»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
00:00 araujo left
timotimo why did this code pass Nil to p6parcel? 00:00
00:01 hypolin joined, hypolin left
timotimo that seems to be an argument to fill the array with Nil? 00:02
if there's nothing in it, huh.
00:07 araujo joined, araujo left, araujo joined
timotimo r: my Mu $res := nqp::list(); nqp::push($res, True); nqp::push($res, Nil); say nqp::p6parcel($res, Nil); 00:08
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«True Nil␤»
timotimo where does it disappear? :\
00:08 dmol left 00:09 rurban1 joined
FROGGS r: say all("5", "4").match(/4/).DUMP 00:09
camelia rakudo-jvm f614e5: OUTPUT«Junction<1>(␤ :$!storage(Array<2>(␤ :$!flattens(True),␤ :$!items(BOOTArray<4>(▶Match<5>(␤ :orig("4"),␤ :from(0),␤ :to(1),␤ :CURSOR(Cursor<9>(:$!ast(Any))),␤ :ast(▶An…»
..rakudo-parrot f614e5: OUTPUT«Junction<1>(␤ :$!storage(Array<2>(␤ :$!flattens(True),␤ :$!items(QRPA<4>(▶Match<5>(␤ :orig("4"),␤ :from(0),␤ :to(1),␤ :CURSOR(Cursor<9>(:$!ast(Any))),␤ :ast(▶Any)…»
FROGGS r: say all("5", "4").match(/4/).perl
camelia rakudo-parrot f614e5: OUTPUT«all(Match.new(orig => "4", from => 0, to => 1, ast => Any, list => ().list, hash => EnumMap.new()))␤» 00:10
..rakudo-jvm f614e5: OUTPUT«all(Match.new(to => 1, hash => EnumMap.new(), ast => Any, list => ().list, orig => "4", from => 0))␤»
timotimo maybe it's just the output of the all that's wrong. 00:11
nope, the storage contains just one thing 00:13
oh 00:14
@values.eager, eh?
huh, weird 00:15
it makes Junction.new(nqp::p6parcel(...), :type($type))
but the signature is *@values
so ... wha?
00:16 ajr_ left
timotimo tries to | the p6parcel 00:17
i don't know how to things, so i just try
clsn_ I think the push is what flattens the Nil. It does not seem to be possible to make Nil an element in a list/array. 00:26
timotimo oh?
nil shouldn't disappear in lists any more
we have () for that
clsn_ Nil definitely disappears in lists. 00:28
r: say [Nil].perl
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«[]␤»
timotimo noooooo! why? :( 00:29
i don't think that's right
geekosaur I thought the re-specced Nil was somewhat NYI in Rakudo 00:30
timotimo might very well be
timotimo looks to try to figure out where that happens
jnthn Trouble is, we relied on that in a lot of places.
So it's a total nuisance of a spec change.
I'd kinda figured we'd just kick it down the road until the next big lists overhaul Pm had in mind, to let us do hyper and race right. 00:31
timotimo it seems simple to make it in ListIter and then i guess everything explodes everywhere 00:32
i'm too impatient :)
00:32 hoverboard left, FROGGS[mobile] left, FROGGS[mobile] joined
timotimo clsn_: i'm sorry :( 00:38
i suppose we need a non-flattening subtype of Nil
00:38 hoverboard joined
clsn_ Should the .orig element of a match be stored as a Str? Currently it's stored as a pointer to whatever the original value was, which might be strange. 00:38
timotimo my class Nil is Nil { } ... :P
i think it's a good idea to have that 00:39
clsn_ I tried making a subclass of Nil; didn't work.
timotimo yeah, the subclass of nil must be invisible from the ListIter, but be used by the Junction construction code
it must not nqp::istype() to Nil :P
jnthn clsn_: Yes, orig means original, as in, "what did we start with". If we're matching against it, it's going to be something that can at least coerce to a string.
timotimo which is problematic
clsn_ Junctions know about smart matching, when they're on the right. But not when they're on the left. And you get weird stuff like this: 00:40
jnthn timotimo: I'd rather not put in hacks...it'll make the real solution harder to do...
timotimo yes
clsn_ r: my $a=(any("a","b") ~~ /./); say $a.gist; say $a.perl
camelia rakudo-jvm f614e5: OUTPUT«｢any("a", "b")｣␤␤Match.new(to => 1, hash => EnumMap.new(), ast => Any, list => ().list, orig => any("a", "b"), from => 0)␤»
..rakudo-parrot f614e5: OUTPUT«｢any("a", "b")｣␤␤Match.new(orig => any("a", "b"), from => 0, to => 1, ast => Any, list => ().list, hash => EnumMap.new())␤»
00:40 pecastro left
timotimo FTR, i find that acceptable 00:41
clsn_ Hmm, I got error messages from the gist I thought.
r: my $a=(any("a","b") ~~ /./); say $a.gist; say $a.Str
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«｢any("a", "b")｣␤␤any(a, b)␤»
clsn_ Yeah, I got weirder results... let me confirm.
timotimo is /./ supposed to give all of it?
r: my $a = ("foobar" ~~ /./); say $a.gist; say $a.Str; 00:42
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«｢f｣␤␤f␤»
timotimo yeah, that *is* weird.
clsn_ Wow. what I was getting only happened when I played with numbers... 00:43
timotimo oh, it *does* autothread that?
clsn_ r: any("4","5") ~~ /\d/;
camelia ( no output )
clsn_ r: any("4","5") ~~ /\d/; say $/
timotimo r: my $a = (any("a", "b") ~~ /{ say $/.CURSOR.orig.perl }./); say $a.gist; say $a.Str;
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«｢any(Failure.new(exception => X::OutOfRange.new(what => "Start of substr", got => 5, range => 0..1, comment => Any)), Failure.new(exception => X::OutOfRange.new(what => "Start of substr", got => 5, range => 0..1, comment => Any)))｣…»
rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«any("a", "b")␤｢any("a", "b")｣␤␤any(a, b)␤»
clsn_ There we are.
Whereas $/.perl is perfectly fine.
r: any("4","5") ~~ /\d/; say $/; say $/.perl
camelia rakudo-jvm f614e5: OUTPUT«｢any(Failure.new(exception => X::OutOfRange.new(what => "Start of substr", got => 5, range => 0..1, comment => Any)), Failure.new(exception => X::OutOfRange.new(what => "Start of substr", got => 5, range => 0..1, comment => Any)))｣␤␤Match.new(to => …» 00:44
..rakudo-parrot f614e5: OUTPUT«｢any(Failure.new(exception => X::OutOfRange.new(what => "Start of substr", got => 5, range => 0..1, comment => Any)), Failure.new(exception => X::OutOfRange.new(what => "Start of substr", got => 5, range => 0..1, comment => Any)))｣␤␤Match.new(ori…»
timotimo ouch.
my brain hurts.
clsn_ r: any("4","5") ~~ /\d/; say $/.perl
camelia rakudo-jvm f614e5: OUTPUT«Match.new(to => 6, hash => EnumMap.new(), ast => Any, list => ().list, orig => any("4", "5"), from => 5)␤»
..rakudo-parrot f614e5: OUTPUT«Match.new(orig => any("4", "5"), from => 5, to => 6, ast => Any, list => ().list, hash => EnumMap.new())␤»
clsn_ Wow. And I can't make that happen for any("a","b") matching against various things. 00:45
see, only happens with the numbers: 00:46
r: any("a","b") ~~ /a/; say $/.gist 00:47
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«｢any("a", "b")｣␤␤»
clsn_ r: any("4","5") ~~ /4/; say $/.gist
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«｢any(Failure.new(exception => X::OutOfRange.new(what => "Start of substr", got => 5, range => 0..1, comment => Any)), Failure.new(exception => X::OutOfRange.new(what => "Start of substr", got => 5, range => 0..1, comment => Any)))｣…»
timotimo where the hell does the 5 come from? if it were -5, i'd guess it's one of the result values for failed cursors, but ...
clsn_ Oh, that I understand.
