»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.
colomon ah, just activating the :radix[ ] form doesn't make it work. problem is in token circumfix:sym<[ ]>, I think... 00:00
lue is trying to fix the fractions of radcalc 00:02
mberends jnthn: suspected regex backtracking bug nopaste.snit.ch/20018 00:04
jnthn mberends: Agree. 00:05
mberends , having found a workaround, goes to sleep
00:06 mberends left
lue hopefully I fixed it, compiling... 00:06
TimToady std: .4 00:07
p6eval std 30146: OUTPUT«ok 00:01 107m␤»
lue std: :16<.4> 00:08
p6eval std 30146: OUTPUT«===SORRY!===␤Malformed radix number at /tmp/7vyfVF06ni line 1:␤------> :16⏏<.4>␤FAILED 00:01 105m␤»
colomon I thought it had to be 0.4?
TimToady metoo
lue just checking :)
TimToady probably parsed it as a method call :/
lue rakudo: say Int.4 00:09
p6eval rakudo 1cf076: OUTPUT«Confused at line 11, near "say Int.4"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
TimToady no, it parsed it as a number
std: 4. 00:10
p6eval std 30146: OUTPUT«===SORRY!===␤Decimal point must be followed by digit at /tmp/jU36jiBhBU line 1:␤------> 4.⏏<EOL>␤FAILED 00:01 105m␤»
TimToady that's what we disallow
lue but python allows it! (oh...)
TimToady so is 4.i (4.)i or (4).i ? 00:12
p6 will always take it as a method call 00:13
sorear is 1.618 a method call? 00:14
should it be? 00:15
TimToady why would it be?
sorear .
TimToady don't be a fundamentalist
colomon TimToady++ # p6l
snarkyboojum enjoyable post from TimToady on the perl 6 mailing list 00:16
colomon: beat me to it :)
colomon snarkyboojum: because I used fewer letters. ;)
lue The tension is unbearable (compiling)
00:16 cls_bsd joined
jnthn TimToady: Wow! 00:18
TimToady: You...kinda said what I feel but thought I should keep to myself. :-) 00:19
TimToady well, someone has to exercise a little moral authority around here occasionally, and I have about as little as anyone...
lue >say :16<16D.4>
1.42578125
.... something went horribly wrong :) 00:20
TimToady :)
colomon but you still got a fractional answer, lue++
it's progress.
TimToady and it's even the right sign
lue I know! I think I broke the first guy's code though :)
jnthn It's OK, I temporarily broke parsing 1 + 2 earlier tonight 00:21
:-)
colomon And I keep on breaking !=
lue I wrote it so that the computation of the fraction is separate, and then I divide it appropriately and add.
jnthn > say (('be','pi') >>~<< ('er','vo')).perl 00:22
["beer", "pivo"]
TimToady looks like you took it as 1.6D4
jnthn > say (('be','pi') >>R~<< ('er','vo')).perl 00:23
["erbe", "vopi"]
lue >say :16<16D>
TimToady yes, almost exactly that
lue 1
jnthn
.oO( mmm....vopi! )
00:23 meppl left
jnthn make spectests and hopes he didn't cause collateral damage. 00:24
colomon jnthn++
lue gah! Without the fraction, it takes the lefmost digit of the number!
cognominal in Perl5, is there a low precedence operator of function application like the $ in haskell?
TimToady actually, 1.6D.4 is what it looks like :)
arnsholt Is the official #parrot on Freenode?
jnthn arnsholt: no, it's on magnet
arnsholt Oh, well. What's one more ircnet? =) 00:25
lue I'm going to kill the original code, seems like :D
(emacs pun)
jnthn lue: If the new code passes more tests, go ahead! :-) 00:26
lue rakudo: Q:PIR{$S0="hello"␤$S1=$S0[5]␤say $S1} 00:28
p6eval rakudo 1cf076: OUTPUT«The opcode 'set_s_s_kic' (set<3>) was not found. Check the type and number of the arguments␤current instr.: 'perl6;PCT;HLLCompiler;evalpmc' pc 972 (compilers/pct/src/PCT/HLLCompiler.pir:538)␤»
cognominal s/Perl5/Perl6/
lue Ah! Did I do something wrong in testing a snippet of PIR? 00:29
(belated) hi cognominal o/
colomon cognominal: how does it work?
snarkyboojum lue: I've been having fun with very similar errors lately :) 00:31
lue rakudo: Q:PIR{$S0="hello"␤$S1=substr $S0 3␤say $S1}
p6eval rakudo 1cf076: OUTPUT«The opcode 'substr_s_s' (substr<2>) was not found. Check the type and number of the arguments␤current instr.: 'perl6;PCT;HLLCompiler;evalpmc' pc 972 (compilers/pct/src/PCT/HLLCompiler.pir:538)␤»
jnthn lue: comma
lue rakudo: Q:PIR{$S0="hello"␤$S1=substr $S0, 3␤say $S1}
p6eval rakudo 1cf076: OUTPUT«lo␤» 00:32
lue rakudo: Q:PIR{$S0="hello"␤$S1=substr $S0, 3, 3␤say $S1}
p6eval rakudo 1cf076: OUTPUT«lo␤»
lue rakudo: Q:PIR{$S0="hello"␤$S1=substr $S0, 3, 1␤say $S1}
p6eval rakudo 1cf076: OUTPUT«l␤»
jnthn \o/
lue substr(STRING, BEGIN, LENGTH) then
jnthn Right.
lue Is it alright if I break your hard work on the :[number] adverbs and have you send just one number, instead of an intpart and fracpart? 00:35
:)
jnthn colomon: BTW, if we've got reduce...and hypers...and Z/X/R/S...does that mean we're now parsing all of the hyperops? :-) 00:36
colomon I think so. We don't actually do anything with S yet other than parse it. 00:37
do you have the non-infix hypers parsing?
lue (Goodbye, old &radcalc code. It'd take too much to modify using one number.)
jnthn colomon: oh, no... 00:38
colomon: I think they may be easier though
colomon: Ah well, gotta save something for tomorrow :-)
colomon :)
jnthn colomon: It occurs to me that those probably desugar to map?
Well
Maybe
As a first cut :-)
colomon well, if they're not hyper. 00:39
jnthn I guess they ain't meant to be lazy.
colomon but that's a good first cut.
jnthn Well
@a>>++
Maybe not quite that simple
colomon oooo, right.
jnthn But maybe
:-)
A first cut can certainly be a thin wrapper around map, anyway. 00:40
lue in PIR, a num is a floating-point, right?
jnthn yees
colomon: Oh, we have that infixy form of reductions to do too I guess. 00:41
colomon errr... so how do I code infix_prefix_meta_operator:sym<!> in such a way that != is not one, but !== and !=== are?
oh, right, I'd already forgotten that one's existence. :) 00:42
jnthn erm
colomon <!before '!=' something?
jnthn <!before '!=' [-<=>]> maybe 00:43
char class of anything that's not =
colomon sounds plausible....
jnthn rakudo: say '$a != $b' ~~ /'!=' [-<=>]/
p6eval rakudo 1cf076: OUTPUT«Confused at line 11, near "say '$a !="␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
lue rakudo: Q:PIR{$N0=3␤say $N0} 00:44
p6eval rakudo 1cf076: OUTPUT«3␤»
jnthn rakudo: say '$a != $b' ~~ /'!=' <-[=]>/
p6eval rakudo 1cf076: OUTPUT«!= ␤»
jnthn rakudo: say '$a !== $b' ~~ /'!=' <-[=]>/
p6eval rakudo 1cf076: OUTPUT«␤»
jnthn colomon: That one.
<!before '!=' <-[=]>>
cognominal colomon, in a haskell sin $ 1 + 5 is equivalent to sin(1+5)
jnthn cognominal: In Perl 6, sin 1 + 5 is also equivalent to sin(1+5) ;-) 00:45
...wait, isn't it in Perl 5 too? :-)
cognominal it works on the left side too
colomon regex assertion not terminated by angle bracket at line 1285, near "[-<=>]> <s" 00:46
jnthn colomon: yes, I got it wrong
See update
<!before '!=' <-[=]>>
00:46 arthur-_ left
colomon ah, I see. :) 00:46
jnthn Sorry
:-)
Not sure what I was thinking. Probably wasn't.
colomon no worries.
cognominal f + g $ 1 + 5 is (f+g) (1+5) # supposing we can add function
colomon is that f composed with g, or the value of f + the value of g? 00:47
cognominal well that means f( 1+5) + g(1+5) according to the addition of functions 00:48
jnthn Oh, it distributes the args amonst the functions? 00:49
sorear jnthn: Is there anything I need to know before killing return_value_helper?
jnthn sorear: I...don't even recall what it does. :-) 00:50
colomon I think the quick answer is you could pretty easily set up Perl 6 to do that, but it doesn't have an equivalent by default.
cognominal no, that the definition of function addition
jnthn sorear: Oh, yes, kill it.
sorear: Before a Parrot Calling Conventions interface, it was a world of pain to not use a hack like that. 00:51
sorear: Now it's not.
er, interface => interface change
colomon rakudo: multi sub infix:<+>(&a, &b) { &a(!@_) + &b(!@_) };
p6eval rakudo 1cf076: OUTPUT«Placeholder variable cannot override existing signature at line 11, near ";"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
cognominal $ is just an operator that the low precedent application of a function to its argument 00:52
*precedence
colomon rakudo: multi sub infix:<+>(&a, &b) { sub { &a(!@_) + &b(!@_) }}
p6eval rakudo 1cf076: OUTPUT«Can not put multi on anonymous routine at line 11, near "}"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
jnthn ouch!
colomon std: multi sub infix:<+>(&a, &b) { sub { &a(!@_) + &b(!@_) }} 00:53
p6eval std 30146: OUTPUT«ok 00:01 115m␤»
jnthn I guess the $*MULTINESS or something is leaking too far 00:54
colomon I'm sure jnthn++ can come up with something better anyway... ;)
Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 70
ok( ?( (1 & 2 | 3) !=3), '& binds tighter than |');
is the line in question.
sorear jnthn: Can you breifly explain what life was like before PCC, and how Blizkost worked *at all* then? 00:55
jnthn probably sub infix:<+>(&a, &b) { sub (|$c) { &a(|$c) + &b(|$c) } } or some such 00:56
sorear: There was PCC
00:56 justatheory left
jnthn sorear: There just wasn't a neat interface to it 00:56
sorear I... don't really understand PCC
jnthn sorear: So it was messing around with structures rather than just some vtable calls.
sorear just a couple magic black boxes :(
jnthn Well, you *can* treat it as a magic box now, for the most part. Which is a relief. :-)
Essentially it's the thing that deals with moving args from caller to callee, and returns in the other direction. 00:57
And those things being passed living in a CallContext data structure (I think that's what it's called now...)
s/data structure/PMC/ 00:58
00:59 clintongormley left 01:00 justatheory joined
dalek kudo: 5edcafa | jonathan++ | src/ (3 files):
Get parsing of hyperops, including transparency of precedence, implemented. Stub in the various &hyper multis. Implement the very simplest case.
 01:02
snarkyboojum rakudo: say <1 2 3>.exists(1) 01:05
p6eval rakudo 1cf076: OUTPUT«Method 'exists' not found for invocant of class 'Parcel'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
lue rakudo: Q:PIR{$S0=substr "Oh", 2, 1␤say $S0} 01:06
p6eval rakudo 1cf076: OUTPUT«␤»
snarkyboojum rakudo: my @a = <1 2 3>; say @a.exists(1)
p6eval rakudo 1cf076: OUTPUT«Method 'exists' not found for invocant of class 'Array'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
lue rakudo: Q:PIR{$S0=substr "Oh", 2, 1␤say $S0=undef}
p6eval rakudo 1cf076: OUTPUT«error:imcc:syntax error, unexpected '=', expecting '\n' ('=')␤ in file 'EVAL_1' line 65␤␤»
lue rakudo: Q:PIR{$S0=substr "Oh", 2, 1␤say $S0==undef}
p6eval rakudo 1cf076: OUTPUT«error:imcc:syntax error, unexpected RELOP_EQ, expecting '\n' ('==')␤ in file 'EVAL_1' line 65␤␤»
jnthn potešit
oops, wrong window
lue rakudo: Q:PIR{$S0=substr "Oh", 2, 1␤if $S0==undef goto fail␤say "yay"␤fail:␤say"boo undef"} 01:07
p6eval rakudo 1cf076: OUTPUT«error:imcc:syntax error, unexpected IDENTIFIER ('undef')␤ in file 'EVAL_1' line 65␤yay␤boo undef␤»
lue rakudo: Q:PIR{$S0=substr "Oh", 2, 1␤if $S0=="" goto fail␤say "yay"␤fail:␤say"boo undef"}
p6eval rakudo 1cf076: OUTPUT«boo undef␤»
lue Now I can avoid keeping track of the length of the string! \o/ 01:08
01:10 astrojp left
TimToady there are no infixy reductions 01:11
jnthn TimToady: Whatcha call the 1, [+] 2,3,4 thingummy? 01:12
er 01:13
01:13 snarkyboojum left
jnthn 1 [+] 2,3,4 01:13
?
TimToady exactly the same as (1 + 2), 3, 4
jnthn Oh.
So it's got nothing to do with reduction ops?
TimToady and 1 [[[[[+]]]]] 2,3,4
jnthn ...phew. :-)
TimToady only an infix in prefix position is a reduction 01:14
jnthn Aye
TimToady and you have to use [] there to disambiguate
jnthn OK, good.
Something I'd read earlier today had led me to think otherwise.
lue why does 1 [+] 2,3,4 not equal 1 + 2 + 3 + 4 ?
jnthn OK, excellent, so we parse all the hyper ops. :-)
erm
meta-ops
:-)
TimToady because it doesn't
jnthn Apart from the few hypers. :-)
01:14 silug joined
jnthn (unary ones) 01:15
colomon To be precise there was misinformation on #perl6 about 1 [+] 2, 3, 4; :)
jnthn colomon: Heh, it wasn't just my brain being mushy then. :-)
colomon but can hardly cast blame, as I fear I probably say something wrong on the channel at least once a day. ;)
TimToady S03:4611
01:16 meppl joined, snarkyboojum joined
TimToady and S03:4196 (emphasis on "in term position") 01:17
any other place, [] is only for disambiguation
jnthn Heh 01:18
It'll be good for obfuscations.
:-)
TimToady @a Z[+=] @b vs @a [Z+]= @b
jnthn Oh. 01:19
TimToady or de-obfuscations
mind you, those might end up doing the same thing
just differently :)
but then there * * * which can be written * [*] * 01:20
in STD, != is distinguished from the other forms by being longer than ! 01:21
so it's simply LTM
colomon if that only worked for me.... :( 01:22
TimToady eventually all the contortions you guys are going through to avoid LTM is going to add up to more than implementing real LTM :)
m6locks how does blessing work with a class that has another class as an attribute? i'm trying to write a constructor for it but return self.bless(self.CREATE(), :attributeClass) complains about type objects being abstract 01:23
jnthn std: pi**time 01:24
p6eval std 30146: OUTPUT«ok 00:01 107m␤»
lue is :16<3.DEAD_BEEF> legal ? (the underscore in the fractional part)
jnthn ...Perl 6 supports censorship of some bad things, at least. :-)
TimToady lue: yes 01:25
colomon lue: yes
jnthn m6locks: That doesn't look immediately wrong
TimToady $<fracpart> = [ '.' <[ 0..9 a..z A..Z ]>+ [ _ <[ 0..9 a..z A..Z ]>+ ]* ]?
jnthn m6locks: What do you mean as "another class as an attribute"?
m6locks: There is a bug in master in that area though.
namely...
rakudo: class Foo { has Str $.a }; say Foo.new.a.WHAT # should be Str() 01:26
p6eval rakudo 5edcaf: OUTPUT«Any()␤»
jnthn You may be hitting that one.
I'll try and fix it soon.
m6locks oh, could be
yes, that was my point
TimToady if you folx are abandoning the intpart/fracpart, it would be easier to parse radix fractions like dec_number does 01:27
and just pass the whole coeff
and then we can allow :16<.D> without requiring the 0 01:28
01:29 nihiliad left
lue Just one string is what I'm programming, with detection of the "." switching over to fractions. 01:29
01:32 nihiliad joined
colomon jnthn: we can say infixish(1) in Grammar.pm? 01:34
TimToady which presumably tells the infix it's in a meta 01:35
okay, testing STD patch that just passes the coeff with a . in the middle 01:36
jnthn colomon: Not at the moemnt 01:37
colomon: I'm not sure what that's needed for.
TimToady: Yes. Is it "just" for error reporting?
colomon see what TimToady just said. But you used it, yes?
jnthn I couldnt' see anywhere it affects teh parse.
No
I just called <infixish>
colomon {} <infixish(1)>
jnthn Oh? 01:38
colomon in token infix_circumfix_meta_operator:sym<« »>
jnthn oh, damm
copy-pasto
(from STD)
It probably doesn't work.
I only checked the non-unicode ones. :-)
colomon: You can copy the stuff from STD probably though to get $*IN_META and add the parameter to infixish
We support those.
colomon I'll keep that in mind. Today I'm just trying to puzzle out how to copy your precedence copying trick. :) 01:39
01:39 TiMBuS left
TimToady it's mostly just for suppressing warnings on [..] and such, which might be mistaken for an attempt to leave out the endpoints of a range 01:39
since the mathematicians seem inclined to write [1..] 01:40
jnthn Ah, OK. :-)
colomon: Just duplicating <O=.Ocopy('infixish')> should do it
colomon yes, I thought so. :) 01:41
except I need to reorder a bit to make it work.
:)
TimToady it also suppresses Unexpected block in infix position
jnthn TimToady: OK, thanks. :-)
We can add it later on, then.
