»ö« | perl6.org/ | nopaste: paste.lisp.org/new/perl6 | evalbot usage: 'perl6: say 3;' or rakudo: / pugs: / std: , or /msg p6eval perl6: ... | irclog: irc.pugscode.org/ | UTF-8 is our friend!
Set by moderator on 20 October 2009.
diakopter TimToady: do you see any downside to emulating the compile-time for each block instead of the whole tree at once? 00:00
thoe more I think about it, the more it seems ok to me.
00:02 dalek left
quietfanatic I want a way to find the class of a method. 00:03
00:03 dalek joined
quietfanatic I mean, the class a method belongs to. 00:03
lisppaste3 Wolfman2000 pasted "Inputs 41, 97, 17, "Hi!" (no quotes) for this RSA code gives 003200330032. I suspect I made a mistake here." at paste.lisp.org/display/90873
quietfanatic Is this possible? 00:04
00:06 middayc_ joined
quietfanatic If so, it will simplify Events to a large degree. 00:07
TimToady diakopter: what do you mean by 'emulating'? 00:20
diakopter I mean, evaluating 00:22
00:23 synth joined 00:24 middayc__ joined
japhb Is there a perl6doc that is A) working, B) reasonably quick, and C) up to date with the spec? 00:25
TimToady diakopter: I still don't understand what you mean. most compile-time evaluation must happen immediately at BEGIN time 00:30
Wolfman2000 rakudo: my Str $fin; $fin = 'AFB'; say $fin; 00:31
TimToady trait handlers run the moment they're recognized and modify they're declarand immediately
p6eval rakudo 7347ec: AFB␤ 00:32
TimToady *their
Wolfman2000 ...I wonder why I'm getting AFBStr() instead of that
00:33 justatheory joined
quietfanatic TimToady: Does a method know what class it's in, ot does that constrain them too much? 00:35
colomon surely self.WHAT would do the job? 00:36
quietfanatic *or
No, I mean given the method as an object, not the code inside the method.
colomon oh 00:37
TimToady no, a method is just a sub that happens to have a reference installed into a metaclass, and could in fact be installed in several at once 00:38
like a Unix inode
quietfanatic I see
TimToady which is also what allows a method to be exported as a multi 00:39
quietfanatic What I need is basically a way, given a method, to find all the objects it would be called on. 00:40
The problem is, that 'would' is not so well defined anymore.
TimToady your brain must be in upside-down 00:41
you keep wanting to follow references backwards :)
quietfanatic Yes well, events are backwards like that. 00:42
You kinda have to think upside-down in order to implement them.
TimToady events usually require a registry of handlers somewhere
quietfanatic I have that
I also need a registry of objects.
TimToady or you think you do 00:43
quietfanatic Simply put, if a routine is like a GOTO then an event is like a COME FROM.
Or a registry of classes that have their registry of objects.
Each object needs to respond to events individually.
And most events only affect one object. 00:44
I also don't want to register a pile of handlers every time an object is created.
TimToady you want the system to do it for you :P 00:46
quietfanatic I want to do as few operations when an object is created as possible.
Ideally, just one push onto an object registry. 00:47
TimToady yes, well, then events are going to be slow, unless you have a Content Addressable Memory sitting around as a co-processor
quietfanatic I understood the first half of that :) 00:48
TimToady but you're basically asking for magic from the system
quietfanatic No it doesn't have to be like that.
TimToady there aren't that many approaches to inverting data 00:49
quietfanatic Each object in a given class will have exactly the same event handlers, so I should only have to register the handlers for the class.
and register the object once for the class
That will be everything required.
00:49 eternaleye joined
TimToady so you need classes that also keep a collection, but what about subtypes? 00:50
quietfanatic you mean subsets, or inheriting classes?
TimToady inheriting
quietfanatic Inheritance is tricky, yes
Or more specifically, inheritance isn't tricky; it's overriding that's tricky. 00:51
00:52 eternaleye joined, middayc joined
quietfanatic One possible approach is to copy all the event handlers from the parents to the children when they are declared. 00:53
and check events only for the immediate instances of each class 00:54
TimToady sounds more like role composition than inheritance
since roles pull in methods but flatten them at composition time
quietfanatic Oh they do?
Hmm
But 00:55
It should ideally work for both role composition and class inheritance.
00:56 yahooooo joined
quietfanatic so if class A does role X, and class B is A, then does the role X get flattened into B as well? 00:57
TimToady well...we've talked about maybe if you "do" a class it takes a snapshot of it and makes an anonymous role
or maybe there's some other way to thread your inheritance through your role structure
quietfanatic In that case roles and classes would be pretty much the same.
TimToady I believe it's specced that a role can supply an 'isa'
quietfanatic Yes...
TimToady it imposes an immutability constraint, yes 00:58
quietfanatic what imposes an immutablitiy constraint on what?
TimToady classes are mutable, roles are not
quietfanatic oh 00:59
TimToady which is why deriving a role from a class would have to snapshot it
but snapshot is what you're doing if you 'copy methods down' 01:00
you wouldn't know to copy down methods that are monkey-typed in to the parent class 01:01
the parent would have to push them
which implies every class has to register with its parent
also, copy-down might not work with private attributes 01:02
quietfanatic That's the other approach
TimToady infrastructure for a parent class might be out-of-place in a child class
quietfanatic How do you mean? 01:03
TimToady suppose a parent method refers to $!foo, the child class will potentially have its own $!foo that is separate
in that sense, a method kinda knows which class its in somehow 01:04
quietfanatic Not necessarily
it could just have a reference to the...wait
the attribute $!foo is seperate for all instances of both classes 01:05
Oh wait you mean
if the parent has $!foo and the child has $!foo each object has two seperate $!foos?
TimToady yes 01:06
quietfanatic In which case the parent method refers to the parent-defined $!foo...
But then what if you copy that method into another class?
TimToady well, depending on the implementation, it might or might not get confused 01:07
quietfanatic does it use that class's $!foo?
TimToady unspecced either way, so erroneous
quietfanatic Hmm
TimToady or at least non-portable
quietfanatic But there's no way of knowing what's in the body of a method when you copy it
TimToady indeed 01:08
quietfanatic in which case copying methods around is almost always a bad idea
TimToady well, if $!foo magically knows the original class, then it would just be a form of delegation, like calling nextsame 01:09
quietfanatic in which case methods may as well know what class/role they belong to.
I guess so
TimToady yes, basically class-of-origin needs to be part of the key under which a $!foo is stored 01:10
a trick people use in Perl 5, btw
quietfanatic Yeah
TimToady $self->{__PACKAGE__ . 'key'} or some such
though that's assuming named classes
quietfanatic Then the method doesn't know where it's from after all, it's the variable that knows
colomon wishes pmichaud would lock down the Iterator interface he wants to use so they could start being implemented... 01:11
TimToady would be more like .WHAT.WHICH
er, no, that's dynamic
well, $?CLASS should be WHICHish
Wolfman2000 rakudo: my $dec = "00650097"; say ($dec.comb(/<xdigit>**4/).map: {chr(:16($_))} ).join(''); 01:12
p6eval rakudo 7347ec: e␤
TimToady well, maybe the internal call knows it
Wolfman2000 ...I don't think that's supposed to be the answer.
quietfanatic Wolfman2000: That looks right to me 01:13
Wolfman2000 ...got it fixed on Feather. original problem: needed a space between : and { by map
Wolfman2000 is finally understanding some of the power of map 01:20
01:21 nihiliad joined
Wolfman2000 I forgot: is there a fast method to convert from integer values to hex? 01:22
rakudo: say ord(65); 01:23
p6eval rakudo 7347ec: 54␤
Wolfman2000 rakudo: say ord(32);
p6eval rakudo 7347ec: 51␤
Wolfman2000 not what I'm thinking...
01:23 jaldhar joined
Wolfman2000 needs dinner. 01:24
diakopter rakudo: BEGIN { a() }; sub a() { say 4 } # does this work?
p6eval rakudo 7347ec: Could not find non-existent sub a␤in Main (file src/gen_setting.pm, line 324)␤
01:28 ashizawa joined
diakopter nqp: a(); sub c() { our sub a() { say(4) } } 01:29
p6eval nqp: 4␤
diakopter sigh.
01:35 nihiliad joined
TimToady yeah, that shouldn't work 01:36
cotto Well, the language isn't called "Exactly the Same As Perl 6". ;) 01:38
diakopter nqp: a(); { our sub a() { say(4) } }
p6eval nqp: 4␤
diakopter std: a(); sub c() { our sub a() { say(4) } }
p6eval std 29167: ok 00:01 105m␤
TimToady oh, wait, I thought that was ng: 01:39
it's okay for nqp
std should know better though
diakopter should it? that seems much farther up the impl ladder
TimToady obviously looking in the package for 'a'
but multi dispatch only looks in lexical scopes these days 01:40
so does only dispatch :) 01:41
01:46 jaldhar joined
ng_feed rakudo-ng: colomon++ 01:49
rakudo-ng: Rework Range and RangeIterator to properly handle Ranges which exclude their ends and non-numeric Ranges.
TimToady ng: a(); { our sub a() { say(4) } } 01:51
p6eval ng 071cdd: 4␤
TimToady well, same problem
calling package subs is a hard habit to break :)
maybe one could argue it's okay as a last resort, maybe 01:52
ng: a(); { our sub a() { say(4) } }; my sub a() { say(-1) } 01:53
p6eval ng 071cdd: -1␤
TimToady that's good
01:56 _eMaX_1 joined, Jedai joined, estrai joined
diakopter ng: a(); { our sub a() { say(4) } }; my sub a() { say(-1) }; my sub a() { say(6) }; 02:02
p6eval ng 071cdd: -1␤
02:03 nihiliad joined
colomon Would it make sense to have a Test.pm function that took an Iterator and a list and checked that the Iterator's output matched the list? 02:03
colomon is struggling to figure out how to test Iterators without a pinned down standard for them, and this seemed like it might abstract the trouble issues out of the test code. 02:04
TimToady std: a(); { our sub a() { say(4) } }; my sub a() { say(-1) }; my sub a() { say(6) }; 02:05
p6eval std 29167: [31m===[0mSORRY![31m===[0m␤Illegal redeclaration of routine 'a' (from line 1) at /tmp/LEQOF88TwA line 1:␤------> [32m4) } }; my sub a() { say(-1) }; my sub a[33m⏏[31m() { say(6) };[0m␤ expecting new name to be defined␤FAILED 00:01 107m␤
TimToady colomon: foo() ~~ (1,2,3) should do that 02:06
list smartmatches are supposed to match the whole thing 02:07
colomon TimToady: That's smart matching an iterator against a list?
afk for sec, got to prep bottle
TimToady if listify any iterator, it should iterate it into a list 02:08
colomon hmm... so the answer is Perl 6 does this automatically for you. 02:10
and the problem is I don't think that's implemented yet in ng. :| 02:11
diakopter rakudo: a(); { our sub a() { say(4) } }; my &a = sub { say(-1) };
p6eval rakudo 7347ec: 4␤
TimToady can you say $iterator.list?
colomon For sure no, since I wrote the iterator and there was nothing for it to derive from. 02:12
Should an iterator have a .list method?
(and should a Range have a .list method?)
TimToady dunno
I didn't write that part :) 02:14
colomon I thought it all sprung fully formed from your head?
;)
TimToady that was Athena, not Perl 02:15
colomon And I suppose you claim you are not Zeus?
