»ö« | perl6.org/ | nopaste: paste.lisp.org/new/perl6 | evalbot usage: 'perl6: say 3;' or rakudo:, alpha:, pugs:, std:, or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by lichtkind on 5 March 2010.
00:02 Khisanth left
colomon rakudo: say NaN == NaN 00:03
p6eval rakudo ab2322: OUTPUT«0␤»
00:06 Khisanth joined
colomon rakudo: say NaN cmp NaN 00:09
lue o hai again 00:10
p6eval rakudo ab2322: OUTPUT«1␤»
lue not a number apparently is not not a number :) [according to NaN == NaN]
rakudo: say NaN ~~ NaN # who knows? 00:11
p6eval rakudo ab2322: OUTPUT«1␤»
diakopter rakudo: say NaN - NaN
p6eval rakudo ab2322: OUTPUT«NaN␤»
diakopter rakudo: say NaN ** NaN
p6eval rakudo ab2322: OUTPUT«NaN␤»
diakopter rakudo: say NaN / 0
p6eval rakudo ab2322: OUTPUT«Divide by zero␤current instr.: 'infix:</>' pc 305405 (src/gen/core.pir:3048)␤»
00:11 am0c left
diakopter rakudo: say 0 / NaN 00:12
p6eval rakudo ab2322: OUTPUT«NaN␤»
ash___ rakudo: say NaN.abs
p6eval rakudo ab2322: OUTPUT«NaN␤»
ash___ rakudo: say (-Inf).abs
p6eval rakudo ab2322: OUTPUT«Inf␤»
lue colomon!: en.wikipedia.org/wiki/James_Anderso...rithmetics 00:14
explains why NaN == NaN fails
rakudo: multi sub infix:</>($a, $b){if $b == 0 {say "Φ";}; else { say $a div $b;}; }; 3/7; 9/0; # transreal numbers! 00:19
p6eval rakudo ab2322: ( no output )
lue ō.o
arnsholt_ rakudo: our multi sub infix:</>($a, $b){if $b == 0 {say "Φ";}; else { say $a div $b;}; }; 3/7; 9/0; # I think...
p6eval rakudo ab2322: ( no output ) 00:20
arnsholt_ Or not. Oh well
00:20 arnsholt_ is now known as arnsholt
lue maybe overload is broken/NYI ? 00:21
rakudo: say 1/0; say "I'm okay!" 00:24
p6eval rakudo ab2322: OUTPUT«Inf␤I'm okay!␤»
colomon rakudo: multi sub infix:</>($a, $b){if $b == 0 {say "Φ";} else { say $a div $b;}; }; 3/7; 9/0; # transreal numbers! 00:28
p6eval rakudo ab2322: ( no output )
colomon :\
rakudo: multi sub infix:<//////>($a, $b){if $b == 0 {say "Φ";} else { say $a div $b;}; }; 3//////7; 9//////0; # transreal numbers! 00:29
p6eval rakudo ab2322: OUTPUT«0␤Φ␤»
colomon lue: my problem at the moment is that somehow Test.pm's is was getting NaN to equal NaN, but I broke it in my local copy, making zillions of tests fail. 00:30
lue why would want it to equal itself? It's an error.
colomon yes, but you need to be able to check and see if that error happened. That's the point of the tests in question. 00:31
ash___ rakudo: say 1/0 ;
p6eval rakudo ab2322: OUTPUT«Inf␤»
ash___ o.0
colomon that's a Rat, btw.
ash___ I thought that was an error and NaN?
colomon nope. 00:32
lue IEEE floating point arithmetic.
rakudo: say 1/-0
p6eval rakudo ab2322: OUTPUT«Inf␤»
lue rakudo: say -1/0
p6eval rakudo ab2322: OUTPUT«-Inf␤»
lue -0 should have produced -Inf, IIRC 00:33
colomon I don't think so, because -0 is an Int, and for Ints, -0 == 0
there's only one zero 00:34
rakudo: say 1 / -(0.Num)
p6eval rakudo ab2322: OUTPUT«Divide by zero␤current instr.: 'infix:</>' pc 305405 (src/gen/core.pir:3048)␤»
colomon rakudo: say 1 / (0.Num)
p6eval rakudo ab2322: OUTPUT«Divide by zero␤current instr.: 'infix:</>' pc 305405 (src/gen/core.pir:3048)␤»
colomon rakudo: say 1.Num / (0.Num)
p6eval rakudo ab2322: OUTPUT«Divide by zero␤current instr.: 'infix:</>' pc 305405 (src/gen/core.pir:3048)␤»
lue rakudo: say (3.5).WHAT
p6eval rakudo ab2322: OUTPUT«Rat()␤» 00:35
lue Is Rat == Floating Point in perl6?
00:36 hercynium joined 00:37 rgrau_ joined
colomon no, Rat is a rational number, int over int. 00:40
decimal numbers are now rational, not floating point.
00:42 supernovus joined
colomon I see. 00:43
00:44 jql joined
colomon Problem is that Rakudo is broken.... or maybe I don't understand the comparison operators. 00:44
Ah, I don't understand the comparison operators.
okay, this is an easy fix.
ash___ ? 00:45
colomon I was thinking that infix:<eq> was equivalent to infix:<cmp> == 0
but it's not, it's infix:<leg> == 0
ash___ eqv or eq? Isn't eq for strings? 00:46
colomon ie infix:<eq> is string equality, not DWIM equality
supernovus I read in the backlogs that := is not going to be implemented any time soon. I would then recommend that the IO::Socket::INET module be rewritten to not use it. Many libraries such as HTTP::Daemon and SCGI depend on the IO::Socket::INET library which is included in the Core, but currently does not work due to the := not being implemented.
lue (with so many comparison operators, no other language compares)
supernovus: who said that?
colomon I overloaded infix:<eq> to be use infix:<cmp> for numbers. that's what broke all the NaN tests. 00:47
because "NaN" does equal "NaN"
even if NaN != NaN
ash___ ah
NaN fails all numeric comparison tests
supernovus lue: Well, I haven't completely caught up yet, so I may be speaking out my arse, but you had asked about it and diakopter said it would take hundreds->thousands of hours to implement... 00:48
lue I don't know much about the guts, but that seemed like a hyperbole to me. 00:49
For $a := $b, you just need to assign $a to $b's location. (masak pointed me to a blog post earlier today, it's in the backlogs and explains better than me) 00:50
supernovus The := thing is one of the few reasons left to stick with the 'alpha' branch. I will continue scouring the backlogs. :-) 00:51
ash___ rakudo: my $a = 1; my $b = \$a; say $b # is that supposed to say capture?
p6eval rakudo ab2322: OUTPUT«Capture()<0x14d22b0>␤»
ash___ alpha: my $a = 1; my $b = \$a; say $b ;
p6eval alpha 30e0ed: OUTPUT«1␤»
lue rakudo: my $a = 1; my $b = \$a; $b = 2; say $a 00:52
p6eval rakudo ab2322: OUTPUT«1␤»
lue rakudo: my $a = 1; my $b = \$a; $b = 2; say $a; say $b;
p6eval rakudo ab2322: OUTPUT«1␤2␤»
ash___ rakudo: my $a = 1; my $b = \$a; $a = 2; say $a; say $b;
p6eval rakudo ab2322: OUTPUT«2␤Capture()<0x3f4e440>␤»
colomon I think there was definitely a misunderstanding on the complexity of implementing binding. 00:53
(in that discussion yesterday)
00:53 [particle] left
lue AFAIK, it should be _that_ hard. 00:54
s/should/shouldn't/
supernovus Well, I wish I knew more about the guts. I'm going to try to get some of my libraries like SCGI working on the new master, then I'll have to find out about adding them to proto. Oh well, I'm off to watch tonight's Doctor Who. Later!
lue me too. 00:55
supernovus: me too! afk
(first me too about the guts, second about DW)
00:55 supernovus left
ash___ this is kinda like := but not exactly: 00:57
rakudo: my $a = 1; my $b = \$a; $a = 2; say $a; say $b[0]; $a = 123 ~ "hello"; say $b[0];
p6eval rakudo ab2322: OUTPUT«2␤2␤123hello␤»
ash___ $b is a capture of $a, so updating $a updates $b 00:58
sorear lue: If you think Rakudo is ever going to be faster than hand-optimized x86 code on qemu, you're crazy
also I don't understand what your deal is with mactel
00:58 [particle] joined
sorear incidentally, Rakudo already implements binding. 01:01
Just not the := syntax.
ash___ sorear: how?
I know you can use a capture to get := like behavoiur
sorear rakudo: sub foo($x is rw, $y is rw) { say $x; $y = 2; say $x; }; my $z = 1; foo($z, $z); 01:02
p6eval rakudo ab2322: OUTPUT«1␤2␤»
sorear within the invocation, $x and $y are aliases
ash___ rakudo: my $x = 1; my $y = \$x; $y[0] = 123; say $x; 01:03
p6eval rakudo ab2322: OUTPUT«123␤»
ash___ thats why that worked, the parameters are captures
although, in alpha you didn't have to do [0] on the capture 01:04
alpha: my $x = 0; my $y = \$x; $y = 123; say $x;
p6eval alpha 30e0ed: OUTPUT«0␤»
ash___ hmm
alpha: my $x = 0; my $y = \$x; $x = 123; say $y; 01:05
p6eval alpha 30e0ed: OUTPUT«123␤»
ash___ maybe in alpha captures are only one directional
sorear What are you talking about? 01:09
ash___ using the $y from "my $x = 0; my $y = \$x;" to update $x 01:10
sorear This has nothing to do with binding.
ash___ I was just pointing out that captures function similar to bindings 01:11
sorear A capture is just an array + hash
You've discovered it's possible to update elements in arrays. 01:12
ash___ rakudo: my $x = 1; my $y = \$x; $y[0] = 123; say $x; # is almost the same functionality as $y := $x; $y = 123;
p6eval rakudo ab2322: OUTPUT«123␤»
ash___ with normal arrays and hashes you can't update the variable that assigned you the value, my $x = 0; @a[0] = $x; @a[0] = 123; doesn't make $x = 123; but you can do that with a capture 01:14
sorear that's because your desugaring is wrong 01:18
my $x = 0; @a[0] := $x; $a[0] = 123 01:19
captures aren't like binding or a substitute for binding
they are implemented using binding
01:49 tylerni7 joined, tylerni7 left, tylerni7 joined 01:54 tylerni7 left, tylerni7 joined, tylerni7 left, tylerni7 joined
dalek kudo: c52c61a | (Solomon Foster)++ | src/core/ (4 files):
Add Numeric versions of the infix:<cmp> family of operators. Move the old infix:<cmp>(Num, Num) operator code to infix:􏿽xAB<=>􏿽xBB(Num, Num) and delete infix:<cmp>(Num, Num). Remove the Rat versions of infix:<cmp>.
 01:59
kudo: 9427875 | (Solomon Foster)++ | tools/test_summary.pl:
Change // to || to keep Perl 5.8 happy.
kudo: 85a7aa4 | (Solomon Foster)++ | src/core/Numeric.pm:
Eliminate Numeric versions of eq, ne, lt, gt, le, and ge -- these should always be string comparisons.
02:00 Psyche^ joined
pugssvn r30593 | colomon++ | [t/spec] Fudge Complex Rcmp tests. Add Real tests for cmp, before, and after. Fix a couple of Complex-valued tests to use is_approx rather than is. 02:02
02:04 Patterner left, Psyche^ is now known as Patterner
lue hello! 02:04
sorear: I don't appreciate that. Who says rakudo can't be faster? I will say again: emulation of a chip (esp. Intel emu. on PPC) is slow for me. 02:07
My deal with mactel is that it's new, and uses Intel.
02:07 chitragupt_1 joined, gurjeet joined 02:10 chitragupt_1 left
lue (When the streets of #perl6 are busy, I ought to ask for details on := and why no-one's implemented yet.) 02:10
02:10 chitragupt_1 joined
colomon So, Dr. Who any good tonight? 02:10
lue yes! The Weeping Angels. 02:13
(part 1 of 2)
02:14 chitragupt_1 left
lue And according to the trailer for Part 2, time is going to disintegrate or something. 02:15
again.