TimToady though I don't know that we currently have any operators beginning with { or ->, but a user might add them 01:42
STD is extensible, y'see...
rakudo: 42 -> { ... } 01:43
p6eval rakudo 5edcaf: OUTPUT«Confused at line 11, near "42 -> { .."␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
TimToady std: 42 -> { ... }
p6eval std 30146: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/fVGRNQesOl line 1:␤------> 42 ⏏-> { ... }␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤FAILED 00:01 109m␤»
lue PIR ifs need ==, not =, right? 01:44
jnthn rihgt.
lue sigh... now to go back :) 01:45
TimToady std: sub infix:<-\>> {...}; 42 [->] 43
jnthn We still need to steal all the two terms in a row good errors.
p6eval std 30146: OUTPUT«ok 00:01 108m␤»
jnthn rakudo: sub infix:<-\>> {...}; 42 -> 43
p6eval rakudo 5edcaf: OUTPUT«Confused at line 11, near "42 -> 43"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
jnthn aw 01:46
We don't do the bracketed form yet
lue now I'm compiling (fingers crossed XoX)
jnthn It's just Xo now
;-)
TimToady don't recognize that operator
actually, I finally removed that deprecation from STD
std: 1 X+X 2 01:47
p6eval std 30146: OUTPUT«Undeclared name:␤ 'X' used at line 1␤ok 00:01 106m␤»
lue It was my best attempt at a fingers crossed person (you can do better perhaps?) 01:48
colomon is really amused by Julie / Julia. 01:49
TimToady ₓ☹ₓ 01:51
lue That was a great movie!
colomon++
jnthn didn't see that one 01:52
lue draws a big breath as he types «./perl6 -e "say :16<16D.4>"» into the terminal (minus the french quotes of course)
oh yeah, I broke the adverbs :D
>say radcalc(16,'16D.4') 01:54
unrecognized/invalid character: . 01:55
current instr.: '&radcalc' pc 11961 (src/builtins/Parcel.pir:78)
called from Sub '_block14' pc 29 (EVAL_1:0)
called from Sub '!UNIT_START' pc 1299 (src/glue/run.pir:23)
called from Sub 'perl6;PCT;HLLCompiler;eval' pc -1 ((unknown file):-1)
called from Sub 'perl6;PCT;HLLCompiler;command_line' pc 1489 (compilers/pct/src/PCT/HLLCompiler.pir:794)
called from Sub 'perl6;Perl6;Compiler;main' pc -1 ((unknown file):-1)
01:56 vamped joined
colomon lue: Having just watched a bit of real Julia Child footage, I'm stunned at how well Meryl Streep has her down. 01:56
lue colomon: Look up. What could be causing that? (the string var is named 'number') 01:57
01:57 justatheory left
colomon hmmmm 01:57
01:58 Sarten-X left
lue should I paste my &radcalc code so far? 01:58
colomon yeah. 01:59
isn't that message coming from radcalc
lue oh, that's an error sent by &radcalc itself :P 02:00
(I should put DON'T PANIC in front of it...)
colomon :)
02:00 snarkyboojum left
m6locks rakudo: my $a, $b = 1; 02:00
p6eval rakudo 5edcaf: OUTPUT«Symbol '$b' not predeclared in <anonymous>␤current instr.: 'perl6;PCT;HLLCompiler;panic' pc 137 (compilers/pct/src/PCT/HLLCompiler.pir:101)␤»
lue I know! I forget to increment my loop variable before jumping to fractions! (when you see the code, you'll know what I mean) 02:01
and so it comes across the . twice!
TimToady std: /\ /
p6eval std 30146: OUTPUT«===SORRY!===␤No unspace allowed in regex (for literal # please quote with single quotes, '#') at /tmp/wQe2LopX01 line 1:␤------> /\⏏ /␤FAILED 00:01 107m␤»
TimToady still slightly LTA...
lue /\/ever give up, TimToady! :) 02:02
colomon has completely broken parsing of the Rop, etc.
TimToady told you it looked kinda like 1.6D.4 :)
colomon oh, maybe it's the action that is broken, not the parsing. 02:03
lue compiling ₓ☹ₓ 02:04
02:04 snarkyboojum joined
lue I just completely rewrote it TimToady (and IMO it's more efficient (less labels) and readable (comments!)) 02:05
Juerd moritz_: /dev/pts is mounted so I don't know what could be wrong.
moritz_: Maybe someone (you?) fixed it?
jnthn Juerd: I think moritz_++ found a way to fix it. 02:06
02:06 M_o_C left
Juerd \o/ 02:06
colomon lue: but does it pass the tests? ;)
jnthn sleep() 02:07
night o/
colomon night!
TimToady o/
lue night-i-o o/
colomon: first it has to work :D 02:08
done compiling! and now...
02:08 Sarten-X joined
lue >say radcalc(16,'16D.4') 02:09
369
the fraction is just added to the integer *sob*
I'm a moron. I didn't add the result of the fraction, but the numerator. :) 02:10
02:13 gfldex joined
TimToady The only way to be a genius is to keep on being a moron until you stop. :) 02:14
lue Fixed it, finger crossing time... 02:15
colomon TimToady++ 02:16
lue OW! I sprained a tendon in my index finger :)
sorear type with your tongue 02:18
TimToady impossible to strain a muscle in your index finger; impossible to strain a tendon in your tongue. tradeoffs... 02:19
hard to cross your tongue though.
lue :D :D :D :D :) :) :) :) :) \o/ \o/ \o/ \o/ \o/ \o/ ^.^ ^.^ ^.^ 02:20
>say radcalc(16,'16D.4')
365.25
Juerd TimToady: Only if you're alone
sorear shifted chars are hard... use your nose for those
TimToady ooh, a significant number
lue yes, the number of days in a year :)
02:20 rcsheets joined
colomon approximately. ;) 02:20
TimToady give or take a gregorian
rcsheets There are 143 people in this channel. 02:21
sorear rcsheets: no, only about 139
Juerd rcsheets: That's not true :)
lue now to fix adverbs, pass spectests, get a commit bit, and commit
02:21 TimToady sets mode: +vv buubot dalek, TimToady sets mode: +vv ilogger2 IRSeekBot, TimToady sets mode: +vv p6eval phenny
Juerd teh botz 02:21
rcsheets ok so my definition of person is broken
lue (or nopaste a patch)
TimToady looks like there are six bots, unless you're one :)
snarkyboojum depends how good the bots are
rcsheets There are 143 IRC users in this channel.
02:21 mauke joined
colomon jnthn++ 02:21
sorear wrong again
144 02:22
colomon > say 3 R* 4 R- 20
8
> say 3 R- 4 R* 20
77
Juerd Irssi said there were 143 "nicks". That's pretty accurate.
lue ...where is the adverb defined? (oh dang it, forgot magnitude O.o)
rcsheets hmm, nicks
ok, fine
i'll agree with nicks
02:22 icke_ joined
colomon > say 1, 2 RZ~ 'a', 'b' 02:23
a1b2
pugssvn r30147 | lwall++ | [STD] pass single coeff to radcalc to make :16<.BABEFACE> easier
r30147 | more awesomization of the no unspace message
02:23 icke_ left
sorear rcsheets: anyways, what brings you here? 02:23
rcsheets necrocow. he's not here. 02:24
TimToady would you like a commmit bit?
sorear who, me?
and to what project?
rcsheets but he thinks he is here. so... i joined, to see for myself.
TimToady we hand out pugs repo commit bits like candy 02:25
TimToady refrains from making joke on "repo man"
lue rakudo commit bit would be nice, as opposed to nopasting patches :) 02:26
TimToady I don't have any of those
lue (I would love a commit byte :) )
TimToady but you can hack on the spec and tests, or even STD
or any of the other projects in the pugs repo
lue I has a pugs commit bit already, it's just rakudo... 02:27
(got to go add in magnitude...)
02:29 rcsheets` joined
lue rakudo: Q:PIR{if 3 == 3 say "hey"} 02:29
p6eval rakudo 5edcaf: OUTPUT«error:imcc:syntax error, unexpected PARROT_OP, expecting GOTO ('say')␤ in file 'EVAL_1' line 64␤»
lue so it has to be a goto...
02:30 necrocow joined
rcsheets` hi necrocow. 02:30
can you see this?
necrocow yes 02:31
02:31 mauke left
necrocow can you see this? 02:31
rcsheets` yup.
02:31 rcsheets` left
sorear wonders what this is about. 02:31
necrocow ok
looks liek giubson was split until a moment ago 02:32
is is true there will be a pure C version of perl6 (like perl5 and before are)
rcsheets i disagree, but anyway now you can ask #perl6 about perl 6.
lue rakudo: say :16<A*2**16>
p6eval rakudo 5edcaf: OUTPUT«655360␤»
02:32 rcsheets left
lue rakudo: say :16<A*2> 02:32
p6eval rakudo 5edcaf: OUTPUT«Malformed radix number at line 11, near "<A*2>"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
lue rakudo: say :16<A**2>
sorear necrocow: why?
p6eval rakudo 5edcaf: OUTPUT«Malformed radix number at line 11, near "<A**2>"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
sorear necrocow: this language is what you make of it 02:33
lue shouldn't A*2 default to A*2**1, and A**2 an ambiguity error?
necrocow sorear: curious
TimToady parrot is "pure C" in that sense
necrocow I've had really bad experiences with parrot 02:34
02:34 astrojp joined
necrocow TimToady: tha'ts not the same, if Perl 6 is weritten in parrot, it's written in parrot, not straight C 02:34
so I'm aksing about Perl6 not parrot
lue TimToady: for magnitude in base conversion, what if someone leaves out the base? The exponent?
colomon in the spec, both parts of *N**M are required
or at least, in the old spec.
both or neither. 02:35
TimToady Perl 5 isn't written in C either, in that case. it's written in the ppcodes that the P5 interpreter uses
lue EMISSINGARGUMENTS then?
TimToady both parts are required in STD too 02:37
std: :16<FACE*42>
p6eval std 30147: OUTPUT«===SORRY!===␤Malformed radix number at /tmp/qMxuFDe1MO line 1:␤------> :16<FACE*⏏42>␤ expecting radint␤FAILED 00:01 107m␤»
TimToady the error is kinda bogus though 02:38
lue I'm going to write in error catchers for &radcalc for that then.
TimToady in the expecting part
02:39 vamped left
TimToady you never get to radcalc in that case, since it can't find the > 02:39
if rakudo follows STD there...
colomon and it does at the moment, at least in this part. 02:40
necrocow TimToady: well the actualy 'perl' program was indeed written in C, I've compiled it myself numerous times. What the 'perl' executable runs is another matter 02:41
TimToady: you seem to be mixing things repeatedly
I just want to know if there will be a striaght 'perl6' written in C, like how '/usr/bin/perl5' is 02:42
lue Just in case you were to use radcalc() :)
necrocow: if someone gets around to it :)
colomon lue: good point.
lue (The adverbs are still broken BTW :) ) 02:43
necrocow lue: I understand that... it would defiantely help if perl6 first existed :p
s/defiantely/definately/
lue It does... as a spec. :D
snarkyboojum ./perl6 exists on my machine :) 02:44
necrocow synlinks don't count :po
or are you using raku or whatever it's called
snarkyboojum certainly not a symlink
necrocow rakudo, sorry
snarkyboojum it's an executable :)
necrocow but it's not a real perl6 though
TimToady necrocow: you insist on making distinctions that are relatively meaningless 02:45
necrocow huh
snarkyboojum for that matter I have a perl6.c on my machine too :D
lue It will be, once it passes the test suite :)
02:45 nihiliad left
necrocow TimToady: you seems to keep making distinctive things to be the same thing which is wrong 02:45
02:45 nihiliad joined
sorear necrocow: explain to me why it even matters 02:45
necrocow TimToady: parrot != perl6, perl6 is (to be?) written in parrot, so what ever parrot is written in is irrevent. Perl5 reads ppcodes, but it is NOT written IN ppcodes, it is written in C 02:46
those ARE important distinctions
sorear (also, TimToady is always right :p)
snarkyboojum the first line of which (perl6.c) is '#include "parrot/parrot.h"' :)
necrocow I tihnk not :p
so there is perl6 in c code then? 02:47
lue Probably the virtual machine bit upsets him. I'm not a fan of it either (can't wait for self-supporting Rakudo)
sorear Perl 5 is not written in C, it's written in an unholy custom macro language
colomon There is certainly C code in Rakudo.
snarkyboojum colomon: was just about to mention that, but once again beaten to it :) 02:48
colomon Perl 6 is a specification, not a particular implementation.
sorear I don't get what's so special about C
it's just a language
TimToady and it obviously confuses everyone who uses it
necrocow sorear: you're write, dispite all the C code, I should have been using javac to build perl5 all these years. How the hell could I have missed that 02:49
I can't believe I was always using gcc
sorear necrocow: "unholy custom macro language" != java
necrocow it's still C ffs
cognominal The important point is that Perl 6, once bootstrapped, will be mostly written in Perl 6 and the back end will be less relevant to the user of Perl 6. That cannot be said of Perl 5.
sorear necrocow: if you want a Perl 6 implementation written in Pig Latin, feel free to write one.
Implementation language is an implementation detail 02:50
necrocow I never saw the benefit of building a compiler with another instance of that compiler... that sometimes made it impossible ot build ghc fir instance
sorear also, you should *never* use gcc to build perl5
you should always use make
necrocow I've never been unable to build perl5 from sources 02:51
sorear perl6 is also built using make
cognominal Things like parsing are written using a manual lexer and yacc for Perl 5 while parsing in Perl 6 is done in Perl 6.
necrocow sorear: huh... I don't use gcc directly, whatever is in the Makefile, which I'm fairly certain uses gcc and such
sorear *plonk*
lue Oi! 02:52
02:52 literal_ is now known as literal
necrocow oh very matuire 02:52
yeah the whole world needs to know you added someone t oyour ignore list
lue Erm, who added you to an ignore list?
necrocow fucking idiot
02:52 necrocow left
lue O WOAH! Language mister! 02:52
m6locks O__o 02:53
sorear being friendly to idiots is so hard
lue rakudo: say :16<A*2**2> #would be 40
p6eval rakudo 5edcaf: OUTPUT«40␤» 02:54
sorear seriously, what's the point of joining and immediately being confrontational
maybe he's just a troll
lue Ah well (IMO, an 'Oi!' was appropriate at that point...)
at least he left volunatarily (now, should we block him from joining again...? Good thing I'm not op :) ) 02:55
02:57 fda314925 joined
lue And for the record, how could I reply to someone on my ignore list? :) 02:57
sorear plonk means I no longer care what you have to say 02:59
TimToady actually, historically (Usenet) it's the sound of putting someone in your killfile 03:02
sorear yes
TimToady oddly, I'm acquanted with the chap who invented killfiles :)
sorear but irssi killlists are broken by design, and I don't use them
lue Are you with the guy who invented kickbans? 03:03
sorear is trying to write a replacement in perl6
colomon TimToady: is there a convenient list_infix operator which isn't a metaop? I need it for a test case...
sorear Z
colomon Hmmm... Z is sort of implicitly a metaop... it's a funky case. 03:04
At least in theory, Z is just Z,
TimToady ...
colomon TimToady: You are a genius. :)
lue let me guess... in PIR, you can't do $N0 pow $N1, it has to be integers...
colomon lue: seems unlikely... pow is easier to implement on floats. 03:05
sorear depends. Is a FPU available?
colomon > say 10 R... 1, 3
13579
in PIR? yes 03:06
lue here's the line I'm trying: $N1 = base pow exponent
it says it expected a newline when I try make :/ 03:07
cognominal what is R... ?
colomon it's ... reversed 03:08
like 4 R- 5 is 5 - 4 = 1
lue rakudo: say :16<A*2.5**3.6> 03:09
p6eval rakudo 5edcaf: OUTPUT«Malformed radix number at line 11, near "<A*2.5**3."␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
colomon 10 R... 1, 3 is 1, 3 ... 10
lue I'm guessing fractional magnitudes are not allowed?
colomon lue: right
lue that makes things easier :)
cognominal clearly Perl 6 is a agglutinative language :)
I just don't yet know all the morphemes 03:10
lue argh! Still i won't let me do base pow exponent.
colomon And you can just keep stacking them... it will work as soon as I spectest and check in. 03:11
you can do stuff like RXR~
lue Is there a PIR expert that tells me why it won't let me use pow, insisting there should be a newline? 03:12
cognominal is ROXOR possible ? :)
colomon only if you define your own O metaop. :) 03:13
pugssvn r30148 | lwall++ | [STD] give better message on missing ** part
cognominal I can't figure where that goes is the grammar
colomon errr, and it would have to be like ROXOR~ or ROXOR+
cognominal *in 03:14
colomon unless we define a R operator that doesn't need an op to follow it.
it would be consistent with X and Z.... ;)
03:14 xomas left
lue I'm about to cry. Hours of coding stopped by an unruly pow opcode... 03:14
Tene sorear: what's wrong with /ignore in irssi? 03:15
pugssvn r30149 | colomon++ | [t/spec] Add tests to make sure precedence is properly carried over from op to Rop.
sorear Tene: it corrupts autolog data
if I'm looking back at my logs in a week, the rage is gone and I've removed the ignore, but the log data is still absent 03:16
Tene nods.
'k
colomon lue: I only see $N2 = pow $N0, $N1 constructions in alpha code....
lue ...
