a to some constructor.
suddenly you have another layer of array.
scalars don't work like that. hashes don't work like that. but arrays do.
yes, I've been upset about this before.
jnthn nom: class Foo { has @.a }; my @a = 1,2,3; Foo.new(:@a).a.perl.say 10:28
p6eval nom d97997: OUTPUT«Array.new(1, 2, 3)␤»
jnthn nom: class Foo { has @.a }; my @a = 1,2,3; Foo.new(:@a).a.elems.say
p6eval nom d97997: OUTPUT«3␤»
jnthn masak: What am I missing? 10:29
masak jnthn: I... I've changed my mind. it's a bug in Niecza. 10:30
rakudo++
perl6: sub foo(:@a) { say @a.elems }; foo
p6eval niecza v9-23-g1a8efca: OUTPUT«1␤»
..pugs, rakudo d97997: OUTPUT«0␤»
masak sorear: ^^
masak submits nieczabug 10:31
10:31 replore left
jnthn :P 10:31
bbkr I cannot find that spec forbids many multi subs with the same signature: rt.perl.org/rt3/Ticket/Display.html?id=71536 - is this ticket still under discussion or should it be rejected? 10:32
masak bbkr: it's probably not spec, no. but I think it's common sense. 10:34
I invoke "things that can be caught at compile-time, should be"
jnthn If the spec doesn't say we must, then it's not a bug. 10:35
It's a worthwhile feature request though.
So, it shouldn't have the bug tag.
But it's an OK ticket.
Perhaps one of those things the optimizer can spot. 10:36
Or maybe we do it at czech time
masak removes the [BUG] tag 10:38
bbkr I agree with moritz here - having multis with the same name and signature is perfectly fine. That can be used in any plugin implementation, for example "role Twiitter { multi method broadcast( Str) {} }; role Facebook { multi method broadcast (Str) {} }; class Social does Twitter does Facebook {}". compile error in this case is harmful. 10:41
benabik bbkr: You expect it to invoke both? 10:43
jnthn bbkr: It's important to distinguish multi methods and multi subs here.
bbkr: The multi method case is potentially useful. The multi sub case can almost certianly never actually work 10:44
benabik: .*/.+ dispatchers probably would
bbkr yes, i expect both methods to be invoked if signature and name matches.
jnthn bbkr: *only* if you used .+ or .* to ask for that. 10:45
bbkr: Otherwise it's an ambiguous dispatch.
masak it's an interesting use case. but yes, the ticket was about multi subs. 10:47
bbkr jnthn: ticket was about signatures in general. you're right about subs. but for methods I should be able to define multis like in my example, and it should throw "ambigous dispatch" for .
but not for .*
jnthn bbkr: Yes, we're only suggesting compile time detection for multi subs, not multi methods. 10:48
masak perl6: sub foo(:@a) { say @a[0]; say "alive" }; foo(:a["bar"]) 10:49
bbkr ok, maybe it should be explicitly mentioned in ticket subject just for clarification?
p6eval pugs, rakudo d97997, niecza v9-23-g1a8efca: OUTPUT«bar␤alive␤»
masak perl6: sub foo(:@a) { say @a[0]; say "alive" }; foo(:a("bar"))
p6eval rakudo d97997: OUTPUT«Nominal type check failed for parameter '@a'; expected Positional but got Str instead␤ in sub foo at /tmp/an_6FTpM6d:1␤ in <anon> at /tmp/an_6FTpM6d:1␤ in <anon> at /tmp/an_6FTpM6d:1␤␤»
..niecza v9-23-g1a8efca: OUTPUT«(timeout)»
..pugs: OUTPUT«bar␤alive␤»
masak sorear: ^^
masak submits nieczabug 10:50
bbkr: yes, that's probably a good idea.
bbkr changed 10:52
masak <== @lunch 10:54
10:55 mj41 joined 11:04 benabik left 11:06 wamba joined 11:16 uasi left 11:38 benabik joined 11:40 Psyche^ joined, Psyche^ is now known as Patterner
tadzik yay, physics passed! \o/ 11:46
jnthn tadzik++
tadzik today's a good day 11:47
moritz \o/
tadzik.physics++
11:48 wamba left
tadzik tadzik.physics.make_immutable 11:48
aaaaand we're done here :P
masak :P 11:51
12:00 woosley left 12:04 cognominal_ left, cognominal_ joined 12:15 wamba joined 12:16 dalek joined, ChanServ sets mode: +v dalek 12:18 aindilis joined
colomon tadzik++ 12:31
phenny colomon: 03:20Z <sorear> tell colomon +1 to doing it all at the C# level.
bbkr nom: say 1.+WHAT # should this be allowed? 12:33
p6eval nom d97997: OUTPUT«===SORRY!===␤Cannot use .+ on a non-identifier method call at line 1, near " # should "␤»
12:33 bluescreen10 joined
jnthn bbkr: .WHAT is a macro, not a method call 12:33
moritz not sure; .WHAT is a macro, not an ordinary methods
s/s$//
jnthn So at best it'd compile, but explode at runtime
(method WHAT not found)
bbkr ok, so this behaves correctly now, right? rt.perl.org/rt3/Ticket/Display.html?id=72818 12:34
jnthn The error ain't so helpful
The ticket is wrong though
It thinks .WHAT is a method call. :) 12:35
bbkr jnthn: that's why i'm asking to confirm it. it's an old ticket. 12:36
jnthn nom: class A { method m { say 1 } }; class B is A { method m { say 2 } }; B.+m; B.*m;
p6eval nom d97997: OUTPUT«2␤1␤2␤1␤»
jnthn Looks like .* and .+ work just fine on actual methods.
So +1 to close.
With a note that WHAT isn't a method call.
bbkr jnthn: which file should I put tests in? i'm not sure if those belongs more to macros or to multi dispatch 12:37
s/those/it/
jnthn bbkr: There already are a bunch of tests for .+ and .*
calling_sets.t or something.
moritz S12-methods/calling_sets.t 12:38
tadzik nice users map
I'd like to see one which lists only the users who said something during the last year or so :)
moritz tadzik: do you want the user list? 12:40
bbkr i've checked calling_sets.t and there is no test which checks "dies_ok {$foo.*MACRO }", IMO worth adding
jnthn bbkr: OK, +1 12:41
bbkr: But it's actually a compile time error
So it'd need eval_dies_ok orso
tadzik moritz: that'd be cool, yes 12:42
moritz tadzik: I'll generate a list of all nicks that spoke in #perl6 this year, should be good enough for that purpose
tadzik moritz: thanks 12:43
moritz tadzik: moritz.faui2k3.org/tmp/active_perl6_nicks.txt
tadzik whoa, that was fast
the starred ones are the ones who spoke? 12:44
moritz tadzik: note that it contains "* $nick" for an action that $nick performed
so you can grep 'em out
tadzik oh, explains a lot
12:44 donri joined
tadzik huh, there are more of them than on the map 12:44
curl moritz.faui2k3.org/tmp/active_perl6_nicks.txt | wcl says '930' 12:45
moritz tadzik: the map was from a snapshot, iiuc
tadzik oh, ok
12:45 satyavvd left
moritz tadzik: and at any given time there are < 200 nicks in here, but over the year a lot more assemble 12:45
tadzik that's interesting
ok, I'll give it a try in some near time 12:46
benabik 774 once I filter out dups.
tadzik I'd rather grep out those from the original list that do not appear on the activity list 12:47
moritz that's probably the better approach
benabik (dups from actions, to be specific)
There are a fair number of nick and nick_ "dups" as well.
tadzik yeah, /tadzik/ appears 5 times :) 12:48
12:49 wamba left
jnthn Probably easy to regex away the ogonky though :) 12:49
moritz mysql knows nothing of identity, just of string comparison
benabik A few minutes of work could probably deal with the easy duplications. 12:50
If anyone really cared. (I don't.)