02:15 chitragupt_1 joined
colomon :) 02:15
02:16 chitragupt_1 left
lue seriously, it happened about 5 months ago (our time) with the last doctor! 02:16
02:17 chitragupt_1 joined
lue and it's not even the season finale, so what the heck is going to happen _*then*_ ? :) 02:17
colomon surprised early season finale!! 02:18
lue no, the episode with stonehe— *cough* a famous monument — hasn't happened yet.
colomon that's what they *want* you to think. 02:21
lue are they saying a trailer for the stonehenge epsiode is an experiment by the BBC to extend DW's "flux time" beyond the series !? 02:22
s/are they/are you/
O.o run. 02:23
colomon rakudo: say sqrt(:x(10)) 02:24
p6eval rakudo ab2322: OUTPUT«No applicable candidates found to dispatch to for 'sqrt'. Available candidates are:␤:(Complex $x)␤:(Any $x)␤␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
lue is getting ready to dabble in 日本語 02:25
spinclad diakopter: pong # finally unbacklogged 02:27
diakopter: i see the wonderful shiny closureness now in perlesque! \o/ diakopter++ perlesque++ 02:28
lue buubot: karma perlesque 02:30
buubot lue: perlesque has karma of 1
02:31 circuitbreaker joined
diakopter spinclad: yeah :) 02:34
02:37 chitragupt_1 left
sorear phenny: tell moritz_ that I've decided I don't like the plan I had earlier for splitting the setting, so I have nothing more for topic/lexical-persistence 02:40
phenny sorear: I'll pass that on when moritz_ is around.
lue does rakudo have an API or equivalent, for people wishing to muck about in its internals? 02:42
colomon not as such. 02:43
sorear it doesn't have a defined API but that shouldn't stop you
colomon I mean, generally you just muck about.
sorear with early-development software like rakudo, the API forms as we standardize mucking 02:44
the mucking comes first
diakopter dukeleto: I'll try to make the May 25 thing
sorear rakudo: say Perl6::Compiler.compile("2+2", target=>"pir");
p6eval rakudo ab2322: OUTPUT«␤.HLL "perl6"␤␤.namespace []␤.sub "_block59" :anon :subid("18_1273372087.95608")␤ .param pmc param_72 :slurpy␤.annotate "line", 0␤ .const 'Sub' $P63 = "19_1273372087.95608" ␤ capture_lex $P63␤.annotate "line", 1␤ get_hll_global $P61, "!UNIT_START"␤ .const 'Sub'
..$P63…
lue because I had a difficult time finding variable creation/initialization
sorear rakudo: my $a; my $b; pir::store_lex__vSP('$b', $a); $b = 1; say $a; 02:46
p6eval rakudo ab2322: OUTPUT«1␤»
sorear rakudo: my $a; my $b; pir::store_lex__vSP('$b', $a); $a = 1; say $b;
p6eval rakudo ab2322: OUTPUT«1␤»
diakopter sorear: should I implement classes next? it's a very straightforward thing, much more so than getting closures right... since runsharp already does most of it for me. 02:48
sorear diakopter: What would I know about priorities for an implementation of Perl 6 I haven't contributed to and don't use in an environment I don't run? 02:51
er
implementation of NQP
02:51 nihiliad joined
sorear oops. 02:51
02:52 JimmyZ joined 02:53 meppl joined
diakopter it's not really meant to implement NQP, though someone could use it for that 02:53
I intend it to be a target language for STD and viv
so STD can translate itself to perlesque
sorear (and then we can rewrite viv in perl6?) 02:54
hmm. interesting concept
NQP is currently an implementation-defined language; I wonder if we should work to draw NQP, Perlesque, NQ-NQP together 02:55
diakopter my other option for "what to do next" was "string eval"
02:55 alester joined
sorear diakopter: How are you going to do that without STD? 02:56
diakopter b/c I consider string-eval highly necessary for getting a Perl 6 compiler/JIT/runtime going in the perlesque edition of STD
sorry, string eval of perlesque code
(not full-blown Perl 6)
I'm not using STD right now to parse perlesque code 02:57
in case that wasn't clear
who is working on NQ-NQP 02:58
oh.
03:05 nihiliad left 03:10 fridim joined 03:12 molaf joined
diakopter thoughts on implementing string eval: at each string-eval callsite in the user code, the emitted IL needs to contain a hardcoded string key that can be used to access a global dictionary/registry of compilation-context restorer callbacks. 03:19
hrm. that won't work for programs emitted to disk. instead of in-memory callbacks then, the eval function itself needs to be emitted to the IL, and the grammar needs to be refactored a bit to allow eval-mode parsing, such that ... 03:23
the types stack and other various global parse state-keepers are programmatically reconstructable 03:24
hrm.
03:25 rv2733 joined
diakopter I'd better work on classes first and let that simmer (fester?) a while. 03:25
03:25 meppel joined
TimToady (it's essentially the same thing as YOU_ARE_HERE in the setting) 03:25
diakopter TimToady alive 03:26
TimToady hides behind a cricket
and continues backlogging
diakopter Times Square is so empty, even the cricket stands out
actually though 03:27
I was thinking of Perl 5-style string eval
JavaScript-style string-eval would be much easier
pugssvn r30594 | lwall++ | [S02] dig out T Rex fossil found by sorear++ 03:28
03:28 meppl left
diakopter TimToady: perlesque passed man_or_boy 03:29
03:29 meppel is now known as meppl
diakopter man_vs_boyt? 03:29
er
man_vs_bot?
ER
man_vs_boy?
sorear: I'm curious how long it takes rakudo to run the man_or_boy.t at the 10 result level... would you mind running it? 03:30
sorear diakopter: rakudo doesn't have closures
diakopter but..
I thought it passed man_or_boy.t 03:31
I'll look again
it doesn't seem fudged
oh, it's not pulled down to spectests 03:32
sry
03:33 BrowserUk joined 03:35 meppl left
TimToady rakudo: sub counter { my $x = 0; return { say $x++ } }; my $c1 = counter; my $c2 = counter; $c1(); $c1(); $c2(); $c2(); 03:35
p6eval rakudo ab2322: OUTPUT«0␤1␤2␤3␤»
TimToady nope
I find it rather amazing that we pass as many tests as we do without that... 03:36
and given that STD uses bazillions of closures...
well, the p5 translation of it does 03:37
TimToady goes back to his cricket
03:37 circuitbreaker left
TimToady alpha: sub counter { my $x = 0; return { say $x++ } }; my $c1 = counter; my $c2 = counter; $c1(); $c1(); $c2(); $c2(); 03:38
p6eval alpha 30e0ed: OUTPUT«0␤1␤0␤1␤»
TimToady okay, just a regression
diakopter perlesque: sub A(int $k,Callable[:(-->int)] $x1,Callable[:(-->int)] $x2,Callable[:(-->int)] $x3,Callable[:(-->int)] $x4,Callable[:(-->int)] $x5-->int) {my Callable[:(-->int)] $B;$B=sub (-->int) {$k-=1;return A($k,$B,$x1,$x2,$x3,$x4)};if $k<=0 {return ($x4()+$x5())};return $B()};sub K(int $n-->Callable[:(-->int)]) {return sub (-->int) {return $n}};say(A(10,K(1),K(-1),K(-1),K(1),K(0)))
p6eval perlesque: OUTPUT«-67␤»
diakopter check it out :D /me boasts
TimToady where did you get that ugly Callable syntax :P 03:39
diakopter I don't know. I almost scrapped the Callable[] and just went with :()
TimToady &:() wouldn't be too bad 03:40
&[] is taken though
diakopter &:() would certainly be better
almost as clean as C#'s Func<TArg0, TArg1, ... , TReturn> 03:41
sorear How do you plan to compile Perl6 code which uses unsigged function arguments? 03:42
diakopter the Perl 6 runtime will be verry different from the perlesque runtime 03:43
the Perl 6 code won't JIT nearly as directly to CIL
it'll JIT to runtime calls
even in perlesque, those subs aren't translated directly to CIL subs 03:44
each is split into a Bind and Exec methods
so the trampoline actually bounces 03:45
CLR Perl 6 may well be just as function-call heavy as Parrot Perl 6 03:46
but it's the function calls and existence of native types in those function calls that'll be the huge efficiency difference, imhwt (in my humble wishful thinking) 03:47
ash___ so far, nq-nqp is very function-call oriented, but the llvm inlines a lot of them
so when you turn on optimizations, + turns into an assembly add in most cases, which is nice 03:48
diakopter ash___: you're the eternity blogger?
sorear where's nqnqp?
ash___ huh?
github
03:48 chitragupt_1 joined
diakopter greaterthaninfinity.com/ 03:49
ash___ yea
thats mine
diakopter oh, infinity, not eternity ;)
sorear ash___: Why aren't you on planet6?
diakopter typo in the README: Not Quiet NQP-rx -> Not Quite NQP-rx
03:50 chitragupt_1 left
diakopter configuratoins 03:50
ash___ ssshh don't say it so loud
diakopter heh
ash___ sorear: i duno?
plus, this was more a thought experiment than anything else
03:50 envi^home joined
sorear diakopter: For fun, I ran man-or-boy.t on rakudo. 03:51
ash___ i don't plan on stopping it, and now i get to work on parrot (as a GSoC) so maybe I can bring some of the llvm's optimizations to parrot to get rid of some method calls
sorear How long did it take? 0.3 s then "Maximum recursion depth exceeded"
diakopter oh; jnthn told me that can be disabled in parrot
at the nqp level maybe 03:52
after all, parrot uses heap frames like smalltalk & perl (& perlesque) 03:53
03:53 meppl joined
diakopter sorear: the runsharp compiler is now fully exposed/available to/from perlesque code. observe: 03:58
perlesque: my AssemblyGen $ag = AssemblyGen.new('foo.exe'); my TypeGen $tg1 = ($ag.Public).Class("Class0"); 04:00
p6eval perlesque: ( no output )
diakopter no output is good, there.
fooey. I need a way to represent type literals... hack hack hack 04:03
04:05 Targhan joined 04:14 BrowserUk left 04:15 JimmyZ left
snarkyboojum alpha runs man-or-boy.t for 5 seconds before 'maximum recursion depth exceeded' :) 04:18
diakopter perlesque: my $ag = AssemblyGen.new('foo.exe'); my $tg1 = ($ag.Public).Class('Class0'); my $m1 = ($tg1.Public).Method(int, 'blah'); my $cg = $m1.AsCodeGen(); $cg.Return(4); $ag.Complete();
p6eval perlesque: ( no output ) 04:19
dalek meta: r255 | diakopter++ | trunk/Sprixel/ (4 files):
[perlesque] enable type literals. I'll need casting/coercion to enable full
04:27 molaf left 04:34 JimmyZ joined, JimmyZ left 04:42 JimmyZ joined
diakopter perlesque: my $ag = AssemblyGen.new('foo.exe'); my $tg1 = ($ag.Public).Class('Foo'); my $m1 = (($tg1.Public).Override).Method(string, 'ToString'); my $cg = $m1.AsCodeGen(); my $o = List[Operand].new(); $o.Add(Exp.new(Operands::StringLiteral, 'blah blah')); $cg.Invoke(Console, 'WriteLine', $o.ToArray()); $cg.Return(Operands::StringLiteral.new('BLAH')); $ag.Complete(); my $foo = Activator.CreateInstance($tg1); say($foo) 04:44
p6eval perlesque: OUTPUT«blah blah␤BLAH␤»
diakopter it 1. creates a new assembly named foo.exe 2. declares a class Foo a part of that assembly 3. overrides System::Object's ToString() method such that calling ToString() on a Foo object prints 'blah blah' and then returns 'BLAH' 04:46
4. Creates a Foo, and calls say() on it, which implicitly coerces it to string (calling the virtual override ToString())
tada.
and I had to add only 2 magical special cases to runsharp :P 04:47
snarkyboojum I manage to get a segmentation fault if I bump up the recursion limit on parrot :) 04:48
but it runs man-or-boy.t further :P
dalek meta: r256 | diakopter++ | trunk/Sprixel/ (2 files):
[perlesque] got class declaration working programmatically in user code, without

  ($ag.Public).Class('Foo'); my $m1 = (($tg1.Public).Override).Method(string,
  'ToString'); my $cg = $m1.AsCodeGen(); my $o = List[Operand].new();
  $o.Add(Exp.new(Operands::StringLiteral, 'blah blah')); $cg.Invoke(Console,
  'WriteLine', $o.ToArray()); $cg.Return(Operands::StringLiteral.new('BLAH'));
  $ag.Complete(); my $foo = Activator.CreateInstance($tg1); say($foo)
snarkyboojum guess we're still a boy :) 04:50
diakopter how long does it take to segv 04:51
snarkyboojum it takes 13 seconds on my laptop 04:52
sorear snarkyboojum: before or after 46422? 04:53
diakopter perlesquel: .
p6eval perlesquel: OUTPUT«blah blah␤BLAH␤real 0.18␤user 0.14␤sys 0.01␤»
diakopter perlesquel: .
p6eval perlesquel: OUTPUT«blah blah␤BLAH␤real 0.17␤user 0.16␤sys 0.00␤»
diakopter ooo. I just realized that adds a 4th stage to the compilation chain
1: Sprixel.exe 2: perlesque.exe (which it builds upon *every* invocation, including from p6eval), 3. the assembly created from p6eval input (emitted every time of course), and now 4. stage 3 can also emit assemblies. 04:55
sidenote: the perlesquel p6eval target just invokes the last stage 3 emitted
snarkyboojum sorear: dies after "ok 4 - man-or-boy test for start value 3" - so something else might well be wrong 04:57
04:57 patspam joined
diakopter phenny: tell pmurias please backlog; all kinds of neat stuff there 04:59
phenny diakopter: I'll pass that on when pmurias is around.