..................... 03:17
......
meep
rakudo: Q:PIR{fail}
p6eval rakudo 5edcaf: OUTPUT«error:imcc:syntax error, unexpected '\n', expecting '('␤ in file 'EVAL_1' line 64␤Could not find non-existent sub fail␤current instr.: '_block14' pc 29 (EVAL_1:0)␤» 03:18
colomon oh, nevermind, pir::pow__NNN is operators.pm
lue rakudo: Q:PIR{$I0=pow 2,3␤say $I0}
p6eval rakudo 5edcaf: OUTPUT«eval_ins: op 'pow_i_i_i' not found␤␤current instr.: 'perl6;PCT;HLLCompiler;evalpmc' pc 972 (compilers/pct/src/PCT/HLLCompiler.pir:538)␤»
colomon NNN means it takes two Nums and returns a Num. 03:19
hmmm.... 03:20
lue rakudo: Q:PIR{$I0=2 pow 3␤say $I0}
p6eval rakudo 5edcaf: OUTPUT«error:imcc:syntax error, unexpected PARROT_OP, expecting '\n' ('pow')␤ in file 'EVAL_1' line 64␤2␤»
colomon rakudo: say pir::pow__III(10,3)
p6eval rakudo 5edcaf: OUTPUT«The opcode 'pow_i_i_i' (pow<3>) was not found. Check the type and number of the arguments␤current instr.: 'perl6;PCT;HLLCompiler;evalpmc' pc 972 (compilers/pct/src/PCT/HLLCompiler.pir:538)␤»
lue so they HAVE to be nums?
colomon maybe?
lue rakudo: Q:PIR{$I0=pow 2.0,3.0␤say $I0}
p6eval rakudo 5edcaf: OUTPUT«eval_ins: op 'pow_i_n_n' not found␤␤current instr.: 'perl6;PCT;HLLCompiler;evalpmc' pc 972 (compilers/pct/src/PCT/HLLCompiler.pir:538)␤» 03:21
lue rakudo: Q:PIR{$N0=pow 2.0,3.0␤say $N0}
p6eval rakudo 5edcaf: OUTPUT«8␤»
lue OK.
03:21 wknight8111 left
lue yes! Compiling... 03:22
snarkyboojum rakudo: Q:PIR{$N0=pow 2.0,3␤say $N0}
p6eval rakudo 5edcaf: OUTPUT«8␤»
lue rakudo: say radcalc(16,'12','.4',3.5,2.2) # radix, int, frac, base, exp 03:23
p6eval rakudo 5edcaf: OUTPUT«283.284102145718␤»
colomon > say 1, 3, 5 RXR~ 'a'..'d'
1a3a5a1b3b5b1c3c5c1d3d5d
> say 1, 3, 5 X~ 'a'..'d'
1a1b1c1d3a3b3c3d5a5b5c5d
lue It allows fractional magnitudes thru radcalc directly then :)
Is that legal?
colomon well, radcalc is just a helper function, I don't know that it's actually spec'd anywhere. 03:24
lue ah well. Until you can use pow on integers, I'm not worrying :)
rakudo: say :16<A*2**3> # as a reference to see if my code works 03:27
p6eval rakudo 5edcaf: OUTPUT«80␤»
03:27 lichtkind left
lue yes! Now to spectest and, if colomon doesn't want to do it, fix the adverbs! \o/ 03:29
colomon er... fix the adverbs? 03:30
(and for sure I'm going to bed as soon as my own spectest run is done.)
lue Yes. I removed a variable. You just need to pass the number as a string, instead of an int string and a frac string :) 03:31
It broke the adverbs though :D
:16<>, :2<>, etc.
colomon oh, you're saying the parsing changed.
or at least, what you use changed.
yes?
lue yes! I completely rewrote radcalc :) 03:32
where are the adverbs then?
colomon but you haven't been able to test the full blow thing because we don't send the right info to you.
hold on, I'll get you a quick patch.
lue I've been using the radcalc() function call to see if things worked. Is something missing? 03:33
colomon once the actions are patched for you, you can run it against S02-literals/radix.t, and check that the 180-some tests we have still work after your change. 03:34
my $intpart := $<intpart>.Str;
my $fracpart := $<fracpart> ?? $<fracpart>.Str !! "0";
my $base := $<base> ?? +($<base>[0].Str) !! 0;
my $exp := $<exp> ?? +($<exp>[0].Str) !! 0;
make PAST::Op.new( :name('&radcalc'), :pasttype('call'),
$radix, $intpart, $fracpart, $base, $exp
);
look for that code in Actions.pm
and add
(after the my $fracpart line)
my $intfrac := $intpart ~ '.' ~ $fracpart; 03:35
and then in the $radix, $intpart, $fracpart, $base, $exp line, replace $intpart, $fracpart, with $intfrac,
snarkyboojum lue: you can get pow to work with Rakudo Int()s paste.lisp.org/display/96699
lue: not sure that's what you were trying to do anyway - but what the hell 03:36
colomon That should make the current parser talk to your version of radcalc so you can run tests on it.
lue so after all modifications, run make spectest, and then nopaste a patch if all goes well? 03:37
colomon don't do a full make spectest (at least at first) just do make t/spec/S02-literals/radix.t 03:38
that will do a short test that just tests your changes.
If that works out, then do a full spectest. :)
and oh, make sure you have the latest t/spec/S02-literals/radix.t, I fudged it today. 03:39
lue make test gets the latest spectest to rakudo, right? 03:40
colomon and then if your code works, we'll have to unfudge some of the tests, including some that have never worked before! :)
lue is running make; make test; make t/spec/more stuff here
I heard make-ing individual tests is bad (oh well) 03:41
colomon lue: TimToady is against it, and it probably is a logical blot on the system, but it works and is very handy. ;) 03:42
lue intermitently afk and !afk whilst compiling
bah well! I can always re-download the file :)
colomon bah, strange failures here, going to bed instead of pushing. 03:44
lue: good luck!
lue NO BUT WHAT IF i need to make a patch? 03:45
I guess the unfudgeing/patching can wait 'til tomorrow if nothing else :)
03:47 unobe joined
lue garh, decimals broke again :/ 03:55
it also doesn't like :16()... 03:56
I'll get to that tomorrow. Later!
sorear lue: make test runs Rakudo-specific tests only
03:56 astrojp left
sorear you want make spectest, which svn clones the pugs repo and runs the P6 spec suite 03:57
04:00 nihiliad left, nihiliad joined 04:06 cognomore joined 04:21 meppl left 04:31 dual left 04:50 xomas joined, xomas left, xomas joined 05:17 abra joined 05:30 nihiliad left 05:31 ShaneC left 05:41 sorear joined
snarkyboojum is wondering how exists could be implemented for Array now that it's not backed by a ResizeablePMCArray (which implements exists_keyed)... 05:43
TimToady bitmap?
what is it implemented with? 05:44
snarkyboojum I don't understand well enough.. perhaps it is :) 05:47
alpha used to be Array <- List <- ResizeablePMCArray
whereas in masters it's Array <- Seq <- Iterable <- Any <- Mu 05:49
05:50 unobe left
snarkyboojum I suspect I'm not making much sense :) 05:50
TimToady whether a container is there in the list of pointers seems not to have much to do with derivation
but I'm no expert in the internals 05:51
05:52 am0c joined
snarkyboojum and Mu is a P6opaque-ish thing ? No idea :) I was just hoping to port alpha's exists() functionality over to master :) 05:56
rakudo: @a = <a b c>; say @a.PARROT 06:03
p6eval rakudo 5edcaf: OUTPUT«Symbol '@a' not predeclared in <anonymous>␤current instr.: 'perl6;PCT;HLLCompiler;panic' pc 137 (compilers/pct/src/PCT/HLLCompiler.pir:101)␤»
snarkyboojum rakudo: my @a = <a b c>; say @a.PARROT 06:04
p6eval rakudo 5edcaf: OUTPUT«Array␤»
sorear As far as I know the Perl6 type ontology has nothing whatsoever to do with representation
06:05 molaf joined, eternaleye left 06:08 eternaleye joined
snarkyboojum my lack of understanding is the problem here :) 06:08
sorear you think YOU lack understanding?
snarkyboojum :) most definitely 06:09
06:09 IllvilJa left 06:16 abra left
snarkyboojum if say @a.PARROT returns 'Array', does that mean it uses an Array PMC? (naivety is bliss :)) 06:16
sorear Where is the master list of valid arguments to --target? 06:18
06:20 cognomore left
snarkyboojum parse, past, post, pir, evalpmc I think 06:26
from line 32 in parrot/compilers/pct/src/PCT/HLLCompiler.pir :) 06:28
afk
06:32 payload joined 06:35 am0c left
TimToady std: /\ / 06:45
p6eval std 30149: OUTPUT«===SORRY!===␤No unspace allowed in regex; if you meant to match the literal character, please enclose in single quotes (' ') or use a backslashed form like \x20 at /tmp/GrxIyYMA4y line 1:␤------> /\⏏ /␤FAILED 00:01 107m␤» 06:46
TimToady std: /\#/
p6eval std 30149: OUTPUT«===SORRY!===␤No unspace allowed in regex; if you meant to match the literal character, please enclose in single quotes ('#') or use a backslashed form like \x23 at /tmp/TinDHTWWpT line 1:␤------> /\⏏#/␤FAILED 00:01 105m␤»
TimToady std: /\␤/ 06:47
p6eval std 30149: OUTPUT«===SORRY!===␤No unspace allowed in regex; if you meant to match the literal character, please enclose in single quotes ('␤') or use a backslashed form like \x0a at /tmp/YOmcrtcG2A line 1:␤------> /\⏏<EOL>␤FAILED 00:02 105m␤»
TimToady std: :16<FACE*2> 06:48
p6eval std 30149: OUTPUT«===SORRY!===␤Base is missing ** exponent part at /tmp/2fnX4fV204 line 1:␤------> :16<FACE*2⏏>␤FAILED 00:01 105m␤»
TimToady zzz & 06:49
06:59 molaf left
sorear snarkyboojum: thanks 07:00
07:01 molaf joined 07:34 abra joined
sorear rakudo: class Bob is Int { }; my Int $a = 0; $a = Bob.new; say $a 07:37
p6eval rakudo 5edcaf: OUTPUT«0␤»
sorear I wonder how the existance of subclasses of primitive types interacts with the concept of unboxing 07:38
07:41 mberends joined
moritz_ Int is not a primitive type; int is 07:52
sorear oh 08:03
rakudo: class Bob is int { }; my int $a = 0; $a = Bob.new; say $a 08:04
p6eval rakudo 5edcaf: OUTPUT«Malformed my at line 11, near "int $a = 0"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
08:07 kaare joined 08:08 kaare is now known as Guest14894 08:09 envi^home joined 08:12 jhuni left 08:13 fridim joined
moritz_ but rakudo doesn't knoe about native/primitive types yet 08:16
08:26 Chillance left
moritz_ phenny: tell pmichaud could you please add a small guide to the nqp-rx README telling people how to update the nqp sources in parrot? it seems to lower the truck number unnecessary as is 08:32
phenny moritz_: I'll pass that on when pmichaud is around.
08:32 moritz_ sets mode: +o sorear
sorear ooh. shiny. 08:34
08:35 Su-Shee joined
moritz_ alpha: say "yes" if "\t" ~~ / <[\t]> /; 08:35
p6eval alpha 30e0ed: OUTPUT«yes␤» 08:36
moritz_ rakudo: say "yes" if "\t" ~~ / <[\t]> /;
p6eval rakudo 5edcaf: ( no output )
08:43 |Jedai| joined
moritz_ phenny: tell pmichaud I've assigned RT #73698 to you; it's blocking me from running the JSON::Tiny test suite, and I need to port JSON::Tiny to proto regexes for the book... - would be nice if you could take a look in your sparse tuits 09:06
phenny moritz_: I'll pass that on when pmichaud is around.
snarkyboojum rakudo: my @a = [1,,3]; say @a.elems 09:17
p6eval rakudo 5edcaf: OUTPUT«1␤» 09:18
snarkyboojum what's happening there?
moritz_ rakudo: my @a = [1,,3]; say @a.perl
p6eval rakudo 5edcaf: OUTPUT«[[1, 3]]␤»
moritz_ std: [1,,3]
p6eval std 30149: OUTPUT«===SORRY!===␤Preceding context expects a term, but found infix , instead at /tmp/83jkkgn6w6 line 1:␤------> [1,⏏,3]␤FAILED 00:01 108m␤»
moritz_ rakudo: my @a = 1,,3; say @a.elems
p6eval rakudo 5edcaf: OUTPUT«2␤»
snarkyboojum cool.. so rakudo should fail?
moritz_ aye
snarkyboojum excellent
09:31 am0c joined 09:35 iblechbot joined
moritz_ phenny: tell jnthn that t/spec/S12-class/attributes.t emits some IMCC syntax errors, probably at compile time (but still runs) and passes all but one test - LHF? 09:38
phenny moritz_: I'll pass that on when jnthn is around.
09:38 colomon joined 09:42 unobe joined 09:45 unobe left
moritz_ rakudo: print 2**20 09:53
p6eval rakudo 5edcaf: OUTPUT«1048576» 09:54
colomon o/ 09:56
09:57 M_o_C joined 09:58 M_o_C left
moritz_ it's c\o/lomon! 10:01
the last test in t/spec/S03-operators/r-metaop.t fails for me
colomon moritz_: actually should be the last two tests that fail. At least, I think I added the tests but haven't pushed the patch yet. :( 10:06
or does it fail hard? 10:07
is 4 R- 5 R/ 10, -2, "Rop gets the precedence of op";
is (10 R... 1, 3), (1, 3, 5, 7, 9), "Rop gets list_infix precedence correctly";
10:08 drbean left
colomon I guess this is a sign the first of those two tests is not sufficient -- I thought it wouldn't work without the patch! 10:11
moritz_ test 29, is 4 R- 5 R/ 10, -2, "Rop gets the precedence of op"; works
colomon nods 10:12
I just tested it on one of my other machines.
you can ignore the #30 fail, it will do so until I get my patch pushed.
moritz_ rakudo: say 3 R/ 9 + 5 # should be 8
p6eval rakudo 5edcaf: OUTPUT«4.66666666666667␤» 10:13
moritz_ that's something we can put in the test file
colomon moritz_++
snarkyboojum rakudo: my @list = <a b c>; @a[0].say;
p6eval rakudo 5edcaf: OUTPUT«Symbol '@a' not predeclared in <anonymous>␤current instr.: 'perl6;PCT;HLLCompiler;panic' pc 137 (compilers/pct/src/PCT/HLLCompiler.pir:101)␤»
moritz_ rakudo: say 9 / 3 + 5
p6eval rakudo 5edcaf: OUTPUT«8␤»
snarkyboojum rakudo: my @list = <a b c>; @list[0].say; 10:14
p6eval rakudo 5edcaf: OUTPUT«a␤»
snarkyboojum rakudo: my @list = <a b c>; @list.exists(0);
p6eval rakudo 5edcaf: OUTPUT«Method 'exists' not found for invocant of class 'Array'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
colomon moritz_: You unfudged a bunch of tests in radix.t that only work on 64-bit platforms. 10:15
moritz_ colomon: I'd like to move S03-operators/foo-metaop.t to S03-metaop/foo.t - what do you think about that? 10:16
colomon: eh - could you please move them to overflow.t?
colomon moritz_: on overflow -- will do.
on S03-metaop -- makes sense to me. 10:17
10:18 rv2733 joined
colomon give me a sec (before reorg) to get the radix.t patches in. 10:18
10:19 drbean joined
moritz_ will do it after breakfast :-) 10:19
colomon is now quietly singing "Overflow" from Porgy and Bess to himself, he just realized. 10:22
dalek kudo: e877b2a | (Solomon Foster)++ | src/Perl6/ (2 files):
Move the <infix_prefix_meta_operator> <infixish> into the <infix_prefix_meta_operator> tokens themselves.