TimToady well, if I get struck by a lightening bolt, you'll know :) 02:17
*lightning
02:19 orafu joined
eternaleye A bolt from the blue turns TimToady's har white! 02:26
*hair
02:30 Rolo joined 02:32 Avada joined 02:33 orafu joined, _eMaX_1 joined, Jedai joined, estrai joined
japhb nqp: class Foo { method foo () { $method := "bar"; self.$method(); }; method bar { say("Hello from method bar"); } }; my $o := Foo.new; $o.foo; 02:34
p6eval nqp: Unable to parse blockoid, couldn't find final '}' at line 1␤current instr.: 'parrot;Regex;Cursor;FAILGOAL' pc 1630 (src/Regex/Cursor-builtins.pir:179)␤
japhb pmichaud: How do I do that ^ ? I need to be able to call a method via a computed name ...
diakopter it's not JavaScript (where you could do self[$method]()).. you'd need eval, I think. 02:37
you need a 'my ' before $method, and '()' after 'method bar' 02:38
also no self.$method, afaik
02:38 TiMBuS joined
Tene japhb: you should be able ot introspect the class, pull the method out of it, and invoke that as a method. 02:39
japhb: $obj.^methods, iirc
japhb Tene, I need to do it in NQP. Which means P6object would have to support that. 02:42
japhb will go spelunking for that in a bit
diakopter, I'm trying to avoid eval, as I'm not entirely sure of the semantics WRT to 'self' through an eval, but I'll keep that in mind as a backup. 02:43
Tene japhb: .^methods == .WHAT.methods or .HOW.methods or something, so you can use that if nqp doesn't support the .^foo syntax 02:44
japhb Tene, OK, thanks, that gives me a couple more places to look. 02:45
I have to go AFK to get kids to bed, but any other ideas appreciated -- I'll backlog. 02:46
02:48 cognominal joined, bluescreen joined, wolverian joined, pnate joined, stepnem joined, p6eval joined, jjore joined
lisppaste3 colomon pasted "colomon's wishlist for ng" at paste.lisp.org/display/90881 02:50
colomon Just in case pmichaud or jnthn feel like working miracles overnight.... :) 02:51
02:52 rfordinal3643 joined
eternaleye japhb: $obj."$methodname" 02:54
Wolfman2000 Back from dinner. 02:55
colomon: You aren't offended that I am probably not able to pull off those miracles, right?
colomon Wolfman2000: I know that I can't. Doubt there are even as many as ten people out there who could... 02:57
Time for bed.
02:59 wolverian joined 03:00 eternaleye joined 03:01 stepnem joined
eternaleye Tene: $obj.^method is $obj.HOW.method( $obj ) 03:01
And .^methods gives a List of Methods IIRC
Wolfman2000 ...and the rakudo bot is down 03:02
eternaleye: Would you know off hand of a built-in function to convert decimal to hex, or is that something I have to do manually?
Tene Wolfman2000: the .fmt method
rakudo: say 14.fmt("%x"); 03:03
03:03 rfordinal joined
Wolfman2000 Tene: rakudo is down 03:03
Tene ;_;
eternaleye Wolfman2000: Is this p6 or p5?
Wolfman2000 p6
eternaleye Ah
03:03 cognominal joined, bluescreen joined, pnate joined, p6eval joined, jjore joined
eternaleye wait, what? 03:04
Did p6eval get voiced before the netplit ended, when it wasn't here?
Wolfman2000 ...not on my screen. 03:05
Tene rakudo: say 14.fmt("%x");
eternaleye Quassel--
Tene rakudo: say "wtf... why is p6eval not responding?" 03:06
eternaleye Wolfman2000: Also, when I disconnected a bit ago, that was due to Postgres dying completely. Don't know it it was a segfault or what.
*if
nqp: say( 1 ) 03:07
Hm, so it's all languages
03:08 orafu joined, _eMaX_1 joined, Jedai joined, estrai joined, elmex joined
p6eval rakudo 7347ec: e␤ 03:10
rakudo 7347ec: wtf... why is p6eval not responding?␤
nqp: 1␤
eternaleye Epic lag is epic
Wolfman2000 I had the same lag.
rakudo: say "Hi!"; 03:11
p6eval rakudo 7347ec: Hi!␤
Wolfman2000 ...yeah, shouldn't take 7 seconds to say hi
03:11 facsimile joined
eternaleye Wolfman2000: It took 5 _minutes_ to follow Tene's instructions 03:11
Tene rakudo: say 15.fmt("%x"); 03:12
p6eval rakudo 7347ec: f␤
Wolfman2000 Tene: thanks for confirming
Wolfman2000 is cleaning up some of his rsa.pl code for his class presentation on Cryptography Monday. 03:13
lisppaste3 Wolfman2000 pasted "Is there a better way than using the range iterator?" at paste.lisp.org/display/90882 03:16
03:21 frew_ joined 03:35 nihiliad joined
Wolfman2000 rakudo: say 5 % 2|3; 03:46
p6eval rakudo 7347ec: any(1, 3)␤
Wolfman2000 ...wasn't what I was after
then again, maybe it is 03:47
rakudo: say "true" if 5 % 2|3;
p6eval rakudo 7347ec: true␤
Wolfman2000 rakudo: say "true" if 8 % 2|3;
p6eval rakudo 7347ec: true␤
Wolfman2000 rakudo: say "true" if 8 % 2&3;
p6eval rakudo 7347ec: ( no output )
Wolfman2000 rakudo: say "true" if 7 % 2&3;
p6eval rakudo 7347ec: true␤
Wolfman2000 rakudo: say "true" if 9 % 2&3;
p6eval rakudo 7347ec: true␤
quietfanatic rakudo: multi infix:<=> (Str $a, Str $b) {...}; my $x = 4 03:48
p6eval rakudo 7347ec: ( no output )
quietfanatic rakudo: multi infix:<=> (Str $a, Str $b) {...}; my $x = 4; say $x 03:49
p6eval rakudo 7347ec: ( no output )
quietfanatic That's a segfault actually
Wolfman2000 quietfanatic: were you able to view my recent pastebin? I'm curious if my loop can be reduced further. 03:50
quietfanatic Wolfman2000: You may be able to do it with an infix operator and the [infix] hyperop 03:51
if $n is known ahead of time
Wolfman2000 $n is known ahead of time
quietfanatic *metaop 03:53
Wolfman2000 quietfanatic: are you saying I would have to make my own multi sub infix? 03:54
quietfanatic you could put it inside you modpow sub
Wolfman2000 ...seems like little point of a custom infix inside a sub if it's only used once
TimToady it would let you write [*%] $base xx $exp 03:55
quietfanatic Yえあh
er
Wolfman2000 check your language settings there
quietfanatic I mean yeah
TimToady input method, rather
Wolfman2000 that too 03:56
quietfanatic somebody made my right-alt key do that...
When I try to implement it Rakudo NPAs though
Wolfman2000 TimToady: how would a custom infix operator make my code smaller/faster? I'm not seeing how. 03:57
TimToady however, the fastest way might be ($res *= $base) %= $mod for ^$exp 03:58
quietfanatic You don't even have to preset $res to 1 because *= will do that for you 03:59
rakudo: say ((my $res *= 2) %= 100) for ^10 04:00
p6eval rakudo 7347ec: 2␤2␤2␤2␤2␤2␤2␤2␤2␤2␤
quietfanatic rakudo: say ((my $res *= 2) %= 100 for ^10)
p6eval rakudo 7347ec: ( no output )
quietfanatic rakudo: my $res; say (($res *= 2) %= 100 for ^10) 04:01
p6eval rakudo 7347ec: ( no output )
quietfanatic Oh I guess 'for' isn;t returning what I expect
rakudo: my $res; ($res *= 2) %= 100 for ^10; say $res
p6eval rakudo 7347ec: 24␤
quietfanatic There. 1024 % 100 04:02
rakudo: (my $res *= 2) %= 100 for ^10; say $res
p6eval rakudo 7347ec: Lexical '$res' not found␤in Main (file src/gen_setting.pm, line 324)␤
Wolfman2000 rakudo: my $res; ($res *= 2) %= 100 for 1..10; say $res
p6eval rakudo 7347ec: 24␤
Wolfman2000 I guess ^10 is a prefix operator meaning 1 .. 10 04:03
quietfanatic Actually it's 0..9 but nobody's counting :)
Wolfman2000 The only thing I'm unsure of...why must $res be prefixed by my outside of the ()?
At best, we've got a two liner
quietfanatic Implied block for the for?
TimToady make the parameter 'is copy' and use that 04:04
zaslon loljnthnhazblogged! jnthn++ 'Failure()<0xb77e0098>': Failure()<0xb77d3924>
quietfanatic I was gonna say that :|
Wolfman2000 TimToady: the only problem is that $res is not even a parameter 04:05
TimToady base is
use that var :)
quietfanatic wait that doesn't work
($base *= $base) squares it every time
You need to keep the original $base too
Wolfman2000 I ended up running my code with the proposed ideas. It did not take 04:06
zaslon loljnthnhazblogged! jnthn++ 'Progress and minor frustration': use.perl.org/~JonathanWorthington/j...7?from=rss
quietfanatic If your $n is a power of two you ought to be able to say "my uint32 $res; $res *= $base for ^$exp", but rakudo doesn't support native types like that 04:08
(replaceing 32 with whatever your size is
TimToady well int declarations would be very fast all around if rakudo supported native types 04:09
quietfanatic I mean a power of two power of two.
zaslon lolmasakhazblogged! masak++ 'Failure()<0xb77fa7f8>': Failure()<0xb77f7e04>
TimToady someday
why is zaslon going nuts? 04:10
04:10 Wolfman2000 joined 04:11 kst` joined
Wolfman2000 quietfanatic: In case my last message didn't get through, the suggested code did not work for me. 04:11
quietfanatic gives more ++es all around?
Wolfman2000: I see. You mean the *= %= one?
Wolfman2000 Yeah
quietfanatic the 'is copy' one (which shuldn't work)?
TimToady it worked here on the channel 04:12
quietfanatic erk!
TimToady oh, yeah, that one won't work
04:12 Wolfman2000 joined
Wolfman2000 ...I got killed. How did I revive? 04:12
Wolfman2000 doesn't recall casting Auto-life
quietfanatic Wolfman2000: the 'is copy' one (which shouldn't work)?
Wolfman2000 quietfanatic: give the example again please. I don't recall trying is copy yet
quietfanatic Wolfman2000: it said something about 'nick collision' when you were killed
TimToady yes, I was hallucinating
quietfanatic Well don't try it cause it doesn't work! 04:13
rakudo: my $res; ($res *= 2) %= 100 for ^10; say $res
Wolfman2000 The only one I recall trying was return ($res *= $base) %= $n for ^$exp;
p6eval rakudo 7347ec: 24␤
04:13 drbean joined
quietfanatic ^ That one works 04:13
I dunno
Maybe you weren't returning $res? 04:14
04:14 Exodist joined
Wolfman2000 quietfanatic: doesn't seem to want to return what I'm asking it to 04:14
quietfanatic Wolfman2000: Yeah that doesn't work. You have to add a third statement 'return $res' 04:15
zaslon lolmasakhazblogged! masak++ 'November 21 2009 -- you mean the media distort the truth?': use.perl.org/~masak/journal/39926?from=rss
quietfanatic That has the 'return' inside the 'for' there
Wolfman2000 ...so it's basically the same as the loop version. Guess there's no avoiding the loop.
quietfanatic yeah. for is always a loop 04:16
Wolfman2000 yeah, re-adding the return statement did it
TimToady the infix form would let you loop using xx 04:17
but it's still a loop
Wolfman2000 may as well keep what I have.
TimToady what you had originally is very readable :)
an FP folk would probably use recursion though :) 04:18
Wolfman2000 TimToady: Considering how this cryptography project is going to involve a new language for practically all of them, may as well impress them with some funny, slightly hard to understand code. :P
Wolfman2000 has a caesar cipher, prime identifier, and simple rsa code implemented now...unsure about going for sha/md 04:19
quietfanatic rakudo: multi modpow ($a, 1) {$a}; multi modpow($a, $b) {modpow($a, $b-1) * $a % 100}; say modpow(2, 10);
p6eval rakudo 7347ec: 24␤
quietfanatic Like that
Wolfman2000 ...you can specify constants as parameters? 04:20
quietfanatic That's actually quite succint. Only two lines
Yeah you can.