05:03 Targhan left 05:07 patspam left
snarkyboojum sorear: what did you mean by before or after 46422? 05:27
05:30 JimmyZ left 05:31 finanalyst joined
sorear snarkyboojum: 46422, committed six hours ago and fixed a few minutes ago, contains a segfault bug, but it's only on a branch (I forgot this for a moment) 05:40
snarkyboojum ah ok 05:44
05:54 JimmyZ joined 06:09 jonrafkind left 06:11 jonrafkind joined, \shade\ is now known as fn 06:12 alester left 06:13 fn is now known as \shade\ 06:19 kaare joined, kaare is now known as Guest48518 06:22 Guest48518 left 06:28 jonrafkind left
pugssvn r30595 | colomon++ | [t/spec] Fix test and unfudge. 06:49
dalek kudo: 118f4aa | (Solomon Foster)++ | src/core/ (5 files):
Overhaul sqrt for Numeric / Real.
 06:51
07:03 gfx joined 07:12 dolmen joined, bakedb__ joined 07:35 swestres joined 07:42 finanalyst left 08:02 plobsing left 08:39 levengli joined
levengli anybody home now? 08:39
08:40 dual_ joined
levengli hi dual_ 08:41
08:41 dual left 08:42 isBEKaml joined, levengli left 08:47 isBEKaml left 08:48 dual joined 08:49 dual_ left, isBEKaml joined 08:51 Su-Shee joined 09:01 gfx left
isBEKaml rakudo: sub subbydo { my $temp=23; return { say $_ + $temp }; }; my $tt = subbydo; my $test1=$tt(23); say $test1; say $tt.WHAT; 09:02
p6eval rakudo ab2322: OUTPUT«46␤1␤Block()␤»
isBEKaml uhhh, it executed alright. Why is a $test1 shown as 1? 09:04
I thought it only contained a ref (kinda) to subbydo?
moritz_ guten Morgen 09:05
phenny moritz_: 02:40Z <sorear> tell moritz_ that I've decided I don't like the plan I had earlier for splitting the setting, so I have nothing more for topic/lexical-persistence
isBEKaml guten morgen, moritz_ !
09:07 finanalyst joined
isBEKaml rakudo: sub subbydo { my $temp=23; return { say $_ + $temp }; }; my $tt = subbydo; my $test1=$tt(23); say $test1.WHAT; say $tt.WHAT; 09:18
p6eval rakudo ab2322: OUTPUT«46␤Bool()␤Block()␤»
isBEKaml I don't understand why $test1 would type to a Bool()... :| 09:19
09:19 masak joined
masak oh hai, #perl6 09:19
isBEKaml masak: hi
moritz_ \o/ 09:21
perlgeek.de/blog-en/perl-6/list-cla....writeback 7 comments - it seems that actually asking the reader a question does encourage people to comment
masak there you go.
[backlogging] diakopter has a screen-time limit? :) 09:25
09:26 Heame joined
isBEKaml that was the return statement in the subbydo doing this.. jeez.. :( 09:27
rakudo: sub subbydo { my $temp=23;{ say $_ + $temp }; }; my $tt = subbydo; my $test1=$tt(23); say $test1.WHAT; say $tt.WHAT;
p6eval rakudo ab2322: OUTPUT«Use of uninitalized value in numeric context␤23␤invoke() not implemented in class 'Boolean'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
isBEKaml rakudo: sub subbydo { my $temp=23;{ say $_ + $temp }; }; my $tt = subbydo; my $test1=$tt(23); 09:28
p6eval rakudo ab2322: OUTPUT«Use of uninitalized value in numeric context␤23␤invoke() not implemented in class 'Boolean'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
09:29 Heame left
isBEKaml odd, I thought the return statement in the earlier subbydo made it Bool(). This error message seems to be very misleading. Can someone over here clarify? 09:32
moritz_ has no idea what's going on 09:33
isBEKaml moritz_: there are two subs. The only difference is, one has a return statement, the other doesn't. With return, it works. Without, the error message is as above. 09:35
moritz_ yes; and I have no idea what's going on 09:36
opinion question: should ^@array actually re-dispatch to @array.keys instead of ^@array.elems ? 09:37
I guess usually the former is meant
isBEKaml if at all, I think the error msg should be "invoke() not implemented in class '<whatever type here>'" I think it can't be Boolean...
moritz_ either way might be dangerous for custom array indexes 09:38
09:42 M_o_C joined
masak [backlog] yes, we perhaps should have a man_vs_bot test. :) 09:51
isBEKaml moritz_: what would be the type of @array here? Any() ? If it contains only non-numeric elements, I think we shouldn't allow ^@array. ( I may be wrong here, though) 09:53
isBEKaml goes looking through backlogs for ^@ discussion 09:55
masak isBEKaml: prefix:<^> should numify its argument. 10:02
isBEKaml: more than a year ago, the spec was more complicated, and said that prefix:<^> on an array should give a list of 0..$n lists of indreasing $n. 10:03
but that extra complexity was deemed unnecessary and a cause of bugs. 10:04
so now it numifies instead, and there was much rejoicing.
isBEKaml masak: yes, that was why I thought we shouldn't allow ^@ on non numeric arrays. 10:06
masak no, I think you misunderstand. 10:07
the length is always numeric, that's all that matters.
both for the old semantics and the present one.
let's say @a = <a b c>; old semantics: ^@a would give ([0], [0, 1], [0, 1, 2]) 10:08
new semantics: ^@a gives (0, 1, 2)
whether the array is 'non numeric' doesn't factor into it.
isBEKaml oh, I see.. I was confused over 0..^ @a. Sorry for the line noise. 10:10
masak 0 ..^ @a works in the same way as given above. prefix:<^> is in every way sugar for 0..^ 10:11
isBEKaml array.length. we could simply have $#array? 10:12
masak er. 10:13
firstly, there's no @array.length. where did you get that?
isBEKaml I was referring to a semantic. I know there's no array.length. So it's 0 .. $#array ? 10:14
masak secondly, the $#array syntax was ugly even for Perl 5. somewhere inside that twisted parser, there must be a rule saying that '#' doesn't indicate a comment if it comes directly after a '$'.
oh, ok.
for a while I thought you were suggesting we reintroduce $#array :)
isBEKaml I won't reintroduce. Unlearn and relearn's fine with me. :) 10:15
masak yes, ^@a means 0..^@a means 0..(@a.elems-1)
rakudo: my @a = <a b c d e>; say @a.elems; say @a.end
p6eval rakudo ab2322: OUTPUT«5␤4␤»
masak I'd say $#a has been replaced by the slightly longer @a.end 10:16
isBEKaml $# was easy to remember. # for count. :) 10:17
now we have a mnemonic for last element. @a.end! :)
10:18 iblechbot joined
masak if '#' really meant count there, how come the number is one less than the count of the elements? :P 10:19
isBEKaml array indexing! :D
masak yes, but set cardinalities aren't usually "zero-based" in the sense that they're off by one from the actual size. 10:21
isBEKaml I was referring to the number and ids that we keep for items in day to day life. "Go over, meet the man in #5."
masak all I'm saying is I like @a.end much better. :)
moritz_ isBEKaml: @array would be positional
isBEKaml all I said it's a good mnemonic. :) 10:22
masak the mnemonic being '# denotes count, so this is the *count* of the elements... minus one'. er. 10:23
isBEKaml er, ppl remember stuff in different ways. room #5, counter #5 and so on. # is simply the last position in an array. remember how we had to grapple with list and scalar contexts in p5 ? assign a list to a scalar, you only get the last element. er, I just said @a.end is a much better mnemonic than $#a. 10:26
masak good, good. :)
we'll leave it at that.
lunch & 10:29
10:29 masak left
isBEKaml moritz_: we could just use @array.keys. opA(T) redispatch to opB(T.something), not opA(T.something). IOW, redispatch to do some other (possibly less costly) operation, never the same one. 10:33
in this case, both look similar.
moritz_ it's not about looks :-)
isBEKaml :)
moritz_ I just meant that ^@array could be made to mean @array.keys, not 0..^@array.elems 10:34
10:34 JimmyZ left
moritz_ but it's probably ETOOMUCHMAGIC 10:34
isBEKaml good, good. you got my point. :)
moritz_ and we should educate people to use @array.keys in the first place, even if it's 4 strokes more
10:35 EXCHIM joined
isBEKaml we probably should. it would also probbly mean saying, "there's one true way" when it's TMTOWDI 10:37
:)
isBEKaml was thinking about a good practices doc as a spill over from java world. 10:38
we could do the same, only in a slightly less intimidating way. :)
rakudo: my @a=<1 2 3 4>; say 0..^ @a; 10:43
p6eval rakudo ab2322: OUTPUT«01␤»
isBEKaml rakudo: my @a=<1 2 3 4>; say 0..^@a; 10:44
p6eval rakudo ab2322: OUTPUT«01␤»
isBEKaml alpha: my @a=<1 2 3 4>; say 0..^@a;
p6eval alpha 30e0ed: OUTPUT«0123␤»
colomon rakodu: my @a= 1 ... 4; 10:49
rakodu: my @a= 1 ... 4; say (^@a).perl
rakudo: my @a= 1 ... 4; say (^@a).perl 10:50
p6eval rakudo ab2322: OUTPUT«0..^[1, 2, 3, 4]␤»
colomon too early
alpha: my @a= 1 ... 4; say (^@a).perl
p6eval alpha 30e0ed: OUTPUT«0..^4␤»
10:54 JimmyZ joined
isBEKaml alpha: sub subbydo { my $temp=23;{ say $_ + $temp }; }; my $tt = subbydo; my $test1=$tt(23); 10:55
p6eval alpha 30e0ed: OUTPUT«Use of uninitialized value␤23␤invoke() not implemented in class 'Boolean'␤in Main (file <unknown>, line <unknown>)␤»
JimmyZ rakudo: my @array = 1, 3, 5; say +@array; 10:56
p6eval rakudo ab2322: OUTPUT«3␤»
JimmyZ rakudo: my @array = 1, 3, 5; say @array.keys;
p6eval rakudo ab2322: OUTPUT«012␤»
10:57 finanalyst left
JimmyZ rakudo: my @array; say ?@array; say +@array; 10:57
p6eval rakudo ab2322: OUTPUT«0␤0␤»
JimmyZ rakudo: my @array = 1; say ?@array; say +@array; 10:58
p6eval rakudo ab2322: OUTPUT«1␤1␤»
JimmyZ rakudo: my @array = 1,2 ; say ?@array; say +@array;
p6eval rakudo ab2322: OUTPUT«1␤2␤»
11:06 pmurias joined 11:07 JimmyZ left 11:18 JimmyZ joined
colomon > my @a= 1 ... 4; say (^@a).perl 11:22
0..^4
isBEKaml colomon: \o/ 11:26
colomon++
11:27 wknight8111 joined
isBEKaml alpha: my @a= 1 ... 4; my @bb= (^@a); say @bb.perl; 11:29
p6eval alpha 30e0ed: OUTPUT«[0, 1, 2, 3]␤»
colomon very easy fix 11:32
:) 11:33
spectesting it right now.
cognominal when I read something like @opstack, I have a itch to add twigils for stack and queue roles as opposed to full aray roles... That would give @>stack and @<queue. But, chaplet of twigils in names will be a sure way to scare more people 11:45
11:45 JimmyZ left
cognominal On the other hand, I find @>op nicer than @opstack 11:45
anyway, anyone will be free to transmogrify Perl 6 if he has the proper skills. 11:47
dalek kudo: c18ef37 | (Solomon Foster)++ | src/core/Range.pm:
Switch prefix:<^> to use +$max internally.
 12:00
12:01 sundar joined
sundar what's the way to get the latest version of Synopses? 12:02
I remember someone saying the web version is usually outdated and it's better to pull it from some repository. 12:03
colomon Web version updates automatically.
I see the page I have open was updated this morning.
but it's worth checking that, because the automatic update was broken for a while earlier this year. 12:04
sundar colomon: oh ok, I now see that the page I'm looking at was updated yesterday evening. thanks. 12:05
colomon the other way to get the latest is to pull them down from the subversion repo
svn.pugscode.org/pugs 12:06
lisppaste3 sundar pasted "Why does the second \s take a + and not a * like the others? (from S05)" at paste.lisp.org/display/98993 12:07
colomon just guessing: because it's between two alphabetic characters. 12:08
sundar colomon: Ok, I'll pull that now for offline reading pleasure. :)
colomon svn.pugscode.org/pugs/docs/Perl6/Spec is more specific; there is a lot of stuff in the old pugs repo. 12:09
12:11 JimmyZ joined
sundar colomon: thanks, that makes sense... TIL ws is another dwimmery in perl6. 12:15
pugssvn r30596 | colomon++ | [t/spec] Refudge a tad, add a couple of tests to make sure prefix:<^> numifies its argument.
colomon mind you, I've never messed around with the :s modifier at all, so I might be completely wrong. :) 12:16
sundar colomon: you were not. I found another part of the Syn where it says <.ws> is "\s+ if it's between two \w characters,", and "\s* otherwise". :) 12:21
12:26 fridim left
JimmyZ rakudo: my @a = 1, 2, 4; say (1...^@a).perl.say; 12:33
p6eval rakudo ab2322: OUTPUT«(1)␤1␤»
JimmyZ rakudo: my @a = 1, 2, 4; (1..^@a).perl.say;
p6eval rakudo ab2322: OUTPUT«1..^[1, 2, 4]␤»
12:36 rv2733 left 12:37 rv2733 joined
colomon JimmyZ: p6eval doesn't have the latest changes yet. 12:37
actually, it's two days behind at the moment. that's a sign of trouble. :( 12:38
12:38 maspalio joined
JimmyZ Yeah 12:39
I am compiling by myself now. 12:40
12:40 maspalio left 12:43 masak joined
JimmyZ alpha: my @a = 1, 2, 4; (1..^@a).perl.say 12:45
p6eval alpha 30e0ed: OUTPUT«1..^[1, 2, 4]␤»
JimmyZ pugs: my @a = 1, 2, 4; (1..^@a).perl.say 12:46
p6eval pugs: OUTPUT«(1, 2)␤»
JimmyZ alpha: my @a = 1, 2, 4; (^@a).perl.say 12:47
p6eval alpha 30e0ed: OUTPUT«0..^3␤»
JimmyZ rakudo: my @a = 1, 2, 4; (^@a).perl.say
p6eval rakudo ab2322: OUTPUT«0..^[1, 2, 4]␤»
JimmyZ ./perl6 -e 'my @a = 1, 2, 4; (1..^@a).perl.say' 12:48
masak cognominal: [backlog] adding twigils for stack and queue sounds like a good idea for a module.