12:50 satyavvd joined
moritz no deduplication necessary, when the list is ony used to filter the nicks in the map 12:51
tadzik jnthn: ogonki :0 12:53
benabik The %:s/^* // dedup might be handy. :-D
tadzik :)
jnthn tadzik: oops, Slovak spelling ;) 12:54
moritz perl -pe 's/^\* //' < nicks.txt | sort -u
benabik moritz: You can spell it that way too. I had opened it in vim. :-D 12:55
12:58 JimmyZ joined
bbkr masak: what is the expected behavior in rt.perl.org/rt3/Ticket/Display.html?id=72820 ? 13:03
moritz std: &1 13:04
p6eval std bb4f150: OUTPUT«ok 00:01 121m␤»
13:05 satyavvd left
moritz std: &foo 13:05
p6eval std bb4f150: OUTPUT«ok 00:01 121m␤»
benabik &1? What's that mean? Does that return something like `-> { return 1 }`? 13:06
moritz why doesn't that produce the same error as foo() ?
benabik: &foo returns a reference to the routine named 'foo'
benabik moritz: Right. But I'm confused about it working on a number.
moritz benabik: so &1 would be a reference to a routine named '1'
benabik: and I'm confused about it not being a compile time lookup 13:07
benabik std: sub 1 { … }
p6eval std bb4f150: OUTPUT«===SORRY!===␤Malformed block at /tmp/SBxx_WovI3 line 1:␤------> sub ⏏1 { … }␤ expecting any of:␤ name␤ routine_def␤ trait␤Parse failed␤FAILED 00:01 117m␤»
bbkr you cannot name routine '1'
benabik std: sub 1 { ... }
p6eval std bb4f150: OUTPUT«===SORRY!===␤Malformed block at /tmp/_zoQ8AtasP line 1:␤------> sub ⏏1 { ... }␤ expecting any of:␤ name␤ routine_def␤ trait␤Parse failed␤FAILED 00:01 117m␤»
moritz nom: Q:PIR { .lex '&1' }; say &1
p6eval nom d97997: OUTPUT«Use of uninitialized value in string context␤Use of uninitialized value in string context␤===SORRY!===␤error:imcc:syntax error, unexpected '\n', expecting COMMA␤ in file '(file unknown)' line 841␤␤»
benabik std: sub '1' { ... }
p6eval std bb4f150: OUTPUT«===SORRY!===␤Malformed block at /tmp/U_Hs6qOCtu line 1:␤------> sub ⏏'1' { ... }␤ expecting any of:␤ name␤ routine_def␤ trait␤Parse failed␤FAILED 00:01 117m␤»
jnthn std: @1 13:08
p6eval std bb4f150: OUTPUT«ok 00:01 118m␤»
benabik moritz: The syntax for .lex is .lex <variable name> <register>
daxim has anyone already written working shapely/eyedrops style P6 programs? code.google.com/codejam/japan/tshirt.html
moritz nom: Q:PIR { .lex '&1' $P0 }; say &1
p6eval nom d97997: OUTPUT«Use of uninitialized value in string context␤Use of uninitialized value in string context␤===SORRY!===␤error:imcc:syntax error, unexpected PREG, expecting COMMA ('$P0')␤ in file '(file unknown)' line 137␤␤»
jnthn moritz: I think it may be parsed as an accidental side-effect of parsing $1, etc
moritz jnthn: sounds plausible-ish
jnthn nom: "foo" ~~ /(f)/; say $0 13:09
p6eval nom d97997: OUTPUT«f␤»
jnthn nom: "foo" ~~ /(f)/; say @0
p6eval nom d97997: OUTPUT«f␤»
jnthn nom: "foo" ~~ /(f)/; say &0
p6eval nom d97997: OUTPUT«f␤»
jnthn heh :)
moritz did that, probably
jnthn++
jnthn b: "foo" ~~ /(f)/; say &0
bbkr so this ticket should be rejected as not a bug?
p6eval b 1b7dd1: OUTPUT«f␤»
jnthn moritz: I think we may have always had it like that, just never realized ;)
moritz ok, I just reused it :-)
jnthn
.oO( let's call it a feature ;-) )
Guess a [ <?{ $<sigil> eq '$'}> <.panic: '...'> ] or so in the right place would catch it though. 13:11
moritz of course it's a nice obfu 13:12
benabik "Nice obfu" is not a good reason to keep it in the language.
moritz and @0 even makes sense, if the capture 0 is quantified
benabik: not? :-)
bbkr i'll copy-paste this discussion to ticket and leave it open 13:13
benabik moritz: :-P
13:14 colomon joined 13:16 pernatiy joined 13:18 tzhs joined 13:20 clairvy joined 13:21 mjreed joined, wolfman2000 left
mls good afternoon! 13:24
bbkr nom: Int.() # parse bug? despite doing nothing useful it should be parsed as Int(), right? 13:26
p6eval nom d97997: OUTPUT«Method 'postcircumfix:<( )>' not found for invocant of class 'Int'␤ in <anon> at src/gen/Metamodel.pm:3023␤ in <anon> at /tmp/8_3SpbH8XF:1␤ in <anon> at /tmp/8_3SpbH8XF:1␤ in <anon> at /tmp/8_3SpbH8XF:1␤␤»
moritz good pm, mls
jnthn o/ mls 13:28
mjreed Query: given a string $x, and an array @w containing a list of other strings, I want to test to see if every element of @w appears as a substring of $x. Is there a cleaner way than this? $x ~~ all( @w.map(-> $w { regex { $w } }) )
mls about 50% of the perl6 startup time is spent in compose 13:29
jnthn mls: That's not so surprising.
mls: We do quite a lot of work in compose.
mls: 50% is a bit more than I'd have expected; any hot-spots of worth knowing about. 13:30
mls: But also see my blog post from yesterday on the serialization stuff
moritz mjreed: I'm sure there's a more concise solution involving index()
jnthn mls: Essentially, the problem will go away because we'll be serializing the composed objects.
mls (compute_mro / c3_merge are the hotspots) 13:31
moritz mjreed: so $x.index(all(@substrings)).defined
jnthn mls: Really?
mls: That's...a little surprising.
moritz rakudo: my @s = <foo bar>; say so 'foobarbaz'.index(all @s).defined
p6eval rakudo d97997: OUTPUT«Bool::True␤»
moritz rakudo: my @s = <foo bak>; say so 'foobarbaz'.index(all @s).defined 13:32
p6eval rakudo d97997: OUTPUT«Bool::True␤»
moritz huh.
mls shall I try to optimize them a bit?
moritz rakudo: say 'foobarbaz'.index('bak')
p6eval rakudo d97997: OUTPUT«Str()␤»
moritz rakudo: say 'foobarbaz'.index('bak').defined
p6eval rakudo d97997: OUTPUT«Bool::False␤»
mls But, as you said, it's probably not worth it because of the serialization changes
jnthn moritz: .defined won't auto-thread there.
moritz jnthn: urks
benabik The junction is defined. 13:33
jnthn right
moritz right
PerlJam moritz: you could do it with a reductio op
er, reduction
moritz I want a way to say that stuff should (or should not) autothread
PerlJam: the old imperative way, yes
jnthn mls: Well, it is in so far as any program that we just compile and immediately run will have to do that.
benabik rakudo: my @s = <foo bak>; say so 'foobarbaz'.index(all @s)>>.defined
p6eval rakudo d97997: OUTPUT«Bool::False␤»
mls ok, I'll have a look at the code... 13:34
benabik rakudo: my @s = <foo bar>; say so 'foobarbaz'.index(all @s)>>.defined
p6eval rakudo d97997: OUTPUT«Bool::True␤»
moritz rakudo: my @s = <foo bak>; say [&&] @s.map({'foobarbaz'.index($_).defined})
p6eval rakudo d97997: OUTPUT«Bool::False␤»
moritz rakudo: my @s = <foo baz>; say [&&] @s.map({'foobarbaz'.index($_).defined})
p6eval rakudo d97997: OUTPUT«Bool::True␤»
moritz does >>. hyper descend into junctions? 13:35
jnthn No
Only into iterables.
mjreed rakudo: my $x = "this is a cat"; my @w = <this cat>; say $x.index(all @w)>>.defined
moritz when I don't see why benabik's solution should work
p6eval rakudo d97997: OUTPUT«all(True, True)␤»
benabik moritz: I wasn't sure if it would, but it appeared to. *shrug*
jnthn moritz: Yeah, I'm confused about that too...
mjreed hm. 2011.04 says "Method 'iterator' not found for invocant of class 'Integer' " 13:36
benabik b: my @s = <foo bak>; say so 'foobarbaz'.index(all @s)>>.defined
p6eval b 1b7dd1: OUTPUT«Method 'iterator' not found for invocant of class 'Integer'␤ in main program body at line 1:src/metamodel/RoleToInstanceApplier.nqp␤»
PerlJam perhaps the "randomization" that rakudo does accidentally listifies the junction
jnthn moritz: My only guess is that HYPER causes it to autothread.
Or something
PerlJam (for >> I mean)
jnthn Yeah...looks like something in the >>. implementation is declared with Any. 13:37
moritz src/core/metaops.pm uses implicit Any all over the place 13:38
13:38 shinobicl joined
jnthn There we go then. 13:38
shinobicl nom: my Int $number = 0 {where 0..23}; 13:39
p6eval nom d97997: OUTPUT«===SORRY!===␤Confused at line 1, near "my Int $nu"␤»
jnthn afk a bit...walk
benabik It working is kinda handy, since it provides a way to force things like .defined to autothread.