Can't introspect them though
(yet)
Wolfman2000 wishes he knew what introspecting was.
quietfanatic rakudo: say :(3).perl
p6eval rakudo 7347ec: :(Int ::TYPE_CAPTURE20 where all({ ... }))␤
quietfanatic blech!
introspecting means...uh...looking at the insides or something 04:21
You notice I can't find a '3' anywhere inside that blechy expression.
I think I'll report that as a bug 04:22
...not that anyone reads my bug reports 04:23
zaslon loljnthnhazblogged! jnthn++ 'Failure()<0xb5ed2774>': Failure()<0xb5ebcf00> 04:24
Wolfman2000 jnthn: looks like you failed to blog
TimToady rakudo: my $n = 100; sub modpow($base,$exp) { sub infix:<*%>($res,$base) { $res * $base % $n }; [*%] 1,$base xx $exp }; say modpow(2,10) 04:25
p6eval rakudo 7347ec: 24␤
TimToady Wolfman2000: ^^^
also, zaslon is going nuts somehow and reporting old blogs 04:26
Wolfman2000 quietfanatic--: maximum recursion depth exceeded
quietfanatic Hoh come one!
Wolfman2000 quietfanatic: You never knew how big $n could be
quietfanatic Haskell wouldn't have complained!
zaslon lolmasakhazblogged! masak++ 'Failure()<0xb78428cc>': Failure()<0xb7837f6c> 04:27
Wolfman2000 The only thing I really don't recall...what's the xx operator? 04:28
quietfanatic list replication
rakudo: .say for 3 xx 5
p6eval rakudo 7347ec: 3␤3␤3␤3␤3␤
quietfanatic contrast with
rakudo: .say for 3 x
p6eval rakudo 7347ec: Confused at line 2, near "x"␤in Main (file <unknown>, line <unknown>)␤
quietfanatic rakudo: .say for 3 x 5
p6eval rakudo 7347ec: 33333␤
Wolfman2000 ...not every day that one of the founders of the language assists someone like this. 04:29
04:29 cls_bsd left
Wolfman2000 TimToady++: Thanks for the insight. 04:30
quietfanatic TimToady: Your modpow above will blow up if given $exp = 0
rakudo: my $n = 100; sub modpow($base,$exp) { sub infix:<*%>($res,$base) { $res * $base % $n }; [*%] 1,$base xx $exp }; say modpow(2,0)
p6eval rakudo 7347ec: 1␤
quietfanatic !!!
Wolfman2000 quietfanatic: is that the blow up you mean?
quietfanatic I guess not
rakudo: sub infix:<*%>($res,$base) { $res * $base % $n }; [*%] 1 04:31
p6eval rakudo 7347ec: Symbol '$n' not predeclared in infix:*% (/tmp/rQ5aW4kRFh:2)␤in Main (file <unknown>, line <unknown>)␤
quietfanatic rakudo: sub infix:<*%>($res,$base) { $res * $base % 100 }; [*%] 1
p6eval rakudo 7347ec: Method '!flatten' not found for invocant of class 'Int'␤in Main (file src/gen_setting.pm, line 324)␤
quietfanatic rakudo: sub infix:<*%>($res,$base) { $res * $base % 100 }; [*%] (1)
p6eval rakudo 7347ec: Method '!flatten' not found for invocant of class 'Int'␤in Main (file src/gen_setting.pm, line 324)␤
quietfanatic rakudo: sub infix:<*%>($res,$base) { $res * $base % 100 }; [*%] [1]
p6eval rakudo 7347ec: ( no output )
quietfanatic rakudo: sub infix:<*%>($res,$base) { $res * $base % 100 }; say [*%] [1]
p6eval rakudo 7347ec: 1␤
quietfanatic I just don't understand what's going on here 04:32
Wolfman2000 Perl 6 has magic
quietfanatic shouldn't it say "not enough positionals given"?
rakudo: sub infix:<?> ($a, $b) {say $a.WHAT; say $b.WHAT}; [?] 1 04:33
p6eval rakudo 7347ec: Method '!flatten' not found for invocant of class 'Int'␤in Main (file src/gen_setting.pm, line 324)␤
quietfanatic rakudo: sub infix:<?> ($a, $b) {say $a.WHAT; say $b.WHAT}; [?] [1]
p6eval rakudo 7347ec: ( no output )
quietfanatic rakudo: sub infix:<?> ($a, $b) {say $a.WHAT; say $b.WHAT}; [?] 1,
p6eval rakudo 7347ec: No result object␤in Main (file <unknown>, line <unknown>)␤
quietfanatic just getting all sorts of weird here
I give up.
Wolfman2000 ...even when I can't contribute to the core Perl 6 code, I can still help...or at least give ideas. 04:34
04:39 gfx joined
pmichaud good evening, all 04:39
Wolfman2000 evening 04:40
04:42 Exodist joined 04:47 envi^home joined
TimToady howdy 04:47
04:49 mlaks joined 04:56 qp_pq joined
qp_pq you bastards, you turned the first name Dijkstra into a verb lol 04:56
just saw masak's talk.pdf 04:57
like "dijkstrifying" :)) 05:00
japhb Any word containing five consecutive consonants ought to get a free pass to be any part of speech it wants. :-) 05:01
TimToady that's not her first name. It's Barbara.
05:05 kaare joined, Nex6 joined
qp_pq TimToady: ?!? who ? 05:06
05:11 astrojp left
TimToady Barbara, Ken Dijkstra's wife 05:13
japhb Where is the syntax for calling subs and methods by computed name discussed in the synopses? I'm failing in attempting to search ... 05:14
qp_pq who is Ken Dijkstra ? 05:21
05:26 beggars joined 05:36 c9s joined 05:44 Avada joined 05:57 Avada joined 06:01 orafu joined 06:11 orafu joined 06:21 orafu joined, carlin joined 06:43 JeffreyKegler joined 06:44 envi^home joined, cognominal joined, bluescreen joined, pnate joined, p6eval joined, jjore joined
TimToady Barbie's husband... 06:46
but people are always confusing them with the other Ken and Barbie... :) 06:48
06:49 meppl joined 06:50 JeffreyKegler left 06:55 justatheory joined
diakopter nqp: Foo::bar(); { module Foo { our sub bar() { say(77) } } } 07:12
p6eval nqp: 77␤
diakopter sigh.
double sigh.
triple sigh.
07:13 jrockway joined
diakopter out of the frying pan, into the center of the earth. 07:13
nqp: Foo::bar(); sub blah() { module Foo { our sub bar() { say(88) } } } 07:14
p6eval nqp: 88␤
diakopter how the blanket can that behavior be mimicked? 07:15
pugs: Foo::bar(); sub blah() { module Foo { our sub bar() { say(88) } } } 07:18
p6eval pugs: 88␤
diakopter I give up. 07:19
07:19 snearch joined
diakopter nqp: Foo::bar(); sub blah() { module Foo { our sub bar() { say(88) } } }; sub blah2() { module Foo { our sub bar() { say(99) } } }; 07:24
p6eval nqp: 99␤
diakopter pugs: Foo::bar(); sub blah() { module Foo { our sub bar() { say(88) } } }; sub blah2() { module Foo { our sub bar() { say(99) } } };
p6eval pugs: 99␤
cotto Should that even compile? 07:33
diakopter std: Foo::bar; sub blah() { my $b := 2; module Foo { our sub bar() { say($b) } } }; sub blah2() { my $b := 4; module Foo { our sub bar() { say($b.WHERE) } } }; 07:34
p6eval std 29167: [31m===[0mSORRY![31m===[0m␤Illegal redeclaration of symbol 'GLOBAL::<Foo>' (from line 1) at /tmp/x5mXWVKies line 1:␤------> [32m }; sub blah2() { my $b := 4; module Foo[33m⏏[31m { our sub bar() { say($b.WHERE) } } };[0m␤Undeclared name:␤ 'Foo::bar' used at line 1␤FAILED
..00:01 107m␤
diakopter whatever compile means, no
but why a module can be declared inside a sub escapes me 07:35
perl -e 'Foo::a(); sub b { package Foo; our sub a { print(44) } }; sub c { package Foo; our sub a { print(55) } };' 07:39
55
whatever. :[
07:57 orafu joined 08:12 drbean_ joined 08:13 Exodist joined 08:14 Su-Shee joined
Su-Shee good morning 08:14
08:17 cls_bsd joined 08:19 cotto_w0rk joined 08:34 am0c joined
Woodi good morning 08:36
08:40 Exodist joined
diakopter perl6.pastebin.com/f1b671708 prototype-object inheritance/derivation in Perl 08:40
cls_bsd m/win 30 08:52
08:57 BinGOs joined
diakopter yeah 09:00
09:04 desertm4x joined 09:25 Exodist joined 09:41 drbean joined
ng_feed rakudo-ng: pmichaud++ 09:56
rakudo-ng: Merge branch 'ng' of git@github.com:rakudo/rakudo into ng
rakudo-ng: pmichaud++
rakudo-ng: Add [eval]. May still want a little bit of testing, but should
09:57 cotto joined 10:18 snearch joined 10:19 Guest37418 joined 10:21 desertm4x_ joined 10:53 bryan[c1] joined
Wolfman2000 *yawn* Morning...I'm up too early. 11:15
11:30 rgrau` joined, mikehh joined
colomon eval \\o/ 11:47
11:55 jayeola joined
jnthn wondes how many tests that may win us back 11:55
11:57 jeremiah joined
jnthn lolitsjeremiah 11:57
jeremiah OH HAI! 11:58
or OH HAIS since we're in Latvia
Wolfman2000 ...how many countries are we in? 11:59
jnthn Most of the time, only one at any point in time, though I was once in three simultaneously. 12:01
Wolfman2000 ...leg on one, other leg on one, butt on third? 12:02
jnthn Something like that. :-)
Wolfman2000 very strange
The master branch of rakudo hasn't been updated for a few days. :( 12:03
colomon jnthn: four tests in rat.t thanks to eval!
Wolfman2000 I know ng is loved by everyone, but...oh well.
colomon Wolfman2000: until pmichaud checked in eval this morning, no one but me had updated ng in days, either.... 12:04
12:05 cotto_working joined
Wolfman2000 eval...right: execute arbitrary code. 12:05
jnthn colomon: You aren't no one! :-)
colomon jnthn: but you guys are the big guns, and I'm just hanging about trying to show you can indeed write some of the core classes in pure p6. 12:07
jnthn I'm very happy that they're being written in Perl 6.
colomon me too, I think that's very importat 12:08
Wolfman2000 colomon: what else are the core classes being written in?
jnthn Wolfman2000: Some bits get done in PIR. But the less of that we have, the better, overall.
Wolfman2000 ...right, the PIR assembly language.
jnthn Yeah. But there's more to it than just writing in PIR
Wolfman2000 I recall seeing some of that in some of the Operator type files
jnthn e.g. we wrap Parrot sub PMCs up into Pelr 6 code objects. 12:09
Introspection tends not to work out so well on Parrot subs.
Wolfman2000 Right...not always easy to introspect multi sub infix:<+>($a, $b) 12:11
jnthn Well, that should have been resolved in master now, IIRC 12:14
But yes, other examples certainly have problems.