JimmyZ 1..^[1, 2, 4]
colomon masak: o/ 12:49
masak \o
colomon I'm hanging with my parents and hunting LHF. 12:51
32,000 passing tests is dangerously close.
12:51 rgrau_ left
masak cheers colomon on 12:51
JimmyZ colomon: Is the result right?
masak I'm drafting up the presentation I'll be giving in a week and some. 12:52
JimmyZ: no, but that bug was reported the other day.
colomon JimmyZ: (^@a).perl.say is for sure wrong, and I've already fixed it.
(1..^@a).perl.say is a much trickier question 12:53
JimmyZ jimmy@ubuntu:~/rakudo$ ./perl6 -e 'my @a = 1, 2, 4; (1..^@a).perl.say'
1..^[1, 2, 4]
jimmy@ubuntu:~/rakudo$ ./perl6 -e 'my @a = 1, 2, 4; (^@a).perl.say'
0..^3
colomon right, that's the correct (fixed) behavior for (^@a).perl.say
masak why is it tricky? I guess it has to do with not being able to numify because of strings.
colomon masak: exactly.
moritz_ it probably deserves a special case 12:54
if $arg ~~ Positional { numifiy }
masak well, I don't think arrays should be whitelisted.
colomon with prefix:<^>, we know we want to numify
masak I think everything but scalar values should be blacklisted.
moritz_ rakudo: say (^'d').perl
p6eval rakudo ab2322: OUTPUT«0..^"d"␤»
moritz_ well 12:55
colomon that's the old behavior, it's 0..0 now.
moritz_ one can always say that if one of the endpoints is ~~ Numeric, the other should be numified too
masak colomon: right, but the *reason* we know we want to numify has to be the same as the reason we know we want to numify 0..^@a
moritz_ rakudo: say (^'d').perl
p6eval rakudo c18ef3: OUTPUT«0..^0␤»
moritz_ colomon++
moritz_ will hold his Perl 6 talk tomorrow 12:56
JimmyZ rakudo: (^'d4').perl.say
p6eval rakudo c18ef3: OUTPUT«0..^0␤»
JimmyZ rakudo: (^'4d').perl.say
p6eval rakudo c18ef3: OUTPUT«0..^4␤»
JimmyZ rakudo: say 'd4' + 4 12:57
p6eval rakudo c18ef3: OUTPUT«4␤»
colomon moritz_: did you just rebuild p6eval, or did it do it on its own?
afk 12:58
moritz_ colomon: I did 12:59
I guess the rebuild is now sufficiently sane that I can enable the cron job again 13:00
I've now enabled it every other hour
15 */2 * * * perl /home/p6eval/rebuild-rakudo.pl >> /home/p6eval/rakudo-buildlog 2>&1 13:02
13:08 sundar left 13:09 TiMBuS left
colomon cron++ 13:20
moritz_++
masak p6eval++
dalek kudo: ee6ec74 | (Solomon Foster)++ | src/core/Pair.pm:
Add infix:<cmp> for pairs.
 13:21
pugssvn r30597 | colomon++ | [t/spec] Unfudge infix:<cmp> tests on pairs, also rewrite two of the tests so they parse correctly. 13:22
colomon with that, I'm showing 31974 passing tests. 13:23
13:24 EXCHIM left 13:25 EXCHIM joined 13:29 M_o_C left
JimmyZ wow, will reach alpha 13:33
13:34 alester joined
masak ...but with different passing tests. 13:35
anyway, only 6 tests left to 32k !
moritz_ 26 13:37
if my subtraction cells haven't left me
JimmyZ rakudo: say 32731 - 31974 13:38
p6eval rakudo c18ef3: OUTPUT«757␤»
moritz_ rakudo: postfix:<k>($x) { 1000 * $k }; say 32\k - 31974
p6eval rakudo c18ef3: OUTPUT«Confused at line 11, near "postfix:<k"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
masak er. that's a seven. right, 26.
colomon rakudo: postfix:<k>($x) { 1000 * $k }; say 32k - 31974 13:39
p6eval rakudo c18ef3: OUTPUT«Confused at line 11, near "postfix:<k"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
moritz_ rakudo: use MONKEY_TYPING; augment class Int { method k ($x) { 1000 * $k }}; say 32.k - 31974
p6eval rakudo c18ef3: OUTPUT«Symbol '$k' not predeclared in k␤current instr.: 'perl6;PCT;HLLCompiler;panic' pc 152 (compilers/pct/src/PCT/HLLCompiler.pir:108)␤»
moritz_ rakudo: use MONKEY_TYPING; augment class Int { method k ($x) { 1000 * $x }}; say 32.k - 31974
p6eval rakudo c18ef3: OUTPUT«Method 'k' not found for invocant of class 'Integer'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
moritz_ weeps
masak \k
rakudo: postfix:<k>($x) { 1000 * $x }; say 32\k - 31974 13:40
p6eval rakudo c18ef3: OUTPUT«Confused at line 11, near "postfix:<k"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
masak rakudo: sub postfix:<k>($x) { 1000 * $x }; say 32\k - 31974
p6eval rakudo c18ef3: OUTPUT«26␤»
colomon masak++
JimmyZ rakudo: sub postfix:<k>($x) { 1000 * $x }; say 32k - 31974 13:41
p6eval rakudo c18ef3: OUTPUT«26␤»
masak oh, even that.
JimmyZ alpha: sub postfix:<k>($x) { 1000 * $x }; say 32k - 31974 13:42
13:42 rurban joined
p6eval alpha 30e0ed: OUTPUT«26␤» 13:42
JimmyZ alpha: sub postfix:<k>($x) { 1000 * $x }; say 32\k - 31974 13:43
p6eval alpha 30e0ed: OUTPUT«Confused at line 10, near "\\k - 31974"␤in Main (file <unknown>, line <unknown>)␤»
13:48 Su-Shee left
moritz_ std: sub postfix:<k>($x) { }; 32k 13:50
p6eval std 30597: OUTPUT«ok 00:01 116m␤»
moritz_ std: sub postfix:<k>($x) { }; 32\k
p6eval std 30597: OUTPUT«ok 00:01 114m␤»
masak std: sub postfix:<k>($x) { }; 32kk
p6eval std 30597: OUTPUT«ok 00:01 114m␤»
JimmyZ rakudo: sub postfix:<k>($x) { 1000 * $x }; say 32kk - 31974 13:51
p6eval rakudo c18ef3: OUTPUT«31968026␤»
JimmyZ rakudo: sub postfix:<k>($x) { 1000 * $x }; say 32\k\k - 31974
p6eval rakudo c18ef3: OUTPUT«31968026␤»
JimmyZ rakudo: sub postfix:<k>($x) { 1000 * $x }; say 32\kk - 31974
p6eval rakudo c18ef3: OUTPUT«31968026␤»
13:55 slavik1 left 13:59 hercynium left, slavik joined 14:00 ReiniUrban joined
moritz_ www.perlmonks.org/?node_id=839112 couldn't resist :-) 14:01
14:01 rurban left, ReiniUrban is now known as rurban, rurban left
slavik moritz_: NICE!!!! 14:02
moritz_: that should be in the book :)
also, what does the 1_000 exactly mean? why not 1000 instead?
moritz_ it's the same
slavik oh
moritz_ rakudo: say 1_000_000 14:03
p6eval rakudo c18ef3: OUTPUT«1000000␤»
moritz_ works the same way in perl 5 too
buubot: eval: 1_024
buubot moritz_: 1024
slavik _ is like a comma? (period in europe)
masak moritz_++
moritz_ yes
slavik good point
moritz_++
moritz_ slavik: but you can use it at any position in a number
rakudo: say 10_24 14:04
p6eval rakudo c18ef3: OUTPUT«1024␤»
slavik I see, so it's only visual?
moritz_ some cultures like to group their numbers in pairs
right
slavik ok, cool
so technically, Perl6 has some features that Java7 was supposed to have ... 14:05
masak the Chinese tend -- at some level anyway -- to group in fours.
JimmyZ: am I right?
moritz_ anyway, I'm fine with allowing _ at arbitrary positions
rakudo: say 1__3
p6eval rakudo c18ef3: OUTPUT«Confused at line 11, near "say 1__3"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
moritz_ std: 1__3 14:06
p6eval std 30597: OUTPUT«===SORRY!===␤Whitespace is required between alphanumeric tokens at /tmp/zrIsnxD0WC line 1:␤------> 1⏏__3␤Two terms in a row at /tmp/zrIsnxD0WC line 1:␤------> 1⏏__3␤ expecting any of:␤ POST␤ bracketed infix␤ infix or
..meta-i…
slavik rakudo say 1_3
JimmyZ masak: I can't follow you
slavik rakudo: say 1_3
p6eval rakudo c18ef3: OUTPUT«13␤»
moritz_ rakudo: say 12_
p6eval rakudo c18ef3: OUTPUT«Confused at line 11, near "say 12_"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
slavik I guess with two _, it thinks there is meaning to it
rakudo: say _12
p6eval rakudo c18ef3: OUTPUT«Could not find sub &_12␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
slavik right, because it's not a number ... 14:07
is it possible with Perl6 to have your own context definitions?
moritz_ yes 14:08
slavik something like: my $feet = context<feet>(convert(30.48cm)) or something ...
moritz_ for example striing context consists only of a method call
to .Str or .Stringy
masak JimmyZ: I mean that 万 is a 'unit', sort of one level above 10. 14:09
slavik 100?
moritz_ so you can define a prefix:<imperial> that calls the .Imperial method
and define that method in all number types that present a value + unit
masak slavik: no, I don't mean like that. we tend to organize things in 10^3 units.
slavik: the Chinese (and the Koreans, and perhaps the Japanese too) tend to go with 10^4. 14:10
slavik moritz_: or have all objects inherit an interface of measurement?
masak: ahh, I see
moritz_ slavik: likely s/inherit an interface/"do" a role/
but of course inheritance would work too
masak so the Chinese don't think of the population of China as being 1.4 * 10^9, but instead 14 * 10^8.
TimToady yes, the Japanese also 14:11
moritz_ which is just as correct :-)
masak aye, but a funny shift of perspective.
I like the pause that both westerners and easterners have to take when converting :) 14:12
JimmyZ masak: yes, but it's old, now most I saw is three , not four.
masak JimmyZ: how would you think about the population of China?
moritz_ "big" *SCNR*
masak JimmyZ: 1.4 * 10^9 or 14 * 10^8?
JimmyZ well, we say 14亿 14:13
masak I rest my case.
slavik moritz_: right ... 14:14
pugssvn r30598 | colomon++ | [t/spec] Unfudge tests. 14:16
colomon 31980 passing tests
JimmyZ masak: actually, we use three or four, depends on the fields. 14:17
masak oh, ok. science does 3 more?
JimmyZ yeah
usually advanced field 14:18
masak makes cultural sense, methinks.
