»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
00:03 SquireOfGothos joined 00:04 dmitri joined 00:05 jjido left
Sgeo Will someone eventually accidentally do something like @foo[$_-1] ? 00:08
(Where {} around tha expression would presumably make it work)
(The $_-1 I mean) 00:09
I think I'd like $_ more if it was more like $^a
I saw a Perl5 vs Perl6 comparison thing, did $a have special meaning in P5?
dmitri Sgeo, yes $a is special 00:10
yoleaux 2 Aug 2015 12:01Z <smls> dmitri: For .pm files, you can assume it's Perl 5 by default, and only Perl 6 if the first non-comment line starts with one of: use v6 unit module class
geekosaur $a and $b are magic, yes
Sgeo dmitri, what does it do?
geekosaur they're always package variables, predeclared, because sort puts the items passed to the comparison function in $a and $b 00:11
not as parameters
dmitri Sgeo, what geekosaur said
geekosaur (an optimization going way back in perl history()
00:11 gfldex_ is now known as gfldex
dmitri yoleaux: That's not always the case, though. I added a special taster to ctags that can figure this out. 00:11
geekosaur basically perl did not build a stack frame for the comparison function, so no way to pass parameters normally
dmitri Speaking of ctags, it now supports Perl 6 00:12
geekosaur that should have been directed to smls, not the bot :)
dmitri heh
geekosaur ...maybe via .tell
dmitri smls: That's not always the case, though. I added a special taster to ctags that can figure this out.
Juerd geekosaur: They're not always package variables. In fact, if you do "my $a;", and then use sort { ... $a ... }, it's not accessing the package variable $a, but the lexical one, and the sorting doesn't do what you want. A warning is emitted. 00:13
geekosaur yeh
it's ugly
Juerd It's not as ugly as it could have been :)
geekosaur (it also predates my variables, so not surprising that things misbehave in that case)
00:14 SquireOfGothos left
geekosaur reminded of the bad old days with local. yeurccccchhh 00:14
Juerd It's more like there's an implicit "our ($a, $b);" at the beginning of your file. 00:15
00:15 cognominal left
Juerd geekosaur: local is *great* for what it does, not so much for being a lexical-ish thing. 00:15
dmitri a few years ago I was writing lots of shell scripts 00:16
I grew to love dynamic scoping
geekosaur yes, but in perl 3 local was all we had!
Juerd dmitri: In Perl 6, dynamic scoping is done with "temp". 00:17
dmitri perl 3... is perl 6 twice as good? :)
Sgeo I thought dynamic was done with $* ?
Juerd Hm, I guess two different things are both called dynamic scoping in my head. 00:18
geekosaur which part is dynamic? :)
Juerd m: my $foo = "foo"; { temp $foo = "bar"; $foo.say; }; $foo.say 00:19
camelia rakudo-moar 720090: OUTPUT«bar␤foo␤»
Juerd m: my $*foo = "foo"; { $*foo = "bar"; $foo.say; }; $foo.say 00:20
camelia rakudo-moar 720090: OUTPUT«5===SORRY!5=== Error while compiling /tmp/KnpcPCZYzX␤Variable '$foo' is not declared. Did you mean '$*foo'?␤at /tmp/KnpcPCZYzX:1␤------> 3my $*foo = "foo"; { $*foo = "bar"; 7⏏5$foo.say; }; $foo.say␤»
Juerd m: my $*foo = "foo"; { $*foo = "bar"; $*foo.say; }; $*foo.say
camelia rakudo-moar 720090: OUTPUT«bar␤bar␤»
Juerd m: my $*foo = "foo"; { my $*foo = "bar"; $*foo.say; }; $*foo.say
camelia rakudo-moar 720090: OUTPUT«bar␤foo␤»
ShimmerFairy Juerd: unless I'm mistaken, in Perl 6 "dynamic scoping" refers to scoping according the the chain of callers (instead of outers), and has nothing to do with 'temp' :)
Juerd m: my $*foo = "foo"; sub x { $*foo.say }; { my $*foo = "bar"; x; }; x;
camelia rakudo-moar 720090: OUTPUT«bar␤foo␤»
geekosaur yes, they're two different kinds of "dynamic" 00:21
Juerd m: my $foo = "foo"; sub x { $foo.say }; { temp $foo = "bar"; x; }; x;
camelia rakudo-moar 720090: OUTPUT«bar␤foo␤»
Juerd These features can be used to achieve the same effect. I don't know when you'd pick which.
ShimmerFairy m: sub x { $*foo.say }; { my $*foo = "bar"; x; } 00:22
camelia rakudo-moar 720090: OUTPUT«bar␤»
ShimmerFairy m: sub x { $foo.say }; { temp $foo = "bar"; x; }
camelia rakudo-moar 720090: OUTPUT«5===SORRY!5=== Error while compiling /tmp/EZ3JiPqnjL␤Variable '$foo' is not declared␤at /tmp/EZ3JiPqnjL:1␤------> 3sub x { 7⏏5$foo.say }; { temp $foo = "bar"; x; }␤»
geekosaur I feel like $* is for control structures and temp is for automatic saving and restoring of values 00:23
ShimmerFairy Juerd: ^ that's a situation where you need $* vars
Juerd Jargon is definitely an issue here: in Perl 5, "dynamic scope" is associated with "local", and in Perl 6, "temp" is documented to be like Perl 5 "local", but "dynamic scope" is something else.
ShimmerFairy (I see the 'temp' keyword for "temporary" things more than dynamic scoping ☺)
Juerd Through Perl 5, "temp" is also associated with the term "dynamic scope". 00:24
ShimmerFairy Juerd: $* vars show up in regexes a lot in my experience, and I believe it's precisely because of the way they can quickly nest in complex ways; it lets you carry info through sub-rules, instead of through inner scopes like lexicals. 00:25
(keep in mind I could be wrong, I'm not too well-versed in dynamic variables. This is just what I've picked up over the years.)
Juerd A shared lexical with use of temp would do the same thing. Only it requires a code layout that has everything in the same file. 00:26
But you could use a global thing and temp, and then I don't really see much difference between the two anymore. 00:27
ShimmerFairy IIRC dynvars also carry into the action methods called by the grammar, which is really nice.
Juerd I know nothing about grammars yet.
Or, more precisely: s/yet/anymore/
00:28 shinobi-cl joined
shinobi-cl hi all... 00:28
Juerd Hello ShimmerFairy
Argh, shi<tab> :(
Sorry, ShimmerFairy and shinobi-cl
Hello shinobi-cl
shinobi-cl Hi Juerd
is there a way to pass a regex as a parameter to a sub?
Juerd Yes. 00:29
m: sub foo ($re) { $re.perl.say }; foo(/bar/);
camelia rakudo-moar 720090: OUTPUT«/bar/␤»
ShimmerFairy Juerd: admittedly, there's also something nice about how $*SPECIAL_INFO looks for information you pass down through calls :3
Juerd should really spend a week, maybe more, relearning Perl 6 00:30
I've learned Perl 6 too early and haven't kept up. 00:31
Wrote hypothetical Perl 6 code in 2002, never wrote anything more than a oneliner with real Perl 6. 00:32
Still should try how much of www.perlmonks.org/?node_id=179755 would work in real life. I expect a lot of rewriting to be necessary. 00:33
ShimmerFairy Juerd: first, there was a time when P6 regexes still looked P5-ish!? O_o (the (?: ... ) is what I noticed). And second, I happen to have an XML 1.1 parser in P6 I wrote not too long ago, I could gist it if you were interested. :) 00:38
skids Well, similar stuff wored well enough for the Grammar::BNF package and ruoso's (not ecosystem yet) Grammar::EBNF package. 00:39
*worked
skids should make sure BNF is still working 00:41
shinobi-cl r: my @a = <1 2 3 4 45 6>; sub regx(Regex $search) { @a.match($search); }; my $x = rx/4.*/; say regx($x);
camelia rakudo-{moar,jvm} 720090: OUTPUT«｢4 45 6｣␤»
shinobi-cl i love perl6 00:42
r: my @a = <1 2 3 4 45 6>; sub regx(Regex $search) { @a.match($search); }; my $x = "4.*"; say regx($x); 00:43
camelia rakudo-{moar,jvm} 720090: OUTPUT«Type check failed in binding $search; expected 'Regex' but got 'Str'␤ in sub regx at /tmp/tmpfile:1␤ in block <unit> at /tmp/tmpfile:1␤␤»
shinobi-cl r: my @a = <1 2 3 4 45 6>; sub regx(Str $search) { @a.match($search); }; my $x = "4.*"; say regx($x);
camelia rakudo-{moar,jvm} 720090: OUTPUT«Nil␤»
shinobi-cl r: my @a = <1 2 3 4 45 6>; sub regx(Str $search) { @a.match($search); }; my $x = "4"; say regx($x);
camelia rakudo-{moar,jvm} 720090: OUTPUT«｢4｣␤»
ShimmerFairy .tell jnthn I'm curious on your thoughts on the Stringy and Unicodey roles as mentioned in S15, I'd like to see them fleshed out finally :) (I have this S32::Str potential rewrite from some while ago featuring the two roles, if it helps: gist.github.com/lue/9941658 ) 00:44
yoleaux ShimmerFairy: I'll pass your message to jnthn.
00:44 mjgardner joined
Juerd ShimmerFairy: No, (?:...) has not existed in Perl 6 regexes as far as I know. That part is Perl 5 :) 00:45
geekosaur not spelled that way 00:46
ShimmerFairy Juerd: I didn't know P5 had rule foo { } back then, then :)
geekosaur but afaik it was always [] in p6
Juerd ShimmerFairy: What I meant was, I wrote Perl 5 code in Perl 6.