PerlJam benabik: except that it isn't really autothreading 13:40
moritz shinobicl: there 'where' must be outside block, for suer
it's auto-hyper-threading :-)
shinobicl nom: my Int $number = 0 where 0 .. 23;
p6eval nom d97997: OUTPUT«===SORRY!===␤Confused at line 1, near "my Int $nu"␤»
shinobicl nom: my Int $number = 0 where {0 .. 23};
moritz nom: my Int $number where (0..23) = 0;
p6eval nom d97997: OUTPUT«===SORRY!===␤Confused at line 1, near "my Int $nu"␤» 13:41
PerlJam nom: my Int where 0..23 $number = 0; # there, that's my permutation :)
p6eval nom d97997: OUTPUT«===SORRY!===␤In "my" declaration, typename Int must be predeclared (or marked as declarative with :: prefix) at line 1, near " where 0.."␤»
PerlJam oh, that's interesting that it complains about Int 13:42
13:42 bluescreen10 left
mjreed I'm still back on the fact that defined() doesn't descend into junctions. How is one meant to test for all-defined, any-defined, one-defined, etc? 13:45
13:48 Shozan left
mjreed I'm also kinda thinking that index called in Boolean context should maybe return true if the string is found and false otherwise, rather than just letting the index boolify and this yielding false for a prefix.. 13:53
PerlJam mjreed: I never have gotten it clear in my head what should/shouldn't autothread. Autothreading is meant to be one of the great features of junctions, so I'd wish that *everything* autothread and that there's be some syntactic marker that says "don't thread this operation"
mjreed Sure. But then you have trouble passing around junctions as junctions. I seem to recall the spec has gone back and forth on this a couple times... 13:56
PerlJam If you can wrap a junction in a non-junctional package, it becomes easy to pass around without invoking autothreading 13:57
13:57 bluescreen10 joined 13:58 daxim left 13:59 daxim joined
masak rakudo: sub contains($big, $small) { defined $big.index($small) }; my @bigs = <food foolish eufooria kungfoo>; say so contains all(@bigs), 'foo' 14:01
p6eval rakudo d97997: OUTPUT«Bool::True␤»
masak rakudo: sub contains($big, $small) { defined $big.index($small) }; my @bigs = <food foolish notthisone kungfoo>; say so contains all(@bigs), 'foo'
p6eval rakudo d97997: OUTPUT«Bool::False␤»
masak mjreed: almost like English!
masak considers writing a blog post called "Don't let the junction get away!" 14:02
mjreed what's "so" doing there? 14:03
moritz extra pionts for writing 'contains' as an infix
masak collapsing the junction.
PerlJam mjreed: emphasis ;)
moritz mjreed: boolean context
masak rakudo: sub contains($big, $small) { defined $big.index($small) }; my @bigs = <food foolish notthisone kungfoo>; say (contains all(@bigs), 'foo').perl
p6eval rakudo d97997: OUTPUT«all(True, True, False, True)␤»
mjreed synonym for ? then?
masak aye.
moritz but with looser precedence
masak like 'and' and '&&'. 14:04
jlaire rakudo: sub infix:<contain>($big, $small) { defined $big.index($small) }; my @bigs = <food foolish eufooria kungfoo>; say so all(@bigs) contain 'foo' 14:06
p6eval rakudo d97997: OUTPUT«===SORRY!===␤Confused at line 1, near "say so all"␤»
jlaire b: sub infix:<contain>($big, $small) { defined $big.index($small) }; my @bigs = <food foolish eufooria kungfoo>; say so all(@bigs) contain 'foo' 14:07
p6eval b 1b7dd1: OUTPUT«Bool::True␤»
moritz jlaire++ # eufooria
jlaire I just copied masak++ :P
jlaire--
masak jlaire++ # honest ;) 14:08
14:08 mjreed left, mjreed joined
moritz masak++ then, and jlaire++ for the infix implementation 14:08
mjreed er, I would still call it infix:<contains>, with singular inflection. Won't always be called with all(). :) 14:09
masak right.
PerlJam or reverse the $big and $small and call it infix:<in> :-) 14:10
mjreed how pythonic!
masak PerlJam: ∈
mjreed 'contains' reads pretty well as a method call on $big, actually.
moritz too general, if you ask me
'contains' is a question you ask a Set, not a Str 14:11
masak rakudo: use MONKEY_TYPING; augment class Cool { method contains($small) { defined self.index($small) } }; my @bigs = <food foolish eufooria kungfoo>; say so all(@bigs).contains('foo')
p6eval rakudo d97997: OUTPUT«Bool::True␤»
masak \o/ 14:12
mjreed Well, I'm not saying add it to core. But as long as you're defining a 'contains', it works as a method call as well as an infix.
jlaire was the parse fail a nom bug or shouldn't it work in b, either?
masak jlaire: I was surprised by it. 14:13
14:14 wtw left
PerlJam #p5p say the oddest things 14:17
jnthn PerlJam: I'm sure we could compete quite well on oddness on #perl6, if we put our minds to it ;) 14:18
masak news at 11!
PerlJam jnthn: in reference to your 6guts mention of reviving the CLR backend for rakudo ... <@Nicholas> and workng Perl 6 on $anything might be subvertable to compile Perl 5 14:20
jnthn ;)
arnsholt jnthn: In re "Bounded Serialization", will that most likely live in NQP or Rakudo land? 14:30
mls jnthn: compute_mro is missing to "trivial" case, i.e. +@immediate_parents == 1 14:32
jnthn arnsholt: NQP for sure.
14:33 sftp joined
shinobicl hi: how do i define a constraint in a variable without using a subset? 14:33
jnthn arnsholt: There's a bigger underlying plan that other languages built on NQP will be able to use this too.
arnsholt Yeah, that's what I was hoping
jnthn In general, instead of writing Compiler/Grammar/Actions, you'll write Compiler/Grammar/Actions/[other thingy we don't have a good name for yet]
Where that thingy handles the "model" of the program, including objects that should exist runtime and compile time. 14:34
arnsholt I've an on-again off-again relationship with a Prolog compiler, and to do things like assert and retract I probably want some kind proper serialisation facility
14:34 uasi joined
arnsholt My current approach is more a long the lines of "if all you've got is a hammer..." 14:35
jnthn "...every problem looks like a PAST tree" 14:36
:)
arnsholt How'd you guess? =)
masak because every problem *does* look like a PAST tree? 14:37
moritz once you've parsed and action'ed it, yes
14:37 mkramer1 left 14:40 Trashlord joined 14:42 thou joined 14:44 thou left 14:48 shinobicl left
masak Unhandled exception: System.InvalidCastException: Cannot cast from source type to destination type. 14:50
auuugh
PerlJam masak: is that niecza's "Null PMC" ? 14:52
dalek ast: 2eddda3 | moritz++ | S02-names_and_variables/fmt.t:
refudge fmt.t for rakudo
moritz nom: printf '%f', 5.6
p6eval nom d97997: OUTPUT«5.600000»
jnthn PerlJam: That'd be NullReferenceException 14:53
moritz is that... expected? normal?
nom: printf '%15f', 5.6
p6eval nom d97997: OUTPUT« 5.600000»
[Coke] rakudo: so say "we all"
p6eval rakudo d97997: OUTPUT«we all␤»
moritz nom: printf '%015f', 5.6
p6eval nom d97997: OUTPUT«00000005.600000»
moritz nom: printf '%.15f', 5.6
p6eval nom d97997: OUTPUT«5.600000000000000»
moritz b: printf '%f', 5.6
p6eval b 1b7dd1: OUTPUT«5.600000»
PerlJam moritz: looks fairly normal to me.
moritz PerlJam: some spectests expect that to be 5.6 14:54
masak PerlJam: no, but I've gotten it twice in two days now, so I'm starting to dislike it.
moritz $ perl -wE 'printf "%f\n", 5.6'
5.600000
moritz wonders what's up with the tests
mjreed $ printf "%f\n" 5.6 14:55
5.600000
even bash agrees.
[Coke] is nom now also rakudo on p6eval?
moritz yes, they are aliases 14:56
PerlJam rakudo: say "hi"
p6eval rakudo d97997: OUTPUT«hi␤»
PerlJam nom: say "hello"
p6eval nom d97997: OUTPUT«hello␤»
14:56 dukeleto left
mjreed b: say "I'm getting too old for this." 14:56
p6eval b 1b7dd1: OUTPUT«I'm getting too old for this.␤» 14:57
14:57 dukeleto joined
[Coke] p6devl: help! 14:57
p6eval: help
p6eval [Coke]: Usage: <(nom|pugs|perlesque|nqprx|yapsi|alpha|highlight|partcl|nqpnet|star|mildew|nqp|niecza|std|rakudo|b)(?-xism::\s) $perl6_program>
masak b: say "...but I still kick nom's butt in some regards"
p6eval b 1b7dd1: OUTPUT«...but I still kick nom's butt in some regards␤»
[Coke] Every time I do that, I'm surprised that partcl is in there. Every time.
PerlJam thought for a second that Coke was invoking a perl 6 devil
[Coke] and there you are!
moritz [Coke]: well, giving partcl a $perl6_program might not be too smart :-)
mjreed I thought maybe p6devl forwarded to the perl6-dev mailing list. :) 14:58
moritz our secret replacement for p6l
mjreed formed by the perl6 cabal.