TimToady interestingly, ancient Greek had the concept of "myriad", which was 10,000
and "myriads of myriads" 14:19
dalek kudo: e113d85 | (Solomon Foster)++ | src/core/Array.pm:
Add proto for unshift so we can use named args.
moritz_ wonders what kind of numbers the greeks measured in "myriads of myriads"
JimmyZ masak: you're right.
moritz_ in great britain the scientists mostly use the SI system too 14:20
TimToady moritz_: in general, they preferred geometry :)
masak moritz_: probably most often "I've told you myriads of myriads of times not to exaggerate!" :P
moritz_ :-)
14:21 alester left 14:22 am0c joined
masak this list might be of interest to someone: gist.github.com/395182 14:22
it's of the 185 t/spec files that Rakudo's t/spectest.data doesn't mention.
moritz_ which tools/update_passing_test_data.pl processes 14:23
masak o rly? I figured some tool would do that :)
moritz_ masak++ #made me look at a particular test file 14:25
pugssvn r30599 | moritz++ | [t/spec] delete a test file that consistet of four tests; two of them bogus, two already present elsewhere 14:26
moritz_ ... which I now deleted :-)
masak \o/
moritz_ does EnumMap preserve key ordering? 14:27
masak don't think so.
if a file comes out as 'some tests passed' by this tool, can there still be reasons not to add it to t/spectest.data?
moritz_ sure 14:28
for example rakudo passes some tests in temp.t
but it doesn't implement 'temp' at all
=> no point in adding it
masak ah.
let.t is the same, but it also runs all tests! o.O 14:29
time to write more tests. :)
pugssvn r30600 | moritz++ | [t/spec] correct some enum tests that relied on hash key order 14:33
r30601 | moritz++ | [t/spec] fix a syntax error 14:34
r30602 | moritz++ | [t/spec] fudge undefined-types.t for Rakudo 14:36
r30603 | moritz++ | [t/spec] unfudges for rakudo, found by autounfudge 14:38
14:38 alester joined
moritz_ uhm 14:40
t/spec/S05-match/make.t
isBEKaml perlmonks.org is down?
moritz_ that looks bogus
isBEKaml I can't connect to it.. :(
moritz_ "4" ~~ / (\d) { make $0.sqrt } Remainder /;
ok($/);
I'd say that the regex doesn't match 14:41
and that $/ is false even though make() was called
correct?
isBEKaml: wouldn't be the first time
... but works for me
isBEKaml hmm, I just wanted to view the page you linked to... 14:42
dalek kudo: 73a3d3e | moritz++ | t/spectest.data:
[t/spec] run another test file
isBEKaml now I can see it.. repeated attempts. :(
pugssvn r30604 | moritz++ | [t/spec] correct make.t - a $/ of a failed match is False, even if make() was called 14:43
isBEKaml rakudo: say 1_00_000; 14:45
p6eval rakudo e113d8: OUTPUT«100000␤»
isBEKaml rakudo: say 1_00_000 + 20_00_000;
p6eval rakudo e113d8: OUTPUT«2100000␤»
isBEKaml exactly like p5 ;)
moritz_ std: my @a; / ||@a /
p6eval std 30599: OUTPUT«ok 00:01 113m␤» 14:46
pugssvn r30605 | moritz++ | [t/spec] fudge sequential-alternation.t for rakudo 14:47
moritz_ rakudo: multi a($a) { 1 }; multi a(:$a) { 2 }; say a(:a) 14:49
p6eval rakudo e113d8: OUTPUT«2␤»
JimmyZ rakudo: say 1_0000
p6eval rakudo e113d8: OUTPUT«10000␤»
moritz_ rakudo: multi a($a) { 1 }; multi a(:$a) { 2 }; multi a(:$a, :$b); say a(:a)
JimmyZ rakudo: say 1,0000
p6eval rakudo e113d8: OUTPUT«Malformed multi at line 11, near "a(:$a, :$b"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
rakudo e113d8: OUTPUT«10␤»
JimmyZ rakudo: say 10,000
moritz_ rakudo: multi a($a) { 1 }; multi a(:$a) { 2 }; multi a(:$a, :$b) { 3 } say a(:a)
p6eval rakudo e113d8: OUTPUT«100␤»
rakudo e113d8: OUTPUT«Confused at line 11, near "multi a(:$"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
moritz_ rakudo: multi a($a) { 1 }; multi a(:$a) { 2 }; multi a(:$a, :$b) { 3 }; say a(:a)
p6eval rakudo e113d8: OUTPUT«Ambiguous dispatch to multi 'a'. Ambiguous candidates had signatures:␤:(Any :a($a))␤:(Any :a($a), Any :b($b))␤␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
JimmyZ rakudo: say 10,000
p6eval rakudo e113d8: OUTPUT«100␤»
JimmyZ pugs: say 10,000 14:50
p6eval pugs: OUTPUT«100␤»
14:50 alester left
moritz_ I guess this should not be ambigious, right? 14:50
t/spec/S06-multi/positional-vs-named.t
JimmyZ pugs: print 1,2 14:51
p6eval pugs: OUTPUT«12»
JimmyZ pugs: say 1,2
p6eval pugs: OUTPUT«12␤»
JimmyZ rakudo: say '000'
p6eval rakudo e113d8: OUTPUT«000␤»
JimmyZ rakudo: say 10,'000'
p6eval rakudo e113d8: OUTPUT«10000␤»
JimmyZ rakudo: say 10,+'000' 14:52
p6eval rakudo e113d8: OUTPUT«100␤»
ash___ rakudo: say 10_000
moritz_ rakudo: eval 'augment class Int { }'
p6eval rakudo e113d8: OUTPUT«10000␤» 14:53
rakudo e113d8: ( no output )
ash___ moritz_: shouldn't that complain about use MONKEY_TYPING?
moritz_ ash___: yes, but the eval() is silent :-)
ash___ ah
moritz_ rakudo: use MONKEY_TYPING; { class Foo { }; augment class Foo { } } 14:54
p6eval rakudo e113d8: ( no output )
moritz_ incredibly stupid
pugssvn r30606 | moritz++ | [t/spec] fudge augment-supersede.t 14:55
moritz_ I was wondering why stuff didn't work in a test file, but worked here on evalbot
I forgot to save the file.
JimmyZ rakudo: eval ' class Foo{ }; augment class Foo { method Str() { 'hi'; }; say ~Foo'
p6eval rakudo e113d8: OUTPUT«Confused at line 11, near "eval ' cla"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
JimmyZ rakudo: eval 'class Foo{ }; augment class Foo { method Str() { 'hi'; }; say ~Foo'
p6eval rakudo e113d8: OUTPUT«Confused at line 11, near "eval 'clas"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
ash___ I love when I do that, I switch to my terminal, hit make, run my program and the new change isn't working for some reason....
silly save button
14:56 Coleoid joined
moritz_ JimmyZ: can't use 'hi' inside a single-quoted string 14:56
ash___ rakudo: eval ' class Foo{ }; augment class Foo { method Str() { \'hi\'; }; say ~Foo' 14:57
p6eval rakudo e113d8: ( no output )
JimmyZ rakudo: eval "use MONKEY_TYPING; class Foo{ }; augment class Foo { method Str() { 'hi'; } } ; say ~Foo" 14:58
p6eval rakudo e113d8: OUTPUT«Useless declaration of has-scoped Method in a module; add our or my to install it in the lexpad or namespace␤Useless declaration of has-scoped Method in a module; add our or my to install it in the lexpad or namespace␤Null PMC access in get_string()␤current instr.: '_block14'
..pc 29…
JimmyZ rakudo: use MONKEY_TYPING; class Foo{ }; augment class Foo { method Str() { 'hi'; } } ; say ~Foo
p6eval rakudo e113d8: OUTPUT«Foo()␤»
JimmyZ rakudo: use MONKEY_TYPING; class Foo{ }; augment class Foo { method Str() { 'hi'; } } ; say ~Foo.new
p6eval rakudo e113d8: OUTPUT«hi␤» 14:59
JimmyZ rakudo: class Foo { method Str() { 'hi'; } } ; say ~Foo
p6eval rakudo e113d8: OUTPUT«Foo()␤»
JimmyZ rakudo: eval "use MONKEY_TYPING; class Foo{ }; augment class Foo { our method Str() { 'hi'; } } ; say ~Foo.new" 15:00
p6eval rakudo e113d8: OUTPUT«Could not find sub 20_1273416224.9741␤current instr.: '' pc -1 ((unknown file):-1)␤»
colomon what does S& mean?
like an all junction, but not threaded?
moritz_ yes 15:01
JimmyZ rakudo: eval "use MONKEY_TYPING; class Foo{ }; augment class Foo { method Str() { 'hi'; } } ; say ~Foo.new"
colomon junctions don't thread now, do they?
p6eval rakudo e113d8: OUTPUT«Useless declaration of has-scoped Method in a module; add our or my to install it in the lexpad or namespace␤Useless declaration of has-scoped Method in a module; add our or my to install it in the lexpad or namespace␤Null PMC access in get_string()␤current instr.: '_block14'
..pc 29…
moritz_ colomon: nope, they don't
pugssvn r30607 | moritz++ | [t/spec] correct S12-class/literal.t
colomon we're skipping 55 tests in rx.t because we don't implement S&. 15:02
JimmyZ rakudo: eval "class Foo { our method Str() { 'hi'; } } ; say ~Foo.new"
p6eval rakudo e113d8: OUTPUT«Could not find sub 19_1273416350.42779␤current instr.: '' pc -1 ((unknown file):-1)␤»
moritz_ colomon: I'm fine with replacing that by a simpler construct
colomon oh?
moritz_ colomon: like 'and' or '&&' if applicable
(watch the precedence)
colomon the tests are like this: 15:03
moritz_ S05-* is not about testing S&. It's about testing regexes
colomon ok eval(q{{ 'abcZ' ~~ /abc\Z/ }}) ~~ Failure S& /reserved/, 'retired metachars (\Z)';
I admit I don't really understand the logic.
moritz_ the return value of the eval should be a failure, and match /reserved/
colomon (I was just thinking of implementing Sop which just does op....)
JimmyZ Could not find sub 19_1273416350.42779? what's that?
moritz_ a LTA error message 15:04
std: print $*OUT: "foo"
p6eval std 30606: OUTPUT«ok 00:01 113m␤»
masak rakudo: {nextsame}() 15:05
moritz_ would gladly throw that syntactic form over board
p6eval rakudo e113d8: OUTPUT«Null PMC access in clone()␤current instr.: '&nextsame' pc 18507 (src/builtins/assign.pir:78)␤»
JimmyZ seems that augment can't in eval
masak moritz_: that use case for that syntactic form is basically the only argument I see for defending it ;P
moritz_ masak: it's only worth defending for Perl 5 convenience, IMHO 15:06
masak perhaps, but even that is not something to be thrown overboard lightly. 15:07
moritz_ not? :-)
pugssvn r30608 | moritz++ | [t/spec] more Rakudo unfudges found by autounfudge 15:08
colomon moritz_++ 15:10
masak no. all other things being equal, I see it as a strength being able to answer affirmatively to the question 'can you still do "print $FILE 1, 2, 3" in Perl 6?', albeit with the slight caveat that we made the syntax more consistent.
colomon doing it the scientific way, instead of just looking manually for LHF...
masak because there are some things we value more than backwards compatibility, such as the never-two-terms-in-a-row rule. 15:11
dalek kudo: ad46ef8 | moritz++ | t/spectest.data:
run two more test files
JimmyZ LTA ? Less Than Awesome?
colomon moritz_: to put it a different way: can you see any reason why those tests need S& instead of just &?
JimmyZ: yes
JimmyZ colomon: thanks 15:12
moritz_ colomon: the scientific (and automated) way has a drawback - it discouraged me from manually looking at test files, which is often also needed
colomon sure, most of my bumps have involved simple patches to Rakudo.
moritz_ colomon: the idea was that the regex match is only carried out when the eval failed... 15:13
colomon: but since a non-Exception return value won't cause an error, I don't think it *needs* to be S&
moritz_ wonders if the S meta op could just set a $*DO_IT_SEQUENTIALLY variable 15:14
colomon moritz_: that's how I'd been thinking about S.
moritz_ and if that is set, the autotreading and hyper dispatcher will do it sequentially
which is currently a no-op, because it is sequentially anyway :-)
colomon: so +1 from me to implementing S 15:15
colomon my only hesitation is that sometimes S means "short-circuit" instead, if I understand the spec correctly. And that's completely different from sequential, and I have no idea how to handle it. 15:16
btw, how many tests did you win back with those two additional files? 15:17
15:17 rgrau_ joined
colomon (first stab at implementing Sop now compiling) 15:20
arnsholt colomon: Sequential evaluation for | and & would essentially be short-circuit semantics
Or at least, that's how I understand it 15:21
colomon do we have tests for this? if no, can we come up with them? 15:22
arnsholt I don't know if there are any tests, but I suppose the examples from S03 are a place to start 15:23
colomon rakudo; say 2 orelse 4 15:24
rakudo: say 2 orelse 4
p6eval rakudo e113d8: OUTPUT«Confused at line 11, near "say 2 orel"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
15:26 dual left
ash___ std: say 2 orelse 4 15:26
p6eval std 30606: OUTPUT«ok 00:01 110m␤»
ash___ do the values in the settings belong to any package? GLOBAL maybe? 15:28
colomon S03-operators/also.t
masak ash___: CORE 15:30
ash___ thanks masak
diakopter moritz_: did you see perlesque passed man_or_boy? 15:31
masak S02 has the details.
colomon ooooo, except for 3 S& 4 S& 5, we pass all the S& tests in also.t
I'm going to add that test file and spectest
isBEKaml what are the flags that can be passed with --trace to perl6? 15:34
the running.pod doesn't flesh them out...