ShimmerFairy: That's the kind of mistakes one makes if there's no parser to tell you you're way off. 00:47
ShimmerFairy Juerd: ah, so that (?: ) in "rule xml :i {" was a mistake in typing? I understand :)
Juerd ShimmerFairy: Note that the *generated* P6 does have [] instead of (?:...)
ShimmerFairy (though there's always :P5 ... :P)
00:48 laouji joined
Juerd ShimmerFairy: Also note that in the second part of the code that does (?:) in Perl 6, I did fix it... So at some point I must have known that it was wrong. 00:48
ShimmerFairy Juerd: interestingly, the (?: ) only appear in your singleton alternation; the <foo>bar</foo> alternation uses [] :)
Juerd That 00:49
geekosaur habits are hard to break?
Juerd Now... should I edit a post from 2002 or leave it be...
ShimmerFairy esp. in 2002 with no P6 implementation, I'd guess :)
Juerd: depends on how obvious it is that it comes from 2002 and the land of no implementation.
Juerd ShimmerFairy: Pretty obvious.
ShimmerFairy Juerd: so I think at most you'd want to add a note at the top that says something like "UPDATE (someday 2015): Just pointing out that this post comes from long before there was even an implementation. Left as-is for historical interest." :) 00:51
Juerd: one more thing, about the <!after>s in your code, and your mention that they should be <!before>s: for PITarget, as an example, I specified token PITarget { <!before :i XML> <Name> } 00:53
(I do recall there were parts of the EBNF grammar I couldn't quite grok, so there were a couple spots where I just went with my best guess. An EBNF converter would do better than me in those spots, I'd imagine :P) 00:54
00:55 kst joined
Juerd I remember that PITarget is weird because the EBNF and the description disagree. The EBNF treats the letters individually, whereas the description only mentioned XML and xml. 00:55
ShimmerFairy: Rewriting things as <!before ...> is really hard, though. 00:56
Perhaps Perl 6 just needs something like EBNF's - operator. 00:57
ShimmerFairy yeah, :i XML was simply a shorter way of writing something like <[Xx]> <[Mm]> <[Ll]> (so I went with the grammar's description)
Juerd - is used in two completely different ways in EBNF by the way
foo - bar sometimes means that foo shouldn't match bar anywhere in it, whereas in other occasions it's anchored to whatever foo matched. 00:58
ShimmerFairy Juerd: I couldn't understand the <foo>* - (<foo>* <bar> <foo>) stuff, looking at the spec again. 00:59
00:59 yqt left
ShimmerFairy (oops, <foo>* for the last one too) 00:59
Juerd That's the version with implicit anchoring
So it allows any number of <foo>, where none of the <foo> is <bar>. 01:00
It's unclear whether this should fail the entire subrule, or just stop whenever bar is encountered.
Off to bed. Good localtime! 01:01
ShimmerFairy CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) for instance I turned into token CharData { [<!before "]]>"> <-[<&]>]* }
♞ Juerd o/
looking at those "implicit anchoring" things, it seems to me like it's just a way of saying "stop when you see this", and not a case for failure (unless you can/want to let backtracking handle that stuff) 01:03
So I think I was right in using <!before> like I showed above :)
Juerd Please put whitespace inside of []. Makes it much easier to read, especially in this case :) 01:04
I think you're right too.
afk! 01:05
ShimmerFairy Juerd: sure, I think most of my rules are more readable than that btw :)
skids m: ("a","b")>>.chars.say; ("a","b")>>.ords.say 01:15
camelia rakudo-moar 720090: OUTPUT«1 1␤97 98␤»
skids hrm, that's broken locally.
TimToady broken how? 01:16
skids The second one gives an empty list.
b2gills m: my \if = "hello"; say '', if if if; 01:17
camelia rakudo-moar 720090: OUTPUT«hello␤»
b2gills m: my \if = "hello"; say if if if;
camelia rakudo-moar 720090: OUTPUT«5===SORRY!5=== Error while compiling /tmp/EEuUJAePmt␤Unsupported use of bare "say"; in Perl 6 please use .say if you meant $_, or use an explicit invocant or argument␤at /tmp/EEuUJAePmt:1␤------> 3my \if = "hello"; say7⏏5 if if if;␤»
skids (or, ((), ()) to be exact)
I only have one very unrelated mod in my codebase... 01:18
TimToady m: my \if = "hello"; say (if) if if; 01:19
camelia rakudo-moar 720090: OUTPUT«hello␤»
TimToady the check for bare say is looking for terminators there, and 'if' is a terminator 01:20
(as are any statement modifiers)
b2gills You mean it looks like the terminator of the same name
TimToady right
just as if you'd said say; 01:21
m: my \if = "hello"; say for if if;
camelia rakudo-moar 720090: OUTPUT«5===SORRY!5=== Error while compiling /tmp/kct0GkcdLN␤Unsupported use of bare "say"; in Perl 6 please use .say if you meant $_, or use an explicit invocant or argument␤at /tmp/kct0GkcdLN:1␤------> 3my \if = "hello"; say7⏏5 for if if;␤»
skids tries to find the more detailed thing like $*PERL 01:24
01:24 rmgk_ joined, rmgk left, rmgk_ is now known as rmgk 01:28 mjgardner left
skids Oh hah. $*PERL.gist--; 01:28
geekosaur you wanted .perl? 01:30
skids yeah it hid itself from me. 01:31
geekosaur just one of those things you have to remember, .gist is often a short version
skids yeah.
geekosaur (not surprising to me given what "gist" means...) 01:32
01:32 mjgardner joined
skids wonders why my version has dots and camelias doesn't (7200905 versus 6.c.3.d.63.b) 01:33
OK my moar is newer.
Heh. $*PERL.Str is even shorter than .gist. 01:35
geekosaur ...and after you said that, I was expecting just "6" or "6.0" 01:36
b2gills m: say ~$*PERL
camelia rakudo-moar 720090: OUTPUT«Perl 6␤»
TimToady it puts dots in where there are letters, 7200905 has no letters 01:38
zostay so, i've updated the P6SGI spec to v0.2.Draft, includes a number of typo fixes and small errors 01:40
i've made the major change of requiring Promise to be used in all cases... this, however, does break all existing implementations of PSGI-like servers and apps in Perl 6 01:41
i have not yet addressed the Channel versus Supply bit yet 01:42
that will require a bit more thought and research.... i'm leaning toward Supply's but they are so flexible it is difficult to see a clear way to a solution
01:44 ButterBalls joined
ButterBalls dSo. 01:44
So.... 01:45
Perl6 can run in a different languages interpreter when compiled...?
IE: .net, JVM, Dalvik
skids JVM yes, not the others yet. Well mono if you count older neicza.
ButterBalls So perl6 threw away the crappy interpreter that 5 had in favour of a well built one by Oracle? 01:46
Does this mean multithreading wont suck anymore? 01:47
skids Yes it is a ground-up rewrite. And multithreading is a top priority.
ButterBalls I just learned perl5 last year. 01:48
I do want to talk about the implications this will have.
By using the JVM the userbase will be much bigger but when that happens malicious programs are wrote in these languages. Perl6 wont be an exception thanks to it's amazing file API's 01:49
O
I
I'm just trying to understand why the JVM?
gfldex because jnthn++ had experience with that vm
skids Neither has Perl 5 been an exception to that, considering it used to e te primary language holding evrything together. 01:50
ButterBalls >so is a loose precedence operator that coerces to Bool. The logical opposite of not, it returns a Bool with a preserved boolean value instead of the opposite one.
gfldex he seams to like moarvm better now tho
ButterBalls I am in love.
my god.
Perl6 has some great choices but I don't agree with all of them. 01:51
geekosaur I would imagine that given jvm support, clr support would not be that difficult (famous last words...)
skids
.oO(OK so my issue is not the moarvm version... maye NQP.)
ButterBalls geekosaur, I would imagine that Dalvik and it's RT library would be used just like Java's 01:52
Imagine it.
Perl6 on Android.
No extra interpreter needed.
I must ask the important question now. Does perl6 support perl5 regex's?
skids Through an adverb on rx//, yes. 01:53
m: say "aa" ~~ m:P5/a(?:a)/ 01:54
camelia rakudo-moar 720090: OUTPUT«｢aa｣␤»
ButterBalls Forgot to add, will perl5 scripts work fine in perl6?
skids Some of them, using Inline::Perl5 01:55
b2gills You may be able to write a small program that will work in both, Perl 5 and Perl 6 but it won't be worth it
ButterBalls It would. 01:56
I normally do my development in Java.
Write once, deploy everywhere.
(I never tie to system dependant behaviours)
b2gills The differences between Perl 5 and Perl 6 are about as different as the difference between C and D
*C++ and D 01:57
ButterBalls M: say "hi"
blargh
why
b2gills lowercase m
ButterBalls Why name it perl then?
Why not something like Perl2
gfldex because it stole not just the name from perl 5
ButterBalls m: say "hi"
camelia rakudo-moar 720090: OUTPUT«hi␤»
ButterBalls m: while(so 2/1==2){say ""} 01:58
camelia rakudo-moar 720090: OUTPUT«5===SORRY!5=== Error while compiling /tmp/v24oNmvJHW␤Undeclared routine:␤ while used at line 1␤␤»
shinobi-cl r: say ((1, 2, 3, 4, 5, 6) ~ (1, 2, 33, 4, 5, 6)) ?? "true" !! "false";
ButterBalls I was wondering if that would work
gfldex and when Perl 6 was first discussed it was much closer to Perl 5 then to Perl 6
camelia rakudo-{moar,jvm} 720090: OUTPUT«true␤»
ButterBalls What's the point of perl6 if perl5 will continue to be mantained 01:59
Why not just split off into an entire new version of perl.
b2gills What's the point when FORTRAN already exists?