The 5 is from the start of the junction.Str.
timotimo r: "5" ~~ /4/
camelia ( no output )
timotimo r: say "5" ~~ /5/ 00:48
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«｢5｣␤␤»
timotimo r: say "5" ~~ /4/
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«Nil␤»
timotimo i apologize for nil-wrongness :(
clsn_ The match is matching against the .Str representation, 'any("4","5")'. But when it tries to print it out it goes all wobbly. 00:49
Now, "a" ~~ any(rx/a/, rx/b/) will give the right Boolean answer, but matches are about more than Booleans. I guess autothreaded matches is a bit much to ask. 00:50
Dunno why that weird stuff only happens with "4" and not with "a". 00:54
I dunno if smart matching between Signatures is really right. It ignores the names of named params. 00:57
(I've been poring over src/core/*pm way too much lately, just out of boredom/curiosity)
psch i've got a working find_symbol in the optimizer now at least, gonna work on trying to get the whole awesome stuff for exceptions in there during the next few days 01:00
clsn_ all().Bool == True is surprising, though correct.
psch and with that, good night #perl6
01:00 psch left
jnthn clsn_: I think smart-matching between sigs is fairly incomplete. Feel free to work on improving it. :) 01:02
clsn_ Thanks; I might even do that.
I note that any("a", "b") ~~ /./ seems to do something akin to the Right Thing, while any("a", "b") ~~ m/./ returns False. Not sure why. 01:05
timotimo oh shit. the substr autothreads, doesn't it 01:06
clsn_ Yeah, taking the substr of a junction of strings gives you a junction of the results. 01:07
01:10 treehug88 joined 02:02 FROGGS left 02:04 mavcunha left 02:07 mavcunha joined 02:23 FROGGS joined
BenGoldberg r: my @a; say [*] @a 02:34
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«1␤»
clsn_ Parameters can .perl-ize to empty strings; this is Not Good. I'm looking at fixing it. But I note that we have SIG_ELEM_HASH_SIGIL and SIG_ELEM_ARRAY_SIGIL to tell if the param was % or @, but there's no SIG_ELEM_CODE_SIGIL for &. Can I just add that? 02:37
r: say :(Int).perl 02:38
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«:(Int )␤»
clsn_ r: say :(Any).perl
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«:()␤»
clsn_ r: say :($).perl 02:40
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«:(Mu )␤»
clsn_ r: say :(@).perl
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«:()␤»
clsn_ You get the idea.
02:41 abnorman left
BenGoldberg r: say :(%).perl 02:48
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«:()␤»
BenGoldberg r: say :(*).perl 02:49
camelia rakudo-jvm f614e5: OUTPUT«===SORRY!=== Error while compiling /tmp/HGvNc4ue80␤Malformed parameter␤at /tmp/HGvNc4ue80:1␤------> say :(⏏*).perl␤ expecting any of:␤ colon pair␤ signature␤»
..rakudo-parrot f614e5: OUTPUT«===SORRY!=== Error while compiling /tmp/H09lV_hq2r␤Malformed parameter␤at /tmp/H09lV_hq2r:1␤------> say :(⏏*).perl␤ expecting any of:␤ colon pair␤ signature␤»
clsn_ r: say :(&).perl
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«:()␤»
03:10 treehug88 left 03:32 rurban1 left 04:05 mavcunha left 04:13 cognominal__ joined 04:16 cognominal left 04:17 xinming left 04:18 Nate848 joined 04:20 Nate848 left 04:26 hoverboard left 04:33 preflex left, FROGGS left, preflex_ joined, ChanServ sets mode: +v preflex_, rurban1 joined, preflex_ is now known as preflex 04:38 rurban1 left 04:39 xinming joined 04:49 hoverboard joined 04:55 smash left 04:56 brother left 05:03 smash_ joined
moritz good morning 05:11
05:20 BenGoldberg left 05:28 [Sno] left 05:31 xinming left 05:34 rurban1 joined 05:35 xinming joined 05:38 rurban1 left 05:40 brother joined 05:47 abnorman joined
lue Could not download module metadata: No such method 'defined' for invocant of type 'Hash' :( 05:50
moritz lue: somehow you manage to get the most curious error messages 05:51
p: say Hash.defined
camelia rakudo-parrot f614e5: OUTPUT«False␤»
lue It's not me, it's panda! :) 05:52
moritz lue: of course. So they all say :-)
lue Trouble is, the message only points to the CATCH block, not the actual error. Hmm... 05:55
r: CATCH { die "Failure to work: $_.message" }; 05:56
camelia ( no output )
05:56 xinming left, raiph joined 06:24 [Sno] joined 06:33 grondilu joined, thou left
raiph i'm aiming at implementing more of the sort spec 06:35
perlcabal.org/syn/S32/Containers.html#line_430
a first step is to add the Ordering type declarations: 06:36
perlcabal.org/syn/S29.html#Ordering
eg: subset KeyExtractor of Code where { .signature === :(Any --> Any) }; 06:37
the --> doesn't work; but i can just leave that off for now, so: 06:38
r: subset KeyExtractor of Code where { .signature === :(Any) }; 06:39
camelia ( no output )
06:39 rurban1 joined
raiph r-j: subset KeyExtractor of Code where { .signature === :(Any) }; say { $^a } ~~ Code; say { $^a } ~~ KeyExtractor; 06:40
camelia rakudo-jvm f614e5: OUTPUT«True␤False␤»
raiph r-j: print { $^a }.perl 06:41
camelia rakudo-jvm f614e5: OUTPUT«Block.new()»
06:41 FROGGS[mobile] left, FROGGS[mobile] joined
raiph r-j: print { $^a }.gist 06:42
camelia rakudo-jvm f614e5: OUTPUT«Block.new()»
raiph r-j: print { $^a }.signature
camelia rakudo-jvm f614e5: OUTPUT«Signature<2144491140>»
raiph r-j: print { $^a }.signature.gist
camelia rakudo-jvm f614e5: OUTPUT«:(Mu $a)»
raiph r-j: subset KeyExtractor of Code where { .signature === :(Mu) }; say { $^a } ~~ Code; say { $^a } ~~ KeyExtractor;
camelia rakudo-jvm f614e5: OUTPUT«True␤False␤»
raiph #perl6: ^^^ i need to go to sleep now but help appreciated, will backlog 06:45
06:55 kaleem joined 07:14 kaleem left 07:19 darutoko joined 07:32 dayangkun left 07:43 FROGGS joined 07:59 syndrum joined 08:08 zakharyas joined 08:09 Piers joined, Piers is now known as pdcawley 08:11 lizmat left, lizmat joined 08:20 abnorman left 08:22 hoverboard left 08:31 REPLeffect left 08:39 rurban1 left 08:40 dayangkun joined 08:41 kaleem joined 08:42 grondilu_ joined 08:45 grondilu left
arnsholt Have any of you people tried LightTable yet? 08:53
(www.lighttable.com/ for those wondering WTH I'm talking about) 08:54
09:00 sqirrel joined
tadzik yeah, I didn't bother to fight with the docs that much 09:06
preflex tadzik: you have 1 new message. '/msg preflex messages' to read it.
tadzik "to start Vim mode, enter :editor somethingsomething"
...enter where? Enter how? No help about that
09:10 rurban1 joined 09:15 rurban1 left
arnsholt Yeah, I just tried to get it to work on my $work machine, but RedHat doesn't have recent enough libs 09:22
09:26 FROGGS[mobile] left 09:29 FROGGS[mobile] joined 09:34 dmol joined 09:40 FROGGS[mobile] left 09:43 FROGGS[mobile] joined 09:50 pecastro joined 09:55 kivutar joined 09:56 ssutch left 09:59 denis_boyun joined 10:00 bob__ joined
bob__ hi 10:00
FROGGS hi bob__ 10:01
10:01 bob__ left 10:10 smash_ is now known as smash 10:12 dayangkun left, rurban1 joined 10:17 rurban1 left 10:18 fhelmberger joined 10:24 grondilu_ left 10:31 grondilu joined
jnthn 'noon o/ 10:35
FROGGS öl o/
jnthn ...bit early for that :P 10:36
FROGGS hehe, true
cuppa o/ then :o)
jnthn yeah, coffee :) 10:37
FROGGS nice earl grey is nice, too
jnthn ooh, yes :) 10:38
FROGGS I even have clotted cream at home :o)
jnthn And scones? :)
FROGGS no :/
jnthn Darn, two segfaults in my last spectest run...