PerlJam partcl: say hi
p6eval partcl 3977a9: OUTPUT«invalid command name "say"␤ while executing␤HLL::Compiler::eval␤ (file "<unknown file>" line 151)␤PCT::HLLCompiler::evalfiles␤ (file "<unknown file>" <unknown line>)␤PCT::HLLCompiler::command_line␤ (file "<unknown file>" <unknown line>)␤main␤ (file "…
PerlJam (just making sure ;) 14:59
[Coke] partcl: puts hi
p6eval partcl 3977a9: OUTPUT«hi␤»
moritz formed by the Perl 6 kobold
masak shudders
15:00 static_perl left
[Coke] partcl: puts [expr sqrt(64)] 15:00
p6eval partcl 3977a9: OUTPUT«Invalid expression␤ while executing␤expr␤ (file "<unknown file>" line 8)␤HLL::Compiler::eval␤ (file "<unknown file>" line 151)␤PCT::HLLCompiler::evalfiles␤ (file "<unknown file>" <unknown line>)␤PCT::HLLCompiler::command_line␤ (file "<unknown file>" …
[Coke] now, that's odd. ah well.
moritz masak: can you confirm my observation that backtraces wrongly contain the name of the current script even for routines executed from modules defined in other files? 15:02
masak moritz: no, I haven't seen that.
haven't been looking for it either.
I try to rely on backtraces as little as possible.
moritz why? 15:03
15:03 thou joined
masak because they're so *booooring* 15:03
[Coke] probably should not try to use JIRA web app shortcuts on every page he goes to. whoops.
masak moritz: when I do the dishes, I use my fingers to feel the surface of plates and cutlery, to see if they're clean. when I code, I do something similar.
moritz masak: you might need to come up with a module that makes backtraces more exciting 15:04
like, inluding animated ASCII art :-)
masak moritz: backtraces are like dishwasher fluid. too strong, too broad, inhuman.
jnthn ...and fairly useful when you're trying to reconstruct what just happened in code you wrote 2 years ago that failed. :) 15:05
masak I'm not saying they're a bad thing.
I'm saying I try to rely on them as little as I can.
TimToady backtraces should be written in HTML5, ovbiously 15:06
and I've always said they should probably go to disk by default, not the terminal
or to a backtrace cacher of some sort
masak +1
benabik Someone wrote a terminal program for OS X that used HTML5… It used JSON to communicate with "command line" apps running in it. It looked odd. 15:07
[Coke] like on OS X, TimToady ?
masak especially in Niecza, they're too long.
but nom is catching up, too.
jnthn \o/
oh, wait..
TimToady [Coke]: how would I know anything about OS X?
jnthn :P
[Coke] TimToady: the same way I would know how you would know anything about OS X?
TimToady did anyone see my speculations at irclog.perlgeek.de/perl6/2011-09-13#i_4419513 15:08
[Coke] but, yes, OS X has a service that tracks dumps for you, so you have a central location to go through your (C) backtraces.
moritz also wondered if backtraces should be gone by default, and only the calling location should be present, much like p5
TimToady I think they should *appear* to be gone, by default :) 15:09
masak benabik: this one? acko.net/blog/on-termkit
[Coke] I think by default, that's fine. I'd really like an easy way to turn on the full backtrace, though.
benabik masak: Looks like.
masak benabik: it's the closest I've come to wanting to leave the text-based CLI model.
moritz TimToady: sure. The full information should always remain somewhere behind the curtain.
TimToady [Coke]: I think you can turn it on retractively if it's been saved somewhere
moritz or rerun with a command line switch 15:10
like currently perl6 --ll-excpetion
masak benabik: I believe with slightly *less* eye candy -- or rather, something like a "feet still on the ground" attitude -- I might go for a project like that.
moritz thought that might become --bt=[none|concise|full]
benabik masak: It looked interesting, but it really needs to be a) cross platform and b) have a rich set of programs supporting it. 15:11
moritz *though
masak benabik: right. it's a bit too much of a reinvention. like Smalltalk and Plan 9.
benabik masak: Mostly agreed, although I tend to use Alfred for basic stuff instead. Launcher bars FTW
15:11 birdwindupbird left
TimToady I know, we'll send all backtraces to gist.github.com :) 15:12
the good ones end up there anyway ;)
moritz
.oO( killing me hardly with this gist )
TimToady "This program has terminated unexpectedly. Do you wish to send the backtrace to github?"
moritz (yes/no/show me already) 15:13
masak benabik: I use Do. when I was on Mac OS X, I used Quicksilver.
benabik: so, yes. I agree.
benabik masak: Alfred is Quicksilver but actively developed. :-D On Linux I haven't ventured past just a term window yet, but I don't live in it much. 15:14
TimToady jnthn, sorear: what do you think of the idea of passing write ownership of mutable objects around via rw binding, and having only one scope at a time be the write owner? 15:15
then a write just fails unless you are the write owner 15:17
jnthn TimToady: Can you define "scope" a little more?
TimToady the lexical scope into which the rw parameter is temporarily bound 15:18
jnthn So, routine-level scope, essentially?
TimToady more or less
kinda fits with only one thread owning write access to an object at a time too 15:19
jnthn hm, it has the benefit of being lexically scoped, which goes well with code analysis.
Yeah, I was just about to ask about that.
TimToady would have to figure out how elements of an array/hash delegate to the container object, I suppose
jnthn Well, I was about to ask about like @huge_thing>>++ 15:20
TimToady but I like that overall direction
moritz my $a = []; sub f($x is rw) { gather { for 1..10 { take $_; $x.push($_) } }; my @b := f($a); # can I modify $a here? or does $f still hold the writing permissions?
jnthn Where we want to break the array over many threads and do data-parallel operations
TimToady: I don't have any immediate reaction either way. I think I need to see a bunch of examples. 15:21
TimToady I think the write token is one aspect of a solution, but there needs to be some kind of dynamic delegation upward too, so that write access can be conferred on children too 15:22
jnthn TimToady: Nothing immediately makes me think "oh, could never work". I like the lexical nature of it, since that's pretty good from an analysis/optimization point of view.
masak sorear: under what conditions should I expect to get "Unable to cast object of type 'Niecza.P6opaque' to type 'Cursor'." ?
jnthn TimToady: But I can imagine there's tricky edge cases.
TimToady there's always tricky edge cases :)
jnthn TimToady: Sure, I put it badly. I more meant, edge cases that users are going to bump in to a lot. 15:23
TimToady anyway, that's the direction I'm thinking currently
jnthn TimToady: The other issue is who is responsible for restoring the old ownership.
s/old/previous/ 15:24
Things like gather/take could be interesting there too, as moritz++ pointed out.
As you may have to set/restore them over coroutine boundaries...somehow.
TimToady but coro/threading is precisely where we'd like better control of that anyway
jnthn Well, there is that. :) 15:25
So long as we don't get ourselves into deadlock issues.
15:26 supernovus joined 15:30 tzhs left
mls jnthn: startup is now 20% faster 15:32
jnthn mls: o.O 15:33
mls++
15:33 thou left
mls gisting... 15:33
diff: gist.github.com/1214122 15:34
(I hope the code is correct, haven't run spectest yet)
three changes: 1) the immediate_parents == 1 check, 2) call self.mro instead of self.compute_mro, 3) the publish_cache change 15:35
same thing for nqp: gist.github.com/1214140 15:39
moritz spectests the rakudo changes 15:40
mls thanks!
jnthn moritz++
masak mls: you, sir, rock. 15:41
mls I just looked at the profile output
benabik mls: The profiling that you got to work in the first place?
mls yes 15:42
masak mls++ mls++
benabik mls++
mls hey, calm down ;)
jnthn mls++ !!!
:)
masak mls: we want to imbue you with positive emotions now, so that you'll do it again :P 15:43
mls That's an old trick ;)
masak but it's working, right? :D
mls you'll see ;)
masak bounces
mls the qregex fail in nqp's "make test" is normal, right? 15:45
jnthn mls: The patch looks good to me, at a first glance.
moritz mls: right 15:46
rakudo spectest is fine so far (S03-sequence)
mls good. Now we just need to wait for the spectest result
15:47 thou joined
supernovus multi postfix:<**> (Numeric $a is rw) { $a *= $a; }; mls**; 15:47
jnthn If startup is 20% faster, should be a little less waiting ;) 15:48
mls unfortunately it's just startup time...
jnthn mls: Yes, but the test run causes 100s of startups :) 15:49
mls now a 20% faster setting compilation would be nice
masak .oO( easy, just compile the setting at startup! oh wait )
jnthn Well, we calculate the MROs during compiling the setting too, so it will have shaved a little off that at least.
moritz t/spec/S12-attributes/delegation.t ............................. Dubious, test returned 1 (wstat 256, 0x100)
mls (actually I can do 6% with my ugly parrot patch...)
masak "Dubious delegation" would be a great name for a rock band. 15:50
mls huh
moritz t/spec/S12-introspection/parents.t also fails
mls debugging...
jnthn The latter one is probably most telling.