ButterBalls Different purposes
shinobi-cl r: say ((1, 2, 3, 4, 5, 6) ~~ (1, 2, 33, 4, 5, 6)) ?? "true" !! "false";
camelia rakudo-{moar,jvm} 720090: OUTPUT«false␤»
gfldex what's the point of c++ if c is continued to be maintained?
ButterBalls r say @_; | y 02:00
gfldex, different purposes
skids ButterBalls: Perl "6" is an entriely new "version" of perl :-)
gfldex exactly
ButterBalls r: say @_; | y
camelia rakudo-{moar,jvm} 720090: OUTPUT«5===SORRY!5=== Error while compiling /tmp/tmpfile␤Cannot use placeholder parameter @_ in the mainline␤at /tmp/tmpfile:1␤------> 3say @_7⏏5; | y␤»
ButterBalls do things like @_ behave like before?
TimToady sometimes
b2gills m: sub p5example { say @_ }; p5example 'hello' 02:01
camelia rakudo-moar 720090: OUTPUT«hello␤»
skids ButterBalls: since you know Perl 5, you may want to start here: doc.perl6.org/language/5to6
gfldex the sane Perl 5 oddities tend to work in Perl 6
ButterBalls So I assume I have access to the entire JVM RT.jar? 02:02
b2gills m: say (<a b> Xx ^9).perl
camelia rakudo-moar 720090: OUTPUT«("", "a", "aa", "aaa", "aaaa", "aaaaa", "aaaaaa", "aaaaaaa", "aaaaaaaa", "", "b", "bb", "bbb", "bbbb", "bbbbb", "bbbbbb", "bbbbbbb", "bbbbbbbb")␤»
skids ButterBalls: I'm not sure where JVM interop is at, implementation-wise. 02:03
ButterBalls "
Space is required immediately after keywords"
no.
just no.
skids Yep.
ButterBalls no.
TimToady there are reasons
ButterBalls I will write an entire cleaning tool to avoid that.
TimToady having to do with extensibility
ButterBalls I will write a tool to prep code so it's valid just to NOT have to add that space. 02:04
my god.
TimToady that's just silly
it's more readable with the space, you're just not used to it
ButterBalls I am coming from java
TimToady you can leave out the parens
ButterBalls while(true) is more common than while (true)
TimToady while true is more better
ButterBalls I find the parens help 02:05
02:05 dayangkun joined
ButterBalls (readability wise) 02:05
TimToady not really, if you linebreak nicely
skids Well, "loop" is better than "while True" :-)
ButterBalls "$ Scalar
The $ sigil is now always used with "item" variables (e.g. $name), and no longer for array indexing and Hash indexing. That is, you will still use $x[1] and $x{"foo"}, but you use them on $x, not @x or %x."
Whoever chose that deserves a pizza. 02:06
$x[<pointer] is something that I love.
TimToady can't eat pizza...
b2gills That was a **very** early decision from what I've read 02:07
ButterBalls I'm trying to understand "&"
so $x = &funcname
TimToady it's now just a noun marker
ButterBalls Just typing $x will trigger &funcnanme?
b2gills & is used the same way you would use it in Perl 5 ( for the most part )
skids Butterballs: no, $x()
b2gills $x would be a reference to the subroutine ( in Perl 5 speak ) 02:08
TimToady just as &foo will not call, but &foo() will
ButterBalls skids, So I can dynamically assign $x to functions by going $x = &func to call $x() to trigger &func
skids Yes.
ButterBalls So $x(params) is valid?
TimToady sure
except we usually call those args
ButterBalls I am loving perl6
TimToady and reserve params for the formals
b2gills m: for &[+], &[-]
camelia rakudo-moar 720090: OUTPUT«5===SORRY!5=== Error while compiling /tmp/eDUefNJa7P␤Missing block␤at /tmp/eDUefNJa7P:1␤------> 3for &[+], &[-]7⏏5<EOL>␤ expecting any of:␤ block or pointy block␤»
ButterBalls These little changes make the language much easier to work with 02:09
skids m: my $x = { $_ .say }; $x("hi"); $x = { 42.say }; $x();
camelia rakudo-moar 720090: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Ceb07PZysG␤Two terms in a row␤at /tmp/Ceb07PZysG:1␤------> 3my $x = { $_7⏏5 .say }; $x("hi"); $x = { 42.say }; $x()␤ expecting any of:␤ infix␤ infix stopper␤ statemen…»
shinobi-cl how can i generate a list starting with 1, then 4, then 7,...up to 64?
TimToady m: say 1,4,7...64
camelia rakudo-moar 720090: OUTPUT«1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58 61 64␤»
b2gills m: for &[+], &[-], &[*] -> &func { say func 4,5 }
camelia rakudo-moar 720090: OUTPUT«9␤-1␤20␤»
ButterBalls Perl6 is about concistency isn't it?
gfldex m: my @a.push(sub {42},sub {43}, sub {44}); say @a>>.();
camelia rakudo-moar 720090: OUTPUT«42 43 44␤»
skids m: my $x = { $_.say }; $x("hi"); $x = { 42.say }; $x(); 02:10
camelia rakudo-moar 720090: OUTPUT«hi␤42␤»
shinobi-cl i see.. I was using " .. " :( Thanks!
skids ButterBalls: It is "strangely consistent" :-)
TimToady ButterBalls: there's certainly more consistency than P5
but you can't be consistent in all dimensions at once 02:11
ButterBalls so % allows you to call a object from a 2d @ by it's name?
so @t["apple", "pear"] = qw("good", bad");
%t["apple"] = "good"?
b2gills %t{'apple'}
skids No. @ is only for positionals. 02:12
ButterBalls b2gills, sorry about syntax but the idea is right?
skids, ok
TimToady you can slice either arrays or hashes, and either can be multidimensional, but then you use ; instead of ,
ButterBalls I'm still trying to understand some Syntax changes but it seems to be pretty straight forward 02:13
How is performance?
Especially for GUI
dalek kudo-star-daily: 9b3538e | coke++ | log/ (9 files):
today (automated commit)
skids Really it is the [ and { and ( that decide what happens, more than the $ @ or %
gfldex performance heavily depends on what you do 02:14
ButterBalls I'm thinking of porting a NES emulator to Perl6
I want to see how feasble it is.
(coming from Java)
b2gills Considering a faithful NES implementation in C/C++ maxes out a modern processor, I wouldn't hold my breath 02:16
ButterBalls b2gills, You think I'm going for perfect hardware accuracy? 02:17
huehuehue
b2gills In Perl 6 normal operators are subroutines, and subroutines are objects, it will require a smart implementation to make it fast enough for an emulator. 02:19
ButterBalls But to be honest the issue with maxing out the CPU ois due to the overhead of the OS
02:19 mjgardner left
b2gills It has more to deal with implementing custom ASICs, and a whole other processor in software 02:20
ButterBalls b2gills, I know
TimToady we're getting faster, but we don't know how fast we'll get eventually
ButterBalls The issue is timing.
b2gills I just read an article about it 02:21
ButterBalls I was implimenting a 6502 (the NES custom A02 one) on a Arduino Uno.
skids Yeah "».ords" broke sometime after NQP 70f65eb.
ButterBalls The issue was the lack of memory so paralell ram was going to be used.
Implimeting a proccesor in software is quite simple. Implimenting it's timing isn't. 02:23
Microsoft cheated though.
FWIU: 360 on One uses a JIT interpreter that hooks every API call and remaps them to the local system API 02:24
The issue though will be that the 360 was cell, not x86.
Cell is known for it's math capabilities.
Even to date the proccessor in the ps3 is stupidly fast at math thanks to the PPU's vs x86 02:25
The issue is you can't really write a JIT in Perl
02:26 shinobi-cl left
ButterBalls b2gills, make sense? 02:26
Instead of interpreting why not boil it down to instructions the host proccessor can simply interpret directly? 02:27
Dolphin does it.
RPCS3 (that's how you spell it IIRC) does it with PS3 games
b2gills I'm currently watching TV 02:29
ButterBalls heh 02:30
skids ButterBalls: do you like working on JIT stuff?
ButterBalls b2gills, TL;DR: JIT = Magic
b2gills The cell processor did seem to be a very interesting design
ButterBalls skids, personally no. 02:31
Just studying
skids Ah. Well if you want to keep track of moarvm's for study purposes, follow brrt++ 02:32
brrt-to-the-future.blogspot.nl/2015...ilers.html
ButterBalls I want Perl6 on Android.
Ideally you shouldn't do a serial multiplication on bigger numbers but break them down to multiple cores to do in paralell 02:33
Well not multiplications but certain pieces of math can be done in parallel much faster 02:34
skids I don't know if dalvik will come first for Android or webassembly. But for now the goal is at least a MoarVM (and maybe JVM) 6.0 RC by Christmas.
ButterBalls For now I'll continue to develop with Perl5 and play with Perl6.
I also use Java, and Python 02:35
b2gills m: say [+] 1..10 # this could be done in parallel
camelia rakudo-moar 720090: OUTPUT«55␤»
ButterBalls WebAssembly should be ignored for now.
It's still to early to call it standardized.
Right now I'm not sure what to do. 02:37
Sgeo I'm a bit of a language exploration addict, although I can't remember when I wrote anything real with any language :(
ButterBalls Sgeo, I am a tinkerer.
If it's wrote in Java I will shred it.
I want to know what makes it tick.
Sgeo But I rejected PErl5 but am fascinated by Perl6
02:38 mr-foobar left, gfldex left
ButterBalls Perl6 on Debian is slow. 02:39
painfully slow.