 10:28
kudo: 6be8f8c | (Solomon Foster)++ | src/Perl6/Grammar.pm:
Proper precedence for !op, Rop, and Sop. jnthn++.
snarkyboojum would this work for Array .exists? (I have a feeling is dodgey - but works i.e. passes tests :)) paste.lisp.org/display/96712 10:31
moritz_ rakudo: my @a = (1, 2); say @a[0] ~~ Proxy 10:32
p6eval rakudo 5edcaf: OUTPUT«0␤»
colomon moritz_: go ahead with the move whenever you're ready.
moritz_ rakudo: my @a = (Proxy, Proxy); say @a[0] ~~ Proxy
snarkyboojum my @a = (1, 2); say @a[-1] ~~ Proxy
p6eval rakudo 5edcaf: OUTPUT«1␤»
snarkyboojum my @a = (1, 2); say @a[6] ~~ Proxy
colomon is non-existence the only reason to return a Proxy? 10:33
snarkyboojum dunno
moritz_ unless you store a Proxy inside it
colomon: will do later on
snarkyboojum but reading the source of Proxy.pir "A Proxy is used to represent non-existent elements in aggregates."
thought it might work ;)
so I take it - that's a no :) 10:34
moritz_ thinks so
snarkyboojum how else could it be done?
moritz_ maybe you need to access the RPA under the hood directly
afk
snarkyboojum I tried, but there's no RPA there anymore in master right?
moritz_ the items must be stored somewhere 10:36
snarkyboojum an Array is a Seq is a Iterable is an Any is a Mu
moritz_: agreed :) but where? 10:37
:)
colomon snarkyboojum: look at postcircumfix:<[ ]> in Array.pir
snarkyboojum colomon: k
colomon that's where the Proxy is returned if the element doesn't exist, I think. 10:38
unless null elem goto have_elem
snarkyboojum so it just checks if values[n] is null or not in PIR 10:39
colomon yeah
snarkyboojum is that the appropriate check then?
colomon I mean, I think -- I'm not an expert on this.
but that would be my thought on how to do it.
snarkyboojum so checking for Proxy is invalid because you could store a Proxy object in the array 10:40
but if you check for an element and it's null Rakudo gives you back a Proxy object :)
how would you discern the difference?
colomon I don't think you can if you just look at [ ]'s return value. 10:41
snarkyboojum colomon: cheers again then 10:42
colomon good luck!
snarkyboojum colomon: it felt like cheating :)
colomon afk # going to try to go back to bed, maybe 10:43
snarkyboojum ?(bed) ;)
11:08 PZt left, meppl joined 11:15 dual joined, snarkyboojum left 11:17 abra left
pugssvn r30150 | moritz++ | [t/spec] re-fudge listquote.t for rakudo 11:23
r30151 | moritz++ | [t/spec] failing Rop precedence test for colomon++
r30152 | moritz++ | [t/spec] unfudgea a >>+<< test for rakudo, jnthn++
r30153 | colomon++ | [t/spec] Move overflowing tests from radix.t to overflow.t.
moritz_ wow, pugssvn seems slow 11:24
colomon it must have just been off, no? 11:25
It didn't have any reports at all yesterday, as far as I can recall. And someone (you, I suspect) was making changes all day long.
moritz_ and I tried to resurrect pugssvn 11:26
another question regarding test file names
currently they are inconsisten
cross-metaop.t vs. r-metaop.t
should we name them cross and reverse, or x and r?
moritz_ votes for the longer names
colomon cross and reverse, I'd say 11:27
pugssvn r30154 | moritz++ | [t/spec] move some meta operator test files to S03-metaops/ 11:29
r30155 | moritz++ | [t/spec] also move hyper op tests to S03-metaops 11:32
11:46 wknight8111 joined
jnthn o/ 11:46
phenny jnthn: 09:38Z <moritz_> tell jnthn that t/spec/S12-class/attributes.t emits some IMCC syntax errors, probably at compile time (but still runs) and passes all but one test - LHF?
colomon \o
jnthn moritz_: Ouch. I'll have to hunt that one down. Maybe we need to skip something we todo. :-)
morning, colomon 11:47
colomon I got Rop precedence working late last night, checked it in this morning.
jnthn++ for figuring out how to do it.
jnthn moritz_: I think I worked out the nqp => Parrot update process, btw. I'll give it a try, and if I don't cause epic fail, I'll document that. 11:48
colomon moritz_: do you want me to patch t/spectest.data to match your new layout? 11:51
11:54 clintongormley joined
jnthn colomon++ # nice! 11:55
colomon jnthn: if you look in the backlog, you can see a R... example, TimToady++'s idea. 11:56
11:56 payload left
colomon I think I just successfully patched my != issue by detecting it in the metaop code and switching it to logical ! == internally. 11:56
jnthn OK. 11:59
There's more tha one way to fix it :-)
colomon obviously a grammar fix would be more elegant, but this way should work, I think. spectesting now. ;)
12:00 Su-Shee left
jnthn colomon: True, it would be. ;-) 12:04
moritz_ colomon: yes, pushing...
dalek kudo: 6a8049b | moritz++ | t/spectest.data:
track name changes of test files
 12:06
kudo: 760b272 | moritz++ | docs/running.pod:
[docs] fixed a typo in running.pod reported in <bugzilla.redhat.com/show_bug.cgi?id=574547>, gerd++
kudo: 547069d | moritz++ | t/spectest.data:
more passing test files
jnthn \o/ 12:08
12:09 Su-Shee joined
m6locks rakudo: my Num $a = 5.22; say $a 12:15
p6eval rakudo 6be8f8: OUTPUT«Type check failed for assignment␤current instr.: '&die' pc 17127 (src/builtins/Junction.pir:399)␤»
m6locks rakudo: my $a = 5.22; say $a.WHAT
p6eval rakudo 6be8f8: OUTPUT«Rat()␤»
moritz_ in ideal Perl 6, you'd constrain $a to Numeric, not Num if you want it to hold any number type 12:16
m6locks so Num is not a number?
jnthn Rakudo only supports unideal Perl 6 so far. :-)
m6locks rakudo: my Numeric $a = 5.22; say $a
jnthn m6locks: Num = Floating point number
p6eval rakudo 6be8f8: OUTPUT«Method 'ACCEPTS' not found for non-object␤current instr.: '&infix:<=>' pc 16823 (src/builtins/Junction.pir:228)␤»
moritz_ the role Numeric is NYI 12:17
m6locks aye, well i guess its not supported then
jnthn m6locks: But 5.22 is not a floating point number in Perl 6, it's a Rat.
m6locks oh, how do i make it a Num
when assigning a value like that
moritz_ with scientific notation, 5.22e0
or 5.22.Num
m6locks i'm cool wit Rat too, I just need to change all the Nums to Rats 12:18
12:25 am0c left 12:29 M_o_C joined
colomon aiiiiiiii! 12:33
jnthn O_O
colomon: ? :-) 12:34
colomon > say (10 R... 1, 3) 12:35
Lexical 'self' not found
that worked fine last night...
jnthn :-/ 12:36
colomon hmmm... that may have been before the Rop overhaul.
jnthn makes the latest
colomon the actions part of it.
oh, hey, good thinking.
jnthn (latest from git, that is) 12:37
Will try it once it's built. :-)
colomon I think it will work with the latest.
so it's a change I've made and haven't pushed.
jnthn rakudo: say (10 R... 1, 3) 12:38
p6eval rakudo 6be8f8: OUTPUT«13579␤»
colomon this damn refactor is going to be the death of me.
jnthn "Tormenting the implementers" 12:39
:-)
It's OK, I've felt that way about plenty of Rakudo refactors and I'm still alive. They're rarely fatal. :-)
12:43 iblechbot left
jnthn > say (10 R... 1, 3) 12:44
13579
colomon: So certainly something you have locally.
colomon yup. 12:45
jnthn colomon: I can glance over the diff if you like.
colomon > say reverseargs(&infix:<...>, 10, [1, 3])
Lexical 'self' not found
our multi sub reverseargs(&op, Mu $a, Mu $b) {
&op($b, $a);
}
actually, can you try just that? 12:46
it would help pinpoint if the problem is that or something with != in the series operator.
jnthn > our multi sub reverseargs(&op, Mu $a, Mu $b) { &op($b, $a); } 12:47
> say reverseargs(&infix:<...>, 10, [1, 3])
Lexical 'self' not found
That's enough.
The backtrace is...weird. 12:48
colomon that's a huge help, then!
jnthn++
jnthn Oh no, maybe it's not
Is ... implemented using gather/take?
colomon yes 12:49
jnthn OK, it's somewhere within that, it appears...
colomon what backtrace did you get? 12:50
jnthn find_lex $P21399, "self" 12:51
$P21400 = $P21399."succ"()
That's where it's blowing up
oh 12:52
$next = { $.succ };
In operators.pm, in infix:<...>, line 363
colomon: note: 12:54
> say ([1,3] ... 10)
Lexical 'self' not found
colomon oh.
moritz_ wrong invocant 12:55
jnthn colomon: I don't know why $.succ and $.pred are being used in a sub though
That can never work.
oh, wait, it's assining a closure...
colomon that's supposed to be defining a closure
jnthn But still... 12:56
It's a closure, not an anonymous method.
Is it just menat to be .succ and .pred?
moritz_ thinks so
colomon I think maybe it is.
moritz_ spectests without that dollar
colomon Oh, it's never been called before because there is infix:<...>($lhs, $rhs)
jnthn colomon: here you go... 12:57
> our multi sub reverseargs(&op, Mu \$a, Mu \$b) { &op($b, $a); }
> say reverseargs(&infix:<...>, 10, (1,3))
13579
colomon so what we're seeing here is something utterly bizarre.
jnthn Well, no, it's just that passing an array means you pass a single item, whereas actually what you really have is a parcel
Also, the missing \s meant that context was being enforced by reverseargs
colomon if there is a single item there (the array), shouldn't it call infix:<...>($lhs, $rhs) ? 12:58
jnthn So the the things had already been put into item context by the time we called &infix:<...> so it dispatched differently.
(@lhs is copy, $rhs) is narrower than ($lhs, $rhs)
colomon rebuilding with \s to see what happens. 12:59
jnthn Anyway, I think the \s are what's missing.
13:06 bbkr left
colomon > say 10 R... 1, 3 13:13
13579
I'd say you're right. :)
13:22 masak joined
colomon o/ 13:22
masak oh hai, #perl6
13:24 M_o_C left
moritz_ \o/ it's masak 13:26
masak \o/
13:26 bbkr joined
moritz_ colomon: in radix.t I get a passed TODO, test 78 - I assume on 32 bit that fails? 13:27
masak enjoys TimToady's "blue sky" email :) 13:29
13:29 justatheory joined
masak it's a bit harsh against p6l perhaps, but it's true. 13:30
13:31 fridim left
colomon moritz_: yes. 13:31
masak jnthn: will you help me dust off my awful Russian too? :) 13:34
he's been standing in a corner for years, collecting all that dust.
13:35 astrojp joined
masak rakudo: "foo:" ~~ / (\S+) ':' / and say "matched $0" 13:36
p6eval rakudo 6a8049: ( no output )
masak I think that's the (known) 'backtracking inside of capture parens' bug 13:37
13:40 MOAR-f00li5h left
masak std: 1 [+] [+] 2, 3, 4 13:40
13:40 f00li5h joined
p6eval std 30155: OUTPUT«ok 00:01 109m␤» 13:40
13:40 f00li5h left, f00li5h joined, iblechbot joined
masak yeah, seems I was wrong about the semantics of '1 [+] 2, 3, 4' yesterday. 13:40
in my defense, it's quite a cute semantics :P 13:41
13:41 muixirt joined
masak std: <-> { [[[<=>]]] } 13:42
p6eval std 30155: OUTPUT«===SORRY!===␤Can't reduce with <=> because structural infix operators are diffy and not chaining at /tmp/QT5tXWtCvt line 1:␤------> <-> { [[[<=>]]]⏏ }␤ expecting prefix_circumfix_meta_operator__S_257reduce␤FAILED 00:02 107m␤»
moritz_ so that's parsed as infix:<[+]> listop:<[+]> ? 13:43
masak oh. :) diffy.
bbkr rakudo: class X { }; class { is X; } #s12 anon class segfaults :(
p6eval rakudo 6a8049: OUTPUT«Invalid namespace key in set_pmc_keyed␤current instr.: 'perl6;ClassHOW;new' pc 3749 (src/metamodel/ClassHOW.pir:27)␤»
masak bbkr: yes. not yet implemented.
bbkr thanks
13:46 abra joined
colomon )P"" 13:47
dalek kudo: d296602 | (Solomon Foster)++ | src/ (2 files):
Reorganization of !op, Rop, Xop, and Zop in preparation for a refactor.
 13:48
kudo: 2067a57 | (Solomon Foster)++ | src/core/metaops.pm:
Switch Mu $a to Mu \$a in the notresults and reverseargs functions, as per jnthn++'s advice.
masak colomon++
13:50 synth left, [synth] left, synth joined, [synth] joined
moritz_ rakudo: say [] ~~ [1] 13:50
p6eval rakudo 6a8049: OUTPUT«0␤»
moritz_ who fixed that?
pugssvn r30156 | moritz++ | [t/spec] a tiny unfudge for rakudo 13:51
colomon is anyone else getting intermittent test failures again? 13:52
moritz_ I get an intermittent pasisng TODO in uc.t
and S04-statements/for.t fails reproducible 13:53
masak twitter.com/arafatkamaal/status/10809395431 # strange/funny realization by someone
colomon I've gotten crashes in uc.t as well.
m6locks haven't built today
colomon I'm also getting sometimes failures in S06-signature/named-parameters.t and a consistent failure in S03-operators/precedence.t. 13:54
If no else is seeing that last one (yet), it may be my fault. Though I really don't know how... 13:55
moritz_ precedence.t is fine here. 13:56
colomon have you pulled yet? ;)
moritz_ no
colomon that makes it sound more like it might be my fault. :(
moritz_ pulls and re-tests 13:57
13:57 justatheory left
bbkr hmm, why this example: "'class X {}; my $x = X; for 1..10000000 { if $_ % 2 {$x .= HOW} else {$x .= WHAT}; }'" is eating the memory? if I call CLASS.HOW.WHAT.HOW.WHAT.HOW.WHAT... it should be constant amount of memory used, because WHAT and HOW methods return the same objects, isn't it? 13:58
moritz_ bbkr: even if not, it shouldn't leak memory,because no references to the original objects are being kept 13:59
bbkr should i fill rakudobug? 14:00
moritz_ so either something is keeping references internally, or parrot's GC is too lazy
bbkr: yes, please
colomon no
it's happening because for eagerly gets the list to process on.
so it tries to build the 10000000 item list before it starts the loop. 14:01
moritz_ oh
and for can't be lazy until we have sink context, right?
colomon That's a separate issue, I think. 14:02
(But then, I admit I don't understand the implementation.)
moritz_ colomon: with your latest changes, test 20 of precedence.t fails
colomon moritz_: yeah, that's what I've been seeing too.
moritz_ nqp: fail('foo')
p6eval nqp: OUTPUT«Could not find non-existent sub fail␤current instr.: '_block11' pc 0 (EVAL_1:5)␤»
moritz_ nqp: fail 'foo'
p6eval nqp: OUTPUT«Confused at line 1, near "fail 'foo'"␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
colomon can the brain trust take a look at that one? I don't have a clue, but it must have happened in one of the last two commits.... 14:03
bbkr colomon: you're right. it used memory for building list; 14:04
colomon as for "for", "map" lets you lazily evaluate its inputs but still eagerly evaluate its results. I don't know why "for" should be any different. But pmichaud actually changed it so "for" eagerly evaluated its inputs, and I'm sure he must have had a good reason. 14:05
14:05 dalek left
colomon afk # loading boxes of books into the minivan... 14:05
14:12 dalek joined
colomon why no t/spec/S03-metaops/cross.t? 14:14
masak std: 64 R... 1, 2, 4
p6eval std 30156: OUTPUT«ok 00:01 108m␤» 14:15
colomon rakudo: say 64 R... 1, 2, 4
p6eval rakudo 6a8049: OUTPUT«1248163264␤»
masak \o/
jnthn masak: You should give your Russian vodka. That'll make everything just fine. 14:16
masak: Apparently it increases fluency.
;-)
Or at least percieved fluency. 14:17
masak I've actually felt my fluency increase from alcohol.
in that particular case, it was Mandarin.
jnthn I'm still trying to work out if it's just perception or if it really actually helps. 14:18
arnsholt I agree. A bit of alcohol does help you loosen up a bit, I think
14:18 Su-Shee left
jnthn It may just be that you fear getting it a tad wrong less. 14:18
And Just Say Stuff. And it's mostly good enough. 14:19
arnsholt Without any actual empirical basis, I think it's related to the loss of inhibition that comes with alcohol
masak jnthn: yeah.
jnthn Of course, I guess this only works up to a point.
:-)
arnsholt Yeah, eventually the loss of brain catches up with you =)
masak aye. inhibition is a necessary part of communication to some extent. :)
colomon I've heard trad musicians say the same thing about drink increasing fluency... 14:21
masak aww, I missed tonight's troll. :/
who would have thought that a person called 'necrocow' would be so hostile? :P
jnthn masak: Maybe you'll get to C him again. 14:22
masak apparently he was very familiar with what '*plonk*' means, at least. I wasn't.
jnthn: :)
necrocow! come back and tell us more about compilers! :) 14:23
pugssvn r30157 | colomon++ | Move cross.t to S03-metaops too.
colomon Yes, and he can spend more time telling TimToady how Perl 5 works.
jnthn Clearly writing compilers in a language that we can all be more productive in is the wrong way to do it.