[Coke] hio. Anyone here getting data from: smolder.parrot.org/app/projects/smoke_reports/5 15:52
moritz mls: with your patch, Str.^parents returns (Str, Cool, Any, Mu) 15:53
mls: should be (Cool, Any, Mu) only
jnthn [Coke]: Haven't been, though it certainly looks useful. 15:54
moritz sometimes does, usually before releases
[Coke] Ok. in fine perl6 tradition, I'm rolling a new one. 15:55
(the administration of that service is a pita, trying to make one that is easier to drop in on a server.) 15:56
jnthn gets curious why t/spec/S06-multi/type-based.rakudo fails on that box but not for him
[Coke] (and gives us more features). Comments welcome on the perl6-compiler thread (that's the right list, aye?)
15:57 Holy_Cow joined, kshannon joined
[Coke] jnthn: if you go to the report and click on "properties" you can see particulars about the test machine 15:57
jnthn [Coke]: Well, it's been run on amd64 and fails consistently. The only pass I see was from a Win32 machine ;) 15:58
15:58 jevin left, Holy_Cow left
masak "nested structural regular expressions" -- sounds familiar somehow. www-cs-students.stanford.edu/~blynn/nex/ 15:58
moritz produces lots of PASSes on amd64, but rarely remembers to submit
jnthn has access to one and should try it at some point 15:59
It's time to start getting The Beast set up tonight too :)
masak The Baest! 16:00
[Coke] ?
jnthn [Coke]: I've got new hardware :) 16:01
masak [Coke]: jnthn can haz a new bad-ass stationary computer.
16:01 jevin joined
jnthn So I can be, like, productive and stuff. :) 16:01
mls jnthn: I think it's because of that compute_mro -> mro change 16:03
jnthn mls: ah. Hmm.
mls the code was self.compute_mro($_)
is the "self" correct? 16:04
I mean, compute_mro caches the result in self.!mro
tadzik say so all(@bigs) contain 'foo' # fantastic poetry :) 16:05
jnthn mls: er, no
mls: That'll be the issue. 16:06
You may want to try $_.HOW.mro($_)
mls yes.
(But I just copied code. So the old code was wrong, too?)
jnthn No, compute_mro cares more for $obj than self, iirc. 16:07
mls but it stores the result in @!mro, so it uses self
masak moritz: I think you will like spikedmath.com/433.html
jnthn mls: Yeah. I guess we got lucky by then overwriting it again with the correct thing later...
mls: So, a good discovery. 16:08
16:08 simcop2387 left, sayu joined
mls so, how to proceed? delete the $class argument from compute_mro and just use self? 16:08
masak moritz: I figured some of them out before I saw the translation below ;) 16:09
16:10 simcop2387 joined
jnthn mls: In your patch where you've turned a self.compute_mro($_) into a call to mro, do it as $_.HOW.mro($_) 16:10
mls: that should hopefully do it. 16:11
mls yes. Testing... 16:12
16:16 jevin left 16:17 molaf joined
mls Hmm, Null PMC access in get_string() 16:18
daxim perl6: one; loop { two; three; four; five; }
p6eval niecza v9-23-g1a8efca: OUTPUT«===SORRY!===␤␤The 'one' listop may not be called without arguments (please use () or whitespace to clarify) at /tmp/InICJnRYFo line 1:␤------> one⏏; loop { two; three; four; five; }␤␤Undeclared routines:␤ 'five' used at …
..rakudo d97997: OUTPUT«Could not find sub &two␤ in block <anon> at /tmp/LIH_GuxKXT:1␤ in <anon> at /tmp/LIH_GuxKXT:1␤ in <anon> at /tmp/LIH_GuxKXT:1␤␤»
..pugs: OUTPUT«*** No such subroutine: "&two"␤ at /tmp/zO6_zkbpCx line 1, column 13-16␤»
16:18 eternaleye left 16:21 jevin joined
tadzik funny. Why "&two"? 16:21
TimToady interesting that niecza lists the undeclared routines in reverse order; so does std
oh, maybe it's alphabetical 16:22
yeah
16:23 bluescreen10 left, stephang joined
TimToady tadzik: I believe that's a run-time error; rakudo still does not check for undeclared subs at CHECK time like niecza does 16:24
but yes, it could strip the &
tadzik no, still. one() should be tried to be called before loop {}, no? 16:25
TimToady one is a built-in, and may not be called without args, as niecza points out
and nothing is called if an error is compile time, or check time 16:26
if niecza says "SORRY!" nothing will be run (except beginish stuff)
perl6: say "alive"; two 16:27
p6eval niecza v9-23-g1a8efca: OUTPUT«===SORRY!===␤␤Undeclared routine:␤ 'two' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 685 (CORE die @ 2) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1136 (STD P6.comp_unit @ 36) ␤ a…
..pugs: OUTPUT«alive␤*** No such subroutine: "&two"␤ at /tmp/B5q4T3MfXQ line 1, column 14 - line 2, column 1␤»
..rakudo d97997: OUTPUT«alive␤Could not find sub &two␤ in <anon> at /tmp/_E7WjWU4GE:1␤ in <anon> at /tmp/_E7WjWU4GE:1␤␤»
TimToady note niecza does not say "alive"
that is corret
*c
16:28 MayDaniel joined 16:30 eternaleye joined
jnthn TimToady: I was pondering have the optimizer detect those, since it has to go over the various called routines anyway. 16:33
er, callsites, I mean.
I guess we could do the mystery approach STD uses also, though. 16:34
TimToady the most import aspect is to check the visibility of post-declared routines from the lexical scope of the call
pmichaud good morning, #perl6 16:35
TimToady howdy
jnthn Right...by the time we're in the optimizer, we know all the declarations.
hej, pmichaud :)
16:35 mj41 left, bluescreen10 joined
[Coke] ho, pmichaud 16:35
TimToady but you have to know all the original scopes too; merely having a declaration out there is no good if it's not visible from the point of the call 16:36
mjreed so we're generally opposed to things that numify to 0 but boolify to true these days, correct?
[Coke] I certainly am!
TimToady it's considered a design smell
btw, .defined should autothread, but .DEFINITE should not 16:37
jnthn TimToady: So .defined should go in Any, not Mu?
pmichaud jnthn: I think .defined in Junction should autothread.
jnthn Oh, we should implement a .defined in Junction...OK.
pmichaud jnthn: I'd still expect to have a Mu.defined, I think.
jnthn wfm
pmichaud is several days out on recent conversations, so modulo my responses with that. 16:38
jnthn So long as we don't go breaking the "auto-threading of method is just a method dispatcher fallback"
:)
TimToady: (know all the original scopes) yes, we very much still do in the optimizer. It has to know exactly what's going to be called, otherwise it's not going to be able to do inlining.
TimToady or candidates for that matter 16:39
jnthn Right. :)
We have that to hand in nom already. :)
mjreed I was asking because we don't have a trivial built-in way atm to do a yes/no "is X a substring of Y". have to either turn X into a regex and smartmatch or call .index.defined. Making .defined autothread makes the latter more palatable, though.
TimToady having the check in one spot does let us list all of the undefined routines together though
jnthn TimToady: Sure - I'd expect the optimizer to note down all of these sorts of things as it goes and then just give a report of anything it found at the end. 16:40
TimToady wfm
jnthn TimToady: It'll no doubt detect a bunch of impossible type-based stuff too. 16:41
Of course, from the user's point of view, they don't much care which stage whines at them. :)
They just want the info.
mjreed as long as the whining includes information meaningful to me... which low-level whinging often doesn't. 16:42
mls how do I create a copy of an array in nqp?
pmichaud mls: deep copy or shallow?
mls shallow is enough
pmichaud my @new := nqp::clone(@old)
mls Thanks! 16:43
jnthn mjreed: Yes, it'd tell you the name of the undeclared thing, where the call was, etc.
16:45 dukeleto left
mjreed Oh, happy programmer's day, in case nobody mentioned it yet... 16:45
mls jnthn: that was the issue with our last change: all classes shared the same @!mro cache afterwards
jnthn mls: oh! 16:46
mls: Well, ouch :)
mls yes ;)
cotto_work mls: how do you regard the sub profiling branch? Is it as good as it'll ever get or do you have further plans? 16:47
mls It's not in a bad state ;). It just needs some cleanups:
int -> correct type, it shouldn't write to stderr but in a file, and similar small changes 16:48
cotto_work also, tests
mls regarding the cla: our legal expert wasn't here today, will try tomorrow... 16:49
yes, tests would be nice
16:49 alvis left
cotto_work mls: ok. Thanks for looking into that. 16:50
mls jnthn: t/spec/S12-introspection/parents.t works again! and startup is still fast!
jnthn \o/
Well, faster ;)
mls gisting...