Sgeo I wonder which lens usescases P6 can meet and which ones it might struggle with
02:41 mr-foobar joined
TimToady ButterBalls: there are several reasons for that, all of which we're hoping to fix in the next month or two 02:41
02:41 noganex_ joined
TimToady well, most of which, anyway 02:42
b2gills Java was slow once too 02:43
02:44 noganex left 02:46 ButterBalls left 02:48 ButterBalks joined
ButterBalks Perl6 hung my computer. 02:49
/yaaaay/
02:50 dayangkun left, dayangkun_ joined
ButterBalks yay 02:50
b2gills I've done that before too 02:53
It is very easy to create an infinite loop `[+] 0..Inf` 02:55
02:56 ButterBalks left
skids .tell timotimo NQP 808ce0fcb broke ("a","b")».ords for some reason. 02:56
yoleaux skids: I'll pass your message to timotimo.
03:05 kaare_ joined 03:14 TimToady left 03:16 TimToady joined 03:38 pc2046 joined
pc2046 hello 03:38
skids o/
03:39 dayangkun_ left
flussence
.oO( it's funny when someone inadvertently reveals their PHP background, by complaining about any other language for not sharing its "unique" built in memory/runtime limits :)
 03:40
03:44 nys left 03:45 dayangkun_ joined
b2gills pc2046: Anything in particular you want to discuss? 03:51
TEttinger pearl earrings 03:52
perl errings
b2gills One reason its not spelled with the "a" is there was already a language called Pearl 03:54
TEttinger heh
Praal
Purl, Pirl, Purrel 03:55
purr wouldn't be a bad name for a convenience library in perl. but I've been watching too much of these darn kitten cams livestream.com/tinykittens/tip 03:57
then again, that implies there's something that perl needs a convenience library for 03:58
pc2046 I have developed perl project for more than 10 years, one technique that I adopt is store all perl processes and components into database, evaluate them and parse the result to the main for display, I am not sure if perl 6 would affect my component based evaluation type of programming 03:59
originally, I want to have central big business-core server/platform, that every perl projects can just reference to the correct version from the central server, either with assign version or upgrade version 04:01
I think to store "pm" on every server is a quite old-fashioned
say, old way is: use ABC::def; new way could be just use 123.123.123.123::ABC::def {v=1,update_local=false}; 04:05
skids pc2046: you should probably be able to make a custom CompUnitRepo that uses your database. 04:10
pc2046 that is a very good idea 04:12
ugexe thats cloudpan
skids pc2046: design.perl6.org/S11.html and design.perl6.org/S22.html are the design docs for packaging. More of S11 is actually implemented than S22 yet. 04:14
pc2046 Skids, your information is useful. 04:18
04:39 skids left 05:01 schmooster joined 05:15 pc2046 left 05:26 mr-foobar left 05:33 xinming joined 05:44 rurban joined 05:53 diana_olhovik_ joined 06:03 AlexDaniel left 06:11 araujo joined, araujo left, araujo joined 06:13 jjido joined 06:23 oetiker left 06:24 oetiker joined 06:25 gfldex joined, rurban_ joined 06:27 llfourn joined 06:41 rurban left 06:42 rurban joined 06:57 domidumont joined 07:00 ShimmerFairy left 07:01 domidumont left 07:02 domidumont joined 07:06 telex left 07:07 jjido left, FROGGS joined 07:08 telex joined 07:11 llfourn left 07:13 ShimmerFairy joined 07:17 darutoko joined 07:19 yeahnoob joined 07:21 labster left 07:26 domidumont left 07:33 mr-foobar joined 07:44 mrf joined
dalek kudo/nom: 1c76b58 | TimToady++ | src/Perl6/Grammar.nqp:
Parse with/without as either control or modifier
 07:49
kudo/nom: b620a37 | TimToady++ | src/ (3 files):
implement with/without modifiers
kudo/nom: 2623525 | TimToady++ | src/ (4 files):
get new orelse/andthen to do assignops
TimToady note that the statement control with/without are still just testing boolean 07:50
FROGGS o/ 07:51
07:51 RabidGravy joined
TimToady the statement modifiers are testing definedness, but still need to topicalize non-block expressions 07:51
jobs for tomorrow... 07:52
TimToady --> sleep &
07:53 yeahnoob left 07:54 mr-foobar left
jdv79 nice 08:04
08:10 llfourn joined 08:11 rurban left 08:14 rurban joined 08:15 domidumont joined 08:16 rindolf joined 08:17 laouji left 08:18 laouji joined 08:21 domidumont left 08:22 laouji left, laouji joined
nwc10 top entry currently on planetpython.org/ is "Python 4 Kids" 08:24
I misparsed that the first time.
08:29 brrt joined
jnthn Yes, that parses two ways... :) 08:38
yoleaux 00:44Z <ShimmerFairy> jnthn: I'm curious on your thoughts on the Stringy and Unicodey roles as mentioned in S15, I'd like to see them fleshed out finally :) (I have this S32::Str potential rewrite from some while ago featuring the two roles, if it helps: gist.github.com/lue/9941658 )
nwc10 jnthn: is it Friday? 08:41
how about now?
:-)
RabidGravy it is now
nwc10 oh gosh, I'm almost halfway to POETS 08:42
jnthn nwc10: Well, going to take on lazy/eager/hyper/race at least somewhat first. 08:43
08:44 rurban left, [Tux] joined 08:45 rurban joined
RabidGravy nwc10, or we could arbitrarily define beer o'clock as 12 and you're almost done ;-) 08:57
nwc10 that late? :-)
08:57 espadrine left
jnthn The term "known infinite" is now resting. Updated everywhere that matters in gist.github.com/jnthn/aa370f8b32ef98e4e7c9 to talk about an iterator considering itself lazy instead 09:00
sergot oh hai #perl6 \0 09:03
jnthn o/ sergot 09:04
RabidGravy erp 09:08
09:13 Sqirrel left, [Tux] left
jnthn yay, my lazy impl worked first time 09:14
Working out a nice really simple test case for lazy was fun
Turns out that the best one is:
RabidGravy nwc10, as a further inducement it's International Beer Day
jnthn my @a = 1, 2;
my @b = lazy @a;
say @b[0]; # 1
@a[1] = 42;
say @b[1]; # 42
Becuase lazy means we don't actually copy from @a until we really need to 09:15
09:15 virtualsue joined, Sqirrel joined
jnthn Note that it's not binding, still assignment, so: 09:15
@a[1] = 0;
say @b[1]; # 42 still, 'cus the bondage is broken once the assignment gets done
I...guess I can't wait to see what evil people actually do with these semantics :P 09:16
ShimmerFairy jnthn: looks nice. I'd hope that in most cases it's not unobvious where 'lazy' things come up :) (though I suppose there's always the possibility of forcing something to be eager if you're paranoid)
nine jnthn: that's neat :)
jnthn ShimmerFairy: The only things that default to lazy are all typically immutable in nature 09:17
09:17 aborazmeh joined, aborazmeh left, aborazmeh joined
nine jnthn: not in a strict sense. A file can be changed while one is iterating with lines() 09:17
09:18 espadrine joined
jnthn nine: "in nature" was my get-out clause for that :) 09:21
nine I thought "typically" was ;) 09:22
jnthn Always have backups!
ShimmerFairy the prudent thing would of course be to do shell("cp $file $file.mine"); for "$file.mine".lines { } :P 09:24
jnthn Darn it. I've an off-by-one somewhere... 09:31
Doing @a[2] actually reifies 4 things, not 3...
oh, it was an off-by-reified error... 09:34
nwc10 order 1 beer, get 2? 09:37
jnthn Well, if you'd reified two beers first, you'd have ended up with 5 when asking for the third one... :) 09:39
And I'm not sure "it's happy hour!" is going to fly in RT :P
09:39 brrt left
jnthn (It got fixed, working on eager now) 09:39
09:40 pRiVi joined
dmitri I am looking for an example of use of "package" keyword in Perl 6, but I cannot find it. 09:40
Is there some code that says package Foo { ... }?? 09:41
jnthn dmitri: We don't tend to use "package" in Perl 6, but rather module/class/grammar
RabidGravy however the BSON module does use it, I was looking at it yesterday 09:42
dmitri jnthn: I realize that -- this is what the docs say. However, it seems that it *can* be done.
BSON?
dmitri looks
jnthn dmitri: It can, but you'll not get various bits of conflict detection 09:43
m: package Foo { }; package Foo { };
camelia ( no output )
jnthn m: module Foo { }; module Foo { };
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/gCzZWGWKvb␤Redeclaration of symbol Foo␤at /tmp/gCzZWGWKvb:1␤------> 3module Foo { }; module Foo7⏏5 { };␤ expecting any of:␤ generic role␤»
RabidGravy but in summary
dmitri RabidGravy: thank you very much!
RabidGravy m: package Foo { our $Bar = "hjs" }; say $Foo::Bar
camelia rakudo-moar 262352: OUTPUT«hjs␤»
dmitri The reason I am looking for it is that I missed this keyword in ctags parser of Perl 6 and I wanted some examples. Now I have them. This is good. 09:45
jnthn Aha :)
RabidGravy jnthn, isn't that "non-confliction" actually a reason to use it, so you can use it in more than one file for instance? 09:48
though
09:48 spider-mario joined
RabidGravy m: package Foo { class Bar {} }; package Foo { class Bar {}} 09:48
camelia rakudo-moar 262352: OUTPUT«===SORRY!===␤Could not locate compile-time value for symbol Bar␤»
RabidGravy is a bit odd 09:49
jnthn RabidGravy: Well, we use it for stubbing int he compiler too
m: class A::B { }; class A { } # this works since A starts out as a package after the first decl 09:50
camelia ( no output )
09:50 virtualsue left 10:00 brrt joined, virtualsue joined
nwc10 [ptc]: do you happen to know what the ticket price was for Pycon Europe? I can't find it online any more 10:00
of course I can *now* :-) 10:01
actually, that's On-Desk rate that I can find. I guess "Standard Rate" was a bit lower than "Personal: EUR 440 (for people enjoying Python from home)" 10:02
[ptc] nwc10: I'll just go check, hang on a tick 10:03
nwc10: 255Euro was the normal price (i.e. not the On-Desk or Early Bird tickets) 10:07
nwc10: hth
brrt what does 'On-Desk' refer to?