FROGGS :(
we need to talk about getstdstream some day 10:39
jnthn r: 42 but role { method sink() { say 'sunk' } }; say 1 10:47
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«sunk␤1␤»
jnthn That doesn't say sunk on Moar. Which is the reason for some of our fail.
10:48 denis_boyun__ joined 10:49 denis_boyun left 10:57 daxim left 11:05 rindolf joined
masak good noon, #perl6 11:10
moritz good masak, #perl6 11:11
or... so :-)
masak would I regret naming a project "capital L with stroke"? :) 11:13
or "Ł" for short.
11:13 kaleem left, rurban1 joined, kaleem joined 11:18 atroxaper joined, rurban1 left, mavcunha joined 11:22 erkan left
tadzik Ł sounds good, what can possibly go wrong 11:30
FROGGS what would an Ł on your forehead mean then? 11:31
something good I suppose
timotimo ohai 11:33
11:33 erkan joined
FROGGS o/ 11:35
syndrum WHAT DOES THE FOX SAY?
masak syndrum: what ever it does say, better not trust Norwegians to provide the answer. 11:36
11:38 raiph left 11:57 tipdbmp joined
tipdbmp If I $*OUT.close; $*OUT.open: 'some-file', :w; say 'write something'; # $*OUT is not automatically closed, I have to $*OUT.close;? Why is that? 12:00
12:00 aindilis joined
masak tipdbmp: that's true for all filehandles. 12:00
tipdbmp: there's no timely destriction in Perl 6. 12:01
timely destruction requires refcounting, and refcounting doesn't mix well with concurrency.
and Perl 6 prefers concurrency.
tipdbmp Hm, I see. Thanks. 12:02
masak destruction*
masak .oO( sorry, I was distructed )
jnthn Also, if we want to run on things like the JVM, we can't demand ref-count semantics of it... :)
...though concurrency is very likely one reason they also don't refcount...so what masak said. :) 12:03
12:10 xinming joined 12:14 rurban1 joined 12:18 rurban1 left 12:35 filly joined 12:39 xinming left 12:45 Alina-malina left 12:46 Alina-malina joined, Alina-malina left 12:56 daxim joined 12:58 Alina-malina joined, kaleem left 13:10 dolmen joined
dolmen Looks like the rendered output of the Unicode sysnopsis is missing : perlcabal.org/syn/S15.html => 404 13:11
masak dolmen++
I have no idea why that might happen, though. :/ 13:12
FROGGS this was recently added 13:13
maybe it must be registered in a job that creates the html
moritz I guess that whoever added the synpsis didn't take care that it also be rendered
masak that sounds likely. 13:17
what steps need be taken to make a synopsis rendered?
is that documented somewhere?
moritz if it is, I wouldn't know where 13:18
but mu:docs/feather/script/update-syn is a good place to start investigating 13:19
13:22 pecastro left 13:23 kbaker_ joined, Timbus left 13:24 Timbus joined 13:28 dayangkun joined, pecastro joined 13:29 dayangkun left 13:30 dayangkun joined 13:31 dayangkun left 13:32 dayangkun joined 13:33 kaare_ left 13:41 ajr joined, ajr is now known as Guest17813 13:42 Guest17813 is now known as ajr_
timotimo r: say -9 div 10 13:57
camelia rakudo-jvm f614e5: OUTPUT«-1␤»
..rakudo-parrot f614e5: OUTPUT«0␤»
moritz p: say (-9) div 10
camelia rakudo-parrot f614e5: OUTPUT«0␤»
moritz r: say (-9) div 10
camelia rakudo-jvm f614e5: OUTPUT«-1␤»
..rakudo-parrot f614e5: OUTPUT«0␤»
13:58 FROGGS[mobile] left, FROGGS[mobile] joined
mathw o/ 14:06
timotimo ah, you're just the person we need
(get it? because your name begins with math)
14:09 daxim left
FROGGS .oO( IMO, u are totally OT Tim ) 14:09
mathw ah but I'm english, we do maths here 14:10
timotimo oh! 14:12
okay then
masak mathw: quick, change your nick to mathsw! 14:26
mathw but that would imply I'm good at maths 14:29
14:31 denis_boyun__ left
masak heh. 14:36
14:37 denis_boyun joined, FROGGS[mobile] left, FROGGS[mobile] joined
grondilu wonders if there is a standard test for math literacy. Like IQ for maths or something. 14:42
14:43 kivutar left 14:45 kaleem joined 14:47 dayangkun left, bluescreen10 joined
hoelzro I don't remember if anything came of this question, but why can't one do for @values → $value { ... } as opposed to for @values -> $value { ... }? 14:47
masak guess we tend to keep within the Latin-1 char space. 14:54
[Coke] is that something we could module-ize? 14:55
hoelzro except for « » vs << >>
on a separate note, consider this code: 14:56
gist.github.com/hoelzro/8355699
lines 14-17 and 32-34 are functionally equivalent, right?
masak [Coke]: I expect so.
[Coke]: ...with care, that is. 14:57
[Coke]: because '->' is also an expression stopper in EXPR, IIRC.
hoelzro: should be.
hoelzro ok, because I get some funny output (gist updated) 14:58
14:58 treehug88 joined 15:00 dayangkun joined 15:01 thou joined
masak p: my %h = foo => <1 2 3>; $h<foo> .= grep: { $_ %% 2 }; say %h<foo> 15:10
camelia rakudo-parrot f614e5: OUTPUT«===SORRY!=== Error while compiling /tmp/ikAPN2ZmS1␤Variable '$h' is not declared. Did you mean '%h'?␤at /tmp/ikAPN2ZmS1:1␤------> my %h = foo => <1 2 3>; $h<foo>⏏ .= grep: { $_ %% 2 }; say %h<foo>␤ expect…»
synopsebot Link: perlcabal.org/syn/S1.html#line_1
masak p: my %h = foo => <1 2 3>; %h<foo> .= grep: { $_ %% 2 }; say %h<foo>
camelia rakudo-parrot f614e5: OUTPUT«␤»
masak yeah, that's wrong, IMO.
but I recognize this bug.
either it or something very much like it has been RT'd before.
timotimo in case somebody here is not on #moarvm and would like a task: check out MoarVM/negative_integer_division and try to make t/spec/S32-num/rounders.t pass 15:11
before the changes in that branch, moarvm had the same wrongness in its integer division as parrot does (you have to use a div (b - 1) if a < 0), so i tried to put that into the division ops of moarvm 15:12
now it fails to create a rational properly in DIVIDE_NUMBERS when you have, for example, -0.5
hoelzro also, are hash keys automatically stringified? or is that just a quirk of Pakudo? 15:13
benabik Hash keys are strings unless specified otherwise.
benabik does not remember the syntax for that though.