:-) 14:24
arnsholt The troll is always right, you know =) 14:25
colomon masak: as for knowing what *plonk* means, apparently you don't spend much time behaving in a fashion to get yourself kicked out of chats... 14:26
jnthn lol...searching Google for "plonk irc" to try and get insight is amusing. 14:27
arnsholt nqp: my @a = (); say @a.WHAT; 14:29
p6eval nqp: OUTPUT«Assignment ("=") not supported in NQP, use ":=" instead at line 1, near " (); say @"␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
arnsholt nqp: my @a := (); say @a.WHAT;
p6eval nqp: OUTPUT«Confused at line 1, near "say @a.WHA"␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
jnthn no listops, use parens with say 14:30
arnsholt nqp: my @a := (); say(@a.WHAT);
p6eval nqp: OUTPUT«Method 'WHAT' not found for invocant of class 'ResizablePMCArray'␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
arnsholt Not what I intended, but it did tell me what I wanted ^^ 14:31
masak colomon: no, to be honest I haven't (and wouldn't consider trying). I guess I used to be more of a groupie on #perl6, before I found my slots of productivity. but never outright maladjusted.
arnsholt nqp: my @a := (); @a.push(1); say(@a); 14:32
p6eval nqp: OUTPUT«1␤»
arnsholt nqp: my @a := (); @a.push: 1; say(@a);
p6eval nqp: OUTPUT«1␤»
14:36 mdxi_ joined
arnsholt nqp: my @a; sub b($x) { @a.push: $x; }; sub a() { @a := (); b("foo"); say(@a); } 14:36
p6eval nqp: ( no output )
arnsholt nqp: my @a; sub b($x) { @a.push: $x; }; sub a() { @a := (); b("foo"); say(@a); } a();
p6eval nqp: OUTPUT«Confused at line 1, near "sub a() { "␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
arnsholt nqp: my @a; sub b($x) { @a.push: $x; }; sub a() { @a := (); b("foo"); say(@a); }; a();
p6eval nqp: OUTPUT«1␤»
arnsholt nqp: my @a := (); @a.push: 1; @a.push: 2; say(@a);
p6eval nqp: OUTPUT«2␤»
mathw Hello 14:39
arnsholt nqp: my @a; sub b($x) { @a.push: $x; }; sub a() { @a := (); b("foo"); }; a(); say(@a);
p6eval nqp: OUTPUT«1␤»
masak mathw!
mathw masak!
colomon o/
14:40 christine joined
mathw guess what masak 14:40
I still can't use the Esperanto direct object properly
masak mathw: it's only been, what, a month or two? 14:41
mathw true
masak are you expecting a miracle? :P
mathw not really
arnsholt nqp: my @a := (); @a.push: 1; @a.push: 2; say(|@a);
p6eval nqp: OUTPUT«12␤»
mathw I just get embarrassed when I forget to even think abou tit
masak that's the only really heavy point of Esperanto grammar. it takes time.
trust me, in seven-eight years it'll be second nature to you :P 14:42
mathw hopefully
14:42 wknight8111 left
mathw how are my named enums coming along? :) 14:42
masak mathw: I'm free this afternoon, wanna help me finish them up? :) 14:43
mathw I'd love to
masak \o/
mathw although I'm not sure if you can expect much in the way of 'help'
masak pushes the branch of the progress so far
jnthn is around to offer moral (and if needed technical) support for named enums too :-)
mathw \o/
masak hold on, gotta rebase and stuff. 14:44
colomon come on, spectest, faster, faster!
masak mathw: github.com/rakudo/rakudo/tree/named-enums 14:45
mathw ooh fancy you
a branch in rakudo's own repo :P
masak mathw: pull that and build, and then try something like 'enum A <b c d>'
mathw: it's only for today, until we make this work :) 14:46
mathw builds
masak too
14:47 dalek left 14:50 dalek joined
mathw masak: it accepts "enum A <b c d>" but not anything I might try to do with it after that 14:53
which bits of code are we going to be concerned with 14:54
masak what's the error message?
mathw ./perl6 -e "enum A <b c d>; say A::b.WHAT;"
Null PMC access in invoke()
current instr.: '_block14' pc 29 (EVAL_1:0)
called from Sub '!UNIT_START' pc 1299 (src/glue/run.pir:23)
called from Sub 'perl6;PCT;HLLCompiler;eval' pc -1 ((unknown file):-1)
dalek kudo: a61c40c | (Solomon Foster)++ | src/Perl6/Actions.pm:
infix_prefix_meta_operator refactor!
mathw called from Sub 'perl6;PCT;HLLCompiler;command_line' pc 1489 (compilers/pct/src/PCT/HLLCompiler.pir:794)
called from Sub 'perl6;Perl6;Compiler;main' pc -1 ((unknown file):-1)
masak hm. 14:55
my build isn't finished quite yet.
but take a look at the partial commit with 'git show' in the meantime.
14:55 am0c joined
jnthn files NPMCA Rakudo bug 14:56
oh, wait..
:-)
moritz_ did you mean: masakbug? :-)
arnsholt nqp: my @paths; sub choose(*@x) { my $v := @x.shift; @paths.push: "dummy"; return $v; }; sub blob() { say(choose("a", "b", "c")); }; blob();
p6eval nqp: OUTPUT«a␤»
arnsholt ARG!
masak aaah. the Rakudo build ran much smoother after I killed Firefox :)
I thought that might be it. 14:57
mathw masak: my git-fu fails at this task
jnthn masak: Yeah, it sucks how firefox gets like that when you have 5 or 6 tabs open. ;-0
masak mathw: just typing 'git show' should be enough.
jnthn: no, that's my two other browsers. :) 14:58
mathw no, git show declares that I merged branch 'master'
moritz_ git show HEAD~2 # try this one
masak mathw: oh, did you pull the named-enums branch into master? :/ 14:59
mathw I didn't try to
I don't grok git branching, really
masak I know the feeling.
mathw but my git thinks I'm on that branch
masak it's ok. you'll probably need to back out somehow when you decide you're done with this experiment. but that's all. 15:00
perhaps a 'git pull --force origin master' will be enough.
mathw bah 15:01
mathw starts again
I really need to learn how to use git properly one day
jnthn Same :) 15:02
With time, I see to screw it up less though. :-)
15:03 tylerni7 joined
masak mathw: starting again would be 'git checkout -b named-enums origin/named-enums'. 15:03
mathw Why is it that I only ever seem to have one USB cable? 15:04
Almost every gadget I own came with one
yet I can find only one
back shortly... 15:05
colomon 28401 passing tests here, BTW. 15:10
moritz_ wow 15:11
masak \o/ 15:12
jnthn tries to remember what the peak was with alpha 15:14
moritz_ 35k?
pugssvn r30158 | moritz++ | [t/spec] now that complex and Rat literals are gone, remove them from S02-literals/numeric.t; this makes the test file rather pointless, but maybe some brave sould will add tests for <2+3i> or so in future 15:17
m6locks rat is gone? 15:18
i just switched all my nums to rat :D
oh well 15:19
jnthn m6locks: No, not in that sense. 15:20
masak mathw: I'll be back in a while too. need to pull the ethernet cord while Firefox boots back up.
jnthn moritz_: I don't think we were passing 35k?
15:20 masak left
colomon m6locks: it used to be that, say 3 / 4 was processed as one number by the grammar. 15:21
moritz_ actually only 3/4 (without the spaces)
colomon m6locks: now it's just the integer 3 divided by the integer 4, which creates the Rat.
m6locks oh ok
moritz_ so 1/2 ** 4 was parsed as (1/2) ** 4, not as 1/(2**4)
colomon and 1/7/7/7/7 was just a nightmare.
moritz_ jnthn: you're right, it was only 32.7k :-) 15:22
m6locks rakudo: say (5.42).WHAT
p6eval rakudo a61c40: OUTPUT«Rat()␤»
colomon no way, we've got 7/8ths of the tests back now?!
jnthn 4.3k-ish to go then.
We've still got a bunch of S05-mass/*.t
moritz_ hash slices 15:24
jnthn Oh. Those. 15:25
Yes.
mathw \o/ 15:26
mathw has masak's partial commit, but no masak
moritz_ currently I'm inclined to remove the fatal 'Use of type object as a value' in the vtables in Mu
jnthn moritz_: Maybe it should be demoted to a warning. 15:27
moritz_ IMHO failing hard on something that's supposed to fail softly is wronger than not failing at all
arnsholt nqp: say("{3}".WHAT) 15:29
p6eval nqp: OUTPUT«Method 'WHAT' not found for invocant of class 'String'␤current instr.: '_block11' pc 0 (EVAL_1:6)␤»
moritz_ jnthn: warning is good, yes
15:31 wallberg joined
jnthn arnsholt: don't expect NQP to handle .WHAT too well in general. 15:31
15:38 masak joined
masak mathw: back again :) 15:39
mathw: while I was offline, I did some further experiments with the patch.
mathw ahah
well I've got the thing up on my screen now
although it's still tryin gto compile it :)
masak mathw: have a look at src/glue/enumeration.pm 15:40
there's a sub there with a few things commented out.
that's the one we want to get right for named enums to work.
jnthn++ for all the help so far.
mathw I've got a sub with one line commented out... 15:41
masak ok.
I put a 'say' statement at the top, and it actually gets called. 15:42
right now I'm rebuilding with an added argument: the scope in which the enum is to be installed. 15:43
it's in that area that my knowledge is a bit weak.
mathw hmm 15:44
my build's blowing up now
masak apparently, the thing to send in on the Actions side is PAST::Op.new( :pirop('get_namespace P') )
mathw ah I know why
arnsholt I think I'm trying to be to clever for my own good ^^ 15:45
masak jnthn: it's now a fact. the PAST::Op.new( :pirop('get_namespace P') ) argument is the one that causes the CREATE_NAMED_ENUMERATION to blow up. :/ 15:46
arnsholt jnthn: Yeah, I sort of guessed. But it does give me the type of the object, even if it's not quite what it's supposed to do =)
jnthn masak: What do you do with it? :-)
Or is it merely passing it that causes that?
masak jnthn: the latter.
mathw oooh dear 15:47
core dump on build
jnthn What's the signature?
masak jnthn: want me to push a state where it blows up?
jnthn Can do 15:48
masak jnthn: github.com/rakudo/rakudo/commit/d26...27bc0023f2
arnsholt nqp: my $s := sub { return "foo" }; my $a := $s(); say $a; 15:49
p6eval nqp: OUTPUT«Routine declaration requires a signature at line 1, near "{ return \""␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
arnsholt nqp: my $s := sub() { return "foo" }; my $a := $s(); say $a; 15:50
p6eval nqp: OUTPUT«Confused at line 1, near "say $a;"␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
arnsholt nqp: my $s := -> { return "foo" }; my $a := $s(); say $a;
p6eval nqp: OUTPUT«Missing block at line 1, near "-> { retur"␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
arnsholt nqp: my $s := -> { return "foo"; }; my $a := $s(); say $a;
p6eval nqp: OUTPUT«Missing block at line 1, near "-> { retur"␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
arnsholt nqp: my $s := -> $x { return "x: " ~ $x; }; my $a := $s("foo"); say $a; 15:51
p6eval nqp: OUTPUT«Confused at line 1, near "say $a;"␤current instr.: 'parrot;HLL;Grammar;panic' pc 500 (src/cheats/hll-grammar.pir:197)␤»
arnsholt nqp: my $s := -> $x { return "x: " ~ $x; }; my $a := $s("foo"); say($a);
p6eval nqp: OUTPUT«No exception handler and no message␤current instr.: '_block15' pc 51 (EVAL_1:0)␤»
arnsholt Anyone more experienced than me have any suggestions on how to do this? 15:52
masak arnsholt: in Perl 6, that's actually correct.
arnsholt That's good at least
masak arnsholt: a pointy block isn't "advanced" enough to return from. 15:53
only subs, methods and submethods are.
arnsholt nqp: my $s := -> $x { say("Alive: " ~ $x); }; $s("foo"); say("Alive"); 15:54
p6eval nqp: OUTPUT«Alive: foo␤Alive␤»
arnsholt nqp: my $s := -> $x { say("Alive: " ~ $x); }; my $a := $s("foo"); say("Alive: " ~ $a);
p6eval nqp: OUTPUT«Alive: foo␤Alive: 1␤»
arnsholt masak: Thanks!
masak no problem. :)
arnsholt I don't really need the return value, but I do need an anonymous sub and then invoke it 15:55
mathw okay, I think I'm finally in roughly the same place as masak
masak jnthn: any thoughts?
arnsholt Hang on, I may actually want a return value
mathw arnsholt: use sub {} then
masak arnsholt: 'return value' and 'returning early' are different things.
arnsholt Oooooh, right 15:56
nqp: my $s := -> $x { say("Alive: " ~ $x); undef }; my $a := $s("foo"); say("Alive: " ~ $a);
p6eval nqp: OUTPUT«Alive: foo␤Null PMC access in get_string()␤current instr.: '_block11' pc 0 (EVAL_1:5)␤»
mathw yes, pointy blocks can't catch return exceptions, but subs can
masak nqpbug! :)
arnsholt I think I'll go with a sub (if I can get it to work), as I'm gonna stick some inline PIR and stuff in there as well 15:57
jnthn masak: You're taking it as a scalar which should be fine...
masak jnthn: this is where I got stuck during the hackathon. it looks like it should work.
and we want to do something very much like this.
jnthn Aye.
arnsholt nqp: my $s := sub () { say "Hello!"; return "pong"; }; my $a := $s(); say($a); 15:58
p6eval nqp: OUTPUT«Unable to parse blockoid, couldn't find final '}' at line 1␤current instr.: 'parrot;Regex;Cursor;FAILGOAL' pc 1664 (src/Regex/Cursor-builtins.pir:179)␤»
jnthn masak: I'm curious what it's trying to invoke.
masak yeah. no idea. 15:59
possibly the sub itself, but then again maybe not.
jnthn Yeah, put passing an extra param shouldn't make the sub unfindable... 16:00
*but
arnsholt nqp: my $s := sub($x) { say "Hello!"; return "pong"; }; my $a := $s(); say($a);
p6eval nqp: OUTPUT«Unable to parse blockoid, couldn't find final '}' at line 1␤current instr.: 'parrot;Regex;Cursor;FAILGOAL' pc 1664 (src/Regex/Cursor-builtins.pir:179)␤»
jnthn masak: Maybe I should grab the branch...
16:00 dual left
masak jnthn: I think we'll need the help, to be honest... :) 16:00
mathw I have simply no idea 16:01
at the moment I get this
./perl6 -e "enum A <a b c d>"
Not enough positional parameters passed; got 2 but expected 3
current instr.: 'CREATE_NAMED_ENUMERATION' pc 307701 (src/gen/core.pir:20547)
arnsholt Bah. I don't think NQP likes me
masak mathw: aye, me too.
mathw thus, your last commit is an attempt to get that third paramet
+er
arnsholt nqp: my $s := sub foo($x) { say "Hello!"; return "pong"; }; my $a := $s(); say($a);
p6eval nqp: OUTPUT«Unable to parse blockoid, couldn't find final '}' at line 1␤current instr.: 'parrot;Regex;Cursor;FAILGOAL' pc 1664 (src/Regex/Cursor-builtins.pir:179)␤»
masak it's also a bit mystifying that it worked previously, when we only had two arguments and still three parameters. 16:02
I mean, when we did that, the sub got successfully called.
right now, I'm trying to add a fourth parameter to see if three arguments can call that sub :) 16:03
mathw I'm trying to comprehend what's going on in that action method 16:04
arnsholt nqp: my $s := sub foo ( ) { say "Hello!"; return "pong"; }; my $a := $s(); say($a);
p6eval nqp: OUTPUT«Unable to parse blockoid, couldn't find final '}' at line 1␤current instr.: 'parrot;Regex;Cursor;FAILGOAL' pc 1664 (src/Regex/Cursor-builtins.pir:179)␤»
masak mathw: maybe I can help with that.
mathw: the outermost 'if' checks if we got a name when parsing.
mathw: the 'else' clause of that 'if' actually works :) 16:05
mathw: (and it just calls a sub that returns an EnumMap)
16:05 iblechbot left
mathw I assume something else knows what to do with that 16:05
masak mathw: well, it's a real Perl 6 object.
jnthn nqp: my $s := sub foo ( ) { say("Hello!"); return "pong"; }; my $a := $s(); say($a); 16:06
p6eval nqp: OUTPUT«Hello!␤pong␤»
jnthn arnsholt: ^
masak rakudo: say (enum <a b c d>).values.perl
p6eval rakudo a61c40: OUTPUT«MapIterator.new()␤»
masak (grrr)
rakudo: say (enum <a b c d>).values
p6eval rakudo a61c40: OUTPUT«0123␤»
masak rakudo: say (enum <a b c d>).keys
p6eval rakudo a61c40: OUTPUT«abcd␤»
arnsholt jnthn: Duh. Parens around the say >.<
masak rakudo: say (enum <a b c d>).WHAT
p6eval rakudo a61c40: OUTPUT«EnumMap()␤»
mathw ah of course
jnthn rakudo: say (enum <a b c d>).kv
p6eval rakudo a61c40: OUTPUT«d3a0b1c2␤»
mathw so what's the named enum maker trying to do
it does some dancing around to get the EnumMap 16:07
which I don't really understand
masak mathw: right. it starts by getting the EnumMap to get the names of the keys.
mathw: then it does something to add those keys to a namespace, like A::a, A::b, etc.
jnthn moritz_: If you're able to build Parrot with ICU, you may find that some of the S05-mass/*.t actually work. 16:08
masak mathw: finally, it installs a piece of code to be executed at loading of the script.
jnthn moritz_: I think pmichaud++ did add back \c[...]
masak mathw: that's the call to CREATE_NAMED_ENUMERATION. that's the one that's failing right now, in some way.
16:08 rv2733 left
mathw I don't understand where @BLOCK comes from 16:09
masak mathw: note the 'our @BLOCK'. 16:10
mathw: that means it's defined elsewhere and we simply get access to it from this method.
mathw yes, I know that
but what is it 16:11
masak no idea. :)
it's defined at the top of Actions.pm.
I guess it holds all the blocks that are eventually serialized to PIR.
or something.
jnthn masak: building 16:12
masak nope, the combination of three arguments and four parameters didn't help :) 16:13
so everything points to PAST::Op.new( :pirop('get_namespace P') ) causing the error in some way.
16:14 pmurias joined
mathw what is get_namespace P intended to do? 16:14
masak get the current namespace, the one we want to install A into. 16:15
Trashlord hey guys
masak Trashlord: hi! o/ 16:16
pmurias hi
Trashlord how's it going?
mathw I'll let jnthn worry about if that's right
masak Trashlord: we're trying to get named enums working :) 16:17
mathw examines CREATE_NAMED_ENUMERATION
Trashlord ahg
-g
mathw so you pass the name of the enum, the EnumMap and the scope to install it into
masak Trashlord: no, it's really more 'ahg' right now :P
mathw however that is eventually achieved :)
Trashlord haha 16:18
masak mathw: it seems that installing is achieved through hash accesses.
mathw then you go over each value, mix EnumMixin into it
masak which is very nice. a namespace is a bit like a Hash.
mathw which is an empty role
or is possibly defined elsewhere
masak no, it's defined above.
and it's empty right now.