12:19 cotto_work joined 12:21 rfordinal left 12:28 zloyrusskiy joined 13:07 ashizawa_ joined 13:16 jaldhar joined 13:17 mjk joined
colomon jnthn: do we not have substr yet in ng? (I'm checking out int.t to see where we stand with it.) 13:18
ng: say substr("hello", 0, 2); 13:20
p6eval ng 7ab131: Confused at line 1, near "say substr"␤current instr.: 'perl6;HLL;Grammar;panic' pc 519 (src/stage0/HLL-s0.pir:336)␤
colomon ng: say "hello".substr(0, 2); 13:21
p6eval ng 7ab131: Method 'substr' not found for invocant of class 'Perl6Str'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤
colomon ng: say "hello".lcfirst 13:22
p6eval ng 7ab131: Method 'substr' not found for invocant of class 'Perl6Str'␤current instr.: 'perl6;Any;lcfirst' pc 181609 (src/gen/core.pir:0)␤
colomon Ug. So we've got a lot less string than it appears in ng, since about half of the defined functions depend on substr, which isn't defined. 13:23
Wolfman2000 colomon: If it matters, I found .sub 'substr' on master/src/builtins/any-str.pir 13:25
Perhaps you can just copy that?
colomon Hmmm... 13:26
ng: say "hello".WHAT 13:28
p6eval ng 7ab131: Str()␤
colomon It definitely looks tantalizingly possible... 13:29
particularly because it appears that there's only one thing in there that really needs to be in PIR. 13:30
pmichaud good morning, #perl6 13:31
colomon Just the person I was hoping to see! 13:32
If I have $S1 = substr $S0, start, len in PIR, how do I do that as one of those fancy pir::XXXX__ calls? 13:33
(and thank you for eval!)
pmichaud I think that pir::substr($str, $start, $len) works. 13:34
if not, then it'd be pir::substr__PPii
colomon ng: say pir::substr("hello", 0, 2);
p6eval ng 7ab131: he␤
colomon \\o/
pmichaud in general, it's pir::opcode(...args...) 13:35
but sometimes the compilers don't know the argument types to use for the opcode, so we have to be explicit
jnthn summon masak
pmichaud thus __PPii would say "return a PMC, first arg is a PMC, second arg is an integer register or integer constant, third arg is a integer register or integer constant" 13:36
jeremiah waves magic wand to summon masak
13:36 masak joined
pmichaud so, did eval() win us many tests? does it actually work? ;-) 13:36
Su-Shee "accio" is the charm to summon something ;)
masak feels groggy
was I summoned or something?
jeremiah lolitsmasak! 13:37
masak lol!
I can haz math algebra parser.
colomon pmichaud: it won us four rat.t tests, but we still need to get Num.Rat working to get the entire file passing. 13:38
pmichaud colomon: does that mean that eval actually works? ;-)
wow.
It took me far longer to write than it should've (parrot--) and when I was getting near the end I wasn't fully awake :-| 13:39
jnthn :-/ 13:40
pmichaud: Yes, it was "fun" enough the first time.
pmichaud: Looks cleanish now, from glancing over the patch, though. :)
colomon pmichaud: I've only tested it on strings which were Rats so far, but it seems to work well there. :)
pmichaud well, at least this time it's moved the lexicals logic into the components where they belong
colomon That's really all anyone ever wanted, isn't it? The ability to eval Rats? 13:41
:)
masak what's this Japanese Twitterer saying? twitter.com/tokuhirom/status/5944222957
pmichaud colomon: that's all it's really good for.
jnthn Yes, it's good to push them into a place where this will work between other HLLs.
masak finally, I can eval my Rats.
jnthn: Rats generally burrow holes in the walls between HLLs. 13:42
colomon One day we'll be able to eval FatRats...
jnthn And our Cat.
pmichaud not only that, but it actually follows contexts instead of subs, which is equally important (since a Parrot sub object might not be able to reference the correct outer context)
masak oh noes! FatRats!
13:44 mikehh joined
masak oh 猫! 13:44
colomon Any objections to my trying to translate the old master substr into a new, mostly p6 ng substr?
13:44 masak joined
jnthn colomon: None from me. 13:44
pmichaud +1 13:45
colomon how do you say optional parameter with specifying a default? I seem to recall there is a ? involved. 13:49
pmichaud sub foo($req, $opt?) { ... }
oh, *with* a default
Juerd Implicit defaults ROCK, pmichaud :)
pmichaud sub foo($req, $opt = 'default') { ... }
Juerd They make coding so much easier :D
colomon s/with/without/ actually! 13:50
so pmichaud++ had it right the first time.
pmichaud oh, *without* a default
colomon danke!
pmichaud sub foo($req, $opt?) { ... } # :D
Juerd Wow, pmichaud has dwimnwis-powers :)
pmichaud note that because of the mis-asked question, the ":D" smiley is now a required part of the syntax :D 13:51
colomon Juerd: I'm not at all surprised. :)
Juerd remembers when the postfix question mark was introduced for this, and was happy about the decision :)
Also, the ! for required nameds 13:52
colomon ng: say "hello".chars
p6eval ng 7ab131: Method 'chars' not found for invocant of class 'Perl6Str'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤
colomon ng: say "hello".codes
p6eval ng 7ab131: Method 'codes' not found for invocant of class 'Perl6Str'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤
masak yes, the current syntax for parameters and arguments is a clear improvement on the original apocalyptic one. 13:53
Juerd S03 says that !< doesn't have a shortcut, but isn't that just >=? 13:54
colomon rakudo: say "hello".chars
p6eval rakudo 7347ec: 5␤
Wolfman2000 colomon++: glad I was able to give you an idea for substr. I have to get going now, however...apparently I (got conned?|agreed) to help sub for a sunday school class today 13:55
masak Juerd: yes.
colomon Wolfman2000++ # for pointing me in the right direction.
rakudo: say substr("hello", 2, -1); 13:57
p6eval rakudo 7347ec: ll␤
colomon Oh, I see. 13:58
gotos, they are fun.
masak Juerd: though I don't consider >= to be the shortcut of !< -- it's not shorter, for one. 13:59
pmichaud hugme tweet rakudoperl rakudo-ng has a working 'eval' again 14:03
hugme hugs pmichaud; tweet delivered
colomon Is .chars correct for getting the number of characters in a string in p6?
14:05 Whiteknight joined
Su-Shee works with my version. 14:06
jnthn colomon: yes 14:07
colomon apparently I need to implement that, too.
:)
dalek p-rx: 23bb989 | pmichaud++ | (3 files):
Add $obj."method"(...) syntax.
p-rx: 9f7db2e | pmichaud++ | src/stage0/ (3 files):
Update bootstrap.
colomon pir::chars(self) doesn't seem to do it.
pir::chars__IP(self) ? 14:08
Nope. 14:09
help?
jnthn nqp: my $x = "lolvodka"; say pir::chars__IS($x); 14:11
p6eval nqp: Assignment ("=") not supported in NQP, use ":=" instead at line 1, near " \\"lolvodka"␤current instr.: 'parrot;HLL;Grammar;panic' pc 519 (src/cheats/hll-grammar.pir:197)␤
jnthn nqp: my $x := "lolvodka"; say pir::chars__IS($x);
p6eval nqp: Confused at line 1, near "say pir::c"␤current instr.: 'parrot;HLL;Grammar;panic' pc 519 (src/cheats/hll-grammar.pir:197)␤
jnthn gha
nqp: my $x := "lolvodka"; say(pir::chars__IS($x));
p6eval nqp: error:imcc:syntax error, unexpected IREG, expecting '(' ('$I17')␤ in file 'EVAL_1' line 12␤Could not find non-existent sub chars␤current instr.: '_block11' pc 0 (EVAL_1:6)␤
jnthn *sigh*
checking the Parrot docs.
colomon danke. 14:12
jnthn nqp: my $x := "lolvodka"; say(pir::length__IS($x)); 14:13
p6eval nqp: 8␤
jnthn =item B<length>(out INT, in STR)
Calculate the length (in characters) of string $2 and return as integer $1.
If $2 is NULL or zero length, zero is returned.
And for .bytes its bytelength
colomon right! jnthn++ 14:14
I should have checked the old any-str.pir to see how chars was defined.
\\o/ 14:15
is export doesn't work yet?
jnthn colomon: I thought it did... 14:16
ng: say uc("wtf")
p6eval ng 7ab131: WTF␤ 14:17
jnthn Otherwise that'd not work.
colomon So did I, but .substr works and substr doesn't.
huh.
jnthn And it's marked "is export"? Hmm.
colomon tell you what. I'll check in what I've got, and we can puzzle it out together. :)
jnthn :-)
14:19 payload joined
ng_feed rakudo-ng: colomon++ 14:19
rakudo-ng: Add chars and substr.
pmichaud + my $len = $given_len || self.chars; 14:20
needs to be
$given_len // self.chars
(since 0 is a valid length) 14:21
colomon okay, changed locally. 14:22
I suppose, since I have chars working now, that could just be $given_len = self.chars for the argument?
pmichaud probably 14:23
colomon Though I think I might like it better this way.
pmichaud although the arg name should perhaps match the spec ($length)
14:24 masak` joined, facsimile joined
colomon given_len becomes length, okay. :) 14:25
pmichaud (building now to see if I can find out why 'is export' isn't working)
colomon k
14:25 arthur-_ joined
colomon it does work for chars, I see. 14:25
pmichaud I wonder if there's another substr already defined somewhere that is causing problems. 14:26
ng_feed rakudo-ng: colomon++
rakudo-ng: Change $given_len to $length to match the spec.
rakudo-ng: colomon++
rakudo-ng: Change || to // as per pmichaud.
pmichaud don't see one. 14:27
colomon has is export been tested with more than one argument, or default arguments?
pmichaud ...shouldn't make a difference, really.
colomon (seeing as how uc and chars work, but substr doesn't....)
pmichaud that could indeed be the difference, but it shouldn't be :) 14:28
colomon :)
pmichaud aha 14:29
jnthn I'm guessing substr is in Any?
pmichaud looks like ng is parsing the "sub" 14:30
i.e., it's parsing as sub str(...)
jnthn !!
pmichaud we had this problem in master also :)
jnthn ng: substr("lol", 1, 1) 14:31
p6eval ng 7ab131: Confused at line 1, near "substr(\\"lo"␤current instr.: 'perl6;HLL;Grammar;panic' pc 519 (src/stage0/HLL-s0.pir:336)␤
colomon ?????????
jnthn oh, it's a parse error!
14:31 pmurias joined
colomon ooooooooo 14:31
pmichaud fixing.
14:33 ejs joined 14:34 snearch joined 14:35 JimmyZ joined
colomon ng: "24".Int 14:38
p6eval ng 7ab131: Method 'Int' not found for invocant of class 'Perl6Str'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤
pmichaud pushed. 14:39
ng_feed rakudo-ng: pmichaud++ 14:40
rakudo-ng: Avoid initial "sub", "method", "submethod" as being declarations.
jnthn pmichaud: We'd not need that after full LTM? 14:41
(just noticing std lacks it)
colomon oh bother, we don't have index yet either. 14:45
jnthn nqp: my $x := "lolvodka"; say(pir::index__ISS($x, "l")); 14:48
p6eval nqp: 0␤
jnthn nqp: my $x := "lolvodka"; say(pir::index__ISS($x, "v"));
p6eval nqp: 3␤
jnthn :-) 14:49
pmichaud jnthn: full LTM should alleviate that, yes.
colomon jnthn: ooo, short and simple implementation, then, eh? 14:51
jnthn colomon: looks like it ;_) 14:52
14:52 rindolf joined
jnthn pmichaud: OK, cool. 14:52
masak` rindolf: hi! 14:53
colomon is there an easy way of checking to see what test files now work (unexpectedly) in ng?
rindolf Hi masak`
masak`: what's up?
colomon I think working substr just bought us a bunch.
(but not substr.t, of course... ;)
masak rindolf: in the lobby of a hotel in Riga, hacking away with jnthn++ and jeremiah++. 14:54
life is good :)
rindolf: how're you?
rindolf masak: we have a Tel Aviv Open Source Club meeting today.
An hour or so from now.
It's a Welcome-to-FOSS meeting.
masak \\o/ 14:55
rindolf No one came to the previous meeting.
jnthn colomon: I thought it may...uc.t for example.
rindolf Except for the lecturer and me.
colomon ucfirst.t and lcfirst.t for sure.
rindolf And his friend from the university.
colomon uc.t doesn't work on my box, no iciu
masak oh, bummer.
rindolf This time it was better publicised.
jnthn colomon: oh. I can check that tomorrow when at home.
rindolf But it's a different meeting.