I just passed numbers and it's dumping the whole trace.. er, it's still doing it.. :| 15:35
pmurias diakopter: hi 15:36
phenny pmurias: 04:59Z <diakopter> tell pmurias please backlog; all kinds of neat stuff there
15:36 JimmyZ left
diakopter pmurias: hi 15:36
15:38 s1n left, gurjeet left 15:39 lichtkind joined, gurjeet joined
isBEKaml Specifically, what do those numbers stand for? perl6 --trace={0,1,4} -e "say 'hi';" 15:39
pmurias diakopter: you intend to also class declaration with class Foo {...} not only through runsharp method calls? 15:40
diakopter yes :)
today
pmurias good
diakopter type literals are cool, too 15:41
you saw the man_or_boy.t results?
pmurias type literals? 15:42
diakopter perlesque: say(int)
p6eval perlesque: OUTPUT«System.Int32␤»
lichtkind mas 15:43
masak: hej
masak o/
lichtkind :)
pmurias t/man_or_boy.t should print out -67
diakopter yes 15:44
oh, I neglected to add TAP stuff 15:45
15:45 isBEKaml left
pmurias does visual studio support TAP integration or does it need the test results in a different format? 15:46
diakopter no idea
I commit to svn, then update from linux VM and test from mono
then fix bugs then commit again
pmurias i see 15:47
masak lichtkind: I saw your post: use.perl.org/~perl6doc/journal/40348
lichtkind masak: and? 15:48
diakopter I've gotten quite good at printf debugging, since I can't really step through or set breakpoints in generated code
pmurias should i fix the man_or_boy.t?
diakopter I can
masak lichtkind: I really hope someone starts such a project. whether from a clean slate or from November doesn't matter that much to me. November has been useful in other ways already.
15:49 finanalyst joined
pmurias fixing 15:49
masak there is still potential to be explored in the area of Perl 6 wiki software, it's just that currently that niche isn't filled with any person with tuits.
lichtkind masak: do you agree that the retrieving DBI data and buitlding html sites is the easy part? 15:52
masak in the sense that it's just a SMOP, yes. 15:53
lichtkind insider answer :)
dalek meta: r257 | pawelmurias++ | trunk/Sprixel/t/man_or_boy.t:
man or boy test produces TAP
 15:54
masak my answer was meant to be read as something akin to this argument: blog.bitquabit.com/2009/07/01/one-w...cker-news/
i.e. some things look trivially easy, but the actual complexity has been invested into making the thing easy to use. 15:55
it might be the lack of something like that which makes SocialText near unusable.
similarly, the RT interface for constructing queries by shunting ANDs and ORs around in a formula tree structure -- obviously designed by a programmer -- is cute in theory, but bafflingly useless in practice. 15:56
16:00 [mark] joined 16:05 [mark] left
colomon 32005 passing tests! 16:07
16:07 molaf joined
masak \o/ 16:07
colomon++
pugssvn r30609 | colomon++ | [t/spec] Fudge for Rakudo.
dalek kudo: 27e05a8 | (Solomon Foster)++ | (3 files):
Add dead simple preliminary implementation of Sop, and turn on the only test we were able to find for it.
 16:08
slavik out of how many?
colomon 39026 that tools/test_summary.pl knows about 16:12
slavik so roughly 80%
but not 100% of the spec has tests, right? 16:13
colomon surely not
slavik how much of spec has tests? best guesstimate :) 16:14
colomon I've got not a clue.
ash___ slavik: you have to consider parts of the spec aren't fully written to, like S15 through S20 are rough drafts 16:15
colomon speaking of which, moritz_++, I think your next LHF challenge should be to write tests for the S metaop. :) 16:16
ash___ really i'd say about 14 through 31 are rough drafts and most of those don't have tests, they might have some, but not many
diakopter pmurias: I overrode your changes; man_or_boy does 1..17 now 16:17
colomon heh, none of those S& tests in rx.t actually work. but I've switched them all to todos instead of skips.
ash___ plus, one of the biggest challenges (that would be awesome to achieve) is to have rakudo run STD.pm6 that would be great
slavik ash___: good point 16:18
dalek meta: r258 | diakopter++ | trunk/Sprixel/ (3 files):
[perlesque] loop form of man_or_boy.t, to test 1..17 starting values
pugssvn r30610 | colomon++ | [t/spec] Move 55 "skip" tests to "todo".
pmurias diakopter: np
16:18 stepnem joined
masak now has a project called 'failure' :) github.com/masak/failure 16:19
it's Perl 5, though.
diakopter masak: lol
masak now I just need to name a project 'disaster' :P
diakopter boondoggle 16:20
deathmarch
masak dancing class &
16:20 masak left 16:23 Ky6uk joined 16:24 patrickas joined 16:28 Ky6uk left 16:31 EXCHIM left
moritz_ last test run: 31998 16:34
passing
16:36 literal left
patrickas moritz_, I always wondered about how many tests does/did Pugs pass and what is the intersection of passing tests between rakudo and Pugs? Is such info easy to get ? 16:36
phenny patrickas: 08 May 11:35Z <masak> tell patrickas that the logotype/logo_32x32.png idea sounds good.
16:37 literal joined
moritz_ patrickas: not sure where such info could be found... apart from running the tests yourself 16:37
arnsholt Pugs probably passed more tests back in the day too 16:38
moritz_ ... when it tracked the spec :-)
arnsholt Yeah
patrickas oh... I was kinda hoping rakudo would be passing more tests than Pugs by now 16:39
moritz_ I'm pretty sure it does 16:41
rakudo: /<[-]>/ 16:42
p6eval rakudo e113d8: OUTPUT«Obsolete use of hyphen in enumerated character class;in Perl 6 please use .. instead at line 11, near "]>/"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
diakopter moritz_: numbers: pugs.blogs.com/pugs/2008/07/pugshs-is-back.html 16:43
moritz_ diakopter: thanks 16:44
rakudo: say 'ab' ~~ /<after a>b/ 16:45
p6eval rakudo 27e05a: OUTPUT«Method 'after' not found for invocant of class 'Regex;Cursor'␤current instr.: '_block48' pc 317 (EVAL_1:128)␤»
moritz_ rakudo: say 'ab' ~~ /<?after a>b/ 16:46
p6eval rakudo 27e05a: OUTPUT«Method 'after' not found for invocant of class 'Regex;Cursor'␤current instr.: '_block48' pc 317 (EVAL_1:128)␤»
moritz_ rakudo: say 'ab' ~~ /<?before a>b/
p6eval rakudo 27e05a: OUTPUT«␤»
moritz_ rakudo: /\A/ 16:47
p6eval rakudo 27e05a: OUTPUT«Obsolete use of \A as beginning-of-string matcher;in Perl 6 please use ^ instead at line 11, near "/"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
moritz_ rakudo: say eval('/\A/') 16:48
p6eval rakudo 27e05a: OUTPUT«␤» 16:49
diakopter how many tests in the entire test suite currently?
moritz_ TimToady: t/spec/S05-mass/rx.t seems to assume that a failed eval *returns* a Failure object - is that correct?
diakopter: about 40k
diakopter so rakudo fails 8K tests. 16:50
is it possible to estimate how many of those 8K tests are of a generated nature?
moritz_ that number sounds deceptively low
diakopter: not many... about 2k or so probably p5 regex tests borrowed from the p5 test suite 16:51
diakopter (that is, how many of those 8K tests appeared after that pugs blog entry)
moritz_ that's not the same question :-)
diakopter it's fairly close
I mean, both of those conditions 16:52
well really, just how many appeared after that pugs blog entry, sorry
using the generated/imported condition as an estimate if necessary
16:52 patrickas left
moritz_ about 20k new tests, of which probably 16k to 18k are not individually hand-written 16:53
diakopter ok. assume 4K handwritten.
seems high
I'll assume 2K
moritz_ well, I've put a lot of effort into it; so has KyleHa++ 16:54
diakopter yes.
oh, I guess 4K is reasonable given some of those are loop-y tests
16:55 patrickas joined 16:57 Intensity joined
diakopter ok, pugs failed 2K/19K tests in July 2008, and today rakudo fails 8K/40K tests (of which 17K (21K minus 4K handwritten/new/passing) are 1. both passing and 2. generated/imported), which leaves 8K out of the July 2008 19K that Rakudo isn't passing 16:59
oops, numbers fail. trying again. 17:00
moritz_ would try to argue otherwise
if pugs passed most of the new/generated tests, minus the 4k new ones
it would pass all but 6k total 17:01
which would be 36k
assuming it did similar spec changes tracking as rakudo does
diakopter why would it pass the new/generated tests but not the 4k handwritten ones?
moritz_ because the generated ones can be made to pass with rather little effort 17:02
diakopter true; good point.
moritz_ so, did we learn anything of value from our little number experiments? 17:03
that pugs was really awesome while it was being actively developed? I guess we knew that before :-) 17:04
that rakudo still has a long way to go? we knew that too
diakopter no; those weren't my points/intents
I don't know. I guess I just don't like all those generated tests, b/c they seem to inflate the numbers, regardless of implementation/comparisons 17:05
moritz_ I also think they should be somehow reduced 17:06
diakopter maybe I could grow to like them someday
moritz_ yet I do see some value; the number of should-be-covered cases does increase quadratically with the number of supported numeric types
17:07 am0c left
moritz_ rakudo: say (+<Int Num Rat Complex Real Numeric>)**2 17:07
p6eval rakudo 27e05a: OUTPUT«36␤»
diakopter yeah. maybe most of the non-edge/corner crosses could be moved to a "stresstest" target 17:08
or something. 17:09
moritz_ not sure; I'll think more about it
afk&
colomon my basic theory on the numeric tests is to decrease the number of values tested while increasing the number of combinations.... 17:10
afk & mother's day dinner
diakopter to clarify/revise my statement, instead of "inflate the numbers", I think more importantly they drown/crowd out the visibility of failing tests that are perhaps much more essential (such as closures, e.g.) 17:15
diakopter shuts up and goes back to Visual Studio to finish implementing class declarations in perlesque 17:16
17:17 snarkyboojum left
patrickas learned that pugs is even awesomer that he though it was! 17:18
audrey++ i guess
17:20 cxreg joined 17:21 am0c joined
slavik pugs is awesome ... but I wonder how close it is to rakudo in terms of features 17:28
is there a perl5 implementation on parrot? 17:30
17:31 M_o_C joined
diakopter slavik: no one is working on one, afaik. 17:36
slavik I have rakudo on my system 17:38
but Data::Dumper cannot be found ...
slava@dogbert:~/src/rakudo/rakudo-2010.03$ ./perl6 -e 'use Data::Dumper; my @a = 1..5; say Dumper @a;' 17:39
Unable to find module 'Data::Dumper'.
current instr.: 'perl6;Perl6;Module;Loader;need' pc 33942 (src/gen/role_pm.pir:0)
slava@dogbert:~/src/rakudo/rakudo-2010.03$ echo $PERL6LIB
.::/home/slava/src/rakudo/rakudo-2010.03/parrot/runtime/parrot/library/
slava@dogbert:~/src/rakudo/rakudo-2010.03$
diakopter oh; you'd need sorear's edition for that, I think.
slavik :(
where is that?
17:39 envi^home left
diakopter maybe his github? 17:39
I don't know
I just know he got a lot of Perl 5 integration (not *implementation*) done for parrot/rakudo 17:40
slavik but the Data/Dumper.pir is there ... 17:41
diakopter ohhh
I have no idea.
I don't use rakudo
slavik diakopter: what do you use?
perlesque?
diakopter I don't *use* anything; I'm developing perlesque
slavik :( 17:42
diakopter but I do weird-stress rakudo here on #perl6 via p6eval fairly often
slavik k
diakopter and std. 17:47
pmurias diakopter: how is the implementation of classes going? ;) 17:48
diakopter got them parsing 17:49
working on vivifying/emitting
pmurias vivifying?
diakopter yeah, the types have to become live during the parse 17:50
as RunSharp.TypeGen that is 17:51
which derives System.Type
pmurias: oh yeah; by "today" I meant "my today" which means by 12 hours from now :D 17:52
pmurias ah
diakopter so, your tomorrow :)
17:56 plobsing joined 17:58 kaare joined, kaare is now known as Guest86603 18:01 Guest86603 left
diakopter oh, and, I get lexical & anonymous classes for free. 18:01
bonus.
well, lexical in the sense its initialization block (and contained methods) close over all their outer lexical scopes, but non anonymous ones will still be globally available after declaration. a minor nit. 18:03
"minor nit" is repetitively redundant 18:04
pmurias class Foo {my $bar=4;method foo {return $bar}}? 18:05
diakopter yeah, and my $baz = 'hi'; class Foo {my $bar=4;method foo {return $bar ~ $baz}} 18:06
so if an anonymous class is declared in a routine/block that runs more than once, each time it runs, it'll create a new anonymous class with its own "private static" variables such as $bar , above 18:08
behavior of a named class whose initializer runs more than is undefined for now...