hoelzro benabik: how do you specify otherwise? 15:14
=(
benabik S09 15:15
my %hash{Any} or {*}
(or whatever type you want to use as the key)
15:15 rurban1 joined
hoelzro nice 15:16
thanks benabik
benabik np
15:21 sqirrel left
Ulti that's neat I didn't realise you could define that type 15:29
15:32 atroxaper left 15:43 SamuraiJack joined
masak lizmat++ # implementing that 15:46
r: my %hash{Str}; %hash<foo> = 42; %hash{42} = "foo"; say "alive" 15:47
camelia rakudo-jvm f614e5: OUTPUT«Nominal type check failed for parameter 'key'␤␤»
..rakudo-parrot f614e5: OUTPUT«Nominal type check failed for parameter 'key'; expected Str but got Int instead␤ in method at_key at gen/parrot/CORE.setting:9101␤ in sub postcircumfix:<{ }> at gen/parrot/CORE.setting:2506␤ in sub postcircumfix:<{ }> at gen/parrot/CORE.setting:…»
masak interesting -- why is the 'expected ...' line missing on JVM?
masak submits rakudobug
benabik masak: Yeah, I only knew about that feature because I remember lizmat++ implementing it. :-D 15:48
15:58 Psyche^ joined 16:03 Psyche^_ left 16:06 gcole_ joined 16:08 gcole left 16:23 [Sno] left, sqirrel joined 16:29 kaleem left
rindolf Hi all. How do I build and install rakudo-jvm? 16:40
I already did «git clone git://github.com/rakudo/rakudo.git» 16:41
colomon look at the README file. ;)
perl Configure.pl --gen-nqp --backends=jvm # copied from the README 16:43
[Coke] also --prefix=/install/path 16:44
then "make install"
colomon never uses --prefix
rindolf colomon: I looked at the INSTALL.
colomon and note that you can say --backends=jvm,parrot and get both
rindolf: …. yeah, kind of weird that INSTALL is a directory and all the installation help you need is in the README. hmm. :\ 16:45
rindolf colomon: I mean INSTALL.txt
colomon oh. which is outright wrong, eh? sigh. 16:46
rindolf colomon: what is? 16:47
colomon INSTALL.txt doesn't seem to mention --backends at all
16:47 FROGGS[mobile] left
PerlJam it's not wrong, just highly parrot-centric and times have changed :) 16:47
16:47 FROGGS[mobile] joined
colomon Ah, I thought --backends was mandatory now. 16:47
PerlJam maybe we should get rid of INSTALL.txt as its relevance is fading and it causes confusion and it's likely to become "more wrong" with time if no one updates it. 16:52
Better to just have the README and update that.
[Coke] +1 for dropping old stuff. 16:53
TimToady feels old and stuffy
[Coke] GONE! 16:54
PerlJam TimToady: we'll try to drop you gently onto plush cushions.
rindolf Is nom a good branch to experiment with rakudo-jvm?
PerlJam rindolf: it's "the" branch unless you're being adventurous. 16:55
[Coke] there is no master, only nom. 16:58
rindolf PerlJam: bye, going.
PerlJam rindolf: happy hacking! :)
16:59 rindolf left
colomon nom is my master now 16:59
17:02 zakharyas left, raiph joined 17:07 SamuraiJack left
TimToady masak: you could name it 七 instead 17:14
17:16 pecastro left 17:27 [Sno] joined 17:29 SamuraiJack joined 17:36 SamuraiJack left 17:37 SamuraiJack joined
raiph r-j: subset Foo of Code where { .signature === :(Mu) }; say { $^a } ~~ Code; say { $^a } ~~ Foo; 17:38
camelia rakudo-jvm f614e5: OUTPUT«True␤False␤»
raiph r-j: say { $^a }.signature 17:39
camelia rakudo-jvm f614e5: OUTPUT«:(Mu $a)␤»
raiph anyone know if i'm making a mistake with the where clause or is this a bug/nyi? 17:40
r-j: subset Foo of Code where { say .signature; .signature === :(Mu) }; 17:41
17:41 ajr_ left
camelia ( no output ) 17:41
17:41 ajr joined, ajr is now known as Guest17988 17:42 Guest17988 is now known as ajr_, SamuraiJack_ joined
moritz r: say { $^a }.signature 17:42
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«:(Mu $a)␤»
moritz r: say { $^a }.signature === :(Mu)
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«False␤»
17:42 SamuraiJack left
moritz raiph: it seems that signatures don't compare as value types in rakudo 17:42
r: say { $^a }.signature eqv :(Mu)
camelia rakudo-parrot f614e5, rakudo-jvm f614e5: OUTPUT«False␤»
17:44 SamuraiJack_ left 17:48 spider-mario joined 17:54 REPLeffect joined 17:55 locsmif joined 18:02 hoverboard joined
[Coke] does perl5 care about spaces after function names on invocation? 18:05
TimToady a little :) 18:06
18:06 denis_boyun left
jnthn back 18:08
raiph: Not sure === on signatures makes sense; ~~ may work out better 18:10
raiph jnthn: yeah, i had come to that conclusion too 18:12
tho of course it doesn't work well atm
++psch ;) 18:13
TimToady p: enum Elems ('A'..'Z'); 18:16
camelia rakudo-parrot f614e5: OUTPUT«===SORRY!===␤No such method 'returns' for invocant of type 'Range'␤»
jnthn Looks like value / AST confusion... 18:18
18:21 ajr_ left 18:23 berekuk joined 18:24 hummeleBop joined
raiph TimToady: does smartmatching of signatures ignore variable names and where clauses? 18:31
18:35 farmG joined
raiph (common sense and research says yes for variable names) 18:43
(so i'm mostly asking about where clauses)
moritz the current enum Actions code sometimes pushes ASTs and sometimes objects onto the same array 18:45
18:46 abnorman joined 18:47 darutoko left 18:53 benabik left
TimToady raiph: where clauses should certainly be enforced, I'd think 18:56
jnthn TimToady: Well, I thin the question is signature/signature, and comparing where clauses is undecidable... :) 19:02
*think 19:03
19:03 rindolf joined
rindolf Hi all. 19:04
Yay! Rakudo JVM is compiled.
And appears to work pretty nicely.
PerlJam rindolf: indeed :)
locsmif But... secretly, it plots to cause havoc :p 19:08
PerlJam
.oO( ... and let slip the dogs of war? )
 19:09
19:09 dmol left 19:11 ssutch joined
rindolf OK, pastie.org/8621484 - this programs gives me a «java.lang.StackOverflowError» 19:15
OK, pastie.org/8621484 - this *program* gives me a «java.lang.StackOverflowError» 19:16
In Perl 5 it runs fine.
# print hyperexp_modulo(3, 3, 1000), "\n"; -this line also runs fine.
I guess I can use a dedicated stack, but it feels hacky.
colomon rindolf: I believe you may be able to expand the stack available to Java? 19:17
rindolf colomon: ah.
colomon (Of course, it may also be a bug.) 19:19
raiph TimToady: well, let me back up. i'm looking at implementing more of the sort spec.
colomon -Xssn
Sets the thread stack size.
raiph TimToady: feather.perl6.nl/syn/S32/Containers.html#sort
colomon rindolf: that might be it?
raiph TimToady: and hence perlcabal.org/syn/S29.html#Ordering 19:20
rindolf colomon: can I pass it to ./perl6 ?
colomon rindolf: no, but if you look at perl6 (in the jvm version) it's a script.
raiph TimToady: so I was thinking we'd declare "subset KeyExtractor of Code where ..." and subset OrderingPair of Pair where ..." and then test if an arg passed to sort matches OrderingPair.
colomon errr, at least in the OS X and I assume Linux versions 19:21
rindolf colomon: yes, I see it now - thanks. 19:22
colomon rindolf: good luck! and please report your results here. :)
moritz ok, this is curious
we kinda use Pairs for enums before class Pair was declared 19:23
rindolf colomon: thanks! It now yields the correct result.