Maybe we'll do a re-run of the previous meeting in two weeks or so. 14:56
pmichaud I'll be attempting spectests shortly :)
jnthn How epic is our fail? :-) 14:57
pmichaud don't know yet. :-) 14:58
but hopefully much less epic by the end of today...? ;-)
jnthn pmichaud: I've been pondering metamodel stuff some more today.
pmichaud: Want to put back role composition in a Good Way.
Been reading Moose source to figure out their model 14:59
Learned a few things.
pmichaud jnthn: +1
jnthn Will do something somewhat along those lines, but with 6isms. :-) 15:00
And probably a bit simpler protocol for now.
rindolf masak: BTW, have you meet Peteris Krumins (IRL) during the Baltic Perl Workshop?
15:03 nihiliad joined 15:07 masak` joined, Psyche^ joined
colomon rakudo: index("hello", ""); 15:19
p6eval rakudo 7347ec: ( no output )
colomon rakudo: say index("hello", "");
p6eval rakudo 7347ec: 0␤
jnthn colomon: Hmm...that looks...odd. 15:21
colomon ng: say min(2, 4);
p6eval ng a9956f: Could not find non-existent sub &min␤current instr.: '_block14' pc 29 (EVAL_1:0)␤
jnthn rakudo: say index("hello", "k") 15:22
p6eval rakudo 7347ec: Substring 'k' not found in 'hello'␤␤
jnthn rakudo: index("hello", "k")
p6eval rakudo 7347ec: ( no output )
jnthn oh, i'ts a fail, not a die. Good.
So above is right too I guess. :-)
colomon clues on how to handle fails in ng would be good, too. :)
(for me, I mean.) 15:23
SHouldn't it be pir::index__ISSi ? 15:25
jnthn Why the i on the end?
colomon int $pos paramete
r
jnthn ISS = returns an int, takes two strings.
oh
maybe there's another variant of the op that does that too
may work then 15:26
colomon pos = index s, substring, pos
is the code in the old Any-str.pir
jnthn ah, then yes 15:27
colomon btw, the old code appears to be somewhat broken? 15:30
at least, it's got two labels which are never used, one of which appears to be an error condition which is not checked.
jnthn That's quite possible.
colomon unless "fail:" is a magic label...
jnthn Don't bring the brokenness into ng ;-)
No, it's not.
colomon any advice on doing 15:31
.tailcall '!FAIL'("Substring '", substring, "' not found in '", s, "'")
in perl 6 and ng?
jnthn fail("Substring '$foo' not found in '$s'"); 15:33
colomon Is that what gets the "non-existant sub &fail called" message? ;) 15:34
jnthn colomon: Yes. ;-) 15:36
colomon then it must be right. :) 15:37
jnthn Well, it won't fail. 15:38
colomon say index("hello", "v") 15:39
Could not find non-existent sub &fail
\\o/
15:41 iblechbot joined
jnthn ng: sub foo { return 42 }; say foo 15:44
p6eval ng a9956f: sh: ./perl6: No such file or directory␤
jnthn aww
ng: sub foo { return 42 }; say foo
p6eval ng a9956f: sh: ./perl6: No such file or directory␤
jnthn EINPATIENT
ng: sub foo { return 42 }; say foo 15:45
p6eval ng a9956f: sh: ./perl6: No such file or directory␤
ng_feed rakudo-ng: colomon++ 15:46
rakudo-ng: Port of index from PIR into (mostly) p6.
jnthn Yay, Perl 6!
ng: sub foo { return 42 }; say foo 15:47
p6eval ng a9956f: 42␤
jnthn ah, we haz some return
So maybe fail ain't so hard now.
colomon bother.
first the test wouldn't run because of substr.
then it wouldn't run because of index 15:48
now it won't run because of fail.
rakudo: say index("hello", "k"); 15:49
p6eval rakudo 7347ec: Substring 'k' not found in 'hello'␤␤
colomon or is that it?
lisppaste3 colomon pasted "The test that's giving me trouble" at paste.lisp.org/display/90898
colomon does fail equal undef or something? 15:50
dalek p-rx: da4d6f4 | pmichaud++ | src/ (2 files):
Refactor regex creation a bit.
 15:52
p-rx: c4b8abd | pmichaud++ | src/stage0/ (3 files):
Update bootstraps.
jnthn colomon: It returns a failure, which is undefined, yes.
colomon makes sense. 15:53
any idea how hard it would be to implement fail? 15:55
jnthn colomon: Probably needs to be done in PIR, but the existing one may be food enough. 15:56
rakudo: Failure
p6eval rakudo 7347ec: ( no output )
15:57 ejs joined
colomon shouldn't we have eval_dies_ok by now? 15:57
(messing around with other tests because I'm clueless on fail...)
jnthn yeah 15:58
well
we have eval
maybe Test.pm needs updating?
Now that eval works?
github.com/rakudo/rakudo/blob/maste...ontrol.pir # for how fail was done before.
colomon you stuck a die line in there. should I just delete it?
that seems naive... 15:59
(deleting the die only, I mean.)
16:00 facsimile joined
jnthn remove the die and uncomment the proclaim call I ugess. 16:01
dobut the regex line can yet be removed.
16:06 am0c joined
cognominal jnthn++ # your blog helps so much to understand about rakudo(-ng)? 16:07
I meant rakudo[-ng]?
colomon jnthn: have to remove the regex in the proclaim line as well, right? 16:08
jnthn colomon: ah, probably, yeah.
colomon: we don't have those back in just yet 16:09
colomon: comment out the original I guess, and put in the simpler one.
colomon already done here. :) 16:11
jnthn Nice :-)
colomon I'm trying length.t, and I love the first message:
Str.length properly not implemented
All right, I'll check in Test.pm 16:12
ng_feed rakudo-ng: colomon++ 16:13
rakudo-ng: Turn eval_dies_ok back on.
16:14 jan_ joined, am0c joined
colomon Ack. It really is frustrating that every fix probably fixes dozens of tests, but we can't count them because something else in the test file fails. 16:17
ng_feed rakudo-ng: colomon++
rakudo-ng: Turn lcfirst.t and ucfirst.t back on.
jnthn colomon: Aye. 16:18
colomon: That's my experience on ng too.
16:19 KyleHa joined
ng_feed rakudo-ng: KyleHa++ 16:30
rakudo-ng: Add some more passing test files
masak` I make the prediction that this interconnectedness of things is worst in the core, where y'all are hacking now. and that it'll get better as those things are in place. 16:31
pmichaud I'm doing regexes now 16:32
jnthn cheers on pmichaud 16:33
colomon masak-prime?! Quick, jnthn, does he have a goatee?
KyleHa++ # getting more tests on line, yay! 16:34
KyleHa *bow* 16:35
colomon That picked us up 50-some tests in the spectest!
pmichaud I'm predicting that we'll move ng to master before the end of Nov.
jnthn colomon: He's bearded, but not quite to the degree I am. :-)
colomon pmichaud: woah.
jnthn pmichaud: Whoa...I'd better get busy! ;-) 16:36
pmichaud jnthn: that makes two of us :)
jnthn Quick...bring beer!
Hey, akshually...I'm sat with a bar like 15 seconds walk away. 16:37
colomon I promise to continue making small improvements to the code while rocking my sleeping son.
jnthn colomon++ 16:38
colomon Probably some while I'm watching him run around and chew on things, too, like this morning. :)
colomon can't concentrate enough watching the boy to get real $work done, but very small-scale hacking with lots of spectests is no problem. 16:41
jnthn nom time! 16:47
pmichaud ugh, trying to follow all of the block/signature logic is really.... difficult. :( 17:10
17:20 dj_goku joined
pmichaud jnthn: ping 17:21
moritz_ re 17:23
17:28 diakopter left
pmichaud jnthn: unping -- I think I found the issue. 17:29
17:35 lichtkind joined 17:38 payload joined
moritz_ ng: class A { }; class A { }; say "alive" 17:48
p6eval ng ebaea1: Class A already registered!␤␤current instr.: 'perl6;ClassHOW;new' pc 2339 (src/builtins/ClassHOW.pir:65)␤
moritz_ ng: use Test; plan 1; eval_dies_ok 'class A { }; class A { }';
p6eval ng ebaea1: 1..1␤not ok 1 - ␤# Looks like you failed 1 tests of 1␤
moritz_ some eval_dies_ok doesn't seem to work
ng: eval ''; say $!.defined 17:49
p6eval ng ebaea1: Method 'defined' not found for invocant of class 'Undef'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤
moritz_ ng: eval ''; say defined($!)
p6eval ng ebaea1: 0␤
moritz_ ng: eval '++'; say defined($!)
p6eval ng ebaea1: Confused at line 1, near "++"␤current instr.: 'perl6;HLL;Grammar;panic' pc 519 (src/stage0/HLL-s0.pir:336)␤
moritz_ ng: my $x = eval '++'; say defined($!) 17:50
p6eval ng ebaea1: Confused at line 1, near "++"␤current instr.: 'perl6;HLL;Grammar;panic' pc 519 (src/stage0/HLL-s0.pir:336)␤
moritz_ ng: my $x = try { eval '++' }; say defined($!)
p6eval ng ebaea1: Null PMC access in can()␤current instr.: 'perl6;Perl6Object;!STORE' pc 1953 (src/builtins/Object.pir:383)␤
moritz_ this looks all so wrong to me
pmichaud: see above (if not yet known)
colomon moritz_: for sure ng's eval_dies_ok does not check for Null PMC, I had to comment that part out this morning because of the regexes. 17:53
moritz_ colomon: that's not what's wrong 17:55
(and that check could also be done with index())
colomon hey, good idea!
moritz_ colomon: the real problem is that 1) errors propagate through eval '', even though they shouldn't
2) empty $! is not a Perl 6 type, and thus doesn't response to .defined
3) try { eval '++' } seems to return a Null PMC access 17:56
s/access$// 17:57
t/spec/S06-signature/optional.t 17:59
passes in ng
but since it uses eval_dies_ok I'm reluctant to add it back in
ng: sub wrong ($a?, $b) { } 18:00
p6eval ng ebaea1: ( no output )
moritz_ ng: sub wrong ($a?, $b) { }; say 'alive'
p6eval ng ebaea1: alive␤
moritz_ ng: say(1, , 2)
p6eval ng ebaea1: 12␤
moritz_ std: say(1, , 2) 18:01
p6eval std 29167: [31m===[0mSORRY![31m===[0m␤Preceding context expects a term, but found infix , instead at /tmp/Q8N5rdvauk line 1:␤------> [32msay(1, ,[33m⏏[31m 2)[0m␤FAILED 00:02 107m␤
moritz_ over the weekend I had two ideas for potential changes to Perl 6 18:03
the first one is rather straight forward:
@list.sort($code) does a Schwartzian Transform if $code has arity (or count) 1
that could be extended 18:04
if $code returns something Positional
the comparison could be made by the first item first
if two first items are equal on the second item
and so on
don't know if it might be overkill, but it might be handy in some cases 18:05
the second idea is a bit weird
I thought that it's a bit ugly to have so many methods defined in class Any
so user defined classes have more than hundred methods by default 18:06
colomon oh, actually index won't work for Null PMC detection until fail is implemented.
moritz_ so what about adding a second class near the root of our type hierarchy
and user defined classes inherit from that class by default
colomon (or rather, it will work great for detecting null pmcs, but fail utterly at detecting their absence.)
pmichaud I think we already have the $code does Positional feature, don't we? 18:07
moritz_ pmichaud: do we? that's news to me
pmichaud our Array multi method sort( @values: Ordering @by )
moritz_ (second root class) and if you want to write a class that acts like a builtin type, you make it inherit Any 18:08
I'm not yet sure if the details work out
or maybe there could be a role that all the builtin objects do 18:09
let's call it Base
so Num ~~ Base, Code ~~ Base etc.