[ptc] brrt: when one turns up to the conference without buying tickets online beforehand 10:08
brrt ah, ok
but they're still pre-selling the tickets? weird...
[ptc] no, the conference is over
brrt oh, right :-) 10:09
10:09 [Tux] joined
[ptc] and they stopped selling normal tickets a couple of days before the conference 10:09
jnthn Giving people a strong incentive to buy their ticket before they show up is rather helpful from an organizational perspective. :)
jdv79 is there any way to interp the type name on the LHS of .? 10:11
my $c = "A"; $c.new kinda thing
jnthn ::($c) 10:14
Not just for LHS of .
jdv79 explains why i coudln't find it whtn looking at method calls docs 10:15
:(
thanks!
jnthn actually wrote a test for my $i = 0; my @a = lazy eager loop { last if $i == 5; $i++ }; 10:16
nine A lazy eager loop?
jnthn Yeah.
nine Where's the difference to my @a = lazy [ 0, 1, 2, 3, 4, 5 ];? 10:17
10:17 eternaleye joined
jnthn gist.github.com/jnthn/aa370f8b32ef98e4e7c9 updated with lazy/eager 10:17
nine: See the comment I wrote above the final test in gist linked above 10:18
nwc10 [ptc]: that was the normal business price? (I'm guessing that your work paid for you) 10:20
10:20 danstoner joined 10:21 Grrrr joined
nine Aah...a lazily started eager loop so to speak 10:22
RabidGravy lazy eager is "really likes the idea of doing it, but in reality can't be bothered"
jnthn "I'm not going to do anything, unless you make me, and then I'll do EVERYTHING just to show you!" 10:24
10:25 kaare_ left 10:31 kaare__ joined
RabidGravy :) 10:31
sounds like me most of the time TBH
jnthn Mostly I'm glad that the test I wrote did exactly what I expected without having to tweak anything
Anyway, I'm happy we have a well-defined meaning of lazy/eager now. I don't know we've had that so clearly before.
10:31 nwc10_ joined
jnthn OK, hyper and race...these will be hard 10:31
jdv79 and awesome!
RabidGravy m: use NativeCall; class Foo is repr('CStruct') { has CArray[num32] $.foo is rw; } say Foo.new(foo => CArray[num32].new); # is that simply NYI or intentional?
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/J_QhMBhHEq␤Strange text after block (missing semicolon or comma?)␤at /tmp/J_QhMBhHEq:1␤------> 3uct') { has CArray[num32] $.foo is rw; }7⏏5 say Foo.new(foo => CArray[num32].new); ␤ expecting an…»
RabidGravy m: use NativeCall; class Foo is repr('CStruct') { has CArray[num32] $.foo is rw; }; say Foo.new(foo => CArray[num32].new); # is that simply NYI or intentional?
camelia rakudo-moar 262352: OUTPUT«Cannot modify an immutable NumTypedCArray[num32]␤ in block <unit> at /tmp/upJjqhtPuB:1␤␤» 10:32
RabidGravy rather
jnthn looks odd to me
RabidGravy or, at option, a bug
jnthn afk for a bit to do design work on hyper/race
10:32 rurban left
jdv79 lets say I want A.new to return a A::B obj - how? 10:40
how best i mean
in p5 i can just bless into A::B
10:41 [TuxCM] left 10:42 tommi joined 10:45 Sqirrel left, araujo left, pyrimidine left, Woodi_ left, pecastro left, ilbot3 left, Khisanth left, schmooster left, silug_ left, vytas left, krunen left, ingy left, bobkare left, f3ew left, freeze left, vike left, aborazmeh left 10:47 nwc10_ is now known as nwc10 10:48 [Tux] left, [Tux] joined 10:49 Sqirrel joined, araujo joined, schmooster joined, pyrimidine joined, Woodi_ joined, pecastro joined, ilbot3 joined, Khisanth joined, silug_ joined, vytas joined, krunen joined, ingy joined, bobkare joined, f3ew joined, freeze joined, vike joined, krakan joined, MilkmanDan joined, felher joined, slavik joined, Guest90936 joined, maddingue joined, ab5tract_ joined, mls joined 10:51 hobbified left, chansen_ left 10:52 hobbs joined 10:53 chansen_ joined 10:57 freeze left 10:59 freeze joined 11:01 Sqirrel left
[ptc] nwc10: no, that was the personal price 11:01
nwc10: I paid 340 for early bird business price, but since work didn't end up paying for me, I converted the ticket into a personal one 11:02
nwc10 ah OK thanks. that's doubly useful
11:03 Sqirrel joined 11:04 rurban joined 11:05 danaj joined
jdv79 'A::B' cannot inherit from 'A' because it is unknown - even with require? 11:07
11:08 rurban_ left
jnthn require is runtime, inheritnace is compile time 11:09
jdv79 no way around that? 11:11
jnthn Use "use"? 11:12
Do the require at BEGIN time?
jdv79 well, the use doesn't work cause its in A 11:13
that's why i tried require
jnthn However you slice it, though, the thing you're inheriting from has to be available to the compiler by the time it sees the "is A"
jdv79 hmm, i thought that sort of cycle would be handled well 11:16
jnthn That Perl 6 does one pass parsing goes pretty deep.
jdv79 painfully 11:17
hope it worth it
*its
jnthn Well, it tends to force you to think out your dependencies better. 11:18
itz [ptc]: *cough* the cricket ;)
jdv79 except it forces architecture patterns but ok
jnthn On hyper/race, my current thinking is that we're going to need some amount of "contaigous" behavior on it
jdv79 just not as perlish as i'm used to
viral ? 11:19
jnthn That is, once you opt in to hyper then things downstream of it will be too
my $max = @climate-files.map(&parse).hyper().map(&normalize).grep(*.continent eq 'Europe').max(*.yearly-average); 11:21
ShimmerFairy jnthn: if that's the case, would there be room for some sort of "I can wait" piece in the stream? (Like, say, @list».foo.bar.baz.«wait-to-do-this )
jnthn I think .hyper() and .race() cannot be quite like .eager() 11:22
We want them to go deep so we can do pipeline-y things
And also we want to be able to do parallel reduce (max, min, etc.) too
Trouble is I don't like the visaul off-by-one I'm seeing too 11:24
In that in the one I just wrote above, the hyper() would parallelize the thing that is to the left of it AND those to the right
I'm wondering if that one should be written 11:25
my $max = @climate-files.hyper().map(&parse).map(&normalize).grep(*.continent eq 'Europe').max(*.yearly-average);
I think previously we might have imagined that 11:26
ShimmerFairy jnthn: honestly, I thought that hyper() only affected everything after it. (would correlate with ». at least)
jnthn hyper for @xs -> $x { } # just @xs.map({...}).hyper.sink 11:27
But that we may want the compiler re-write to be
hyper for @xs -> $x { } # just @xs.hyper().map({...}).sink
ShimmerFairy: Yes, that would be your first guess reading the code, and I think we probably want to make that how it is
It *is* a departure from earlier thinking, where hyper and race were just forms of eager 11:28
The trouble with putting it on the end is it doesn't work out for things that leave the monad
Like reduce
ShimmerFairy I think it'd make more sense for .hyper and .race to say "from this point on, things run in parallel" (as though it were actually .hyper(stuff) instead of .hyper.stuff, in a sense)
jnthn Yeah
Until you call something that doesn't know how to play within the paradigm 11:29
So basically .hyper() and .race() return a HyperSeq, and multi-dispatch picks parallel implementations of .map, etc.
JimmyZ thinks it is DWIM # about race/hyper contagious 11:30
jnthn And that way it can pick a parallel implementation of .reduce too, for example
ShimmerFairy jnthn: would this affect the fact that you sometimes end up with @list».foo».bar».baz , or would it just affect the use of .hyper ?
jnthn ShimmerFairy: It depends what >>.foo is meant to return 11:31
ShimmerFairy: If it can return a HyperSeq than we'll get the opportunity to pipeline.
Which TimToady++ really wants
(And me too!) 11:32
ShimmerFairy Personally multiple ». has always felt a bit strange to me, and almost seems like it should be a sign of multidimensionality (that is, the example line I gave maybe sorta reads like for @list { for $_.foo { for $_.bar { ... )
jnthn Cache locality is everything.
Well, with what we're discussing here, it becomes sugar for .hyper().deepmap(*.foo) 11:33
Writing a parallel deepmap will be a headache-inducing...
ShimmerFairy wonders if HyperSeq would look a bit Junction-y in spots...
jnthn No 11:34
It'll be incredibly boring
Like Seq is
[ptc] itz: please remember that I'm a New Zealander ;-) 11:35
jnthn (New Zealand)++ # beautiful scenery AND craft beer 11:36
ShimmerFairy jnthn: Yeah, I suspected we wouldn't suddenly be introducing a new non-Any type at this point, I was just reminded of autothreading for some reason :)
[ptc] itz: it's nice to see the Aussies lose for once :-) They were too powerful a force in cricket for so long
jnthn: Summer Ale is *very* yummy 11:37
jnthn Also, hyper whenever $supply -> $x { ... } really, really wants the $supply to be affected for a clean desugar
So making hyper for work the way I suggested will make it be exactly the right operation on hyper whenever.
I...guess hyper/race are statement prefixes... 11:38
Only trouble is what hyper @foo.map(*.bar) means, which if it just calls .hyper on the end result will not be good.
[ptc] itz: but 60/10 in one innings in test cricket is a whipping! I believe only the Kiwis have played worse...