I little bit, maybe ;)
jnthn mls: Does the delegation.t one pass also
mls yes 16:51
pmichaud afk, errands
mls jnthn: gist.github.com/1214298 16:54
note the strategic nqp::clone() ;)
spectesting...
jnthn ah, yes :) 16:55
mls (btw, the nqp patch was not affected, as it doesn't cache the mro)
16:55 envi left 16:57 packetknife joined
mls argh, instants-and-durations hangs... 17:01
(stupid 32bit host...)
17:02 alvis joined 17:03 dakkar left 17:06 Chillance joined 17:10 mjreed left, localhost joined 17:11 daniel-s left
supernovus So, my SCGI library now has PSGI support. Now to get it working with apache's mod_scgi again. 17:24
mls jnthn: spectest complete, only the 32bit tests seem to fail
jnthn mls: Great, thanks...will apply later today. 17:25
mls thanks!
afk -> home
[Coke] really hopes versions in perl6 don't suck. 17:26
17:30 packetknife left
flussence [Coke]: you were asking about TAP files the other day, do you still need them? 17:37
17:42 dukeleto joined, stephang left
[Coke] I could use a few, sure. 17:43
juicy ones with failures and skips and things! ;_)
flussence have you got access to feather3?
[Coke] er, ;)
just feather.
17:45 molaf left
flussence hm, I'm not sure how to get them to you... 17:45
moritz: ping?
17:45 benabik left
[Coke] you have access to feather3 but not feather? odd. 17:46
tell you what - just hang on to them, if you can, and I'll have you try to submit them via the webterface when that works.
flussence yeah, I only needed it to poke tryrakudo when it stops working...
oh, most of these are getting uploaded to smolder.parrot.org too, I just happened to save them all locally 17:47
[Coke] I no longer have access there ;)
no big deal. I'm days away from doing something with them anyway, and I can always generate my own if needed.
thanks.
Util rakudo: class Wo { method look ($_:) { say $_.WHAT, ' ', self.WHAT; }; }; my $w = Wo.new; $w.look; 17:48
p6eval rakudo d97997: OUTPUT«Wo() Wo()␤»
Util rakudo: class Wo { method look { say $_.WHAT, ' ', self.WHAT; }; }; my $w = Wo.new; $w.look;
p6eval rakudo d97997: OUTPUT«Any() Wo()␤»
Util Didn't the Specs (at some point in the past) say that methods auto-topicalize `self`, so that you can always say .method from within a method?
Am I just imagining that history?
flussence [Coke]: depends how many you need... I have the whole nom branch here :)
17:49 pernatiy left
masak Util: you are not imagining that history. 17:51
Util: it was changed to the current state in 2005. I have a blog post about why that was a good idea.
flussence rakudo: class Wo { method look ($_:) { say .WHAT; }; }; my $w = Wo.new; $w.look; # curious... 17:52
p6eval rakudo d97997: OUTPUT«Wo()␤»
masak Util: strangelyconsistent.org/blog/the-do...test-sigil
flussence rakudo: class Wo { method look { say .WHAT; }; }; my $w = Wo.new; $w.look;
p6eval rakudo d97997: OUTPUT«Any()␤»
flussence no surprises here, good :) 17:53
Util masak: tyvm 17:54
17:54 wamba joined
[Coke] flussence: ah, nifty. Might be fun to use that as a load to get data over time stuff working. 17:55
17:58 JimmyZ left 18:04 clairvy left 18:05 uasi left 18:10 zby_home joined 18:20 JodaZ joined 18:28 daxim left, mj41 joined
Util `for` is eager in Perl 5, so this is safe: 18:35
my @GLOBAL_ARRAY = 1,2,3; sub_which_mutates_the_global_array($_) for @GLOBAL_ARRAY;
`for` is lazy in Perl 6, so to be safe, I must do this:
my @GLOBAL_ARRAY = 1,2,3; my @temp = @GLOBAL_ARRAY; sub_which_mutates_the_global_array($_) for @temp;
Is there some single-statement syntax to make `for` eager?
jnthn Note that for in sink context is lazy. 18:37
gah
Note that for in sink context is eager.
18:38 pernatiy joined
PerlJam Util: modulo what jnthn said, you can use "eager" to force eagerness. :) 18:40
Util PerlJam++ That is what I was looking for. 18:41
jnthn: Interesting. I cannot think of example syntax, though. 18:42
thanks to you both 18:43
jnthn std: eager for 1..10 { } 18:44
p6eval std bb4f150: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/Tb7wJVdRED line 1:␤------> eager for 1..10 ⏏{ }␤ expecting infix or meta-infix␤Parse failed␤FAILED 00:01 120…
jnthn Thought not...
eager map ... works though.
Util std: .say for eager 1..10 18:45
p6eval std bb4f150: OUTPUT«ok 00:01 120m␤»
18:47 sayu left 18:52 daxim joined
masak Util: all the 'eager' would do in that context would be roll out the range into a list in memory before iterating over it. 18:53
the merits of that are questionable, though irrelevant in the case of 1..10 :P
18:54 wallberg joined 18:55 mjreed joined, frhodes joined 18:56 zby_home left
mjreed how do I do a destructuring bind? Something like my ($head, @rest) := @somearray.. 18:57
jnthn *@rest
mjreed Ah, I was putting the star on the wrong side. Thanks. 18:58
jnthn b: my @a = 1,2,3,5; my ($head, *@rest) := @a; say $head; say @rest
p6eval b 1b7dd1: OUTPUT«1 2 3 5␤␤»
jnthn er, hm
mjreed hm.
indeed.
jnthn rakudo: my @a = 1,2,3,5; my ($head, *@rest) := @a; say $head; say @rest
p6eval rakudo d97997: OUTPUT«===SORRY!===␤Cannot use bind operator with this LHS at line 1, near " @a; say $"␤»
jnthn Yeah, didn't tackle that bit in nom yet. 18:59
mjreed it seems like the syntax for that would be my ($head, @rest) := *@a; 19:02
jnthn no.
The LHS there is parsed as a signature.
There's no prefix:<*>
mjreed Hw do you pass the elements of an array as individual parameters to a sub? 19:03
TimToady prefix:<|> 19:04
Util masak: The 1..10 is bad example; in the backscroll, you will see I was looking for `eager` to do this: 19:06
mjreed so "my ($head, @rest) := |@a;" would seem to be the non-sub binding analogue..
Util my @GLOBAL_ARRAY = 1,2,3; sub_which_mutates_the_global_array($_) for eager @GLOBAL_ARRAY;
masak hey! #phasers!
Util Whoops!
masak Util: as jnthn pointed out *that already does what you want*.
Util: for loops are lazy (because maps are lazy), but sink context eagerifies them. 19:07
Util masak, (and anyone else who is questioning my sanity), @GLOBAL_ARRAY is just a placeholder for *anything* that I am trying to pre-read into memory before the `for` gets it. 19:08
I should have used $fh.lines for an example instead. 19:09
Thanks.
19:09 molaf joined
masak Util: sink context eagerifies for loops. 19:10
masak should perhaps change his way of saying this, since it hasn't worked so far
Util but /me was absent the day they taught sink context :( 19:11
masak Util: the example, even with the modifications you propose, is not lazy. unless it's at the last statement of a routine and some other things hold.
19:11 im2ee joined
im2ee Hello! :) 19:11
masak im2ee: cześć! 19:12
jlaire im2ee: terve
im2ee masak, witaj.
tadzik hej
im2ee jlaire, terve! :)
tadzik, hej! :)
Util masak: So, like in Perl 5, it is always safe to say this? : 19:13
for IO.readdir -> $file { clone_the_file_into_the_same_dir($file) }
masak Util: unless it's the last statement of a routine and some other things hold, yes. 19:14
Util Can you point me to the doc? I am writing example code, and I need to know those "other things" to be sure I don't err; this *is* being done as the last statement of a routine. 19:15
masak rakudo: for 1..10 { .say }
p6eval rakudo d97997: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
masak rakudo: sub foo { for 1..10 { .say } }; my @a := foo(); 19:16
p6eval rakudo d97997: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
masak hm. should that really evaluate?
b: sub foo { for 1..10 { .say } }; my @a := foo();
p6eval b 1b7dd1: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
TimToady afaik rakudo still cheats on sink context 19:17
masak Util: "other things": the call isn't in sink context.