more than *once
that'll allow for parametric class generation, methinks 18:09
18:11 iblechbot left
diakopter rakudo: class Foo {my $bar=4;method foo {return $bar}; method baz {$bar=7}}; my Foo $f.=new; say $f.foo; $f.baz; say $f.foo 18:12
p6eval rakudo 27e05a: OUTPUT«4␤7␤»
18:12 gurjeet left
diakopter rakudo: class Foo {my $bar=4;method foo {return $bar}; method baz {$bar=7}}; my Foo $f.=new; say $f.foo; $f.baz; say $f.foo; $f.=new; say $f.foo; my Foo $g.=new; say $g.foo 18:13
p6eval rakudo 27e05a: OUTPUT«4␤7␤7␤7␤»
diakopter yeah, that's the behavior of rakudo
alpha: class Foo {my $bar=4;method foo {return $bar}; method baz {$bar=7}}; my Foo $f.=new; say $f.foo; $f.baz; say $f.foo; $f.=new; say $f.foo; my Foo $g.=new; say $g.foo 18:14
p6eval alpha 30e0ed: OUTPUT«4␤7␤7␤7␤»
diakopter <whew>
locals in the class block are private static
pugs: class Foo {my $bar=4;method foo {return $bar}; method baz {$bar=7}}; my Foo $f.=new; say $f.foo; $f.baz; say $f.foo; $f.=new; say $f.foo; my Foo $g.=new; say $g.foo 18:15
p6eval pugs: OUTPUT«␤7␤7␤7␤»
diakopter weird
$bar was never set to 4
pugs: class Foo {my $bar=4; say($bar); method foo {return $bar}; method baz {$bar=7}}; my Foo $f.=new; say $f.foo; $f.baz; say $f.foo; $f.=new; say $f.foo; my Foo $g.=new; say $g.foo
p6eval pugs: OUTPUT«4␤␤7␤7␤7␤»
diakopter weird. 18:16
diakopter will drop the issue
mimicing rakudo is plenty at this point 18:17
18:18 ^aristotle joined
^aristotle alpha: my @a = <<a b c>>; my %h; %h{@a} 18:19
p6eval alpha 30e0ed: ( no output )
^aristotle alpha: my @a = <<a b c>>; my %h; %h{@a}.join(" ")
p6eval alpha 30e0ed: OUTPUT«Use of uninitialized value␤Use of uninitialized value␤Use of uninitialized value␤»
^aristotle alpha: my @a = <<a b c>>; my %h; %h{@a}.join(" ").say
p6eval alpha 30e0ed: OUTPUT«Use of uninitialized value␤Use of uninitialized value␤Use of uninitialized value␤ ␤»
^aristotle alpha: my @a = <<a b c>>; my %h; [=] (%h{@a},1); %h{@a}.join(" ") 18:20
p6eval alpha 30e0ed: OUTPUT«Syntax error at line 10, near "[=] (%h{@a"␤in Main (file <unknown>, line <unknown>)␤»
^aristotle alpha: my @a = <<a b c>>; my %h; [=] %h{@a}; %h{@a}.join(" ")
p6eval alpha 30e0ed: OUTPUT«Syntax error at line 10, near "[=] %h{@a}"␤in Main (file <unknown>, line <unknown>)␤»
18:20 M_o_C left 18:24 ^aristotle left
lue hello! 18:29
diakopter lue: hi
18:29 snarkyboojum joined
diakopter perlesque: class Foo { method bar (--> str) { return 'hiyo' } }; say(4); 18:30
p6eval perlesque: OUTPUT«4␤»
diakopter heh. it doesn't actually do anything yet :)
just recognizes the syntax
dalek meta: r259 | diakopter++ | trunk/Sprixel/ (4 files):
[perlesque] parsing class declarations & their methods; parameterized the sub
 18:32
diakopter perlesque: class Foo { method bar (--> str) { return 'hiyo' }; say('declared Foo') }; say(4); 18:34
p6eval perlesque: OUTPUT«declared Foo␤4␤»
dalek meta: r260 | diakopter++ | trunk/Sprixel/src/Perlesque/Declarations.cs:
[perlesque] make class declaration block a lexical scope
 18:36
lue I still wish there were some sort of resource where you could learn about rakudo's guts, the bits with nqp and pir. 18:38
diakopter pmurias: see that last diff if you're curious about how to make a block a lexical scope. it's a two-line change
lue (and I don't wish to keep badgering people; learning everything that way takes forever :D)
arnsholt lue: I think the best way ATM to learn NQP and PIR is simply to do something with them 18:41
The PIR docs are decent (the PDDs are quite useful), and NQP is pretty much a thin layer on top of PIR
lue just some fun programming exercises? 18:42
arnsholt And if there's anything you need to know, src/NQP/Grammar.pm in nqp-rx is pretty comprehensible
Whatever is the best way you learn programming languages pretty much
lue I delved into PIR a while ago, not that hard. What I wish existed were Rakudo's "API" or equivalent.
Ah well, time to find my pith helmet :)
arnsholt I've got my Prolog on Parrot project, which I've learnt quite a bit from already
lue I can't find src/NQP/Grammar.pm :/ 18:46
diakopter in the rakudo sources 18:47
er
snarkyboojum parrot src somewhere probably?
diakopter github.com/perl6/nqp-rx/blob/master...Grammar.pm
lue not in [rakudo]/src or [rakudo]/parrot/src (can't find the NQP folder)
colomon NQP is in Parrot, not Rakudo
diakopter lue: it's in nqp-rx sources 18:48
moritz_ only a compiled version of nqp-rx is shipped with parrot
diakopter see the link above...
lue ok. I ought to download it then :)
snarkyboojum ah ok :)
lue (nqp-rx)
diakopter anyone here know about parrot's inheriting-hashes? 18:51
18:51 M_o_C joined
patrickas moritz_: are you still updating the proto web manually or is it ina cron already ? 18:51
moritz_ patrickas: cron job
patrickas and does the cron automatically pulls the latest update script from github ? 18:52
moritz_ it should, yes
patrickas ok ... I guess I have to be a bit more careful when pushing :-) 18:53
moritz_ btw the "has tests" detection is broken
json has tests
patrickas that's masak's work :-) 18:54
sorear slavik: You need to write use Data::Dumper:from<perl5>;
slavik: different languages have disjoint module namespaces
moritz_ in fact it thinks that no project has tests at all
patrickas or lack of ... actually I don't see anything in the code related to it... I guess he just added the pngs and relied on me to add the code. I'll do that next...
snarkyboojum footer on proto.perl6.org looks a bit funny left aligned and outside the 700px column 18:55
sorear slavik: there are two projects to support "perl5" as a Parrot language. PONIE, a port of the P5 compiler to generate Parrot code (is dead, needs >5 man-years to revive), and blizkost (autothunks Parrot calls into libperl.so, mostly works) 18:56
slavik sorear: blizkost makes use of parrot can call arbitrary C libraries thingy? 18:57
snarkyboojum perhaps put it in a footer div which is styled similarly to head (in terms of width margin etc?) 18:58
moritz_ snarkyboojum: code is in the proto repo in web/index.tmpl
snarkyboojum moritz_: aye :)
slavik jnthn: blizkost ... близкость? 18:59
or sorear
lue should peek in docs/ once in a while *cough* 19:07
19:09 dual joined
sorear slavik: no. blizkost is a Parrot-API extension; it does not use the FFI 19:13
slavik what does FFI stand for?
sorear foreign function interface
aka "parrot can call arbitrary C libraries thingy"
slavik also, does blizkost come from the slavic word to mean nearby?
I see 19:14
moritz_ yes
slavik :)
finally, my russian is worth some random trivia
:P
sorear jnthn speaks Slovak 19:15
I don't
slavik :( 19:17
moritz_ I think it says as much in the README
slavik moritz_: bah at you and reading READMEs :P 19:18
sorear in any event, you don't need Data::Dumper in Perl6 19:19
the functionality is built in
slavik .WHAT?
or .perl?
sorear rakudo: say %*ENV.perl
p6eval rakudo 27e05a: OUTPUT«Method 'key' not found for invocant of class 'String'␤current instr.: 'perl6;EnumMap;_block21197' pc 504340 (src/gen/core.pir:76818)␤»
sorear rakudo: say Cool.^methods.perl
moritz_ :-)
p6eval rakudo 27e05a: OUTPUT«[{ ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ... }, { ...
..}, { …
moritz_ well, it doesn't handle routines yes
rakudo: say [{a => 2}, 4].perl
p6eval rakudo 27e05a: OUTPUT«[{"a" => 2}, 4]␤»
sorear neither does Data::Dumper, really
sorear ponders writing a rakudo decompiler 19:20
moritz_ just store a pointer to the original source code
slavik since all code get's compiled to pir/pbc ... what would happen if you compile language1 on parrot and use a decompiler for language2 on the bytecode? 19:21
would be interesting methinks 19:22
instance language translation
19:22 jonrafkind joined
sorear language translation at the semantic level is a lot easier than you might think 19:22
slavik you mean source to source?
sorear I mean source to source-that-runs-the-same 19:23
most of the complexity in a program like p2c is involved in preserving formatting and comments
slavik ahh 19:24
sorear: formatting and comments aside, compiling wins the optimizations from the compiler, so you can compile/decompile to see how to better write code or something 19:25
19:27 M_o_C left
arnsholt slavik: I think applying a decompiler to a different language would result in weird stuff 19:31
In the best case just a lot of inline PIR
lue erm,.... YOU_ARE_HERE is seemingly useless. 19:32
arnsholt From my understanding, it isn't 19:33
It's where user code is inserted to be run
sorear YOU_ARE_HERE is very black magic
it's what makes the setting work
in topic/lexical-persistence, it's also what makes the REPL wor 19:34
k
lue ok. 19:36
19:37 ash___ left
lue YOU_ARE_HERE.pm containing the line YOU_ARE_HERE; must be deep magic indeed :) 19:37
sorear if you get rid of that line, the stage 2 compiler will be non-functional
resuming lexical scopes across compiler runs is very tricky 19:38
arnsholt I think jnthn claimed responsibility for YOU_ARE_HERE
lue rakudo: YOU_ARE_HERE; # just curious 19:39
p6eval rakudo 27e05a: ( no output )
diakopter rakudo: say YOU_ARE_HERE;
p6eval rakudo 27e05a: OUTPUT«Null PMC access in type()␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
diakopter rakudo: say YOU_ARE_HERE.WHAT;
p6eval rakudo 27e05a: OUTPUT«Null PMC access in find_method('WHAT')␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
sorear rakudo: say 2+2; YOU_ARE_HERE;
p6eval rakudo 27e05a: OUTPUT«4␤»
sorear rakudo: YOU_ARE_HERE; say 2+2; 19:40
p6eval rakudo 27e05a: OUTPUT«4␤»
sorear rakudo: say $x; my $x; YOU_ARE_HERE;
p6eval rakudo 27e05a: OUTPUT«Any()␤»
sorear see? magic.
slavik is @*IN all the input lines?
sorear I've tied the lexical scope in a knot
diakopter rakudo: say .lines 19:41
p6eval rakudo 27e05a: OUTPUT«Method 'lines' not found for invocant of class ''␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
diakopter rakudo: say lines
p6eval rakudo 27e05a: OUTPUT«No applicable candidates found to dispatch to for 'lines'. Available candidates are:␤:(IO $filehandle, Any :bin($bin) = { ... }, Any :enc($enc) = { ... }, Any :nl($nl) = { ... }, Any :chomp($chomp) = { ... })␤:(Str $filename, Any $limit = { ... })␤␤current instr.: '_block14'
..pc 29 …
19:41 Patterner left
sorear it's considerably more awesome in lexical-persistence though ;) 19:41
lue rakudo: say @*IN
p6eval rakudo 27e05a: ( no output )
sorear attempts to get people to try it 19:44
19:46 kensanata joined 19:48 alester joined
lue wait, there's a fail command? 19:49
colomon lue: not only that, but for a while there it didn't work. ;) 19:51
mantovani guys can I ask something here ?
19:51 Psyche^ joined, Psyche^ is now known as Patterner
colomon mantovani: yes. 19:52
mantovani all right.
19:52 finanalyst left
mantovani I tryng solve some problems and when the number is very big I have "inf" 19:52
19:52 finanalyst joined
mantovani paste.scsys.co.uk/43329?tx=on&w...Format+it! 19:53
can I do something like big int 19:54
colomon mantovani: that's perl 5, there is a separate channel for that, #perl
but isn't it just
use BigInt;
?
in 5.10, I mean.
mantovani I can't
lue colomon: the irony :)
colomon I've not done that in a long time, but I believe it's something like that.
mantovani if I do the result is very diferent
I did, bigint and bitnum(bignum lock my pc) 19:55
ok I'll try #perl5
thanks
lue just #perl ought to work.
lue gets a warm feeling as he finds src/builtins/assign.pir . Hope still exists for := 19:56
sorear := cannot be implemented there 19:57
:= must be a parser special
colomon feels like he should know a shortcut for that equation, but nothing is coming to mind....
lue I know. My current goal is to find variable initilization.
sorear the implementation is trivial if TimToady would just get back to me on what it means
19:57 alester left
sorear colomon: that's the closed form of the fibonnaci sequence 19:58
colomon sorear++
sorear it can be reexpressed in a purely integer form using the LDU decomposition of the generating matrix
however, I forget the form 19:59
20:04 justatheory joined
lue how does trace work in Rakudo, if it even works yet? 20:09
20:13 justatheory left
sorear --trace is a PCT argument 20:14
it gets parsed and then fed into Parrot as trace flags
lue ah.