and then for '24'.sin to work, there's a method sin in role Base
moritz_ kinda likes that idea 18:10
Wolfman2000 Back from giving my sunday school lesson and having lunch. ...seems like I unexpectedly caused some more development in the ng branch. 18:11
pmichaud 17:55 <moritz_> 2) empty $! is not a Perl 6 type, and thus doesn't response to .defined
for this we just need to fix !FAIL
for that we probably should figure out how Failure works 18:12
moritz_ that sounds LTT (Less Than Trivial) 18:15
colomon moritz_: which is exactly why I'm happy to sit back and like pmichaud and jnthn sort it out. 18:16
18:18 ejs joined
pugs_svn r29168 | moritz++ | [t/spec] test for Nil() coercion ans the sink statement prefix 18:18
r29169 | moritz++ | [t/spec] prefix + calls .Numeric, ~ calls .Stringy (see r28502, r28502)
r29170 | moritz++ | [t/spec] tests for prefix:<~>/.Stringy vs .Str
Wolfman2000 rakudo: multi sub infix:<^>(Num $x, Num $y) { return $x ** $y; }; say "10^2 = {10^2}"; 18:19
p6eval rakudo 7347ec: push_pmc() not implemented in class 'Sub'␤in Main (file <unknown>, line <unknown>)␤
Wolfman2000 std: multi sub infix:<^>(Num $x, Num $y) { return $x ** $y; }; say "10^2 = {10^2}";
p6eval std 29167: ok 00:01 110m␤
Wolfman2000 ...sure, std likes it
colomon Isn't infix ^ already in there somewhere, with a different meaning? 18:20
moritz_ Wolfman2000: that's the (LTA) error you get when overloading operators that aren't in the setting yet
colomon rakudo: say 4 ^ 10
p6eval rakudo 7347ec: one(4, 10)␤
moritz_ colomon: it is (junction)
colomon has it been moved to the setting yet?
moritz_ apparently not
Wolfman2000 apparently not
colomon ng: say 4 ^ 10 18:22
p6eval ng ebaea1: Confused at line 1, near "say 4 ^ 10"␤current instr.: 'perl6;HLL;Grammar;panic' pc 519 (src/stage0/HLL-s0.pir:336)␤
colomon infix:<^> isn't implemented at all in ng yet. :) 18:23
Wolfman2000 Since ng is apparently the new upcoming master, I'll just wait for it to show up as the master before I try to figure out the PIR for it
Wolfman2000 could still probably use a PIR tutorial. 18:31
moritz_ there's a book 18:32
Wolfman2000 free?
moritz_ yes
in the parrot repo somewhere
you can also buy a hard copy
Wolfman2000 I have the parrot repo: I required it for compiling rakudo 18:33
moritz_ docs.parrot.org/parrot/latest/html/ scroll down to "PIR book"
Wolfman2000 nice
pugs_svn r29171 | moritz++ | [t/spec] more Str -> Stringy changes 18:55
r29172 | moritz++ | [t/spec] typo in fudg 18:56
moritz_ moritz--
more typos
colomon What's Stringy? 19:00
pmichaud it's the "stringification" method, I think. 19:01
(haven't read the updated spec yet)
moritz_ the abstract role that Str and Buf implement
and prefix:<~> converts to anything that's Stringy
colomon Is it actually implemented anywhere yet? 19:02
pmichaud nafaik
it's really new
like <7 days
moritz_ it's implemented in the test suite :-)
at least partially 19:03
19:03 revdiablo joined 19:04 Nex6 joined
moritz_ phenny: tell qp_pq thanks for your suggestion, but I consider working code to be better than pseudo-code (and working code is also pseudo code simulatneously) 19:05
phenny moritz_: I'll pass that on when qp_pq is around.
moritz_ phenny: tell qp_pq also pseudo code can be used for explaining algorithms, but the goal of the book is to teach Perl 6, not abstract concepts 19:06
phenny moritz_: I'll pass that on when qp_pq is around.
19:09 synth joined 19:11 masak joined
s1n pmichaud: ping 19:12
19:14 snearch joined
jnthn pmichaud: pong....if you've still issues. 19:15
19:15 snearch_ joined
moritz_ masak: should I add gge to hugme? 19:16
masak moritz_: yes, please. :)
19:19 hugme joined
moritz_ hugme: list projects 19:19
hugme moritz_: I know about book, gge, hugme, json, november, nqp-rx, perl6-examples, proto, svg-matchdumper, svg-plot, temporal-flux-perl6syn, tufte, web
19:19 IOOIIIOIIO joined
masak hugme: add mberends to gge 19:20
hugme hugs mberends. Welcome to gge!
moritz_ masak: just curious - why did you port PGE and not NQP-RX?
masak moritz_: when I started porting PGE, nqp-rx didn't exist. :) 19:21
moritz_ that's a good point :-)
jnthn A long-held sekrit. ;-)
masak moritz_: besides, though I know that nqp-rx is way cooler, I'm still fascinated by PGE. :)
yep. I started this in early September, I think.
moritz_ hugme: add moritz to gge
hugme hugs moritz. Welcome to gge!
jnthn Wow!
masak checks the git log
moritz_ doesn't plan contributing right now, but wants to save him the trouble of changing the git URL later on 19:22
masak moritz_++ # good reason :)
moritz_ tries the 03-optable.t to see how glacial it really is 19:23
IOOIIIOIIO noob question: is rakudo known to SEGV on fedora12, or have I broken something? 19:24
moritz_ IOOIIIOIIO: what do you do to get it segfaulting?
IOOIIIOIIO perl6 -e 'say "hello"'
moritz_ that... shouldn't segfault 19:25
IOOIIIOIIO or pretty much anything really
masak ouch.
moritz_ IOOIIIOIIO: is that a rakudo you built yourself?
masak IOOIIIOIIO: tell us a bit about your Rakudo and Parrot versions.
moritz_ IOOIIIOIIO: or installed from rpm?
IOOIIIOIIO no, I just installed the rpm
19:25 pnu joined
moritz_ IOOIIIOIIO: then you should open a bug in the fedora bug tracker 19:26
IOOIIIOIIO it's upgraded from fedora11 via yum, so something might be a little crooked in the libs
moritz_ well, that's the task of a package manager to take care things aren't that screwed up
IOOIIIOIIO rpms are rakudo-0.0.2009.09_1.6.0-2.fc12.i686, parrot-1.6.0-1.fc12.i686 19:27
yeah, I'll bug the fedora folks. Or maybe I'll just build it from source. 19:28
moritz_ IOOIIIOIIO: even if you build from souce you should tell the fedora folks
62.24user 0.49system 1:57.33elapsed 53%CPU for 03-optable.t in gge 19:29
IOOIIIOIIO yeah, I will, but building from source will let me play with perl6 sooner :-)
and I might get to find out what's broken along the way... 19:30
moritz_ IOOIIIOIIO++ # good attitude
IOOIIIOIIO yeah, well it might be my fault :-)
masak IOOIIIOIIO: when I don't check myself, I have the opposite approach: "haha, I'm going to find this bug in Rakudo/Parrot"... those are often the times when it turns out to be my fault. :P 19:35
19:41 mikehh_ joined 19:42 synth joined 19:47 synth joined 19:53 riffraff joined 20:03 zaslon joined
zaslon lolmasakhazblogged! masak++ 'What you can do with GGE::OPTable that you couldn't without': use.perl.org/~masak/journal/39929?from=rss 20:03
masak zaslon: you are one slow cookie. :)
zaslon Sorry, I don't understand that command
moritz_ mask++
IOOIIIOIIO ho hum. Just built my own rakudo, and it works just fine. Not that it proves much. I'll be off to fedora-bugs and see if anyone else has seen the problem. 20:04
or maybe I'll play with perl6 first :-)
masak IOOIIIOIIO: fall for the temptation! :)
IOOIIIOIIO: as long as it doesn't prevent you from going to fedora-bugs. 20:05
20:06 carlin joined 20:10 spinclad_ joined 20:11 mikehh__ joined
ng_feed rakudo-ng: (jnthn)++ 20:15
rakudo-ng: Fill out metamodel details somewhat. No doubt this will need and get tweaks and additions, but it'll serve as a first cut.
jnthn I'll implement the above spec in the next couple of days. We can fill out and go from there. :-) 20:16
mathw yay 20:18
moritz_ YaY, something nobody reads :-)
20:19 mikehh__ joined
mathw wow 20:19
github formats pod
github++
jnthn moritz_: That's fine, less complaints. ;-)
mathw fewer :P 20:20
jnthn mathw: Nah, somebody will half complain, surely.
mathw hmm
jnthn It's entirely possible if I really screwed it up that the complaints will be continuous. 20:21
masak now I know why most of the Rakudo people are working on ng; it's much nicer to run the spectests there...
mathw masak: I suspect it's not just that
masak mathw: oh, it is. believe me. :) 20:22
mathw jnthn: I assume that Attribute figures out if it's private or not, and thus if it needs to install accessor methods, by looking for the ! or . in its name? 20:23
jnthn No 20:24
moritz_ jnthn++ # meta model
jnthn By whether accessor is set to true.
In the call to .new
All attributes are $!foo under the hood.
But the Attribute is responsible for generating the accessor if that is true. 20:25
mathw Then how?
argh
sorry
I had massive lag
I'm an idiot, :accessor is right there in the spec
jnthn Yeah, you should make less mistakes like that. :-P 20:26
masak (making less mistakes)++
mathw I guess I'm not really awake
jnthn You got fewer sleep than needed last night? 20:28
mathw no I've just not been running on a full tank for a while
and I'm back to work tomorrow, I think that's the main reason 20:29
I don't want to go back, and I've just been moping around today
jnthn Aww. :-(
mathw I've had two weeks off
and I've been interviewing for a job that's much better, so it's not the best situation 20:30
20:30 masak joined
mathw ah well 20:30
20:30 mikehh joined
mathw I shall survive it :) 20:30
jnthn Hope the interviews land you somewhere better.
mathw I have stocked up on chocolate
jnthn \\o/
mathw so far it's promising :)
jnthn Cool :-)
mathw just got the last round to go
ng_feed rakudo-ng: masak++
rakudo-ng: Added infix:<!~~>.
mathw where they pay for me to go visit them
masak++ 20:31
masak: although... does it work? :P
masak mathw: for me, it does.
mathw that'll do
masak mathw: I exercised it really well before I pushed.
mathw you're much meaner to Perl 6 than I am
jnthn masak ran all the spectests. He told me so. 20:32
;-)
mathw :)
masak all of them! several times!
mathw I'm sorry my ambitions to do useful things for Rakudo didn't pan out, but those job interviews were distracting (and unexpected)
But I'm glad masak is
masak mathw: ...distracting and unexpected? :)
moritz_ so how did they go? 20:33
mathw masak: yes
masak mathw: boo!
mathw moritz_: actually they were quite fun
masak :P
jnthn mathw: That's fine - it sounds like a long-term win for you if it all works out. :-)
mathw Damn well should be
20:34 diakopter joined
mathw :D 20:34
Just as long as they don't keep me waiting too long for their decision
moritz_ hasn't reached the interviews stage yet 20:35
mathw moritz_: are you looking?
moritz_ mathw: yep. Finishing my Diploma in mid December
mathw aaah
good luck!
moritz_ thanks 20:36
mathw I wasn't looking, but they said hey, want to talk to us about a groovy job?
masak what happens on (0..Inf).pick ?
mathw So I did self.think for about pi seconds
and said yes
moritz_ masak: I don't think there's a meaningful answer to that
mathw masak: hmm
jnthn Wasn't that a bit irrational? 20:37
masak mathw++
mathw jnthn: when has rationality anything to do with it? :P
masak moritz_: no, me neither.
moritz_: but what happens?
mathw masak: I would say, it might return a number, or it might sit there forever until your stack explodes
moritz_ masak: fail("I can't make up my mind how to pick something from an infinite list") 20:38
mathw moritz_: that would be a sensible approach, I think
moritz_ s/list/range/
mathw it could do oh look it's an infinite list
jnthn mathw: It was a bad joke about pi :-P
mathw rand(Inf)
jnthn: I know :)
jnthn :-P
moritz_ I'm sure jnthn didn't want to pi()ss off anybody :-) 20:39
mathw actually I thought about it for a little longer than that
including checking to make sure it was genuine
one does get job spam
jnthn Yeah 20:40
I get loads, but thankfully lots of it is crazy obvious.
masak rakudo: say rand() * Inf
mathw yeah
p6eval rakudo 7347ec: maximum recursion depth exceeded␤in Main (file <unknown>, line <unknown>)␤
jnthn "Want to work from home?" - I already do...