colomon \o/
rindolf++
moritz we do this by checking that a certain piece of AST would evaluate to a pair
but we don't actually evaluate it; we extract key and value separately
19:23 berekuk left
colomon rindolf: btw, is that an implementation of the built-in expmod? 19:24
19:28 farmG left
moritz hates boostrapping issues 19:28
19:30 sqirrel left
rindolf colomon: maybe. 19:30
colomon: well, it's a hyperexp_mod for projecteuler.net/problem=188 19:31
colomon: it's a bit different.
colomon ah, yes
rindolf colomon: thanks for the karma bump. :-)
19:32 hummeleBop left, treehug88 left
rindolf colomon: bitbucket.org/shlomif/project-eule...at=default - this is my Perl 5 solution. 19:34
19:36 treehug88 joined 19:37 berekuk joined
rindolf colomon: OK, expmod can make my life a bit simpler in the second part of the function (after the recursion). Thanks. 19:46
colomon: pastie.org/8621582 - new version. 19:54
19:55 treehug88 left
rindolf colomon: I should note that running the program ends up consuming at least 12.2% of my 8GB of RAM (I'm on x86-64 Linux). 19:56
19:56 jnap joined
rindolf The same program on perl 5 consumes 0.1% of RAM. 19:58
jnthn r: my $b = BagHash.new("a", "b", "b"); my $a = $b.grab; my @a = $b.grab(2); say $b.total; say $b.elems;
camelia rakudo-jvm f614e5: OUTPUT«0␤2␤»
..rakudo-parrot f614e5: OUTPUT«0␤0␤»
rindolf jnthn: hi, what's up? 19:59
jnthn rindolf: hi...not much, just got a few moments now to look into some failing tests 20:00
rindolf jnthn: that's great. 20:01
moritz pokes dalek 20:05
20:07 ajr joined, ajr is now known as Guest94449
timotimo well, that nap kind of took longer than i expected 20:07
20:08 Guest94449 is now known as ajr_, treehug88 joined
tadzik naps are goood 20:08
20:09 dalek left 20:10 dalek joined, ChanServ sets mode: +v dalek
rindolf puts the nap in napkin. 20:12
diakopter lol 20:15
20:16 fhelmberger left
diakopter at least you didn't put the nap in nappie 20:16
20:17 berekuk left 20:19 berekuk joined
masak or in kidnapping. 20:20
TimToady has put kin in a nappie, however 20:21
20:23 jnap left 20:25 pecastro joined, jeffreykegler joined 20:28 pmurias joined 20:30 berekuk left
pmurias masak: what sort of project do you want to name Ł? 20:31
20:31 FROGGS[mobile] left, FROGGS[mobile] joined 20:33 berekuk joined 20:38 jeffreykegler left
rindolf Any ideas how I can make this program faster and/or consume less memory - pastie.org/8621582 - the Ints can be signed 32-bit ones too. 20:39
20:42 pecastro left
arnsholt Anything that doesn't need arbitrary precision can be made into int rather than Int 20:42
Int overflows automatically to arbitrary precision, while int is platform native integers (long on JVM) 20:43
timotimo right, that'll do a big difference
arnsholt That should help with memory as well, since there'll be less objects in play
But the JVM is generally happy to eat your memory, mind
jnthn Not sure it'll actually be an improvement, though. 20:44
20:44 jnap joined
jnthn At present, meta-ops like *= probably box, if they work at all... 20:44
arnsholt Ah, right
timotimo good point; they may very well not even work with int rather than Int 20:45
arnsholt I think that's a sacrifice rindolf can live with though
rindolf Can I also say that the function returns a single int value? 20:47
arnsholt You can, but I'm not sure if the compiler uses that information yet
timotimo it might, but not in a thoroughly worthwhile manner :|
rindolf timotimo, arnsholt : well, for posterity - how can it be done? 20:48
timotimo --> int 20:49
or --> Int
or returns Int (or int)
20:50 dmol joined
arnsholt Or even my int sub foo(...) 20:53
clsn_ jnthn: Taking your advice; I'm all over smart-matching Signatures, and also fixing the .perl representation of Parameters.
rindolf timotimo: two dashes and a great than sign? 20:56
timotimo: two dashes and a greater than sign?
20:56 FROGGS[mobile] left, FROGGS[mobile] joined
clsn_ At first I thought I was mistaken, but now I'm thinking we really do need a $SIG_ELEM_CODE_SIGIL flag. Otherwise, how can I tell between :(Int &) and :(Callable[Int] $)? Have to ponder. 20:57
jnthn There isn't a difference...
preflex jnthn: you have 1 new message. '/msg preflex messages' to read it.
rindolf pastie.org/8621753 - this programs gives me «Cannot modify an immutable value in sub hyperexp_modulo at euler_188.p6:15» 20:58
clsn_ Welllll... I think :(Int @) would be different from :(Positional[Int] $) in that one would have the array sigil flag set. I'm thinking in terms of .perl; how to make a string that exactly (as near as possible) generates the "same" Signature. 21:00
rindolf pastie.org/8621753 - this programs gives me «Cannot modify an immutable value in sub hyperexp_modulo at euler_188.p6:15» - why? And how can I fix it?
s/programs/program/ 21:01
jnthn clsn_: Yes, the @ case is different, for sure
clsn_ But the & case is not? 21:02
jnthn clsn_: Well, it depends if you insist on producing something identical to what the user typed or just semantically equivalent.
@ and % are special because they imply flattening semantics
clsn_ I would think .perl should strive for the former.
21:02 dayangkun left
jnthn Well, so far as is reasonable... 21:02
If you think getting & right is worth trawling through a couple of thousand lines of binder code that you might affect, you can add it. :) 21:03
21:03 denis_boyun joined
clsn_ Well, it's something to consider. I won't go changing World.nqp in my first patch. :) You think it would entail all that? 21:03
jnthn Well, the sigil flags are used in signature binding. 21:04
clsn_ (do I have to fork rakudo on github and all, or can I just use a local repository and send in patches or something?)
Mmm... yeah, something to consider.
jnthn Which is implemented in C for Parrot, Java for the JVM and NQP for MoarVM...thus why I'm a bit hesitant. :)
21:05 skids joined
jnthn You could just send in patches also. But pull requests can be a little easier. 21:05
[Coke] r: my $a = 2; my $b = 3; my $c = 4; ($a *= $b) %= $c;
jnthn But I don't mind processsing git format-patch output :)
camelia ( no output )
[Coke] r: my int $a = 2; my int $b = 3; my int $c = 4; ($a *= $b) %= $c;
camelia rakudo-jvm 263aa0: OUTPUT«Cannot modify an immutable value␤ in block at gen/jvm/CORE.setting:16313␤␤» 21:06
..rakudo-parrot 263aa0: OUTPUT«Cannot modify an immutable value␤ in block at gen/parrot/CORE.setting:16590␤ in block at /tmp/x1CqcyfAxN:1␤␤»
[Coke] rindolf: ^^ ints vs. Ints.
jnthn < jnthn> At present, meta-ops like *= probably box, if they work at all...
[Coke] I assume it's one of the remaining intbugs.
jnthn :)
Right
clsn_ Yeah, I see what you mean by all the bind stuff... OK, well, we'll see what I come up with. Can't really work on it more until tomorrow night, but should get something together soon.