TimToady: oh. that would explain it. 19:18
TimToady niecza: sub foo { for 1..10 { .say } }; my @a := foo();
p6eval niecza v9-23-g1a8efca: OUTPUT«Potential difficulties:␤ @a is declared but not used at /tmp/I17hkYIkBJ line 1:␤------> sub foo { for 1..10 { .say } }; my ⏏@a := foo();␤␤1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤»
Util masak: OK, thanks!
masak rakudo: my @a := gather for 1..10 { .say; take }; say "alive"
p6eval rakudo d97997: OUTPUT«alive␤» 19:19
masak rakudo: my @a := gather for 1..10 { .say; take }; say "alive"; say @a[4]
p6eval rakudo d97997: OUTPUT«alive␤1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤Nil␤»
masak Nil!?
rakudo: my @a := gather for 1..10 { .say; take }; say "alive"; say '@a[4] = ', @a[4]
p6eval rakudo d97997: OUTPUT«alive␤1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤@a[4] = Nil␤»
masak this is wrong, right? 19:20
b: my @a := gather for 1..10 { .say; take }; say "alive"; say '@a[4] = ', @a[4]
p6eval b 1b7dd1: OUTPUT«alive␤1␤2␤3␤4␤5␤@a[4] = ␤»
masak oh!
rakudo: my @a := gather for 1..10 { .say; take $_ }; say "alive"; say '@a[4] = ', @a[4]
p6eval rakudo d97997: OUTPUT«alive␤1␤2␤3␤4␤5␤@a[4] = 5␤»
masak <-- very forgetful
PerlJam rakudo: my @a := gather for 1..10 { .say; .take }; say "alive"; say '@a[4] = ', @a[4] 19:21
p6eval rakudo d97997: OUTPUT«alive␤1␤Method 'take' not found for invocant of class 'Int'␤ in block <anon> at /tmp/cP1aes3Fv8:1␤ in method reify at src/gen/CORE.setting:3649␤ in method reify at src/gen/CORE.setting:3554␤ in method reify at src/gen/CORE.setting:3554␤ in method gimme at src/…
masak TimToady: what do you think of nom's tendency to give Nil back from indexings that didn't find anything. (as opposed to Any, that Beijing gives back)
TimToady rakudo: my @a := do for 1..10 { .say; $_ }; say "alive"; say '@a[4] = ', @a[4] 19:22
p6eval rakudo d97997: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤alive␤@a[4] = 5␤»
TimToady niecza: my @a := do for 1..10 { .say; $_ }; say "alive"; say '@a[4] = ', @a[4]
p6eval niecza v9-23-g1a8efca: OUTPUT«1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤alive␤@a[4] = Any()␤»
TimToady Any seems a bit better to me if there's any chance it'll be used as a placeholder 19:24
(in a list)
masak I agree. 19:26
TimToady well, it depends on what the default value for the array is 19:27
but most arrays default to Any 19:28
19:28 mberends joined
TimToady more precisely, most arrays default to their type object, and untyped arrays have a type object of Any 19:30
perl6: my Int @array; say @array[1].WHAT
p6eval pugs: OUTPUT«Scalar␤»
..rakudo d97997, niecza v9-23-g1a8efca: OUTPUT«Any()␤»
19:30 kshannon left
TimToady I think that should probably be Int 19:31
jnthn TimToady: Agree
TimToady but we want to be able to set a default separate from the type
mathw (wakes up) That seems sensible 19:33
masak math! 19:34
er, mathw! \o/
mathw has been busy with new job 19:35
been learning F# and C#, wishing for Perl 6
masak P#
TimToady smile when you say that
masak P# :D 19:36
19:36 araujo joined
mathw actually been wishing for anything with macros, .NET seems to grow boilerplate randomly :( 19:36
19:36 benabik joined
mathw would like to congratulate the nom developers on its promotion 19:39
19:39 thou left
masak its promotion to... still being nom :P 19:40
mathw well yes 19:41
but it is now the default branch
which is a promotion :)
PerlJam masak: but now it's *the* nom
masak :P
19:42 kshannon joined
mathw We're using Mercurial at $newjob, which is confusing me a bit because it's almost the same as git, but not quite... 19:42
masak mathw: having played around with nom in the past week, I find it less ready to replace b than I thought. but it seems to be closing the gap quickly enough.
mathw masak: at some point the leap had to be taken, it's not like releasing a big new version of a popular app and leaving most of the functionality out. 19:43
Nobody ever does that, after all
masak surely not
pmichaud good afternoon, #perl6 19:44
mathw oh hai pmichaud 19:45
masak pmichaud! \o/
just the Perl 6 List Expert I was looking for :) 19:46
pmichaud: btw, #phasers!
pmichaud yeah, I just got in from @errands 19:47
mathw on stun!
(sorry)
masak peow peow
19:47 frhodes left, kshannon left 19:48 kshannon joined 19:49 MayDaniel left 19:51 frhodes joined, mberends left 19:59 jevin left 20:02 jevin joined 20:03 skangas left 20:20 eiro joined
supernovus hmm, that's odd, if I run a script on the command line, it works fine, if I try to run it as a CGI script, it bails with "Method 'exists' not found for invocant of class 'Hash'". That's new one oO 20:20
eiro hello
20:21 soh_cah_toa joined
jnthn supernovus: Are you running an installed version, ooc? 20:21
supernovus yup 20:22
tadzik hello eiro 20:23
masak salut, eiro.
eiro: bonne chose vous êtes venu. du peuple polonais et finlandais ont été la prise sur le channel. :) 20:25
tadzik es polonais 20:26
20:27 wamba left
arnsholt hands tadzik a t 20:29
I think it's yours =)
tadzik est. Like ist :)
yeah, it's me t
arnsholt Like ist, indeed. They're even etymologically related =D 20:30
masak tadzik: och, nie mówił o *tobie* :)
tadzik :P
arnsholt I think the slavic copula is *Hes as well 20:31
masak unless you're Russian, then it's ''
arnsholt Quite, quite
I need to learn some kind of Slavic language 20:32
Old Church Slavonic might be useful ^_^ 20:33
jnthn It looks pretty if nothing else :)
arnsholt Yup. And considering my Master's work used data from a Latin corpus, I can even make a case that I might even benefit from knowing it 20:34
Historical linguistics for the win =) 20:35
masak Ѧ!
supernovus Okay, so IO::Socket::INET works if I telnet into it, and manually send commands. If I use Apache or lighttpd to connect to it, they don't seem to send the data until the connection times out (or at least that seems to be the symptoms.) Anyone have experience with sockets? 20:36
arnsholt Yeah, I should learn to read those things as well. I already know the Greek alphabet, so it should be mainly a matter of applying myself
.u Ѧ
phenny U+0466 CYRILLIC CAPITAL LETTER LITTLE YUS (Ѧ)
masak supernovus: sounds like something to do with buffering.
arnsholt: latin, greek, and cyrillic. learning them all is like looking at cubistic art :P 20:37
arnsholt Well, I don't know actual Greek (modern or ancient). I just know the letters of the alphabet 20:38
masak aye.
arnsholt I learned half of it from doing maths too, TBH
masak same here.
nodnod
arnsholt But not sure I see how it's like cubism 20:39
20:39 uasi joined
arnsholt You trying to say Classics aren't a useful real-world skill? ;p 20:39
masak I would never claim that. 20:43
I learned Cyrillic before Greek. with each Greek letter I learned, I understood more about the Cyrillic ones. the perspective gained was almost like in cubism, with more observer angles than usually possible ;) 20:45
20:48 kaare_ left
mikemol rosettacode.org/mw/index.php?title=...ldid=91998 20:48
masak rakudo: say '1' + 1 20:51
p6eval rakudo d97997: OUTPUT«2␤» 20:52
masak this has little or nothing to do with "strong typing", at least in Perl 6.
...where the operators themselves more often than not contain implicit type conversions.
arnsholt masak: Ah yes, I see what you mean now
Sanskrit is mighty useful in that context as well. Lots of fun stuff going on with the verbs 20:53
masak d'oh, gotta learn Sanskrit.
arnsholt It's pretty cool. Thankfully I can just ask the missus in most cases if I need some kind of Sanskrit trivia answered =) 20:54
masak a Sanskrit missus? oh mama. 20:57
arnsholt Oh yeah =D 20:59
Incidentally, it turns out Sanskrit classes are an excellent way to meet people interested in Sanskrit =D 21:00
supernovus masak: Hmm, any idea how to work around buffering issues? I never came across these problems in ng or alpha. Always new and exciting things to find when new versions come along :-)
im2ee Uhh, it's time for me. Good night! :)
21:01 im2ee left 21:02 MayDaniel joined
masak im2ee: dobranoc! 21:02
supernovus: I'm sorry, I have absolutely no idea. maybe mberends will know, he's done a bit in that area. 21:03
supernovus masak: okay, I'll ask mberends when he's around. 21:04
21:05 mberends joined
masak :) 21:05
he's around. 21:06
supernovus I'm cleaning up my old projects, I'm going to pull half of them from the active list, as I have no intentions of resurrecting them. Exemel, SCGI, WebRequest (soon to be renamed WWW::App) Temporal::Utils and Flower will likely be the only survivors of my project purge.