(because I wanted to get this to work first)
mathw then you stick it into the outer scope under the key name 16:19
or, when it's not commented out and wrong, that is
masak nod.
mathw and then you install the enumeration itself
and presumably that class will also be filled out
masak aye.
mathw what needs to go in those 16:20
masak I'm not 100% sure yet.
but S12 goes into enough detail.
std: Enumeration
p6eval std 30158: OUTPUT«Undeclared name:␤ 'Enumeration' used at line 1␤ok 00:01 107m␤» 16:21
lue ohai masak, mathw, Trashlord, and friends o/ 16:22
masak lue: \o 16:23
lue Now to fix the adverb form of base conversion :)
Trashlord sup
masak lue: planning to attack RT again today? I liked the way you found a closeable yesterday :)
lue masak: did you see my stupendous results with the radcalc function?
masak lue: no, I must have missed that. 16:24
lue I will, in addition to fixing the :16<> type base conversions. (when I has permissions to close bugs that is...)
masak lue: until you have permission, just task someone who does with closing the ticket. they'll often happily oblige :) 16:26
16:26 hercynium left
lue Alright (still trying to find the failure in the adverbs...) 16:27
16:28 Psyche^ joined
mathw is reading S12 and getting an idea of what needs to go in EnumMixin 16:28
although I need to check on what the anon enum creator's doing 16:29
masak mathw: it's in src/glue/enum.pir. it just builds an EnumMap. 16:30
16:30 Patterner left, Psyche^ is now known as Patterner
lue what do $< [name here] > mean/do (including the angle brackets) ? 16:30
mathw rakudo: my $e = enum <a b c>; a.WHAT.say; 16:31
p6eval rakudo a61c40: OUTPUT«Could not find non-existent sub &a␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
masak lue: it's short for $/< [name here] >
mathw doh
masak lue: that's a hash lookup into $/.
mathw wake up
masak mathw: right. an anon enum doesn't install anything.
mathw: it's literally just an EnumMap object.
mathw yeah 16:32
anyway, what I'm thinking about is, once things do install
S12 says that if you define an enum Day <Mon Tue> etc, then Mon.WHAT should say 'Day'
jnthn back, sorry
phone
masak jnthn: it's ok. we held back all brilliant ideas for how to solve this until you were back. :P 16:33
lue And what does $/ hold, I wonder...
mathw lue: the current match object
jnthn masak: Hmm. It makes it to CREATE_NAMED_ENUMERATION but may be dying in signature binding. :-/ 16:34
masak lue: may I recommend S05 as reading at this point?
jnthn: sounds very reasonable.
jnthn rakudo: pir::get_namespace__P()
p6eval rakudo a61c40: ( no output )
jnthn rakudo: my $x = pir::get_namespace__P(); say $x;
p6eval rakudo a61c40: OUTPUT«perl6␤»
lue (I broke radcalc in coding radcalc I believe, back to look (stupid PIR...)) :)
jnthn rakudo: my $x = pir::get_namespace__P(); say $x<IO>;
p6eval rakudo a61c40: OUTPUT«2␤»
jnthn rakudo: my $x = pir::get_namespace__P(); $x<&lol> = sub { say "rofl" }; lol; 16:36
p6eval rakudo a61c40: OUTPUT«Cannot assign to readonly value␤current instr.: '&die' pc 17127 (src/builtins/Junction.pir:399)␤»
lue rakudo: Q:PIR{if 1==1 and 2==2 goto win␤say "boo"␤win:␤say "yay"}
p6eval rakudo a61c40: OUTPUT«error:imcc:syntax error, unexpected PARROT_OP, expecting GOTO ('and')␤ in file 'EVAL_1' line 64␤boo␤yay␤»
mathw masak: what I'm thinking is that I don't think using a mixin role can achieve S12's semantics for the enum values
lue sigh, I guess you can't do that either :/
mathw err
keys
jnthn rakudo: my $x = CREATE_HASH_LOW_LEVEL(pir::get_namespace__P()); $x<&lol> = sub { say "rofl" }; lol;
p6eval rakudo a61c40: OUTPUT«Invalid namespace key in set_pmc_keyed␤current instr.: 'perl6;Proxy;!STORE' pc 14414 (src/builtins/Associative.pir:22)␤»
masak mathw: ok.
mathw err
I've confused myself now
16:37 yves joined
jnthn rakudo: my $x = CREATE_HASH_LOW_LEVEL(pir::get_namespace__P()); say $x<IO>; 16:37
p6eval rakudo a61c40: OUTPUT«2␤»
jnthn rakudo: my $x = CREATE_HASH_LOW_LEVEL(pir::get_namespace__P()); say $x<lolnotexisting>;
p6eval rakudo a61c40: OUTPUT«␤»
jnthn Hmm.
mathw masak: the thing that makes me think this is that enum Day <Mon>; Mon.What eq "Day"; Mon ~~ Int; Mon.perl eq "Day::Mon"; Mon.key eq "Mon"; 16:38
I'm mostly just guessing, but I think it might be necessary to construct a new type at this point 16:39
jnthn masak: Hmm. Treating Parrot namespaces as hashes is...not working out quite as I hoped it may. 16:40
masak jnthn: what do you propose we do?
m6locks rakudo: my @eSa = 1.08635005541779435134e02, 6.57024977031928170135e00; 16:41
p6eval rakudo a61c40: OUTPUT«Multiple Dispatch: No suitable candidate found for 'multiply_float', with signature 'PNP->P'␤current instr.: 'perl6;Perl6;Actions;_block6075' pc 227527 (src/gen/perl6-actions.pir:12321)␤»
jnthn Burn an effigy of Parrot for not coping with subclasses of String, maybe.
lue I know what I did rong :) [compiling...]
mathw port Rakudo to the .NET DLR :P
m6locks what's with that multiple dispatch
jnthn mathw: Oh boy can I not wait to start on the .Net backend.
mathw: But Rakudo * first.
TimToady if parrot would just give us compact arrays of native ints, we could do the rest of strings ourselves
m6locks i want my numbers accurate ;___; 16:42
mathw TimToady: I loved your message on p6l.
jnthn TimToady: Apart from the bit of code I just discovered...will only work if passed a *Parrot* string.
*sigh*
masak: This means that writing this stuff in Perl 6 is going to be...tricky...until I can work out some solution. 16:43
masak jnthn: no wonder it felt tricky! :P
jnthn masak: Sorry. :-(
masak: I...didn't expect the NameSpace PMC to contain...this... :-/
masak jnthn: don't be. it was my idea to try to implement named enums.
m6locks rakudo: my @eSa = 108.635005541779435134, 6.57024977031928170135;
p6eval rakudo a61c40: OUTPUT«PAST::Compiler can't compile node of type BigInt␤current instr.: 'perl6;PCT;HLLCompiler;panic' pc 137 (compilers/pct/src/PCT/HLLCompiler.pir:101)␤»
masak jnthn: I still would really like to see us implement large parts of the enum spec for R*. 16:44
jnthn masak: Me too.
masak: I'm afraid you've just hit on something I'm actually going to have to think about though. :-(
masak mathw: seems your best option for Form.pm in the short term is to do what I do: run off alpha for the time being.
jnthn Rather than having an instant answer for.
masak jnthn: it's ok. 16:45
jnthn mathw: What bits of named enums are you using?
mathw: Are you essentially just using them as constants?
mathw jnthn: yes
jnthn Or do you rely on .pick and all that fancy?
masak: Maybe we could as a stop gap literally compile enums to something like: 16:46
enum A <a b c>; => module A { constant a = 1; constant b = 2; constant c = 3; }
Or some such
masak jnthn: that would be better than nothing, methinks.
jnthn Yeah
Just build up a PAST tree that does it. 16:47
lue colomon told me to run make [spectest file here] to run certain spectests :)
masak tries that
jnthn It's Not Good and very temporary
colomon alpha: my @a; say (1, 2, 3) >>+>> @a
jnthn But it's better than nothing.
mathw it would cover my use case :)
jnthn And should unblock mathw++
lue (for the record, this is the only method that's worked for me)
masak aye.
p6eval alpha 30e0ed: OUTPUT«Use of uninitialized value␤Use of uninitialized value␤Use of uninitialized value␤123␤»
jnthn colomon: Wow!
colomon that's alpha. 16:48
jnthn colomon: Er. I think alpha had that wrong. ;-)
Plz to doing better in alpha. :-)
colomon what should it do, though?
jnthn Oh, for empty?
Oh hmm
colomon (I've got the guts of hyper torn up at the moment...)
jnthn There isn't a last element to use I guess. :-/
jnthn checks So3 16:49
lue I guess I should add support for things like :10('0b1092') #despite it being marked todo in the spectest 16:50
jnthn colomon: Hmm...I don't see a clear answer in the spectests akshually. 16:51
mathw S03 is also fuzzy on it
it talks about too-short on the pointy end, not empty 16:52
jnthn Yeah
You can't duplicate a non-existent last element. :-)
mathw and the too-short extension case is clearly nonapplicable to the empty list
thus I propose that this should cause the compiler to replace one work-related file on the user's computer with a lolcat 16:53
lue rakudo: say :16<'0o10'> # should be 8
p6eval rakudo a61c40: OUTPUT«Malformed radix number at line 11, near "<'0o10'> #"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
jnthn mathw: Well, if TimToady++ will spec it that way, wfm.
;-)
mathw :P 16:54
jnthn colomon: Maybe alpha's response isn't so insane after all in some senses, but we really only want such a warning once, if at all. :-)
masak hm. if I do decide to go to NPW on Iceland, I'll submit a talk and call it "Perl 6: Prince of Parsia" :) 16:55
jnthn oh my.
masak or, 'Parsea", p'haps.
then I can lecture with a towel wound around my head! :D
mathw \o/ 16:56
man
jnthn :-D
mathw I wonder if I can get a grant to go, on the grounds that missing such a sight would be the end of my ability to enjoy life
mathw writes to the NHS 16:57
masak mathw: there'll probably be photos...
16:57 diakopter joined
jnthn probably. ;-) 16:57
mathw masak: but I have to see it in 3D super-HD with a free-roaming viewpoint
lue mathw: Try the IMU. They were able to swindle some cash with Hypercalculus, last I hear.
16:58 Lorn_ left 17:00 dual joined
lue working on RT while figuring out how to complete the radcalc function 17:03
rakudo: sub x{};sub x{1};say x
p6eval rakudo a61c40: OUTPUT«Null PMC access in type()␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
lue rakudo: sub x{};sub x{1};say x; say "It didn't die, just warned us!" 17:04
p6eval rakudo a61c40: OUTPUT«Null PMC access in type()␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
lue rakudo: class A { my $.x = 7; say $.x }
p6eval rakudo a61c40: OUTPUT«Lexical 'self' not found␤current instr.: 'perl6;A;_block45' pc 303 (EVAL_1:136)␤»
lue hey, it doesn't cause a parrot segfault anymore (it seems), so would #57882 be considered closed? 17:06
masak looks
colomon > say (1..5) >>+>> 1 17:07
23456
> say (1..5) >>+>> (1,2)
24567
17:07 Maddingu1 joined
masak lue: seems the segfault is only half of it. 17:08
lue: I don't remember if this was ever discussed on p6l, and if so what the conclusion was.
masak tries to dig up a p6l thread
lue alright, just checking (due to my sometimes very literal nature. The bug sent in was a segfault, not a failure in general. :) )
moritz_ if it was discussed on p6l, there was likely no conclusion :-)
lue Would it be wise for me to join p6l? #prediction: yes 17:09
moritz_ doesn't feel wise enough to assess wisdom
std: /<foo::bar>/ 17:10
p6eval std 30158: OUTPUT«ok 00:01 108m␤»
moritz_ rakudo doesn't parse this => lots of fallout in regex tests 17:11
lots of other regex tests blow up because rakudo doesn't allow named regexes outside a grammar
masak lue: here: www.mail-archive.com/perl6-language...29047.html 17:12
lue: would also be interesting to hear what jnthn has to say about it.
colomon rakudo: say &infix:<+>()
p6eval rakudo a61c40: OUTPUT«No applicable candidates found to dispatch to for 'infix:<+>'. Available candidates are:␤:(!whatever_dispatch_helper , !whatever_dispatch_helper )␤:(Int $a, Int $b)␤:(Rat $a, Rat $b)␤:(Rat $a, Int $b)␤:(Int $a, Rat $b)␤:(Complex $a, Complex $b)␤:(!whatever_dispatch_helper ,
..Any $rh…
moritz_ colomon: we need those for [+] of 0-item list, right? 17:13
lue rakudo: jnthn.say(jnthn.thoughts) 17:14
p6eval rakudo a61c40: OUTPUT«Could not find non-existent sub &jnthn␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
colomon moritz_: right
lue I'll leave it open then :) 17:15
17:17 p6eval left, p6eval joined
diakopter perlito: say 'O HAI' 17:18
p6eval perlito: OUTPUT«O HAI␤»
diakopter (it's the perl5 backend) 17:19
masak \o/
diakopter++
fglock++ 17:20
perlito: say [+] 1..5
p6eval perlito: ( no output )
moritz_ diakopter: if you patched p6eval, please commit the changes into the pugs repo
perlito: class A { has $.x; method foo { say $.x } }; A.new(x => 3).foo 17:21
p6eval perlito: ( no output )
pugssvn r30159 | diakopter++ | [p6eval] and stuff
moritz_ diakopter++ 17:22
diakopter hrm, it needs a custom munger sub
diakopter goes to fix
17:24 nihiliad joined
lue rakudo: $/<a><b>; say "Alive!" 17:25
p6eval rakudo a61c40: OUTPUT«Method 'postcircumfix:<{ }>' not found for invocant of class ''␤current instr.: '!postcircumfix:<{ }>' pc 13740 (src/builtins/Code.pir:120)␤»
lue Huh?
diakopter I think < > is treated the same as { } 17:26
lue rakudo: if 5 > my $a = 4 { say $a } 17:28
p6eval rakudo a61c40: OUTPUT«Use of type object as value in numeric context␤current instr.: 'perl6;Perl6Exception;throw' pc 14682 (src/builtins/Seq.pir:77)␤»
masak right. postfix <> delegates to postfix {}
rakudo: if 5 > (my $a = 4) { say $a } 17:29
p6eval rakudo a61c40: OUTPUT«4␤»
17:30 p6eval left, p6eval joined
lue Should that bug be closed, or is the problem no parens? (62472) 17:32
masak looks 17:33
it can be closed when it has a better error. 17:34
in my view, it doesn't yet.
moritz_ std: if 5 > my $a = 4 { say $a }
p6eval std 30159: OUTPUT«ok 00:01 108m␤»
masak it's *syntactically* correct :)
moritz_ rakudo: warn('foo') 17:37
p6eval rakudo a61c40: OUTPUT«foo␤»
mathw is lured away by the prospect of sausages
lue DON'T PANIC! But do you really need to declare a variable in an if's conditional? Seems pretty faulty...
17:37 p6eval left, p6eval joined
lue there's your error 17:37
diakopter oh, no, I just restarted it
oh, nm
17:37 envi^home left
mathw lue: you'll inevitably find somebody who does 17:38
masak lue: I think I discovered that ticket while actually writing some actual code. but no, it's not always a Good Thing to do... :)
17:38 nihiliad left
mathw Most of the time I'm fairly sure it's a bad idea though 17:38
17:38 nihiliad joined
mathw That doesn't mean there isn't a good use for it 17:38
lue that, IMO, should be illegal in the spec. This would guarantee a result, and that's pretty... interesting. (convice me otherwise) 17:39
moritz_ it's specced what it should do.
there's nothing unclear about it
the variable is visible in the rest of the current lexical scope
and the declaration returns the variable 17:40
and the precedence of the item assignment operator is also specced
lue Well then, maybe a warning? I don't forsee this as a common task...
moritz_ why a warning? it's not like the code looks harmless, but does something unpredictable
mathw if you add a warning ,anybody who does use it has to put up with a warning, so you're effectively forbidding it 17:41
lue Eh, I just don't like it :)
mathw you either have it or you don't
moritz_ if you write a declaration inside an if-statement, you do it deliberately. That's not a typo :-)
mathw well, you're forbidding it or encouraging the ignoring of warnings which is a Bad Thing
masak I can easily give the rationale for a declaration in a while condition.
lue Alright! You convinced me otherwise on warnings/errors
but I still don't see why you'd want to do it that way :) 17:42
17:42 p6eval left, p6eval joined
lue (I bet I will once I need to use it...) 17:42
diakopter ahah.
perlitobug
it doesn't like empty statements 17:43
17:44 p6eval left, p6eval joined, p6eval left 17:45 p6eval joined
masak removes the named-enums branch on github 17:45
jnthn masak: (class vars) can probably be made to work, just a pain.
17:45 p6eval left, p6eval joined
masak jnthn: cool! 17:45
jnthn Well 17:47
Less of a pain than other things that need doing. :-)
diakopter perlito: class A { has $.x; method foo { return $.x } }; my $a := A.new(x => 5); say $a.foo 17:50
p6eval perlito: OUTPUT«5␤»
moritz_ perlito: class A { has $.x; method foo { return $.x } }; my $a = A.new(x => 5); say $a.foo 17:51
p6eval perlito: OUTPUT«*** Error in assignment operation: infix<=> not implemented; use infix<:=> instead at /home/p6eval/Perlito/lib5/MiniPerl6/Grammar.pm line 39.␤»
moritz_ yikes
just like nqp
diakopter perlito: class A { has $.x; method foo { return $.x } }; say A.new(x => 5).foo
p6eval perlito: OUTPUT«Can't call method "foo" without a package or object reference at (eval 3) line 16.␤A=HASH(0x825df0)␤»
moritz_ perlito: class A { has $.x; method foo { return $.x } }; say (A.new(x => 5)).foo
p6eval perlito: OUTPUT«5␤»
diakopter (but aliasing to $a first, works)
oh
moritz_ parsing issue
diakopter nice
moritz_ kp6 had that too 17:52
diakopter nice, to you, I mean
ok, here, I hacked in something that recognizes if the eval input starts with a class
o wait, hang on 17:53
easier solution
moritz_ why?
diakopter b/c perlito requires a Main class
17:54 p6eval left, p6eval joined
diakopter perlito: class A { has $.x; method foo { return $.x } }; class Main { say (A.new(x => 5)).foo } 17:54
p6eval perlito: OUTPUT«5␤»
moritz_ rakudo: Q:PIR{ $P0 = get_hll_global '&warn'␤$P0("foo") } 17:55
p6eval rakudo a61c40: OUTPUT«foo␤»
moritz_ rakudo: Q:PIR{ get_hll_global $P0, '&warn'␤$P0("foo") }
diakopter ok. if you don't define any classes, it's wrapped in 'class Main {' . $inp . '}' but if you define your own class, you must explicitly write out your own Main class, too.
p6eval rakudo a61c40: OUTPUT«foo␤»
17:55 abra left
moritz_ wonders what went wrong locally 17:56
jnthn masak: Any luck with the neamed enums hack?
pugssvn r30160 | diakopter++ | [p6eval] helper wrapper for perlito, for convenience
masak jnthn: no, I looked at it and didn't see how to create the PAST structure.