ShimmerFairy I feel like hyper and race are more "deliberate" than eager/lazy. eager/lazy feel like something you'll naturally encounter often (and perhaps not always need to think about), while hyper/race feel like they'll only usually happen when you specifically ask for it. 11:39
jnthn Oh, for sure... >>.foo is just a succinct way of asking for it :)
Maybe for now we conservatively only support hyper/race statement prefixes on for and whenever... 11:40
I think most people setting up pipelines will do it with the methods 11:41
11:41 spider-mario left 11:43 jferrero joined
ShimmerFairy »» for race, obviously :P (I did stumble upon an old specs revision today that mentioned »» as a possible deepmap-ing hyper) 11:44
jnthn It's funny that hyper and race share almost exactly the same code until the end when you're about to hand back a buffer, and with hyper you have to check if it's the right sequence number :) 11:45
Anyway, will ponder some more over lunch, but I think we're sorta converging on something workable.
jdv79 ha. the OO hierarchy seems to work as i wanted. i forgot to use the parent in the child:(
jnthn jdv79: phew...so it's not as bad as you feared... :) 11:46
ShimmerFairy jnthn: so, hyper is just .race(:who'sin1st) ? :P
jdv79 my assumption about the error being related to the deps was not 100% corret ^H 11:47
11:47 laouji left
jdv79 jnthn: seems so. nice. 11:47
brrt jnthn: we want to do SIMD over hyper, don't we? 11:50
11:51 [TuxCM] joined 11:56 timotimo left 12:02 Sqirrel left 12:10 dayangkun_ left 12:12 Sqirrel joined 12:16 Sqirrel left 12:21 Sqirrel joined
[Coke] jnthn: typo of "Corret" in one place in the glr gist. 12:22
12:25 TEttinger left 12:28 timo joined, timo is now known as Guest16283, Guest16283 is now known as timotimo 12:34 cognominal joined 12:36 lucasb joined 12:38 yqt joined
lucasb m: multi f($,$,$,$,$,$) {}; f 12:47
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/nhKVACBKCS␤Calling f() will never work with any of these multi signatures:␤ ($,,,,,)␤at /tmp/nhKVACBKCS:1␤------> 3multi f($,$,$,$,$,$) {}; 7⏏5f␤»
lucasb lost the anonymous scalars somewhere down the road 12:48
12:48 pmurias joined, dmitri left
pmurias hi 12:49
tony-o hi 12:54
12:54 AlexDaniel joined
RabidGravy erk 12:55
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff7b210c6 in MVM_frame_find_lexical_by_name () from /home/jonathan/.rakudobrew/moar-nom/install/lib/libmoar.so
itz [ptc]: :) 12:58
12:59 rurban left 13:04 beastd left, beastd joined 13:11 zakharyas joined
sergot can you see any LHFs in the RT? 13:17
I'm looking for an easy task to work on 13:18
are you aware of one?
lucasb m: my \x = 1; x\ .say 13:24
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/rktUk5i_V7␤Variable '&x' is not declared␤at /tmp/rktUk5i_V7:1␤------> 3my \x = 1; 7⏏5x\ .say␤»
lucasb Unspaces doesn't mix with terms. this affects 'self\ .method' as well 13:25
m: class C { method f { self.say } }; C.new.f 13:28
camelia rakudo-moar 262352: OUTPUT«C.new␤»
lucasb m: class C { method f { self\ .say } }; C.new.f
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/cDjU2HwYpT␤Variable '&self' is not declared␤at /tmp/cDjU2HwYpT:1␤------> 3class C { method f { 7⏏5self\ .say } }; C.new.f␤»
13:31 andreoss joined 13:34 yqt left
[Coke] sergot: no tickets are tagged LHF at this time. 13:35
When I'm digging for stuff, I sometimes try the LTA errors - sometimes making them awesome enough is easy enough.
searching for things with Parrot in the title and seeing if they work ok on Moar or JVM... 13:36
m: my Int a 13:37
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5===␤Type 'Int' is not declared. Did you mean 'int'?␤at /tmp/dzhAgkCmKQ:1␤------> 3my Int 7⏏5a␤Malformed my␤at /tmp/dzhAgkCmKQ:1␤------> 3my Int 7⏏5a␤␤»
[Coke] m: my Int a;
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/kTKug1Rgoj␤Malformed my (did you mean to declare a sigilless \a or $a?)␤at /tmp/kTKug1Rgoj:1␤------> 3my Int a7⏏5;␤»
[Coke] ^^ RT #120831; ISTR jnthn fixed a bug like this recently that needed to allow for an end of file, not just an end of statement. 13:38
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=120831
jnthn sergot: rt.perl.org/Ticket/Display.html?id=125733 is an easy error message fix
13:38 raiph joined, skids joined
[Coke] m: uniname("\0").say 13:39
camelia rakudo-moar 262352: OUTPUT«NULL␤»
13:43 rurban joined 13:47 Sqirrel left 13:49 Sqirrel joined 13:52 mjgardner joined
tony-o m: ''.^can('uniname').say; 13:53
camelia rakudo-moar 262352: OUTPUT«uniname␤»
andreoss m: "\1".uniname.say 13:54
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/iXnbMrmjGC␤Unrecognized backslash sequence: '\1'␤at /tmp/iXnbMrmjGC:1␤------> 3"\7⏏051".uniname.say␤ expecting any of:␤ double quotes␤ term␤»
andreoss m: "\0".uniname.say
camelia rakudo-moar 262352: OUTPUT«NULL␤»
13:54 brrt left
tony-o m: '👩‍👩‍👦 '.uniname.say; 13:54
camelia rakudo-moar 262352: OUTPUT«WOMAN␤»
andreoss isn't \1 just 1? 13:55
tony-o no
\1 is 'start of heading'
'1' is \31 iirc 13:56
x31 ..
timotimo m: say "1".encode.perl.say
yoleaux 02:56Z <skids> timotimo: NQP 808ce0fcb broke ("a","b")».ords for some reason.
camelia rakudo-moar 262352: OUTPUT«utf8.new(49)␤True␤»
timotimo :o
oh crap, i broke something?
tony-o x31 == 49 :-)
skids Well, only for bleeding edge people.
tony-o timotimo: it's bc of the 2 'say's 13:57
timotimo no, i meant what skids showed me
tony-o oh ^
timotimo we obviously didn't have a test in roast that'd exercise this code path
tony-o m: "\x31".uniname.say;
camelia rakudo-moar 262352: OUTPUT«DIGIT ONE␤»
tony-o m: "\x1".uniname.say; 13:58
camelia rakudo-moar 262352: OUTPUT«START OF HEADING␤»
tony-o andreoss ^^
13:58 Sqirrel left
tony-o m: "\xFE".uniname.say; 13:58
camelia rakudo-moar 262352: OUTPUT«LATIN SMALL LETTER THORN␤»
skids timotimo: as big as roast is, it still has holes :-) 13:59
tony-o that isn't a nice thing to say about someone skids 14:00
skids tony-o: sorry, roast :-)
timotimo skids: can you give me a one-liner that shows ("a", "b")>>.ords breaking?
m: say ('a', 'b')>>.ords
camelia rakudo-moar 262352: OUTPUT«97 98␤»
timotimo this is the expected output?
skids Yes, camelia does not have that nqp yet.
timotimo of course
skids With it you get a blank line -- ((),()) actually. 14:01
timotimo oh lord
i'm about to have a built rakuod on this machine
skids But only for ords. And maybe some other thing. But not for .chars at least. 14:02
timotimo ... what? %)
skids m: say ('a', 'b')>>.uninames 14:04
camelia rakudo-moar 262352: OUTPUT«LATIN SMALL LETTER A LATIN SMALL LETTER B␤»
skids Yeah, things that return lists are broken. 14:05
So probably more a problem in ">>." than ".ords"
timotimo mhm, likely 14:09
... there aren't even "while"s in there :\ 14:11
14:12 AlexDaniel left 14:17 brrt joined
lucasb what does 'rpa' stands for in the core? 'returned parcel' or something? 14:20
14:22 FROGGS left
timotimo nah 14:22
something PMC Array
[Coke] resizable 14:23
timotimo ah, yeah
from when we primarily used parrot as our backend
[Coke] original parrot nomenclature.
PMC == "parrot magic cookie"
sort of an primitive VM object.
lucasb oh, thanks timotimo and [Coke]!
Woodi_ hi #perl6 :) 14:24
timotimo hello Woodi_
RabidGravy and there I was thinking that a libsamplerate binding would be nice and easy 14:27
hahahahaahahahahahahaha!
nine Every time you break object encapsulation in Perl 5 code, an Inline::Perl5 developer's kitten dies... or so
Woodi_ jnthn: .hyper don't need to parallelize to the end, eg. .map().map().map() 2 first can work concurently but 3rd can use values from all places in the array. so before 3rd map() values on "one side" can be inserted into buffer and on the "other side" iterated sequentially 14:30
timotimo how would the last .map be able to use values from everywhere? 14:32
Woodi_ also I wanted to say: "instruction pipeline is more valuable" in CPUs (.map() instructions can be put there and data can flow) but realy we want all cases :) static instructions in cache but also small "static" data with long .map() sequence... 14:33
timotimo: something needs to collect them :)
14:33 AlexDaniel joined
timotimo huh? 14:34
but map works only with chunks of the thing on the LHS
Woodi_ is .map().map like my @b = @a.map(); my @c = @b.map(); .... ? 14:35
timotimo kind of
but more like my $in_between := @a.map(); my @result = $in_between.map()
there's not really an assignment step in between when using .map().map() 14:36
Woodi_ yes, but if 3rd map wants to see all data then something must to bufer them. or programmer can use 2 lines of code... :) 14:37
jnthn Indeed
.map doesn't return data
.map returns something that knows how to provide data
Woodi_ details :) 14:38
jnthn Details you don't know about, so you spout meaninglessly, yes... 14:39
14:42 skids left
pmurias nine: why does breaking object encapsulation cause trouble for Inline::Perl5? 14:45
timotimo no need for a second line of code either, you can just map.map.eager.map to have the result of the first two maps eagerly evaluated and thus buffered
14:45 xfix joined
jnthn timotimo: Yeah, that works; I do wonder if we want a .sequential() or so that drops you back to sequential processing without having to buffer up everything though. 14:48
This bit is really, really tricky to design and get right...