finanalyst rakudo: my regex an { <digit>+ }; my $x='qw=345 er=12'; $x ~~ m/<an>/ 20:20
p6eval rakudo 27e05a: OUTPUT«Method 'an' not found for invocant of class 'Regex;Cursor'␤current instr.: '_block70' pc 668 (EVAL_1:241)␤»
finanalyst alpha: my regex an { <digit>+ }; my $x='qw=345 er=12'; $x ~~ m/<an>/
p6eval alpha 30e0ed: OUTPUT«Unable to find regex 'an'␤in regex »
finanalyst std: my regex an { <digit>+ }; my $x='qw=345 er=12'; say ?($x ~~ m/<an>/) 20:21
p6eval std 30610: OUTPUT«ok 00:01 115m␤»
pmurias diakopter: do you think writing a "fake" replacement for the part of runsharp that Sprixel uses so that it generates C# would be hard? 20:22
finanalyst rakudo: my regex an { <digit>+ }; my $x='qw=345 er=12'; say ?($x ~~ m/<an>/)
p6eval rakudo 27e05a: OUTPUT«Method 'an' not found for invocant of class 'Regex;Cursor'␤current instr.: '_block70' pc 690 (EVAL_1:243)␤»
sorear finanalyst: it's spelled <&an> for "my" regices
diakopter pmurias: explain more?
pmurias gathers his words 20:23
finanalyst rakudo: my regex an { <digit>+ }; my $x='qw=345 er=12'; say ?($x ~~ m/<&an>/) 20:24
p6eval rakudo 27e05a: OUTPUT«Confused at line 11, near "say ?($x ~"␤current instr.: 'perl6;HLL;Grammar;panic' pc 501 (ext/nqp-rx/src/stage0/HLL-s0.pir:327)␤»
lue rakudo: my regex xyzzy { <digit>+ }; my $x='qw=345 er=12'; say $x ~~ m/<xyzzy>/ 20:25
p6eval rakudo 27e05a: OUTPUT«Method 'xyzzy' not found for invocant of class 'Regex;Cursor'␤current instr.: '_block70' pc 679 (EVAL_1:22420114)␤»
pmurias diakopter: how hard would it be to write a version of runsharp that produces C#/something human readable as it's output would be
diakopter very hard. it would be much easier to write a decompiler
sorear what does runsharp produce currently? 20:26
pmurias like c.Try(); being implemented like method Try {$.code .= 'try {'}
diakopter it builds a stream of CIL
but it is designed to emit to a stack-based assembly language
which C# is definitely not 20:27
pmurias: I do have the benefit of using Reflector on windows...
pmurias: do you have a windows installation anywhere? Reflector is free
20:28 alester joined
diakopter pmurias: otoh I could spend some time trying to get [some version of] Reflector to work in mono 20:28
(on linux)
20:30 molaf left
diakopter probably the best thing to do is spend some time trying to get Reflector, since it already has an excellent decompiler 20:30
runsharp is quite an achievement; there's no way perlesque could have progressed so quickly [even close] without it. 20:33
pmurias generating C# would have worked 20:35
diakopter no
well in the mono case, yes, now it would have, since mono has string-eval for C# 20:36
but that's quite recent
.NET doesn't have that
it needs runtime codegen and loading of those types/routines in the same environment
which runsharp provides 20:37
theoretically though, on .NET it could emit the C# and call out to the Microsoft csc.exe compiler, and then load the resulting .dll or .exe ... but that would be very tricky to get right 20:39
arnsholt Sounds a bit like implementing "eval" in C 20:40
diakopter right.
20:40 pmurias left 20:42 pmurias joined
diakopter pmurias: (see the backlog) 20:44
pmurias looking... 20:48
dalek meta: r261 | diakopter++ | trunk/Sprixel/src/ (3 files):
[perlesque] a bit of incremental progress on classes.
 20:50
diakopter sorear: the author of runsharp is a Stefan
moritz_ it's quite a common name $here :-) 20:55
20:56 finanalyst left
moritz_ rakudo: sub a(@a, @b) { say @a === @b }; my @r = 1, 2, 3; a(@r, @r) 20:57
p6eval rakudo 27e05a: OUTPUT«1␤»
moritz_ rakudo: sub a(@a, @b) { say @a === @b }; my @r = 1, 2, 3; a(@r, (1, 2, 3)) 20:58
p6eval rakudo 27e05a: OUTPUT«0␤»
moritz_ rakudo: say 1/2 === 1/2 20:59
p6eval rakudo 27e05a: OUTPUT«0␤»
moritz_ I guess that should be True, no?
if Rat is a value type
21:01 orafu left, orafu joined 21:02 alester left 21:04 vdo joined
moritz_ rakudo: say (1, 2, '3') eqv (1, 2, '3') 21:06
p6eval rakudo 27e05a: OUTPUT«1␤»
moritz_ rakudo: say (1, 2, '3') eqv (1, 2, 3)
p6eval rakudo 27e05a: OUTPUT«0␤»
21:07 alester joined 21:08 pmurias left, dolmen left 21:16 skangas left 21:24 dolmen joined
snarkyboojum moritz_: proto.per6.org has gone from 141 HTML validation errors to 0 :) 21:25
lue I'm going afk for a bit, and leaving you a blog post, IIUC: rdstar.wordpress.com/ 21:32
21:33 dolmen left 21:41 plobsing left 21:42 meppl left 21:50 alester left
diakopter rakudo: class Foo {...}; class Foo {...} 21:51
p6eval rakudo 27e05a: ( no output )
diakopter std: class Foo {...}; class Foo {...}
p6eval std 30610: OUTPUT«ok 00:01 110m␤»
diakopter "unnecessary re-pre-declaration ..." 21:56
:)
sorear shrug, it's useful in autogenerated code
diakopter looks like perlesque will need to require full predeclaration of classes 21:58
(including fields & method signatures)
since it's truly single-pass
21:58 bakedb__ left
diakopter so, I can just co-opt class predeclaration like so: 21:59
21:59 jjore left
diakopter class Foo {...; has str $str1; has List[PrototypeChain[string,P6object]] $chain; method (str $first, Foo $another_foo) {...} }; 22:01
or whatever the member syntax is
:)
(haven't gotten to those yet; I'm still on methods)
22:07 kensanata left
diakopter std: class Foo {...; has Str $str1; has Int $chain; method (Str $first, Foo $another_foo) {...} }; 22:07
p6eval std 30610: OUTPUT«ok 00:01 112m␤»
diakopter I'm not certain whether this will be a problem
(requiring full predeclaration) 22:08
22:11 lichtkind left
diakopter looks around for TimToady 22:12
22:14 jjore joined
diakopter obviously compilers targeting perlesque would need to be more-than-one-pass 22:16
snarkyboojum patrickas: just noticed, something about the styling of the dl and children is causing the height of the content div to be computed incorrectly 22:17
22:17 PacoLinux left
diakopter rakudo: class Foo {...; has Str $str1; has Int $chain; method (Str $first, Foo $another_foo) {...} }; say Foo.new 22:18
p6eval rakudo 27e05a: OUTPUT«too few positional arguments: 1 passed, 2 (or more) expected␤current instr.: 'perl6;Failure;new' pc 15527 (src/builtins/Seq.pir:77)␤»
diakopter interesting
rakudo: class Foo {...; has Str $str1; has Int $chain; method new (Str $first, Foo $another_foo) {...} }; say Foo.new
p6eval rakudo 27e05a: OUTPUT«too few positional arguments: 1 passed, 2 (or more) expected␤current instr.: 'perl6;Failure;new' pc 15527 (src/builtins/Seq.pir:77)␤»
diakopter rakudo: class Foo {...; has Str $str1; has Int $chain; method bar (Str $first, Foo $another_foo) {...} }; say Foo.new
p6eval rakudo 27e05a: OUTPUT«too few positional arguments: 1 passed, 2 (or more) expected␤current instr.: 'perl6;Failure;new' pc 15527 (src/builtins/Seq.pir:77)␤»
snarkyboojum patrickas: i.e. box model height is computed to be 260px roughly - makes putting divs with margins and padding after the content div problematic 22:19
22:20 jjore left
snarkyboojum patrickas: anyway - don't know how to "fix" it, or even if it makes sense to :) 22:20
patrickas strange! 22:23
snarkyboojum: are you trying to fix the footer ?
snarkyboojum patrickas: nope - was going to put a footer div in there below the content one, but I've noticed you've put the paragraph in the content div
patrickas: just mentioning the issue I see with the page 22:24
patrickas Oh ok ...well if it is not affecting anything specific I guess no need to spend much time on it :-) 22:25
snarkyboojum patrickas: it has to do with the dl, dt, dd CSS - if that's ripped out then the height for the content div is computed correctly - but don't know enough to fix it
patrickas: yeah - thought I'd mention it anyway
patrickas: site validates now too o/ 22:26
patrickas I saw that thanks :-)
snarkyboojum patrickas: the link on the tests badge is currently broken too btw 22:27
patrickas how are you calculating the heights ? is it a cross browser issue ?
I have no idea what the link is supposed to point to .... 22:28
snarkyboojum patrickas: just looking at the computed height in firefox and chrome 22:29
patrickas: using firebug/chrome dev tools etc
22:42 ethel left 22:44 hercynium joined
jnthn rakudo: class Foo { ...; has $.x; } 22:49
p6eval rakudo 27e05a: ( no output )
jnthn rakudo: class Foo { ...; has $.x; }; Foo.new
p6eval rakudo 27e05a: OUTPUT«too few positional arguments: 1 passed, 2 (or more) expected␤current instr.: 'perl6;Failure;new' pc 15527 (src/builtins/Seq.pir:77)␤»
jnthn Heh. I guess the ... gets take to mean the whole lot is a stub. D'oh. 22:50
diakopter I think that's what it's supposed to do...?
jnthn diakopter: And ignore the rest of the declaration silently?
That feels a tad wrong.
diakopter that's what I presumed from my reading of S12 22:51
jnthn We do need a better error for when you stub a class that never finishes getting defined.
Heh. Maybe I read S12 and implemented what it said then. ;-)
22:52 patrickas left
diakopter If the class body begins with a statement whose main operator is a single prefix:<...> (yada) listop, the class name is introduced without a definition, and a second declaration of that class in the same scope does not complain about redefinition. 22:52
jnthn oh, cool
Not a bug \o/
diakopter therefore, I can co-opt it to allow full predeclaration of attribute/method sigs 22:53
since it seems perlesque will need to require it.
which is fine for me... STD's translation of itself isn't going to be generating classes...
I mean. 22:54
STD won't need to generate perlesque classes to translate itself
jnthn oh noes...my flight is delayed to 2am 22:56
diakopter eww
jnthn I already did a 5 hour bus journey to get to an airport that's not in the ash cloud so I can fly home. :-) 22:57
22:58 plobsing joined 23:08 jjore joined 23:17 jjore left 23:18 circuitbreaker joined, mariano joined
diakopter jnthn: in your experience programming in general, if a routine is written as recursive, is it usually a static routine, or are recursive instance routines just as common? 23:19
sorear: u2
well, anyone
but hilite you
arnsholt A recursive instance is no problem 23:21
A tree-traversal method for example, should be an instance method on the tree
diakopter my question is whether to make instance methods also trampolined. since a "method" can't be returned as a closure in perlesque, I guess it doesn't make sense to do it.
arnsholt: well, my interviewers at a certain large software company would disagree with you... they seemed obsessed with rewriting recursive algorithms as iterative ones 23:22
arnsholt Well, some problems are best expressed iteratively
And others recursively
23:23 jjore joined
arnsholt Also, IME some people are a bit allergic to recursion 23:23
diakopter these were tree-traversal algorithms, mostly.
arnsholt In that case, was there a good rationale? 23:24
diakopter it was an interview question
arnsholt Ah, right
diakopter so there's no rationale
:)
arnsholt 'Cause IIRC you'd have to maintain an explicit stack to transform non-tail recursion into iteration 23:25
diakopter right
"don't use the language's callstack, maintain your own"
arnsholt In which case you'd probably be better of with the OS/VM stack I think
diakopter sure, but they're too limited for some applications. 23:26
arnsholt True, true
diakopter in any case, Perl 6 routines will be expressed as perlesque closures anyway, not methods, so it doesn't matter.. 23:27
jnthn diakopter: I think many of the ones I write don't use instance data. Actually I find C# shoehorns me into things being methods when actually I just want subrutines.
diakopter well yeah
jnthn So it's hard to say in that sense.
I often find myself having a public method that does some prep for the recursion and then calls some over thing that's private and recurses (so the fact it's implemented recursively isn't part of the interface). 23:28
diakopter I was worried about potentially overflowing the CLR stack in perlesque code, but it can't happen in the 5th stage (Perl 6) now that I've thought through it.
jnthn Not that that's especially relevant to the question though... 23:29
23:29 rgrau_ left
jnthn Probably fairest to say is that I do have some that rely on instance data, but most don't. 23:29
diakopter ... and if a perlesque method needs to get recursive, it can be rewritten as a recursive sub, so it'll be trampolined, if there's any danger of hitting stack limits. 23:30
ok. I'm satisfied that methods don't need to be trampolined, for my purposes. 23:31
they can always be in the future.
phenny: tell pmurias it's going to be another few days to finish classes/methods. They're somewhat more complex than I initially thought. But it's all worked out in my head now, so I'm quite certain I can finish them soon. 23:32
phenny diakopter: I'll pass that on when pmurias is around.
23:44 lestrrat is now known as lest_away
dalek meta: r262 | diakopter++ | trunk/Sprixel/ (4 files):
[perlesque] a bit more progress
 23:46
23:49 bluescreen joined 23:50 bluescreen is now known as Guest2257 23:53 mariano left, Guest2257 left 23:54 Guest2257 joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!