[Coke] jnthn: yes, but he may not know what box implies.
jnthn ok
[Coke] plus also that scrolled up real quick and I didn't see it. :) 21:07
jnthn [Coke]: Sure, but "may not work at all" was clear ;)
rindolf [Coke]: I don't understand - are "int"s immutable?
jnthn clsn_: Sounds good. Certianly .perl on a Paramter shouldn't ever be returning an empty string.
rindolf I mean "int" variables. 21:08
jnthn rindolf: So far, "int" has been implemented mostly to be useful to us for use in built-ins, rather than "ready for the typical user"
rindolf jnthn: ah.
jnthn rindolf: So if you use them you'll run into all sorts for the next couple of months, I'm afraid...
clsn_ Yeah... I originally fixed it so that nameless params always had a bare sigil, so :(Int) became :(Int $), but I think you got too many $s that way. I think I can change it just not to blank out the Any if there's no name, and so on.
jnthn (Fairly high on my todo list to sort those out...)
rindolf jnthn: run into all sorts of what? 21:09
skids
.oO(all sorts of WTF)
 21:10
jnthn rindolf: Places where they don't work like non-native types would.
rindolf: Especially anywhere they need to be passed by reference
clsn_ Yeah, I see what you mean; my int $a behaves very differently from my Int $b. Try setting it equal to Int, or to 19299084908429308087834834598345
r: my int $b=19299084908429308087834834598345; say $b 21:11
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«1169738125069744585␤»
jnthn Well, yeah, it's not gonna be big enough to hold that :)
clsn_ Exactly. Whereas an Int is.
Huh, I got a different value when I tried locally.
21:11 treehug88 left
skids 32 bit vs 64 bit machine, prolly. 21:11
clsn_ Mm, good point.
[Coke] rindolf: if you'd like some six style guide suggestions, I have a few, (though I know you were trying to keep the same code working in both, earlier)
21:12 berekuk left 21:14 berekuk joined
rindolf [Coke]: six style guide? 21:15
[Coke]: what is that?
[Coke] Perl 6 style. (e.g.: parens around if/while conditionals are unneeded) 21:16
rindolf [Coke]: ah, that.
[Coke]: keep the same code working in both what?
[Coke] perl 5 and perl 6
rindolf [Coke]: I wasn't. 21:17
[Coke]: I took my Perl 5 code and translated it to Perl 6.
21:17 dayangkun joined
[Coke] ok. 21:17
clsn_ If six is python3 * python2, shouldn't perl5 and perl6 be thirty? 21:20
timotimo :) 21:21
21:21 ilbot3 joined
clsn_ OK, really have to go idle now. Will resume work on Signatures and Parameters tomorrow night. 21:22
[Coke] 30perl² 21:23
21:25 raiph left 21:32 xenoterracide left
japhb__ wonders if perl³ would be a solid block of perl 21:32
dalek kudo/moar-support: dd0b9a0 | jnthn++ | src/vm/moar/ (2 files):
Implement various missing phaser-related ops.
 21:38
jnthn ooh, dalek's back
That fixes "No registered operation handler for 'p6takefirstflag'" and "No registered operation handler for 'p6setpre'"
21:41 hoverboard left 21:42 xenoterracide joined 21:43 hoverboard joined
[Coke] yah, that's another 42 or so possibles. 21:48
21:48 raiph joined
jnthn Seeing if I can crack theS05-capture/caps.rakudo.moar one 21:49
21:52 ilbot3 left
lue r: say (1,2,3).Buf 21:54
camelia rakudo-jvm 263aa0: OUTPUT«No such method 'Buf' for invocant of type 'Parcel'␤␤»
21:54 FROGGS[mobile] left
camelia ..rakudo-parrot 263aa0: OUTPUT«No such method 'Buf' for invocant of type 'Parcel'␤ in block at /tmp/oC6AQyukr5:1␤␤» 21:54
21:55 FROGGS[mobile] joined
lue r: say Buf.new((1,2,3)) 21:55
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«Buf:0x<01 02 03>␤»
lue I assume there's a reason I can't use .Buf on a Parcel, which Buf.new() gladly takes? 21:56
PerlJam no one wrote it? :) 21:57
lue I'd've thought there was a Cool.Buf or something though. 21:58
moritz how would you convert a Num to a Buf?
or a Match, for that matter?
21:58 locsmif left
moritz and please in a way that is consistent with Int.Buf 21:59
PerlJam very very carefully
moritz++ (for actually *thinking* about the question unlike myself :) 22:00
lue moritz: Ah, that would be it. It's just that I expect $bufobj[range].Buf to Just Work™. 22:01
22:04 BenGoldberg joined
lue Now that I think about it, we don't really have a "just integers" list type, do we? $bufobj[range] would need to return a Buf itself in my case. 22:04
22:05 treehug88 joined 22:07 skids left
grondilu r: my $buf = Buf.new: (^256).roll(5); say $buf[^3].WHAT; 22:11
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«(Parcel)␤»
grondilu r: my $buf = Buf.new: (^256).roll(5); say $buf.subbuf(0, 3).WHAT
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«(Buf)␤»
lue I'm not quite sure I like $list-type[range] -> Parcel , though I'm sure there's a good reason. 22:12
I would use subbuf, but I think it's too awkward for list subscripting :P
grondilu subbuf is the analogous to substr, because a buf can not just be seen as a list of integers, but also as a binary string. 22:14
(that's how I understand it, anyway)
lue grondilu: Yeah, but I never see bufs as "integer view of strings", but rather "low-level array", i.e. what you use for binary data.
22:15 denis_boyun left
grondilu well you don't, but you could. 22:15
lue :) I still think Buf[] returning an un-.Buf-able thing is a smell. 22:16
22:17 kaare_ joined
grondilu well, Str[] does not exist, so I guess Buf[] can't refer to the "string" side of buffers, but rather to the "array" side. So Buf[] should return an array (but I guess a Parcel would do), not a Buf. 22:17
if you want a buf out of a buf, use subbuf, same as you use substr to get a string out of a string. 22:18
lue grondilu: to my knowledge, every list's [] returns a Parcel already.
grondilu there you go then, it does make sense that Buf[] returns a Parcel, then. 22:19
lue r: Buf.new((1e5, Match.new())); # moritz: this is what I think Parcel.Buf should do on non-integer elements.
camelia rakudo-jvm 263aa0: OUTPUT«This representation can not unbox to a native int␤ in method new at gen/jvm/CORE.setting:5048␤␤»
..rakudo-parrot 263aa0: OUTPUT«This type cannot unbox to a native integer␤ in method new at gen/parrot/CORE.setting:5060␤ in method new at gen/parrot/CORE.setting:5050␤ in any at gen/parrot/Metamodel.nqp:1478␤ in block at /tmp/qAnMpz96vv:1␤␤»
lue If Buf has to be the integer-y string type, then I'd like a type that's designed specifically for binary data. (i.e. Buf without the Stringy stuff.) 22:21
dalek rl6-roast-data: b507320 | coke++ | / (3 files):
today (automated commit)
grondilu Blob, maybe? I don't thing Blob does subbuf, does it?
lue Blob is just immutable Buf 22:22
[Coke] moar: 94.76%
lue \o/
grondilu Well, isn't a string just binary data, in the most generic sense? I mean it's like that in P5, iirc.
22:23 mavcunha left
lue Everything's binary data :) I just don't want features on my binary data array weird because "we expect string-ish binary data." 22:23
grondilu what kind of features would you like to avoid, exactly? 22:24
grondilu tries to remember what is a string, exactly 22:25
lue Well, my only issue at the moment is that Parcel.Buf doesn't exist, or alternatively Buf[] returns a Buf.
But this isn't the first time I've been bitten when trying to use Buf because my "binary data" concept conflicts with everyone else's "integery string view" concept. 22:26
dalek p: 2b7389d | jnthn++ | src/vm/moar/QAST/QASTCompilerMAST.nqp:
Cope with duplication of QAST::Block in the tree.