timotimo i'm not sure i understand that?
the first image that appears in my head is "turn non-cooperating code into gather/take"
nine pmurias: because subclassing P5 classes in P6 is kind of a stunt that works surprisingly well...until someone from outside the base class tries to look at the object's internals 14:51
pmurias: as if it wasn't already complicated enough. Perl 5 documentation tells you for example that an object is a blessed reference and it looks like that's true. But deep down it isn't. It's a reference to a blessed thingy. I was quite surprised when I found out that. 14:54
timotimo jnthn: or do you mean some kind of mechanism to signal that only parts of a sequence are supposed to be eagerified?
jnthn timotimo: eager blocks until the previous stage has produced everything 14:57
14:58 skids joined, telex left
jnthn timotimo: If you still want to consume values as they become available from the parallel computation, but that you want to be sure you're on a single thread, then eager is kinda overkill 14:58
.lazy works too, but that may be overkill in the other direction
timotimo ah 14:59
jnthn Also, the way this is turning out, .sequential() is probably going to be a one line method to provide... I bet TimToady'll want it called soemthing else though.
timotimo so it'll give you something more like pipeline semantics, but on a single thread using batching instead of communicating threads or something 15:00
PerlJam call it .seq and confuse everybody ;)
timotimo basically a moving buffer/window thingie that gets pushed along?
15:00 telex joined
jnthn PerlJam: heh, or call HyperSeq a Par :P 15:01
15:03 Possum joined 15:07 rurban_ joined
itz m: class A { has $.foo; has $.foo}; 15:07
camelia rakudo-moar 262352: OUTPUT«===SORRY!===␤Package 'A' already has an attribute named '$!foo'␤»
nine pmurias: or not! It's really the reference that's blessed. Looking at the code, it's just newSVrv's documentation that's clearly wrong. Sending a doc patch...
itz shouldn't that be $.foo in the error or am I missing something? 15:08
jnthn itz: Attributes are always called $!foo 15:09
itz: The . just means "and also generate an accessor" 15:10
itz ah
jdv79 complexity ftw
TimToady isn't awake yet, but that kinda sounds like what I was proposing to call "silo" 15:11
15:12 vytas left
awwaiid ICFP contest today! icfpcontest.org 15:12
GREAT place to show off a new language. Actually pretty much the perfect place.
15:12 xfix left
jdv79 what does this silo do? 15:14
15:14 vytas joined
nine Wow, newSVrv's docs have been wrong since they were written in 1994 15:17
15:17 diana_olhovik_ left 15:19 captain-adequate joined
RabidGravy something weird that someone may able to explain to me 15:19
pmurias nine: when I create 2 references to the same hash, both change class when I call bless on one of them 15:21
15:21 pierrot_ is now known as pierrot
nine pmurias: oh yes, that's true. And is the reason, why my first attempt at implementing this failed back then. 15:22
RabidGravy if I do something like "class A is repr('CStruct') { has CArray[num32] $!foo; submethod BUILD() { $!foo := CArray[num32].new }}" and then pass that a new one of thiose to a native sub it segfaults trying to write the array
but something like "class A is repr('CStruct') { has CArray[num32] $!foo; submethod BUILD() { my $foo := CArray[num32].new; $!foo:= $foo }}" works fine 15:23
any clue as to why?
nine pmurias: so what I wrote earlier is true after all. It's the thingy that's blessed, even if you have to pass the reference to sv_bless. 15:24
RabidGravy yeah 15:28
perl -e 'my $a = 1; bless \$a, "Foo"; print $a,", \$a'
1 Foo=SCALAR(0x1c676e0)
15:30 pmurias left
dalek osystem: 23c4882 | (Steve Mynott)++ | META.list:
add HTML::Restrict
 15:36
RabidGravy Hmm libsamplerate produces 0.2% too few frames when doubling the samplerate 15:48
15:53 Psyche^ joined 15:58 Psyche^_ left 15:59 cognominal left 16:00 zakharyas left
RabidGravy resorts to 'ok ($out-frames-total / ($in-frames-total * 2)) <1, "got the expected total number of frames (approximately)";' 16:02
16:04 Sqirrel joined 16:17 brrt left 16:21 nys joined
tony-o who is stmuk ? 16:25
itz me 16:27
timotimo didn't know that 16:28
tony-o does that HTML::Restrict just filter out bad html tags, does it kill all the child nodes or place them in the parent?
itz I think it might wrongly kill child nodes which I only noticed a few mins ago 16:29
tony-o should have a :save-the-children flag 16:30
itz :)
tony-o that way it's fun to say to other people
16:33 rurban left
RabidGravy or :geldof 16:35
tony-o geldof ? 16:37
itz UK charity figure
tony-o ah 16:38
itz a bit like the vim editor
b2gills m: 'a' ==> { .say }() # does it make sense to allow this to work without the parens 16:39
camelia rakudo-moar 262352: OUTPUT«a␤»
16:41 brrt joined, rurban_ left
tony-o m: 'a' ==> { .say; }; 16:44
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/wO8SKf8llE␤Sorry, do not know how to handle this case of a feed operator yet.␤at /tmp/wO8SKf8llE:1␤------> 3'a' ==> { .say; }7⏏5;␤»
16:46 mr-foobar joined
sergot jnthn: thanks! 16:57
jnthn: I will start there then :) 17:00
TimToady I don't think I like contagious hyper if it's user-visible, cuz it's adding another dimension to the ambiguity of dispatch, which was already bad enough with 'nodal' 17:05
in @a».foo.elems, is that .elems supposed to be the number of elements that came through from @a, or the number of elems from each return from .foo, or the number of hyper batches? 17:06
if we want something contagous, it should have both a beginning and an end
*gious
much like we took the original placeholder proposal with its arcane scope rules, and said, "It's always just to the curlies" 17:07
jnthn TimToady: Aside from any side-effects foo may do, it's exactly equivalent result wise to @a.deepmap(*.foo).elems 17:08
TimToady so for now I think hyper should only be contagious to the next hyper
17:09 eam joined
timotimo how did my change to nqp::while break deepmap? :( 17:09
the things i changed were - potentially - how passing the condition to the block will be handled, and what nqp::while returns 17:10
eam I'm playing around with perl6 and I don't understand why perl6 -e'say ("a", "b" ...^ * gt "y").pick' prints one character "a".."y" like I'd expect, but if I say gt "z" I get two random letters
jnthn TimToady: If you'd prefer @a>>.foo to mean @a.hyper().deepmap(*.foo).sequential() it's fine too
timotimo the bodies of the nqp::while blocks in deepmap aren't blocks with arguments, so no argument passing needed, and the return value would just get sunk'd
TimToady I just don't want it magically turning into @a.deepmap(*.foo.elems) 17:11
17:11 havenwood joined
jnthn Oh, that wasn't even slightly what I was suggesting 17:11
TimToady okay, was probably misunderstanding something in the backlog then
17:12 brrt left
timotimo double-checks it's actually that exact commit that breaks it 17:13
eam nevermind, I figured it out - "za" is the first sequence gt "z" 17:14
"yz" actually 17:15
timotimo oh damn it, it *is* that commit
dalek rl6-roast-data: e1a0985 | coke++ | / (9 files):
today (automated commit)
17:15 domidumont joined
TimToady jnthn: another more credible wrinkle is if you have ».foo.map: -> $a, $b {...} then you risk $a being at the end of one batch and $b at the beginning of another, so you can't pipeline batches quite independently 17:17
tony-o what order does » operate on a list?
is it doing a map reduce essentially? 17:18
TimToady any order it likes, as long as it returns results in the original order
tony-o ah okay, i was consistently getting the same order from it and was curious. ty
TimToady I think race is more like map-reduce 17:19
tony-o m: ("a".."z")»{$_}.join.say;
camelia rakudo-moar 262352: OUTPUT«abcdefghijklmnopqrstuvwxyz␤»
timotimo tony-o: we intentionally throw the items around, but it's deterministic sticc 17:20
m: ("a".."z")>>.say
camelia rakudo-moar 262352: OUTPUT«z␤x␤v␤t␤r␤p␤n␤l␤j␤h␤f␤d␤b␤y␤w␤u␤s␤q␤o␤m␤k␤i␤g␤e␤c␤a␤»
timotimo m: ("a".."z")>>.say
camelia rakudo-moar 262352: OUTPUT«z␤x␤v␤t␤r␤p␤n␤l␤j␤h␤f␤d␤b␤y␤w␤u␤s␤q␤o␤m␤k␤i␤g␤e␤c␤a␤»
tony-o m: race (1..5) { $_; }
camelia rakudo-moar 262352: OUTPUT«5===SORRY!5=== Error while compiling /tmp/IfAgrj4FtL␤Unexpected block in infix position (missing statement control word before the expression?)␤at /tmp/IfAgrj4FtL:1␤------> 3race (1..5)7⏏5 { $_; }␤ expecting any of:␤ infix␤ …»
tony-o timotimo: that's what i was experiencing ^ thanks for confirming
TimToady race is NYI, barely :) 17:21
jnthn TimToady: I'm just not going to attempt to parallelize maps where the block's arity ain't 1 for now
TimToady: But yeah, there's no very easy way to handle that.