17:58 ChanServ sets mode: +o diakopter, diakopter was kicked by diakopter (MEDDLING))
moritz_ rakudo: say Q:PIR{ get_hll_global $P0, '&warn'␤$P0("foo")␤ %r = 3 } 17:58
p6eval rakudo a61c40: OUTPUT«foo␤Null PMC access in set_integer_native()␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
moritz_ rakudo: say Q:PIR{ get_hll_global $P0, '&warn'␤$P0("foo")␤ %r = box 3 }
jnthn masak: PAST::Stmts node, and then loop over the hash thingies and make a PAST::Op.new( :pasttype('bind'), PAST::Var.new( :name($enum_element_name) :namespace($enum_name) :scope('package') ), PAST::Val.new( :value($enum_value) ))
p6eval rakudo a61c40: OUTPUT«foo␤3␤»
jnthn Or some such
masak jnthn: oh! ok. 17:59
jnthn Yes, it's a sick hack.
:-)
masak jnthn: I'll try it while making supper.
jnthn Yeah
I'm going out for a wander, and I'll either come back having found supper and eaten it, or with stuff to cook. :-)
Adventure! 18:00
;-)
lue bye jnthn o/
colomon o/
jnthn back later :)
masak o/ 18:02
lue (I just need to add in support for things like :2<0x98> and then unfudged radix.t ought to work :) ) 18:03
(and I need to unfudge radix.t)
masak go lue! \o/ 18:05
colomon \o/ 18:07
lue just went to get some noms (lunch noms) now back to actually do what I said 18:08
18:12 PZt joined 18:14 jaldhar left
masak & # noms 18:15
18:15 masak left
lue rakudo: Q:PIR{$I0=1 and 0␤if $I0 = 1 goto fail␤say "yay"␤fail:␤say "darn"} 18:17
p6eval rakudo a61c40: OUTPUT«error:imcc:syntax error, unexpected PARROT_OP, expecting '\n' ('and')␤ in file 'EVAL_1' line 64␤error:imcc:syntax error, unexpected '=' ('=')␤ in file 'EVAL_1' line 65␤yay␤darn␤»
18:17 abra joined
lue rakudo: Q:PIR{$I0=1␤if $I0 = 1 and 2=2 goto win␤say "fail"␤win:␤say "win"} 18:19
rakudo: Q:PIR{$I0=1␤if $I0 == 1 and 2==2 goto win␤say "fail"␤win:␤say "win"} 18:20
p6eval rakudo a61c40: ( no output ) 18:21
rakudo a61c40: OUTPUT«error:imcc:syntax error, unexpected PARROT_OP, expecting GOTO ('and')␤ in file 'EVAL_1' line 65␤fail␤win␤»
moritz_ lue: PIR does not support infix operators
lue sigh :) 18:25
The more I do this, the more I'd love to port this to P6 afterwards :)
moritz_ why not do it Perl 6 straight away 18:28
lue Because the method was set up for me in pir, and I don't know how in P6 :D 18:30
So I'll do it in PIR first, then study up on P6, and then port it :) 18:31
18:31 jaldhar joined 18:38 jaldhar left, jaldhar joined
lue OK! Time to compile... 18:39
18:40 synth left, [synth] left 18:41 synth joined, synth left 18:43 synth joined 18:44 synth left, sorear left 18:45 sorear joined
dalek kudo: 9780625 | (Solomon Foster)++ | src/core/metaops.pm:
Rework hyper so that it properly handles cases with differing numbers of elements on each side.
 18:46
lue rakudo: Q:PIR{$S0=substr "hello there",5␤say $S0} 18:47
p6eval rakudo a61c40: OUTPUT« there␤»
18:47 nihiliad left
lue just out of curiosity, can you tell rakudo (make) to compile for a different CPU chip? Ex. On an intel machine, tell it to compile for PPC 18:51
moritz_ nope
18:51 abra left, nihiliad joined
moritz_ however the generated parrot byte code is platform independent 18:51
18:52 Psyche^ joined, snarkyboojum joined
moritz_ (in theory - no idea how well it works in practice) 18:52
18:54 snarkyboojum left 18:55 Patterner left, Psyche^ is now known as Patterner 18:56 ruudhaya joined
lue YYYEEEESS! All of radcalc is programmed in, now to spectest (with fudged radix.t) :) 18:56
18:56 ruudhaya left
dalek kudo: 0d10ff6 | moritz++ | src/core/IO.pm:
enable printf() again
 18:58
kudo: 3c9f65b | moritz++ | src/core/operators.pm:
small fix in series operator
19:00 rv2733 joined
lue misc.rakudo returned Dubious... 19:02
(S03-operators)
rakudo: say :16<2_F_A_C_E_D> 19:10
p6eval rakudo a61c40: OUTPUT«3124461␤»
lue bah! I left in an endless loop scenario (thank you spectest) 19:11
*sob* why do fractions fail me now? 19:24
19:27 pmurias left
lue TWO periods? :16<16D.4> is sent to radcalc as 16D..4 ! 19:34
problem with the adverbs *grumble* 19:42
(First big thing I've ever worked on in Rakudo :) )
19:43 snarkyboojum joined 19:44 nobyx joined
jnthn back 19:44
lue++ # persistence! 19:46
lue I can't tell Actions.pm to C<say> things so I find out where the double . comes in :( 19:49
I'm guessing it's in $<fracpart>, but I'm not sure :) 19:50
19:52 iblechbot joined
jnthn lue: pir::say($thing); # will work 19:52
lue in Actions.pm? 19:53
jnthn Yes
lue It started failing around when I added in support for things like :16<0o17>. I'm trying to get all of S02-literals/radix.t to pass :) 19:55
colomon moritz_++ # fixing my wee series bug 19:57
19:57 Maddingu1 is now known as Maddingue
lue Either I fix it Actions.pm (the correct way) or program a workaround in radcalc (the cheat way). I would prefer Actions.pm 19:58
19:58 molaf left
jnthn lue: Yes, fixes are preferable to workarounds. :-) 19:58
They save pain and wtfs later. :-)
lue (aliens to Earth say WTZ :) ) 20:00
colomon I thought it would be WTB? 20:01
20:01 snarkyboojum left
lue What The Zark 20:03
froody. The extra dot is a result of it being with $<fracpart>
if it's always like that, then I can remove the dot we force in ( ~ "." ~ ) 20:04
colomon lue: can you paste the grammar you're using from fracpart now? 20:05
lue the whole thing that sends off to radcalc (in Actions.pm)? 20:06
colomon no, the bit you've changed in Grammar.pm 20:07
Or both parts, maybe!
oh, I see. yeah, take out the extra ~ "." ~ in Actions.pm. 20:08
That was my mistake, apologies.
20:09 jaldhar left
lue SINCE WHEN DID I NEED TO CHANGE GRAMMAR.PM? /o\ 20:09
...sorry :)
colomon I thought you were changing it this morning. 20:11
It does need to be changed a bit eventually, but that's a low priority thing. 20:12
20:12 clintongormley left
lue there we go. fractions work again. (and another run of (fudged) radix.t) 20:13
all pass! now to unfudge (let's see if that's easy) 20:14
20:15 Guest14894 left
lue unfudging is just removing all the #?rakudo stuff, right? 20:19
20:20 jaldhar joined
jnthn lue: yes 20:20
lue alright! 20:21
sorear hello jnthn 20:22
pugssvn r30161 | lue++ | [t/spec] unfudged radix.t for rakudo 20:23
lue Test fails at :60[12,3,64] and stuff like that. I'll leave that for someone else. I'm done :) 20:28
How do I create a patch, again?
colomon lue: commit your changes locally, then use git patch. 20:29
lue git commit ?
colomon git commit, yes. 20:31
git commit -a
Is the easiest way to commit all your changes at once.
jnthn sorear: on phone, moment :-) 20:32
lue then git patch, and nopaste said patch! 20:33
erm, git patch doesn't work :( 20:34
colomon should have left a file in the directory, with a name something like 0001 - your commit message 20:37
but you might need something like
git patch -1
lue 'patch is not a git-command' :(
and no 0001 file :( 20:38
colomon sorry, it's git format-patch
git format-patch -1
will do it, I think. 20:39
lue there it is!
colomon \o/ 20:40
lue now to nopaste! 20:43
20:44 dukeleto joined
lue bah, paste failed. Try Again... 20:48
OK... something's wrong with nopaste :( 20:49
colomon use gist.github.com
nopaste hasn't worked the last three times I tried it. 20:50
:(
jnthn back 20:51
colomon o/
jnthn sorear: oh hai
lue gist.github.com/339561 there it is
you will notice I'm much more liberal with comments than most :) 20:52
arnsholt Anyone in here familiar with Parrot's continuations? 20:53
jnthn arnsholt: mumble
arnsholt: I think I've understood how they work at some point before now. :) 20:54
arnsholt Heh
In that case, if you could take a look at this: nopaste.snit.ch/20025
Each time I call fail() in blob() choose() should backtrack and choose the next element in the list passed to it 20:55
bkeeler Afternoon folks
lue afternoon!
arnsholt Which it does, on the -inside-, but for some reason the return value is always the same
lue (has anyone looked at my patch? :) )
arnsholt Is there something wrong with my code, or is it an actual bug? 20:56
colomon lue: I started looking at it, then got distracted thinking of noms.
jnthn lue: DON'T PANIC! Someone will look at your patch
lue just checking :)
arnsholt Those dastardly noms, always getting in the way =)
jnthn arnsholt: Huh....I have to use my brain... :-) 20:57
jnthn tries to work out what the code does
arnsholt It's supposed to be a non-deterministic searc (I'm fiddling around with doing Prolog for Parrot) 20:58
21:01 payload joined
colomon Did I say this shortly ago? Unobe has a patch to make @a[*-1] work (which it does) but it blows up a bunch of spectests. 21:01
arnsholt jnthn: It's adapted from a Scheme source if you'd find that easier to digest 21:02
jnthn arnsholt: Probably not :-) 21:03
So essentially it does:
say chosen
.return (chosen)
And that's what populates city in
city = 'choose'("la", "ny", "bos")
say city
>
arnsholt Yah
jnthn ?
grr, keybarod!
...that looks very weird.
lue Happens to me too, jnthn. Especially things are being sent to IRC :) 21:04
colomon lue: that is some well-commented code there.
arnsholt The key is the bit above
jnthn It has the right value and returns it...but then the thing it returns to gets a different value. :-/
arnsholt Yeah. I just had an idea
I'm gonna try and remove the .tailcall
lue I developed the habit when beginning my Python NES emulator :)
jnthn oh.
arnsholt: Tail calls do *certainly* have bugs.
We work around one of them in Rakudo. 21:05
sorear Why do tail calls have bugs?
arnsholt No joy, unfortunately
jnthn sorear: The one that bites me is, if you do a call into PIR from C, and that bit of PIR does a tailcall, the return value handling gets messed up and you get a null back instead of what the thing that was tail called to returns. IIRC. 21:06
Probably not the case for arnsholt++ 21:07
But wouldn't surprise me if more bugs were lurking. :-/
arnsholt Yeah. But I get the same result without the tailcall
jnthn arnsholt: Aww. It does look odd. May be worth trying to get it down to a smaller example (though what you have is quite small) and ask some Parrot folks. 21:08
sorear any hope of working TCO in R*?
lue rakudo: my $num = 12345; printf "%d is %d digits long", $num, sub($s, @args is rw) { @args[2] = $s.elems }, 0;
jnthn TCO = ?
p6eval rakudo 0d10ff: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1664 (ext/nqp-rx/src/stage0/Regex-s0.pir:907)␤»
colomon tail call opt, I assume.
arnsholt jnthn: Yeah, I've got some names from the people in #parrot, so right now I'm just waiting for them to show up
jnthn Oh
sorear tail call optimization, where anything evaluated in a returning context automatically becomes a tailcall
jnthn For R*? No. 21:09
Very unlikely.
21:09 smash_ joined
smash_ hello everyone 21:09
moritz_: ping
jnthn We've enough stuff to make *work*, let alone optimize. :-)
sorear it lets you write loops recursively without blowing the stack, and is mandated by r5rs
arnsholt So I figured I could ask in here as well, see if anyone were familiar with the continuations
jnthn sorear: Yeah, I understand the concept, just didn't recognize the acronym. :-) 21:10
sorear: I liked doing that in ML. Write recursive, get iterative memory usage. :-)
We may be able to do it in Perl 6 but it probably screws up CALLER
So it's a bit harder to know when you can do it, I guess. 21:11
arnsholt Hmm. That's one for TimToady methinks
sorear CALLER needs to be rethought anyways in the long run 21:12
since INLINING will screw it up
and that's a slightly important long-run optimization
lue rakudo: ... 21:14
jnthn sorear: Aye
p6eval rakudo 0d10ff: ( no output )
21:15 hudnix joined
lue rakudo: ??? "Hello"; say "<<just a warning" 21:15
p6eval rakudo 0d10ff: OUTPUT«Stub code executed␤<<just a warning␤»
lue rakudo: !!! "Hello"; say "<<just an error" 21:16
p6eval rakudo 0d10ff: OUTPUT«Stub code executed␤current instr.: '&die' pc 17127 (src/builtins/Junction.pir:399)␤»
lue jnthn: remember ... !!! ??? ? Would that be implementable in the new rakudo now ? 21:17
colomon rakudo: say (1, 2, 3) >>~>> 'a' 21:18
p6eval rakudo 0d10ff: OUTPUT«1a2a3a␤»
21:22 rgrau` joined
jnthn lue: I thought you just tried it out? :-) 21:22
colomon: Nice! :-)
colomon rakudo: say (1, 2, 3) >>~>> () 21:23
p6eval rakudo 0d10ff: OUTPUT«No applicable candidates found to dispatch to for 'infix:<~>'. Available candidates are:␤:(Any $a, Any $b)␤␤current instr.: '_block27127' pc 304349 (src/gen/core.pir:19496)␤»
lue the arguments aren't accepted though :(
colomon jnthn: I figured the [op] approach made sense here. But that's easy to change if people feel that is the wrong approach.
lue rakudo: say "foofoofoo".subst("foo", "bar", :x(1..2)) 21:24
p6eval rakudo 0d10ff: OUTPUT«barbarfoo␤»
lue rakudo: say "foofoofoo".subst("foo", "bar", :x(5..6))
p6eval rakudo 0d10ff: OUTPUT«barbarfoo␤»
jnthn colomon: ooh, good point
colomon: Yes, let's try that, until it's discovered to have issues, or the spec goes another way. 21:25
colomon: Maybe I'll take a look at ++<<@a and @a>>++ this evening.
21:25 Chillance joined
lue rakudo: my $a = $b; my $b 21:26
p6eval rakudo 0d10ff: ( no output )
lue rakudo: my $a = $b;
p6eval rakudo 0d10ff: OUTPUT«Symbol '$b' not predeclared in <anonymous>␤current instr.: 'perl6;PCT;HLLCompiler;panic' pc 137 (compilers/pct/src/PCT/HLLCompiler.pir:101)␤»
lue that's still a problem :)
rakudo: class A { my $.b; method x { $!b = "b" } }; A.new.x 21:29
p6eval rakudo 0d10ff: OUTPUT«No such attribute '$!b'␤current instr.: 'perl6;A;x' pc 398 (EVAL_1:173)␤»
jnthn s/my/has/ 21:30
lue just killing bugs on RT :) 21:31
rakudo: class A { has $.b; method x { $!b = "b" } }; A.new.x
p6eval rakudo 0d10ff: ( no output )
moritz_ smash_: pong 21:33
smash_ moritz_: greetings, is it me of only registed users can see the copenhagen annouce in rakudo.org ? 21:34
lue rakudo: class A does Int {}
p6eval rakudo 0d10ff: OUTPUT«Method 'postcircumfix:<[ ]>' not found for invocant of class ''␤current instr.: 'perl6;ClassHOW;add_composable' pc 4039 (src/metamodel/ClassHOW.pir:186)␤»
jnthn smash_: I don't see it while not logged in. 21:35
21:35 wknight8111 joined
smash_ jnthn: yeap, me neither 21:35
moritz_ smash_: I seem to have broken something...
lue afk 21:36
moritz_ but I have no idea what :(
21:37 jhuni joined
smash_ moritz_: sorry can't help you, i don't have any privileges there 21:38
moritz_ and I don't see any options to make it public or so 21:39
I can try to re-post
if that helps...
21:39 molaf joined 21:40 alester joined
jnthn rakudo: my @a = 1,2,3; @a>>++; say @a; 21:41
p6eval rakudo 0d10ff: OUTPUT«Method 'isa' not found for invocant of class 'Undef'␤current instr.: 'perl6;Perl6;Actions;_block5849' pc 225685 (src/gen/perl6-actions.pir:0)␤»
moritz_ I created another announcement node on rakudo.org
that doesn't show up eitehr
and now both show up... wtf? 21:42
colomon woah, lue actually made each error message start with DON'T PANIC! :) 21:43
smash_ moritz_: yeao, seeing two annouces now
moritz_ better than none I think
jnthn colomon: Yeah! :-)
smash_ moritz_: yes, better
jnthn rakudo: my @a = { a => 1}, { a => 2}, { a => 3}; say @a>><a>; 21:44
p6eval rakudo 0d10ff: OUTPUT«123␤»
jnthn wow!
moritz_ w00t
colomon wow, I don't think I knew what was even possible in the spec!
jnthn ...I accidentally the win when I implemented parallel dispatch, evidently.