:-)
mathw the genuine stuff is properly written
heh
masak o.O
masak submits rakudobug
mathw also, I find 'earn as much as £5 an hour' to be quite offputting
jnthn ng: say rand() * Inf
p6eval ng ebaea1: too many arguments passed - 0 params expected␤current instr.: '&die' pc 7545 (src/builtins/control.pir:18)␤
mathw I earned £5 an hour when I was 17
moritz_ mathw: but you didn't pay taxes back then :-) 20:41
mathw true
scowl scowl mutter mumble
moritz_ somebody offered me a telecomute jobs at the OTRS developers 20:42
mathw I'm not sure I'd lik ethat 20:43
the option to work from home sometimes is nice
but there's something good about going to an office and being with everyone in the same room to discuss
moritz_ aye 20:44
and separating $work from $home
mathw yes, which is something I'm quite bad at
20:45 mikehh_ joined
masak pmichaud: ping 20:48
jnthn pmichaud: ur doin it rong!
;-)
rakudo: my $x = Any.new; say $x.perl 20:51
p6eval rakudo 7347ec: Any.new()␤
jnthn o: my $x = Any.new; say $x.perlng
ng: my $x = Any.new; say $x.perl
p6eval ng 218aee: Method 'Iterator' not found for invocant of class 'Any'␤current instr.: 'perl6;Perl6Object;' pc -1 ((unknown file):-1)␤
mathw hmm 20:53
jnthn: looking at the source code, there's method perl in Any-list.pm, and it assumes it can call map on self 20:54
and renders like it's a list 20:55
hmm
I guess it can call map, because that's also defined in Any-list 20:56
but map is presumably not working properly in this case
why is map defined in Any?
I don't understand this
20:57 masak` joined
moritz_ mathw: so that you can doo <foo bar>.map and <foo>.map 20:57
mathw: even though <foo> returns a Str, not a List 20:58
mathw it does?? hmm
why
it should be a one-element list, shouldn't it
although it seems things have been arranged so that you can treat it that way
jnthn masak++ and I agree that .perl for List being in Any is wrong. map does belong there, however. 20:59
moritz_ agreed 21:00
mathw so what should .perl in Any do?
masak fixing it now.
moritz_ mathw: nothing
mathw: there's a .perl in Object, iirc
masak mathw: not be in Any-list
mathw I guess it simply cannot know
moritz_ which should be the default .perl returning $type.new(%attribs.perl)
mathw you need to be something that knows it's a list to start doing listy things in .perl
21:01 astrojp joined
jnthn mathw: Right, thus why it almost certainly belongs in List. 21:03
mathw mmm
and we don't seem to have a List.pm 21:04
masak I'm on it. 21:05
and learning more about ng in the process. :) 21:06
mathw is too
RACE!
although I've got to update parrot first
masak mathw: good luck, punk. 21:07
ng_feed rakudo-ng: masak++
rakudo-ng: moved .perl from Any-list to List
mathw pfft
having commit access is cheating
:P
masak I guess.
moritz_ mathw: submit a few more patches, send in a CLA...
masak but I didn't make the rules.
mathw :D 21:08
moritz_: that's the plan
hey guess what
masak: you did exactly what I'm halfway through doing 21:09
mathw is happy about that
masak mathw++ # obviously a guy with good judgment
mathw except you didn't put the vim filetype hint in the file, as many of the others in core/ have 21:10
jnthn meh, who uses vim.
;-)
mathw me
although I seem to be lacking perl6.vim at the moemnt
which is odd
moritz_
mathw as i used to have it
jnthn Yes, amusingly I think masak++ does too ;-)
masak mathw: oh!
yeah. 21:11
right now, in fact.
but I don't care much about the difference between Perl 5 and Perl 6 syntax highlighting.
moritz_ does
mathw it's not the most complicated method ever :)
especially since you just cut and pasted it
masak mathw: and yet you only got half-way. :P 21:15
mathw nonsense
I was about to build it
but I had to update parrot
masak ok :) 21:16
21:17 pmurias joined
mathw aaaaargh 21:18
I updated parrot
and it still thinks its too old
did I remember to run make install on parrot 21:19
apparently not
and that doesn't help 21:20
how odd
21:21 masak joined
dalek p-rx: 19d5ef4 | pmichaud++ | (3 files):
Convert 'from' to 'p' (:p/:pos).
 21:24
p-rx: ba1f213 | pmichaud++ | (7 files):
Add 'c'ontinue option to regexes.
p-rx: e395215 | pmichaud++ | src/Regex/Cursor.pir:
Change check for :rule type.
p-rx: 88e6e26 | pmichaud++ | src/stage0/ (4 files):
Update bootstrap.
ng_feed rakudo-ng: masak++ 21:25
rakudo-ng: added vim setting line to the new file
21:25 masak joined
Wolfman2000 moritz_: I did some reading of the PIR after having just woken up from a nap. PIR makes me wish this was assembly instead of the real gunk 21:28
moritz_ :-)
mathw PIR is so much higher level it's silly to call it assembly 21:29
Wolfman2000 If you want, I'll try playing around with it Monday night after my presentation.
mathw except that it is, in a sense
Wolfman2000 This way I can also help with some of the operator overloading
jnthn ng: my @x = 1,2,3; @x.push(4); say @x.perl; 21:30
p6eval ng 440921: Method 'push' not found for invocant of class 'Array'␤current instr.: '_block14' pc 29 (EVAL_1:0)␤
ng_feed rakudo-ng: (jnthn)++ 21:31
rakudo-ng: We now pass mixed-placeholders.t (and are only a push function off passing slurpy-placeholders.t, it appears).
rakudo-ng: (jnthn)++
rakudo-ng: Various fixes to placeholder parameters.
mathw jnthn: we don't have .push in ng yet? 21:32
jnthn mathw: Seems not. 21:35
It wasn't quite right in master though, irrc.
*iirc
Wolfman2000 jnthn: I checked for push in any-list.pir on master, but...didn't find it
jnthn So it's not a copy-paste job.
Wolfman2000: It'd be in Array
Wolfman2000: List is immutable.
Wolfman2000 So not in builtins then 21:36
jnthn maybe in master it'd live in src/classes/Array.pir
mathw it's in src/classes/Array.pir
I'm looking at it wondering if it can be ported fairly straight across
although hopefully into src/core/Array.pm so we don't have to do all that signature stuff
pmichaud no, it cannot.
mathw pmichaud: somehow I'm not surprised 21:37
pmichaud well, arrays now are "isa RPA" instead of "hasa RPA"
mathw ahah
Wolfman2000 RPA?
pmichaud and I'm thinking that push/pop/shift/unshift should all be written in terms of splice anyway
jnthn pmichaud: Other way round?
pmichaud jnthn: right, other way round (thanks)
jnthn :-) 21:38
pmichaud they're now hasa RPA instead of isa RPA
mathw Wolfman2000: ResizablePMCArray, I believe
pmichaud has been fighting regex integration into ng all day :-(
Wolfman2000 mathw: ugh...I am NOT a fan of those.
jnthn pmichaud: Aww...it turned out harder than expected?
mathw pmichaud: in maste rit does use splice, but it does a lot of faffing about first
pmichaud jnthn: yeah -- trying to get the methods to be standard subclass of Method has been a pain
mathw ah 21:39
it goes through args, packs them all into Perl6Scalar, then splices args onto the end of self
jnthn pmichaud: Ah, how so?
pmichaud jnthn: just that it doesn't seem to work right "out of the box"
Method 'ACCEPTS' not found for invocant of class 'Regex'
but if I add a ".sub 'ACCEPTS' :method" to the Regex class, that doesn't help. 21:40
jnthn That's decidedly odd.
pmichaud are we still allowed to write methods in PIR ?
jnthn I believe so. 21:41
I mean, Code.new is writte in PIR for example.
pmichaud right.
jnthn Amongst many other bits and pieces.
pmichaud I saw that.
Wolfman2000 pmichaud: In perl 6? I wonder why we would want to.
pmichaud I'll try it again.
Wolfman2000: because sometimes we can't do things in Perl 6 that can be done in PIR 21:42
in this particular case, we have to be able to set the caller's $/
Wolfman2000 If I remember right, $/ is a capture variable
...and Perl generally doesn't allow you to write that
pmichaud Wolfman2000: I'm not sure that's true, but assuming that it is -- that'd be why we'd have to do it in PIR instead of Perl 6 :) 21:43
jnthn pmichaud: OK - feel free to push the not-quite-working code if you want me to peek. 21:44
pmichaud: Nothing comes to mind about why that would not work.
pmichaud nopasting
gist.github.com/240739 21:45
jnthn looking 21:47
pmichaud changes pushed
ng_feed rakudo-ng: pmichaud++ 21:48
rakudo-ng: Merge branch 'ng' of git@github.com:rakudo/rakudo into ng
rakudo-ng: pmichaud++
rakudo-ng: Updates with attempt to get Regex.ACCEPTS to work.
rakudo-ng: pmichaud++
rakudo-ng: Add initial version of regex matches.
21:49 IllvilJa joined
jnthn pmichaud: Will I need Parrot version bump for this? 21:49
mathw realises that his PIR might work better if he didn't put semicolons on all the line endings 21:50
Wolfman2000 ugh. I just took a look at the push sub in master. Why must it use instructions I haven't seen yet in the tutorial when other subs use the push command? 21:51
mathw because you need to read more of the tutorial?
Wolfman2000 So future chapters cover setprop, :slurpy, :subid, set_signature_elem, etc? 21:52
mathw set_signature_elem is part of Rakudo
I believe
jnthn It is. 21:53
mathw it's part of the block that creates the method's Perl 6 signature
mathw is trying to define push in Perl 6... using inline PIR
this is not good for my night brain
but it's kind of interesting
Wolfman2000 mathw: I have a feeling you may want to start with copying what's in master and then trying to optimize. 21:54
pmichaud please don't do that
the version in master is *wrong*
Wolfman2000 ...nevermind
pmichaud like "not even close" sort of wrong
because we've fundamentally changed the way lists and arrays are stored
mathw Which is why I'm not 21:55
I am learning more about PIR
just trying to persuade the compiler to swallow it
pmichaud jnthn: yes, you need Parrot version bump
mathw so even if I don't come up with anything that works, it's good
21:55 frew_ joined
jnthn pmichaud: OK, doing so. 21:56
pmichaud: Nothing looks immediately wrong with the patch though :-S
ng_feed rakudo-ng: (jnthn)++ 21:57
rakudo-ng: Parse various junctional operators.