 22:27
jnthn That's S05-capture/caps.rakudo.moar
22:27 hoverboard left 22:28 mavcunha joined
grondilu to me a string is an unstructured binary dataset. It is therefore binary data in its purest form. Instead of willing to avoid the stringy aspect of Buf, you should embrace it, imho. 22:28
lue grondilu: That's totally opposite of how I view strings. They're a very specialized class of binary data to me, and definitely shouldn't inspire design decisions on something more generalized. 22:29
22:29 jnap left
grondilu though I should have mentioned that I was talking about strings as in P5, not as P6's Str. 22:30
22:31 rurban1 left
lue Ah. Strings in Perl 6 are a bit more complex, what with encoding systems and all. *cough* S15 *cough* :) 22:31
22:31 abnorman left
lue fwiw I do see value in having an integer-y view of strings, I just don't want that to be the only low-level array type I have available to me. 22:32
grondilu so instead of being a integerish string-like thing, you would like buffers to be an integerish array-like thing. 22:34
lue Considering you get Bufs out of IO methods when you use :bin, yes. 22:35
22:35 kbaker_ left
lue (if a new type was created for this purpose, that's what :bin would work with of course) 22:35
r: buf8.new(0x100) 22:36
camelia ( no output )
lue O.o Huh, I could've sworn that buf was the place where rakudo cared about bit-size.
grondilu but aren't strings the natural thing that is supposed to come out of IO? If so, a :bin should be an adverb to a "special case" of string. Thus Buf would be closer to a string than to an array. 22:38
lue grondilu: I don't think Strs should be treated as anything near the bytes level. They're pretty high level to me. 22:39
lue &
22:41 rindolf left
grondilu They are. That's why for byte level there is Buf. But Buf still shares some similarities with strings. I mean, Str is the implementation of strings in Perl 6, but they are only a special case of what a string is in a more abstract way. 22:41
22:43 abnorman joined
grondilu It's clearer in P5, I think. I like how in Perl 6 there are two different types for binary strings (Buf) and "textual strings" (Str), but the downside is that it kind of hides the common aspects of both concepts. 22:44
I mean, they really are both strings.
in perl 5 it can be a bit confusing that they are both the same kind of object, but at least it makes it clear that they are indeed the same kind of object. 22:46
22:46 rurban1 joined
grondilu thouh I guess you could say that a buffer is not at all a string, but rather it's an array of integers, while a string is an array of characters. 22:52
lue Remember, I'm fine with an integer view of strings --- that's probably helpful to somebody. I just want a low-level array type whose design is not influenced by a subset of binary data types. 23:00
jnthn lue: When it's implemented, that'll be called "my int @a" :) 23:02
lue :) Could IO.meth(:bin) not return a Buf, but an Array[int8] or similar then?
jnthn No, I think Buf is right there 23:03
lue (when I use :bin, I specifically do not want to treat the file's contents as strings. But that's just me.)
jnthn Well, Buf ain't a string, it's something you can choose to turn into one, or at least try to
lue jnthn: yeah, my issues merely come from when I run into a Buf design choice that was informed by "integer view of string" 23:04
jnthn lue: Which one, ooc?
lue But all this truly started because Buf[$range].Buf doesn't work. Which is only tangentially related to my history with Buf's design choices. 23:05
r: say Buf.new(1,2,3,4,5)[2..4].Buf;
camelia rakudo-parrot 263aa0: OUTPUT«No such method 'Buf' for invocant of type 'Parcel'␤ in block at /tmp/u0wuH_Gfke:1␤␤»
..rakudo-jvm 263aa0: OUTPUT«No such method 'Buf' for invocant of type 'Parcel'␤␤»
jnthn lue: What's wrong with .subbuf for that? 23:06
lue r: say Buf.new(Buf.new(1,2,3,4,5)[2..4]);
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«Buf:0x<03 04 05>␤»
lue jnthn: feels too clunky and string-like for my low-level-array-ing :)
grondilu r: say .[^3].Array given my Array @a = ^10; 23:07
camelia rakudo-jvm 263aa0: OUTPUT«Type check failed in assignment to '@a'; expected 'Array' but got 'Int'␤ in method REIFY at gen/jvm/CORE.setting:8337␤ in block at gen/jvm/CORE.setting:7220␤ in method gimme at gen/jvm/CORE.setting:7628␤ in method at_pos at gen/jvm/CORE.setting…»
..rakudo-parrot 263aa0: OUTPUT«Type check failed in assignment to '@a'; expected 'Array' but got 'Int'␤ in method REIFY at gen/parrot/CORE.setting:8363␤ in method reify at gen/parrot/CORE.setting:7239␤ in method gimme at gen/parrot/CORE.setting:7654␤ in method exists_pos a…»
grondilu r: say .[^3].Array given my Array $a = ^10;
camelia rakudo-jvm 263aa0: OUTPUT«Type check failed in assignment to '$a'; expected 'Array' but got 'Range'␤␤»
..rakudo-parrot 263aa0: OUTPUT«Type check failed in assignment to '$a'; expected 'Array' but got 'Range'␤ in block at /tmp/ju9CFQO_xn:1␤␤»
grondilu hum 23:08
lue jnthn: the real issue here though, is that either Buf[] should return a Buf (consistent with .subbuf) and/or Parcel.Buf works ( consistent with Buf.new(Parcel) )
All that more philosophical posturing above is just a result of my idea of Buf conflicting with everyone else's. 23:09
grondilu as I wrote Buf[] returning Buf is not consistent with .subbuf, since subbuf is analogous to substr and Str[] does not exist 23:11
lue grondilu: If that inter-type consistency must be maintained, then I'd be just as fine with Parcel.Buf
Though unless someone can point me to the part of the spec that explains why otherwise, I think one could argue that ListType[] should return a ListType, not a Parcel. 23:12
grondilu r: say .WHAT, .[^3].WHAT given my $ = ^10; 23:15
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«(Range)(Parcel)␤»
23:15 dmol left
lue I'd also like to point out that I'd prefer Buf.push to Buf ~= , but that's not as important :) 23:15
grondilu I see reasons why FooType[] should not necessarily return FooType, but it's hard to explain. 23:16
lue r: my $a = Buf.new(1,2,3); $a ~= "hello"; say $a; 23:17
camelia rakudo-jvm 263aa0: OUTPUT«java.lang.RuntimeException: java.lang.StackOverflowError␤␤»
..rakudo-parrot 263aa0: OUTPUT«maximum recursion depth exceeded␤current instr.: 'print_exception' pc 138097 (src/gen/p-CORE.setting.pir:58679) (gen/parrot/CORE.setting:10848)␤called from Sub 'Stringy' pc 213188 (src/gen/p-CORE.setting.pir:88441) (gen/parrot/CORE.setting:1014)␤ca…»
dalek kudo/moar-support: 9f6c362 | jnthn++ | src/ (3 files):
Fix various cases of where clauses.
lue Regardless of how you think Bufs should operate, I think we can agree that that's a bug :)
grondilu well, the fact that you prefer Buf.push to Buf ~= is an other example showing you reluctance to consider buffers as anything looking like a string. 23:18
23:18 pmurias left
grondilu Yet to me a Buf is a binary string, so it's just fine to use stringish notation 23:18
23:18 spider-mario left
lue grondilu: It's the closest thing I have to a low-level array right now, I'd use Array[int8] in a heartbeat if it wasn't just decoration at this point. 23:19
r: my int8 $a = 0xFF; say $a; $a++; say $a;
camelia rakudo-parrot 263aa0: OUTPUT«255␤Cannot modify an immutable value␤ in sub postfix:<++> at gen/parrot/CORE.setting:4335␤ in sub postfix:<++> at gen/parrot/CORE.setting:1751␤ in block at /tmp/dTQyN9JflE:1␤␤»
..rakudo-jvm 263aa0: OUTPUT«255␤Cannot modify an immutable value␤ in sub postfix:<++> at gen/jvm/CORE.setting:4324␤␤»
lue r: my int8 $a = 0xFF; say $a; $a += 1; say $a;
camelia rakudo-parrot 263aa0: OUTPUT«255␤Cannot modify an immutable value␤ in block at gen/parrot/CORE.setting:16590␤ in block at /tmp/5WkQ8o2pPI:1␤␤»
..rakudo-jvm 263aa0: OUTPUT«255␤Cannot modify an immutable value␤ in block at gen/jvm/CORE.setting:16313␤␤»
lue r: my int8 $a = 0x100; say $a;
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«256␤»
23:19 pmurias joined
jnthn lue: It doesn't pay attention to the sizes there just yet 23:20
lue jnthn: which is why Array[int8] is just decoration to me :) . I'd use it instead of Buf if I could, because I don't want an integer view of strings. 23:21
grondilu
.oO( Maybe we could create a "Bring" type, novlang for Binary String :-) )
 23:22
lue :)
23:23 syndrum left, rurban1 left, FROGGS[mobile] left, FROGGS[mobile] joined
lue By the way, if we insist on :bin returning a Buf in IO, could it *please* be renamed :buf ? When I say "binary data", I'm pretty sure I didn't mean "strings in number form". 23:23
grondilu why not?