It's rather cool. :-) 21:45
Well, it's only syntactic sugar for .postcircumfix:<{ }>('a')
So by the time postfixish sees it, it's just a method call.
So the standard parallel dispatch path works. :-) 21:46
It is quite a cool discovery though. :-)
moritz_ wonders if we have tests for that...
dalek kudo: 70dd723 | (David Romano)++ | src/core/Array.pm:
[src/core] indentation fix
kudo: af16e57 | (David Romano)++ | src/core/operators.pm:
[src/core] Define ne in terms of !eq
21:48 Enlik joined
jnthn moritz_: I suspect not. :-) 21:48
Hmm
Guess it works for {..} and [...] postcircumfixes too just automatically. :-)
21:49 Enlik left, Enlik joined
moritz_ hm, lue has unfudge radix.t without patching rakudo - right? 21:52
colomon I'm testing his patch right now.
so if he committed changes to radix.t, then yes. 21:53
moritz_ he did
jnthn Talk about confidence in patch acceptance. :-)
21:54 snarkyboojum joined
colomon I'm going to run out to pick up noms (spicy beef with peanuts, woo-hoo) while the test runs, so even if it works I'll be a bit slow to push. 21:55
jnthn colomon: Ooh, sounds nice. :-)
jnthn had beef with curry and veg tonight :-)
21:56 Enlik left
colomon from this place, it's usually heavenly. 21:56
pugssvn r30162 | moritz++ | [t/spec] fudge S04-statement-modifiers/while.t for rakudo
r30163 | moritz++ | [t/spec] test parallel dispatch on postcircumfix operators
21:56 Enlik joined, payload left 21:57 Enlik left, Enlik joined 21:58 pugssvn left
moritz_ thinks that pugssvn-- behaves very weirdly 22:02
22:02 pugssvn joined
moritz_ I just regenerated the locales on feather3 in the hope that it helps... 22:03
lue I wanted to test an unfudged radix.t against my patch, and didn't feel like a simple cp when I'd upload it anyway :)
moritz_ lue: you can test it locally unfudged without committing
lue: commiting tests that fail for everybody else is considered rather rude here. Plesae don't do that again 22:04
lue colomon: you know me! I'm going to start each message with DON'T PANIC! (you can change if you're really against it)
mortiz_ alright :/
22:05 eternaleye left
sorear pcc_hackathon isn't merged yet :( 22:05
22:06 nobyx left
pugssvn r30164 | moritz++ | [t/spec] test that .HOW is readonly 22:06
jnthn sorear: I think merging it would also make Rakudo fail to build.
sorear: I asked on list for a patch to fix up Rakudo. 22:07
lue I'll work on improving bits of radix conversion in a while (including a port over to P6) :)
sorear jnthn: a feature I need for Blizkost (slurpy return signatures in Parrot_pcc_invoke_sub_from_c_args) is only available in the branch 22:09
jnthn sorear: Aww. :-(
sorear it's ok, #parrot thinks I should be writing an entire C library to wrap libperl.so, then wrapping that with the NCI, then writing a binding layer in PIR 22:10
jnthn ...
That...wasn't the plan I'd had in mind for Blizkost. 22:11
And sounds like a chunk more work.
sorear Because, according to chromatic, the NCI code generator is the only thing that should be making PCC calls
jnthn er, wtf? 22:12
sorear I probably just failed to communicate again
jnthn I thought the whole reason for having a unified model for PCC and the pain we all went through to move to one was so there _was_ an interface to do those things.
Perhaps. 22:13
Rakudo relies plenty of the ability to work with the PCC interface, for its multi dispatcher and signature binder, and I've never heard complaints about those along the same lines.
pugssvn r30165 | moritz++ | [t/spec] in Test::Util, use Test; in order to have sub skip() available 22:14
jnthn moritz_++ # That allows the tests that use Test::Util to have a chance of working again, I guess? 22:15
moritz_ jnthn: currently these tests emit
ok 1 - # SKIP test died: Not enough positional parameters passed; got 2 but expected 4
jnthn Oh. :-/ 22:16
moritz_ oh, it relies on $^a and $^b which seems not relieable 22:17
next problem:
rakudo: say $*EXECUTABLE_NAME
p6eval rakudo 0d10ff: ( no output )
moritz_ alpha: say $*EXECUTABLE_NAME
p6eval alpha 30e0ed: OUTPUT«../rakudo-alpha/perl6␤» 22:18
jnthn Probably just NYI. 22:19
moritz_ aye
jnthn Should be an easy "copy from alpha" job.
moritz_ it's in src/builtins/globals.pir in alpha 22:21
and in master @*INC is set up in src/Perl6/Compiler.pir
what would you prefer? globals.pir?
jnthn But it where e.g. $*PROGRAMENAME is set up 22:22
moritz_ it's not, in master
jnthn Don't think we have a globals.pir in master
moritz_ rakudo: say $*PROGRAMNAME
jnthn oh, I though mberends++ did that.
p6eval rakudo 70dd72: ( no output )
jnthn oh.
rakudo: say $*OS
p6eval rakudo 70dd72: OUTPUT«linux␤»
moritz_ rakudo: say %*ENV.perl
p6eval rakudo 70dd72: OUTPUT«Method 'key' not found for invocant of class 'String'␤current instr.: 'perl6;EnumMap;_block14104' pc 408062 (src/gen/core.pir:58659)␤»
jnthn :-/
22:23 eternaleye joined
sorear rakudo: say ::GLOBAL.keys.perl 22:23
p6eval rakudo 70dd72: OUTPUT«Null PMC access in find_method('keys')␤current instr.: '_block14' pc 29 (EVAL_1:0)␤»
sorear what's the correct syntax for dumping all $*?
moritz_ src/cheats/setup-io.pm probably
jnthn rakudo: say $*PROGRAM_NAME 22:24
p6eval rakudo 70dd72: OUTPUT«/tmp/sQndXSIIeZ␤»
jnthn ah
Underscore :-)
moritz_: src/glue/run.pir
lue is off to correct some of his errors in fixing radix conversion 22:26
22:26 IllvilJa joined
lue colomon: did you upload the patch yet? I'm off to correct my eagerness in unfudgeing radix.t :) 22:27
moritz_ jnthn: trying to add $EXECUTABLE_NAME in run.pir... 22:28
jnthn +1 22:29
pugssvn r30166 | moritz++ | [Test::Util] simplify code a bit 22:32
jnthn > my @a = 1,2,3; @a>>++; say @a; 22:36
234
> my @a = 1,2,3; ++<<@a; say @a;
234
moritz_ now say @a>>++ 22:37
and say ++<<@a
jnthn cheeky :-P
> my @a = 1,2,3; say @a>>++;
123
> my @a = 1,2,3; say ++<<@a;
234
moritz_ impressed
pugssvn r30167 | moritz++ | [Test::Util] further simplification; now runs on Rakudo
dalek kudo: 12b6d27 | moritz++ | src/glue/run.pir:
re-instatiate $*EXECUTABLE_NAME
moritz_ anyway, bed time 22:38
jnthn :-)
moritz_ 'night
jnthn moritz_++
night o/
lue night o/
jnthn spectests 22:39
22:41 huf left, huf joined
jnthn Anyone know if the S03-operators\precedence.rakudo test fail is an existing one? 22:51
jnthn is sure he's heard about it before...
lue rebooting to try and get things working again. back in a bit! 22:53
23:03 diakopter joined, iblechbot left 23:04 Enlik left
jnthn
.oO( /kick diakopter ABSENCE )
 23:04
dalek kudo: 92745eb | jonathan++ | src/ (3 files):
First cut of prefix and postfix hyper-operators. Parsing should all be fine, implementation probably needs tweaking for nested arrays.
 23:06
23:06 rgrau` left
diakopter jnthn: ;P 23:07
rakudo: say [+]++[+] 1 23:21
p6eval rakudo 12b6d2: OUTPUT«2␤»
jnthn std: say [+]++[+] 1 23:22
p6eval std 30167: OUTPUT«===SORRY!===␤Prefix requires an argument at /tmp/EHGW94kCDp line 1:␤------> say [+]++[+⏏] 1␤FAILED 00:01 108m␤»
jnthn Hm.
:-/
diakopter std fail?
jnthn Or rakudo fail :-)
diakopter oh, no. yeah. 23:23
jnthn I'm not sure what it'd mean if it did work.
diakopter seems like it parsed it entirely right-associative, with which I don't see a problem
er, prefix.
jnthn starts to see what it means :-) 23:24
[+](++([+]1))
Which...gives 2. :-)
diakopter rakudo: say [+](++([+]1)) 23:25
jnthn Well, that's just an obfuscator's dream. :-)
p6eval rakudo 12b6d2: OUTPUT«2␤»
jnthn I'm not sure why STD doesn't like it.
diakopter rakudo: say +[+]++[+]+1++ 23:26
p6eval rakudo 12b6d2: OUTPUT«2␤»
diakopter std: say +[+]++[+]+1++
p6eval std 30167: OUTPUT«===SORRY!===␤Prefix requires an argument at /tmp/RlYzOf8ZvO line 1:␤------> say +[+]++[+⏏]+1++␤FAILED 00:01 108m␤»
23:26 kst joined
diakopter mebbe std doesn't like nested reductions (without parens or other brackets) in general 23:27
std: say +[+]++([+]+1++)
p6eval std 30167: OUTPUT«ok 00:01 106m␤»
diakopter std: say +[+]++ [+]+1++ # o_O 23:28
p6eval std 30167: OUTPUT«ok 00:01 106m␤»
23:30 crythias joined 23:31 wallberg left
diakopter ö_ö 23:31
rakudo: say +[+] ++ [+]+1++ 23:32
p6eval rakudo 12b6d2: OUTPUT«2␤»
diakopter jnthn: O_O
jnthn :-/
diakopter std: say ++ 4
p6eval std 30167: OUTPUT«ok 00:01 106m␤»
diakopter std: say 4 ++
p6eval std 30167: OUTPUT«===SORRY!===␤Postfix found where infix expected (omit whitespace?) at /tmp/GTTMNkVqnY line 1:␤------> say 4 ⏏++␤ expecting any of:␤ bracketed infix␤ infix␤ infix or meta-infix␤FAILED 00:01 107m␤»
jnthn rakudo: say ++4 23:33
p6eval rakudo 12b6d2: OUTPUT«Cannot assign to readonly value␤current instr.: '&die' pc 17140 (src/builtins/Junction.pir:399)␤»
jnthn rakudo: say 1++
p6eval rakudo 12b6d2: OUTPUT«Cannot assign to readonly value␤current instr.: '&die' pc 17140 (src/builtins/Junction.pir:399)␤»
jnthn ...how on earth you're getting away with it above, I've no idea. :-)
diakopter where'd the intervening ++ go, then
yeah
jnthn Well, --target=parse is your friend if you want to see what Rakudo's parsing it as. :-)
diakopter and ... I didn't realize whitespace was allowed between prefix...
jnthn Yes 23:34
Even between the << bit in a hyper one.
diakopter ok, so it's pre-not-quite-fixed
jnthn std: my @a; say ++<< @a;
p6eval std 30167: OUTPUT«ok 00:01 106m␤»
jnthn :-)
preclose
diakopter prenear
jnthn preish 23:35
oh noes, $*LEFTSIGIL gets tweaked inside EXPR. :-/
diakopter std: my @a; say -<< @a;
p6eval std 30167: OUTPUT«ok 00:01 106m␤»
23:35 lue joined
jnthn rakudo: my @a = 1,2,3; say (-<< @a).perl 23:36
p6eval rakudo 12b6d2: OUTPUT«Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 11␤current instr.: 'perl6;Regex;Cursor;FAILGOAL' pc 1664 (ext/nqp-rx/src/stage0/Regex-s0.pir:907)␤»
diakopter dunno what prefix - does
jnthn rakudo: my @a = 1,2,3; my @b = -<< @a; say @b;
p6eval rakudo 12b6d2: OUTPUT«Confused at line 11, near "my @b = -<"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
diakopter rakudo: my @a = 1,2,3; my @b = -<< @a; say @b; 23:37
p6eval rakudo 12b6d2: OUTPUT«Confused at line 11, near "my @b = -<"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
jnthn negation
diakopter rakudo: my @a = 1,2,3; my @b = +<< @a; say @b;
p6eval rakudo 12b6d2: OUTPUT«Confused at line 11, near "my @b = +<"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
diakopter rakudo: my @a = 1,2,3; my @b = !<< @a; say @b;
p6eval rakudo 12b6d2: OUTPUT«Confused at line 11, near "my @b = !<"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
jnthn oh
23:37 JimmyHoff left
jnthn wonder if the bot is up to date 23:37
rakudo: my @a = 1,2,3; say ++<<@a; # know this one works locally
p6eval rakudo 12b6d2: OUTPUT«Confused at line 11, near "say ++<<@a"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
jnthn ah
ETOOSOON.
lue colomon: ping
colomon pong 23:38
just looking at the test results from your patch
diakopter pugs: my @a = 1,2,3; say ++<<@a;
p6eval pugs: OUTPUT«<Error><Error><Error>␤»
jnthn Wow!
diakopter pugs: my @a = 1,2,3; say -<<@a;
p6eval pugs: OUTPUT«-1-2-3␤»
lue I'm off to correct my eagerness in unfudgeing all of radix.t :/ 23:39
diakopter "and that's when she travelled forward in time..."
colomon at least one of the tests, you're right and the test is wrong. 23:40
jnthn That puts a new slant on test fail.
lue I am?
colomon is(:8('0b1110'), 0o14, ':8(0b1110) converts from decimal');
should be Oo16 for the answer, I believe. 23:41
1110 bin is 14 decimal, not 14 octal
diakopter yeah but
colomon actually all those answers are wrong in that section.
diakopter it's supposed to use the string value of the number
no
it's spec 23:42
colomon diakopter: can you be more specific about what the spec says?
lue I plan on porting radcalc o'er to P6 when it's a good time to do so (it would be less code than PIR :) ) 23:43
colomon the :16() tests definitely assume the inner base overrides the radix.
lue rakudo: say :16(0b110) # 6
p6eval rakudo 12b6d2: OUTPUT«6␤»
diakopter I don't know. when I implemented them in sprixel, I understood it. but I have no certainty that radix.t hasn't been changed since then. 23:44
lue One thing I left out is the :60[34,59,2] type stuff :/
jnthn oh no!
I...just pushed a patch that included something it wasn't meant to. :-/
diakopter o dears 23:45
lue rakudo: say :60[59] # should be 59
p6eval rakudo 12b6d2: OUTPUT«Malformed radix number at line 11, near "[59] # sho"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
lue perl6: say :60[59] # should be 59
p6eval pugs: OUTPUT«59␤»
..rakudo 12b6d2: OUTPUT«Malformed radix number at line 11, near "[59] # sho"␤current instr.: 'perl6;HLL;Grammar;panic' pc 500 (ext/nqp-rx/src/stage0/HLL-s0.pir:328)␤»
..elf 30167: OUTPUT«Unknown rule: rad_number␤It needs to be added to ast_handlers.␤ at ./elf_h line 2850␤»
lue colomon: considering you're testing my patch, would it be aggravating to you if I started switching over to P6 code? 23:46
dalek kudo: 53707fb | jonathan++ | src/Perl6/Grammar.pm:
Start to scatter various $*LEFTSIGIL bits around, in prep for item assignment. This covers all cases in STD apart from those in EXPR.
 23:47
lue (it'd take a while, but...)
jnthn The above patch accidentally also enables the check for item assignment. I'm spectesting now to see if there's fallout.
No point undoing it if it works.
Tene or if it can be made to work. 23:48
colomon lue: if you do that, let me work on getting the fudging correct for radix.t.
jnthn Tene: I didn't realize that I'd also saved the bit that checked it - I'd wanted to make sure adding the various assignments to $*LEFTSIGIL hadn't broken anything. :-)
Tene nods. 23:49
jnthn And then was prepping to spectest that as the next step. :-)
.oO( I accidentally the item assignment. )
lue will do colomon
jnthn rakudo: my $a = 1, 2, 3; say $a; 23:50
p6eval rakudo 92745e: OUTPUT«1 2 3␤»
jnthn That's what's wrong and what the item assignment changes should fix.
diakopter colomon: ... nor if radix.t was changed, whether it was changed correctly
jnthn lol 23:51
> my $a = 1, 2, 3; say $a;
1
\o/
Let's see what it makes of the test suite. :-)
colomon lue: your patch looks solid -- at least, the test failures are not at all your fault. lue++ 23:53
sorear jnthn: RTLD_GLOBAL is in.
jnthn sorear++ 23:54
\o/
diakopter rakudo should rebuild in 6 minutes..
should start rebuilding I mean
23:54 dual left
sorear I was about to say 23:54
colomon eval('2«1.1*:2<10>**:2<10>»') ????!? 23:55
23:55 dual joined
lue 6 minutes? That would not only be fast, but quite serendipitous 23:55
jnthn colomon: huh?! 23:56
:-)
colomon std: 2«1.1*:2<10>**:2<10>»
sorear 6 minute rebuilds for rakudo would be extremely awesome
p6eval std 30167: OUTPUT«ok 00:01 105m␤»
lue It's a french quote test :) (I know...)
(I didn't do it!) 23:57
23:57 justatheory joined
lue .sub '&radcalc' in pir would be multi method radcalc( in p6, correct? 23:57
23:59 diakopter left

Powered by Cro and the Raku® Programming Language.

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

Thank you!