mberends depubbed (pity, $work tomorrow)
jnthn mberends: aww. 21:11
mberends supernovus: IO::Socket::INET initially had similar lack of buffer flushing in alpha and beijing, and each was patched after the initial implementation. Not sure, but perhaps the fixes are not portable enough between Rakudo branches. 21:14
21:16 MayDaniel left
jnthn I suspect the gist of them will be though. 21:16
mberends yep, nom passes spectests, although networking is very hard to test thoroughly 21:17
supernovus Ah, well, that likely explains the issue. I submitted a pull request with some changes that make the accept() method work properly in nom (still no buffer flushing though, I'm not sure where to find that.) 21:18
jnthn Did anyone apply supernovus++ pull request?
jnthn is just in the middle of setting up his new dev machine at the moment... :)
mberends jnthn: with Linux this time? ;) 21:20
jnthn :P
mberends :Positive \o/
jnthn Good luck with that :)
I do have 16 GB of RAM to do something with though :)
So I can certainly do a Linux VM :)
Pondered, but didn't feel like the shock therapy of just installing that. ;) 21:21
mberends :)
cotto_work jnthn: is your boolification work in nqp any more complicated that it sees to be or is it just a way to specify how a 6model thingy gets turned into 1/0?
supernovus dual booting is always fun... games are on windows, work is on linux, keeps me undistracted during the day ;-) 21:22
cotto_work s/sees/seems/
jnthn cotto_work: It's basically that.
cotto_work: It's a circularity saw in a sense.
cotto_work: I wanted something that would (a) let me handle a bunch of common cases efficiently and (2) be easy to port. 21:23
er, 1 and 2 :)
cotto_work: Turns out it gave us a ~15% performance win :) 21:25
cotto_work nice find
Is that from avoiding nested runloops?
21:26 bbkr1 joined
jnthn cotto_work: Yes, but even if we didn't have to enter a nested runloop it'd still be faster for the common cases too. 21:26
21:27 dukeleto left, wamba joined 21:29 frhodes left, frhodes joined 21:30 frhodes left, frhodes joined 21:31 frhodes left, frhodes joined, skangas joined 21:37 wallberg left
jnthn Ugh. 21:39
.\miniparrot.exe -Iruntime/parrot/include config_lib.pir > runtime/parrot/include/config.fpmc
NMAKE : fatal error U1077: '.\miniparrot.exe' : return code '0xc0000005'
...that's a segv. :(
Hmm, I see your chosen INTVAL isn't the same size as your pointers. Parrot 21:40
should still compile and run, but you may see a ton of warnings.
...that's a giveaway... 21:41
diakopter new windows trouble?
jnthn diakopter: 64-bit windows trouble.
cotto_work Parrot doesn't do well in that situation
jnthn cotto_work: Yeah...wonder why it ends up configuring that way.
21:44 benabik left, mj41 left
diakopter jnthn: yeah, last I tried parrot/rakudo on 64-bit windows, I guess I downloaded the 32-bit parrot installer. 21:45
21:45 packetknife joined
jnthn cotto_work: Turns out configuring with --intval="long long" helps 21:47
21:48 benabik joined, frhodes left, frhodes joined
bbkr1 nom: my $::x = 3 # segfaults for me 21:48
p6eval nom d97997: ( no output )
bbkr1 nom: my $::x = 3; say "alive";
p6eval nom d97997: ( no output )
cotto_work jnthn: yeah. That's the workaround. 21:49
jnthn cotto_work: Not quite sure how to update the configure probe.
cotto_work jnthn: what should it do? always use intval=long long on windows x64? 21:50
21:50 jevin left
jnthn cotto_work: "if it's Windows x64 on msvc, use long long" sounds like a sane starting point. 21:50
21:51 jevin joined
cotto_work jnthn: ok 21:51
21:53 thou joined 21:55 clairvy joined
masak 'night, #perl6 21:56
21:56 clairvy left
masak keep it real :) 21:56
21:56 clairvy joined, masak left, bluescreen10 left
dalek osystem: 0df35d1 | (Timothy Totten)++ | META.list:
Removed my old dead projects, and added WWW::App.
 21:57
21:57 kfo_ joined
flussence
.oO( keep it rat? )
 21:57
21:59 Limbic_Region joined
cotto_work jnthn: how does this work: nopaste.snit.ch/81313 22:00
bbkr1 what happened to ClassHOW type? are tickets related to it, such as rt.perl.org/rt3/Ticket/Display.html?id=73142 , still valid?
cotto_work jnthn: actually, I don' think that's sufficient 22:01
jnthn bbkr1: It's all under Metamodel:: these days 22:03
22:03 supernovus left
jnthn cotto_work: Can try in a moment 22:03
cotto_work: Just getting a Rakudo build at all right now...and setting everything else up on here.
(new machine)
bbkr1 jnthn: thanks, 22:05
22:06 donri left
cotto_work jnthn: try this one instead: nopaste.snit.ch/81314 22:06
it's possible it'll break something 22:07
22:08 benabik left
jnthn Will try when this spectest run is done :) 22:12
240 wallclock secs. Not bad. :) 22:15
pmichaud: ^^ Who wins? ;)
(spectest time) 22:16
22:16 frhodes left 22:17 frhodes joined
jnthn cotto_work: For some reason the patch doesn't seem to apply. :S 22:22
22:23 Limbic_Region left
cotto_work jnthn: to parrot master? 22:24
jnthn yeah
not sure what's up with that, it looks like it should.
cotto_work ditto 22:25
dalek kudo/nom: cca1166 | jnthn++ | t/spectest.data:
Label a few more tests as needing ICU.
 22:27
jnthn Commit from $new-machine \o/ 22:28
tadzik nice nice
jnthn: you run spectest in 240 secs?
cotto_work jnthn: win64-workaround branch 22:29
jnthn tadzik: yes 22:31
tadzik jnthn: that's... quite unusual :)
what kind of cpu is that beast?
jnthn Some quad core i7 :)
I ran it with --jobs=4
diakopter jnthn: try --jobs=8 22:32
jnthn cotto_work: Just tried the branch. Doesn't help :(
cotto_work: Gives:
init::hints - Load platform and local hints files...Use of uninitialized value in string eq at config/init/hints/mswin32.pm line 68.
.............done.
cotto_work: And then still has the INTVAL/pointer different sizes error 22:33
diakopter: trying :)
let's see if this slows things down
:)
no, machine still pretty responsive :) 22:34
diakopter heh
jnthn I wonder if it's doing HT...
diakopter run all your dev area on a ramdisk :) 22:35
jnthn Already doing it on an SSD... :)
22:35 packetkn_ joined
jnthn Got OS, compiler, and build all on the SSD 22:35
This seems to have the added bonus that it's *very* quiet in here now :)
diakopter: Gets it down to 220. 22:37
diakopter hrm
22:38 packetknife left
sjohnson perl consistently blows my mind with awesomeness every day 22:48
22:48 aloha joined
cotto_work jnthn: I'm pretty sure I'm not fixing this in the right place. I'll need to revisit this when I'm fresher. 22:49
22:49 packetkn_ left
jnthn cotto_work: OK, np. I have a workaround so I won't be blocked. 22:53
diakopter 32-bit parrot installer? 22:54
22:57 aloha left
jnthn diakopter: No, one can just pass --intval="long long" to the Configure script and the build works out. 22:58
23:08 f00li5h joined 23:10 clairvy left 23:15 mattp_ left, mattp_ joined 23:18 wamba left 23:23 tokuhirom joined, bbkr1 left 23:24 bbkr1 joined 23:25 aloha joined 23:29 supernovus joined
supernovus mberends: I'm guessing HTTP::Server::Simple hasn't been ported to nom yet? 23:30
mberends supernovus: correct, but only because of tuit shortage 23:31
supernovus Ah, okay, I just updated my new WWW::App library so it can use either SCGI or HTTP::Server::Simple::PSGI as backends, but my attempt to test with the latter led me to the conclusion that it wasn't working in nom :-) 23:32
tadzik mberends: may be Buf shortage as well :)
mberends tadzik: ouch! 23:34
tadzik yeah, sometimes we have no bufs :) 23:35
jnthn OK, seems I've got my dev environment on my shiny new hardware ready to go. :) 23:36
mberends there is a round buf somewhere in fifo implementations :)
jnthn Too bad it's time to sleep now rather than hack...
jnthn rests...on-site $dayjob during the day, but should get some Rakudo time in the evening, and I've got all day Thursday for it too :) 23:38
night all o/
tadzik 'night! 23:39
I should 'night too
TimToady mikemol: response to paddy at rosettacode.org/mw/index.php?title=...did=120438 23:40
supernovus I'm going to call it a night as well. WWW::App using SCGI is ready for action when IO::Socket::INET flushes its buffers, and will work with HTTP::Server::Simple::PSGI once that's running under nom. It should (not tested) work fine with both under the "ng" branch right now, in case anyone is still running that branch (and wants to write web apps for it?) 23:44
23:46 molaf_ joined 23:49 molaf left 23:52 whiteknight joined 23:55 packetknife joined 23:56 abercrombie joined 23:58 Chillance 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!