I mean, why couldn't "binary data" mean "strings of numbers"? 23:24
lue Because string data ⊂ binary data. Thinking of binary data as a really funny-looking iso-8859-1 string isn't always useful.
(and for me, never is) 23:25
grondilu that's how strings are in P5, though. The most general string in P5 is a binary string, isn't it?
jnthn lue: Buf has nothing to do with iso-8859-1 23:26
lue jnthn: I know, I only said "iso-8859-1" because it's the most common encoding that won't break on some of the higher-valued bytes.
grondilu strings are not just text. After all a string is just something that has a length..
23:27 raiph left
jnthn grondilu: The point is to make a distinction between a bunch of bytes, which it doesn't mean anything to manipulate as if they were text, and things that are text and can be sanely treated as such. 23:28
lue I reject the general notion that strings are "just another view of numbers", I always consider them higher-level.
23:28 bluescreen10 left
jnthn Str is text. Buf is bytes. 23:28
You shouldn't even assume Str is implemented as an array under the hood. 23:29
lue
.oO(Strings are just a built-in interpreter of some kinds of binary data.)
jnthn: should Parcel.Buf work exactly the same as Buf.new(Parcel)? I just want $buf[$range].Buf to work. That's how this all started :) 23:30
grondilu jnthn: Str is text indeed, but I was talking about strings as a general concept, for which a certain vocabulary (such as concatenation, substirngs...) makes sense. That's Buf borrows some of these concepts to Str (such as ~ or substr (translated in subbuf)) 23:31
s/That's Buf/That's why Buf/ 23:32
jnthn grondilu: Yes, sometimes being able to re-use concepts that make sense is useful
Concatenation can be sanely defined on two bunches of bytes. It's things like lc, chars, etc. that cannot.
lue To tell the truth, I'm not quite sure why open("file", :bin).read($number) giving me a low-level array of numbers is such a seemingly hard concept. :/ 23:33
grondilu the problem is that Str in Perl 6 only covers one part of the general concept of string.
I mean that Str really is textual strings. 23:34
timotimo r: say -1 gcd 2
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«1␤»
timotimo moarvm gives -1 for that, that's wrong right?
r: say nqp::gcd_i(-1, 2)
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«1␤»
jnthn timotimo: Consistency would be good, yes :)
timotimo that may be the cause of the rat failure. 23:35
lue timotimo: wxMaxima gives me 1, so I think MVM is wrong there :)
jnthn grondilu: Yes, it's true we took the name Str for that. otoh, most of the time when you say "string" to somebody, they think of textual data, so it's not so bad naming wise...
lue I would really love it if Buf wasn't so tied to Strs, "Buffer" does not imply string data to me :/ 23:36
I'd much prefer if Buf was the native equivalent of Array, much like int and Int .
grondilu jnthn: good point. Indeed when people say string they usually mean "text". 23:37
timotimo should i just abs() both arguments to the gcd algorithm?
jnthn lue: It's only tied in that some of the same operations are defined on Buf and Str. But + is defined on Complex and Int, but I don't hear you complaining those are too tied :)
timotimo: That could work. Worth checking out what we do on Parrot for gcd_I given that's also done with libtommath 23:38
lue Complex and Int are both mathematical concepts. Buf and Str are *not* both string concepts.
grondilu lue: this is arguable.
23:38 treehug88 left
grondilu to me Buf and Str *are* both string concepts 23:39
timotimo oh, interesting
lue Buffers (the general concept). are. not. Strings (the general concept). This is getting maddening. I just want open("file", :bin).read() or slurp("file", :bin) to give me data that functions like an array.
Not like a funny perspective on strings.
grondilu why do you say it's funny when it's how binary data has been treated in P5 for ages? 23:40
IIRC, when you open a binary file in P5, you don't get an array. 23:41
lue Perl 5 is wrong. C/C++ is more right here, treating strings as totally separate from all other kinds of data.
23:41 pmurias left
jnthn Time for me to get some rest. Got some flying to do tomorrow, so won't be online much at all... 23:42
'night o/
lue I just want to do binary IO with objects that don't operate in the string-verse.
♞ jnthn o/
23:43 skids joined
timotimo i see no special handling for abs-ing big integers before passing it to libtommath 23:44
and in fact gcd_I on moar is also correct 23:45
lue timotimo: but... -1 < 1 , so isn't 1 the "greatest" common denominator? :) 23:46
timotimo yes.
that's why i'm fixing moar to return 1 as the gcd of -1, 2
i'm still a bit puzzled as to what's going on. i think i need to put some debug prints. 23:47
jercos The first prime is 2, what if the 0th prime were -1? :p 23:48
lue I just wrote '00' because I was too lazy to write out '0x00' (but the extra zero helps me in keeping the nybbles straight), and I got a bunch of reminders that I'd enjoy no perl5-warnings-seriously
jercos -2 gcd -10 suddenly equals -2?
timotimo r: say -2 gcd -10 23:50
camelia rakudo-parrot 263aa0, rakudo-jvm 263aa0: OUTPUT«2␤»
timotimo phew. i thought i was doing it wrong :)
23:52 smls joined
grondilu lue: I don't want to insist too much, but I'd just like to remind you that the word "string" is a metaphor when it's used to describe the machine representation of text. A string is initially a long, flexible piece of rope. This metaphor does not only make sense for text. 23:52
lue grondilu: most people associate "string" with textual data, I do the same. Imagine 'Str' was instead 'Txt', and maybe you'll see why I hate .subbuf and Buf ~= and the general design choices for Buf. 23:54
"string" is synonymous with "text" these days, so I need *very* good arguments to treat the word "string" as anything more. 23:55
In fact, I think (but am not sure that) "stream" has become the general term nowadays. 23:57
smls lue: Maybe the default interpretation of the word "string" absent qualifier is "string of characters", but when the given definition is explicitly "string of bytes" , I don't see any room for confusion
grondilu I'm used to consider string as not necessarily text for two reasons. The fist one is that I've been doing so in Perl 5 for quite some time. The second one is that when I learned some bioinformatics on rosalind, I got to read bout string algorithms, where the mathematical definition of strings was quite general: en.wikipedia.org/wiki/String_(computer_science)
smls (if you don't mind me joining the discussion... :)
lue smls: yes, but "string of bytes" is to be handled way differently than "string of characters", and conflating the general notion of "string" with textual "string"s when designing Bufs is incorrect. 23:59
grondilu smls: exactly, "string of bytes" or "binary string" are perfectly valid expression. Again, the string metaphor is valid not just with text.
lue The Str type is *specifically* designed for "strings of characters", to be explicit. Buf could be described as a "string of bytes", but to me that's equivalent to calling Array a "string of objects".

Powered by Cro and the Raku® Programming Language.

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

Thank you!