21:58 Su-Shee left
pmichaud jnthn: yeah, I'm a bit stumped by it. I mean, it clearly knows that it's a Regex. 22:01
mathw oh
jnthn pmichaud: Right.
mathw "syntax error, unexpected $end" has to be one of the least enlightening things I've seen all day
jnthn pmichaud: New build takes a while here. :-S
mathw but it's bedtime, so I'm sure when I get back someone else will have done push right and I can see how :)
night all 22:02
jnthn night, mathw
pmichaud ...wtf? 22:05
imports = split ' ', 'PAST PCT HLL Regex _dumper'
that can't be right
(from Perl6::Compiler.pir)
oh, I guess it is right. 22:06
ummmmmm, okay, I know what's wrong then
jnthn oh :-/
pmichaud The Regex from nqp is overwriting and losing the Regex locally
jnthn *nod*
Ouch :-(
pmichaud well, there's actually nothing *in* the Regex space itself that is needed -- just Regex::Cursor 22:07
jnthn Can you just import that?
pmichaud well, it's a bit harder since it's not a top-level namespace, but I suspect I can find a way to do it. 22:11
or perhaps the real answer will be to simply subclass the parrot classes and use the rakudo ones. We'll have to do that anyway. 22:12
I'll put in a cheat for now.
jnthn k
22:13 enigmujo1 joined 22:18 enigmujo left
quietfanatic According to my dictionary, the aforelinked Japanese twitterer was saying "Perl 6 is too powerful; it seems kind of like Trunks" 22:18
Wolfman2000 quietfanatic: Perl as a Dragon Ball [Z] character?
quietfanatic where Trunks is, I think, the name of a Dragon Ball Z character.
Wolfman2000 Why don't we just put programming languages in other media?
22:19 facsimile joined
quietfanatic Wolfman2000: Perl 6 is more like the kid with the huge power level who can't control it yet 22:19
moritz_ aye
Wolfman2000 ...there's only one way to do this.
moritz_ we need to develop idioms to guide poeple
Wolfman2000 <Vegeta> IT'S OVER -- 22:20
                             
                              
                             
                       
                             
quietfanatic That almost says something...
I just see a bunch of [0016] boxes
Wolfman2000 It's supposed to be 9000
In boxes
22:21 masak joined
zaslon lolmasakhazblogged! masak++ 'November 22 2009 -- think globally, act globally': use.perl.org/~masak/journal/39930?from=rss 22:21
masak :)
zaslon: you're fast again!
zaslon Sorry, I don't understand that command
22:21 cognominal joined
masak zaslon: are you sorry you don't understand this command either? :) 22:21
zaslon Sorry, I don't understand that command
masak thought so
22:22 Avada joined
pmichaud ouch, this is turning out to be a real pain. 22:23
moritz_ doing what? 22:24
masak regexes?
pmichaud I need to get ['parrot';'Regex';'Cursor'] to appear as ['perl6';'Regex';'Cursor']
(namespaces)
masak ah.
pmichaud I've never done importing of a second-level namespace before.... and parrot's builtins don't quite handle it.
moritz_ thought that was what type maps do
not importing, but mapping 22:25
pmichaud type maps aren't that sophisticated either, alas. But really this is simply that I need one namespace to appear in two places.
(which we do for a number of other top-level namespaces)
oh, perhaps it's just faster to do it the right way than the cheat 22:26
pmichaud goes to work on that :-(
nope, that's not going to be any quicker 22:27
*sigh* 22:28
Tene :(
pmichaud the fact that Parrot doesn't allow invocables besides Sub is _really_ annoying.
at least, not good ways of doing it.
jnthn pmichaud: Give them an "invokable" role.
pmichaud: If that isn't working, I think it's a Parrot bug.
pmichaud jnthn: that doesn't help me in NQP-rx, which is where the root problem really lies. 22:29
jnthn Code should have 'em though.
Oh OK.
;-(
pmichaud what I *really* want is to be able to have the regex methods that nqp-rx (or PGE, for that matter) compiles be classed different than "Sub"
anyway, that problem isn't going to be solved this decade.
(where decade ends in about 40 days)
maybe I just "cheat" for now and go with Perl6Regex as the internal class name. 22:30
22:31 cognominal_ joined
jnthn night o/ 22:33
cognominal_ night \\o 22:34
22:49 ashizawa_ joined
pmichaud argggggggh 22:51
okay, regexes now run, but it's a pain to get $/ to be set properly, because &infix:<~~> is imposing its own $/
so, this begs the question... what is responsible for setting the caller's $/ ? Is it .match, Regex.ACCEPTS, or &infix:<~~> ? 22:52
moritz_ so you need to implement some form of OUTER::?
pmichaud not OUTER 22:53
caller
moritz_ right
pmichaud and we have to know how many callers to go up
or we we have to implement 'lift'
which I'm not quite ready to do
moritz_ ouch.
pmichaud eh, I guess all of them could do it. 22:56
each one is responsible for setting its callers $/
that can work.
t/00-parrot/10-regex.t ............. ok 23:01
moritz_ pmichaud++ 23:02
pmichaud I'm sure there are lots of broken corner cases.
hugme tweet rakudoperl rakudo-ng now has (limited) regexes 23:04
hugme hugs pmichaud; tweet delivered
ng_feed rakudo-ng: pmichaud++
rakudo-ng: Merge branch 'ng' of git@github.com:rakudo/rakudo into ng
rakudo-ng: pmichaud++
rakudo-ng: Enable regexes, at least a limited form of them. I'm sure there
moritz_ I was about to say "it doesn't have regex support until you push", but you really did push first :-) 23:05
pmichaud yes, I did.
my sequence these days is becoming "push, tweet"
moritz_ it should be 'test, push, tweet' :-)
pmichaud oh, I did the test already, yes. :) 23:06
cognominal_ pmichaud++ 23:08
moritz_ pmichaud: S02-whitespace_and_comments/one-pass-parsing.t fails 23:10
pmichaud in ng? I'd not be surprised :) 23:11
moritz_ it used to pass
pmichaud I'd still not be surprised :)
looking
moritz_ but maybe that's because eval didn't work properly back then
FWIW I'd be fine with commenting it out of spectest.data again
pmichaud oh, sure
moritz_ it might have been a bogus pass anyway
pmichaud we don't recognize 'regex' yet. 23:12
the only regexes that we really handle properly are /pattern/
moritz_ that explains a lot :-)
pmichaud (we used to recognize 'regex', but it was the NQP form of 'regex' and not the rakudo one)
so yes, taking it out of spectest.data is wisest for now
moritz_ pushes 23:13
ng_feed rakudo-ng: moritz++
rakudo-ng: re-disenable one-pass-parsing.t, it was a bogus pass before
moritz_ from S05-metasyntax/changed.t 23:14
# (The /s modifier is gone.)
eval_dies_ok('$str ~~ m:s/./', '/s modifier (as :s) is gone');
that's so wrong, isn't it?
m:s exists, it's just something else today, no? 23:15
(:sigspace)
pmichaud we've never parsed it correctly, though
moritz_ I know
but the test as Perl 6 is wrong
pmichaud ummm. I dunno off the top of my head.
moritz_ asks ack 23:16
ack confirms :-)
Wolfman2000 Merge branch 'ng' of git@github.com:rakudo/rakudo into ng <-- does this mean ng is now master?
moritz_ Wolfman2000: nope 23:17
Wolfman2000: it means that two people had different versions of "ng", and merged them
it didn's say "... into master" :-)
s/s/t/
nqp: token { a }; say 'alive'; 23:18
p6eval nqp: Confused at line 1, near "token { a "␤current instr.: 'parrot;HLL;Grammar;panic' pc 519 (src/cheats/hll-grammar.pir:197)␤
moritz_ nqp: token { a }; say('alive');
p6eval nqp: Confused at line 1, near "token { a "␤current instr.: 'parrot;HLL;Grammar;panic' pc 519 (src/cheats/hll-grammar.pir:197)␤
moritz_ nqp: grammar Foo { token { a } }; say('alive');
p6eval nqp: Unable to parse blockoid, couldn't find final '}' at line 1␤current instr.: 'parrot;Regex;Cursor;FAILGOAL' pc 1652 (src/Regex/Cursor-builtins.pir:179)␤
pmichaud tokens want a name
moritz_ nqp: grammar Foo { token b { a } }; say('alive');
p6eval nqp: alive␤
moritz_ nqp: grammar Foo { token b { \\A } }; say('alive'); 23:19
p6eval nqp: Obsolete use of \\A as beginning-of-string matcher;in Perl 6 please use ^instead at line 1, near " } }; say("␤current instr.: 'parrot;HLL;Grammar;panic' pc 519 (src/cheats/hll-grammar.pir:197)␤
pmichaud lta error there -- I'll fix it.
moritz_ nqp: grammar Foo { token b { \\z } }; say('alive');
p6eval nqp: Obsolete use of \\z as end-of-string matcher;in Perl 6 please use $instead at line 1, near " } }; say("␤current instr.: 'parrot;HLL;Grammar;panic' pc 519 (src/cheats/hll-grammar.pir:197)␤
moritz_ nqp: grammar Foo { token b { \\Z } }; say('alive');
p6eval nqp: Obsolete use of \\Z as end-of-string matcher;in Perl 6 please use \\n?$instead at line 1, near " } }; say("␤current instr.: 'parrot;HLL;Grammar;panic' pc 519 (src/cheats/hll-grammar.pir:197)␤
moritz_ adds a space to that error message 23:21
pmichaud $ ./perl6 23:22
> say "abcdefabcdef".match(/d/).from
3
> say "abcdefabcdef".match(/d/, :c(5)).from
9
lambdabot <no location info>: parse error on input `,'
<no location info>: parse error on input `)'
moritz_ pmichaud++
that will make Str.split(Regex) *so* much easier
pmichaud yes, I had to rework the logic anyway so went ahead and did that. 23:23
dalek p-rx: a10da9f | moritz++ | src/Regex/P6Regex/Grammar.pm:
add a space to the "Obsolote use" error message
 23:24
moritz_ loves small optimizations that make programs more usable 23:25
for example gqview, and image viewer, has the "rename" context menu item 23:26
and if your file is named foo.jpg, then the 'foo' part is already marked
so you can go ahead and type a new name without havin to re-type the extension
Juerd Windows Explorer does that too :) 23:27
moritz_ the last Windows Explorer I worked a lot with was that of Win 98 SE
I don't think it did that back then :-)
Juerd Heh :)
pmichaud okay, got 'eval' working today, got simple regexes working. 23:28
ng_feed rakudo-ng: pmichaud++
rakudo-ng: Eliminate the "Perl6Regex" cheat, enable :c(ontinue) option in regexes.
pmichaud That's an earned break for an hour or three, I guess. :) 23:29
pugs_svn r29173 | moritz++ | [t/spec] remove a bogus test; the :s (sigspace) modifier exists, it just has a different meaning today
moritz_ indeed :-)
pmichaud oh, I suspect we can implement :g(lobal) too pretty soon. 23:30
moritz_ does that make sense inside a regex? 23:33
pmichaud oh, not in a regex, no 23:34
but we can do "foo".match(/pat/, :g)
moritz_ you mean in .match?
pmichaud yeah
and then use that directly inside of split, subst, etc.
(instead of repeating it several places as we do now)
Juerd Is .match preferred to ~~? 23:35
Stylewise
moritz_ not really
pmichaud no, probably not
moritz_ it's more of a stop-gap
pmichaud no, it's spec
~~ delegates to .ACCEPTS, which then does .match
Juerd I see
moritz_ it makes passing of pairs easier than ~~, for which you need more parsing goodness
a specced stop-gap :-)
pmichaud well, even with ~~ it'd be rx :g /pat/ 23:36
or ~~ m :g /pat/
which puts the :g on the 'rx' or 'm', not on the smartmatch
moritz_ right 23:37
Juerd Are those spaces required or can you do m:g/pat/ too?
moritz_ works without spaces too 23:38
std: 'a' ~~ m:g/pat/
p6eval std 29172: ok 00:01 106m␤
Juerd keeps forgetting that you can just ask the bots
moritz_ notices that "git pull" says "Already up-to-date." while "git push" says "Everything up-to-date" 23:42
inconsistent
cognominal_ I guess that's a requirement for the Turing test :) 23:43
moritz_ is not sure if it'd pass a Turing test :-) 23:44
23:44 spinclad_ joined, frettled joined
moritz_ s/it/he/ 23:49
hugme: tweet rakudoperl rakudo-ng now passes 454 spectests 23:53
hugme hugs moritz_; tweet delivered

Powered by Cro and the Raku® Programming Language.

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

Thank you!