17:22 mr_ron joined, cognominal joined
jnthn Whatever you do. 17:22
I think what I've got will be flexible enough API wise to find a way to cope reasonably well. 17:23
dinner; bbiab &
timotimo perl6 -e 'say deepmap(-> $a { (1, 2).map(->$a {$a}) }, ("a", "b"))' - comes up empty 17:24
perl6 -e 'say deepmap(-> $a { my @ = (1, 2) }, ("a", "b"))'
1 2 1 2
TimToady on a slightly different subject, I know Fortress did a lot of work in list structures that could be split N ways easily, so I'm wondering how much of a bottleneck with left-to-right splitting of batches at the head of the pipeline...
havenwood How to you limit to just the `foo(bar)` or `bar.foo` style? My googlefu is failing me.
How do* you 17:25
timotimo what do you mean?
RabidGravy method or sub? 17:26
sub or method
havenwood I'd swear there was a way to do something like `strict OO`, my memory fails me.
TimToady why would you want to do that?
17:27 diana_olhovik joined, andreoss left
mr_ron Anyone know a p6 module with good examples of exception classes and handling? 17:27
TimToady there are very good reasons for both early binding and late binding
havenwood TimToady: I don't want to use it. Just recalled there being a way like `use oo;` but can't find it for the life of me. 17:30
TimToady: Just playing around, mere curiosity. 17:31
TimToady you're probably thinking of S12's use oo :closed :final;
17:32 domidumont left
TimToady but that's just knobs on OO, not a stricture 17:32
17:33 diana_olhovik left 17:39 coffee` joined 17:41 llfourn left
TimToady jnthn: on bikeshedding .sequential, some starting ideas: .reform .linear .serial .queue .fifo .hypo 17:44
well, and .silo 17:46
but maybe .seq is okay 17:47
17:47 plicease joined
TimToady
.oO("Why are they calling string eq there?")
 17:48
colomon just booked his plane tickets to Switzerand.
17:51 Sqirrel left 17:53 Sqirrel joined
jdv79 Your search - nodal site:perlcabal.org/syn/ - did not match any documents. :( 17:59
18:11 espadrine left 18:15 pRiVi is now known as Tisch_der_Sieger 18:22 havenwood left 18:27 [TuxCM] left 18:33 espadrine joined
raiph jdv79: 6 matches in design.perl6.org/S03.html 18:37
jdv79 i noticed that 18:41
the google site search on design.perl6.org aint so hot
yay for good ole grep 18:42
18:56 [TuxCM] joined 19:04 [TuxCM] left 19:06 AlexDani` joined 19:07 AlexDaniel left 19:16 [TuxCM] joined
dalek kudo/nom: c048e75 | lizmat++ | src/core/Pair.pm:
Naively implement pair()

As discussed at irclog.perlgeek.de/perl6/2015-08-06#i_11014717 .
Please note that I don't think it's a good idea to also have a named variant for this: if you want to use named parameters, please do Pair.new()		 19:25
lizmat $ 6 'say pair(42,666)' 19:26
42 => 666
please note I also see a use for pair() in those cases you want to be sure to pass a Positional Pair, rather than a named parameter
aka frobnicate( pair("a",42") )
aka frobnicate( pair("a",42) ) # rather :-) 19:27
jnthn TimToady: Any feeling in exception semantics for hyper/race? 19:28
*on
TimToady well, we'd like each dataflow to succeed or fail separately, so we don't blow up the rocket just because one sensor was bad 19:32
jnthn That's fine if folks fail, but what if they die? 19:33
Uh...that sounded more morbid than intended. :)
19:35 mjgardner left
dalek ast: 9940e5e | lizmat++ | S04-statements/if.t:
Remove superstitious parens from if tests
 19:35
TimToady for now that's a DIHWIDT, I suspect; I don't recall that we've specced any kind of distributed try
jnthn *nod* 19:36
19:37 llfourn joined
jnthn Well, I guess the more specific question I shoulda asked is: do we just throw the first exception we become aware of, or if we manage to get concurrent failure to we package them up into some kind of aggregate exception? 19:37
In .Net they do the latter in the name of preventing information loss, but I've rarely seen that actually happen in the wild. 19:38
And it makes exception handling harder, 'cus now your when blocks won't match
So refactoring to parallel becomes more of a job.
19:42 llfourn left
dalek kudo/nom: 1a3f1a2 | TimToady++ | src/ (2 files):
make andthen/orelse topicalizers

This also fixes the with/without statement modifiers to topicalize correctly.		 19:42
ecs: 8268851 | TimToady++ | S03-operators.pod:
change orelse to topicalize to $_, not $!

  (Also makes 'without' use $_ for its topic.)		 19:44
ast: 9ba702f | lizmat++ | S04-statements/with.t:
Add basic "with" test (todo for now)
lizmat off for some MI:4 19:45
TimToady off for lunch
19:50 diana_olhovik joined
AlexDani` sergot: some time ago you asked for LHFs. Well, I don't mark my tickets as LHFs because most of the time I have no idea what the problem is, but at the same time I've never submitted anything that is extraordinarily hard 19:50
19:50 AlexDani` is now known as AlekDaniel 19:51 AlekDaniel is now known as AlexDaniel
AlexDaniel sergot: so maybe filtering by [email@hidden.address] will give you a decent list of LHFs 19:52
yoleaux 6 Aug 2015 05:00Z <skids> AlexDaniel: The message for "..." still has to be fixed; it is a separate issue ("..." implicity tries to "return" from the mainline code, not just sink a Failure) What was fixed was there is no longer a different backtrace when a Failure is sunk in the last statement.
AlexDaniel oh ok
and most of my tickets are just LTAs 19:53
m: .say for 1...NaN; 19:58
camelia rakudo-moar c048e7: OUTPUT«(timeout)1␤0␤-1␤-2␤-3␤-4␤-5␤-6␤-7␤-8␤-9␤-10␤-11␤-12␤-13␤-14␤-15␤-16␤-17␤-18␤-19␤-20␤-21␤-22␤-23␤-24␤-25␤-26␤-27␤-28␤-29␤-30␤-31␤-32␤-33␤-34␤-35␤-36␤-37␤-38␤-39␤-40␤-41␤-4…»
AlexDaniel mwahaha :)
m: .say for Inf..0 19:59
camelia rakudo-moar c048e7: OUTPUT«(signal XFSZ)-9223372036854775808␤-9223372036854775808␤-9223372036854775808␤-9223372036854775808␤-9223372036854775808␤-9223372036854775808␤-9223372036854775808␤-9223372036854775808␤-9223372036854775808␤-9223372036854775808␤-922337203685…»
20:02 [particle] left 20:11 jjido joined 20:12 jjido left 20:13 rindolf left
TimToady m: .say with 42; 42 andthen .say 20:24
camelia rakudo-moar 1a3f1a: OUTPUT«42␤42␤»
TimToady m: .say without Rat
camelia rakudo-moar 1a3f1a: OUTPUT«(Rat)␤»
TimToady m: .say with 0; .say with Nil
camelia rakudo-moar 1a3f1a: OUTPUT«0␤»
20:25 darutoko left
TimToady m: with 0 { .say } # now to make this work 20:25
camelia ( no output )
20:37 llfourn joined 20:41 llfourn left 20:44 diana_olhovik left 20:51 yqt joined
jdv79 is it possible to use a role to share a new and BUILD method amongst consuming classes? 20:54
it seems to only partially work
TimToady role BUILDs are not yet composed properly 21:01
skids jdv79: the workaround has been to abuse the attribute default value initialization logic. 21:07
21:10 skids left
jnthn 'fraid I ain't gonna get the hyper/race all the way there today 21:11
(Headache + bureaucratic distractions... :/) 21:12
Think I'm close to a first working example though.
So probably tomorrow.
TimToady be better! 21:13
jnthn Thanks!
The headache's already gone, by now I'm just tired, and that's not a good state to juggle a lock and a pair of condvars :)
TimToady m: say "jnthn++" without $*headache 21:15
camelia rakudo-moar 1a3f1a: OUTPUT«jnthn++␤»
jnthn ooh :) 21:16
I see you left the bit involving hack on code-gen until the end. :)
TimToady working on that bit now, yeah 21:17
I can generate with and without opcodes, but they still behave like if/unless :)
just need to figure out the bits where it evaluates the conditional so as to pass the original object but evaluate with .defined 21:18
21:18 lucasb left
RabidGravy proselytises a CPAN author who was keen for me to check out his module to port it to Perl 6 ;-) 21:18
first time I've actually been "spammed" about a module I think
jnthn TimToady: I'll be you'll have an easier time doing it on Moar first
TimToady also still need to force passing of the conditional 21:19
well, that's what I'm doing
jnthn TimToady: JVM is a stack machine which makes code-gen harder (unless you're at university writing a compiler for a tiny language, and then the stack machine looks wonderfully elegant to make code for :P)
Yeah, we do stash the conditional away 21:20
TimToady I see the bind, but I need to sneak in a defined after that
jnthn right 21:21
The two places you'll need to tweak are those near resolve_condition_op 21:22
There's this: 21:23
push_op(@ins, 'decont', $decont_reg, @comp_ops[0].result_reg);
$decont_reg now holds the thing to test
So if you just do a findmeth on that and then shove in a call, you can even write the result of the call back into the same register and leave the code the same 21:24
Also, there's a conditional "@comp_ops[0].result_kind == $MVM_reg_obj"
TimToady so that one is the evaluation, and is post-bind to the temp? 21:25
jnthn Yes
That one is literally at the point where we spit out the jump
TimToady right
jnthn We've already stashed the condition away by that point if we need it
Anyway, earlier on you have: 21:26
@comp_ops[0] := ...
Twice
I'd pass :want($MVM_reg_obj) to the two calls that assign to it
That demands an object register
Then you can delete the condition and its else branch that I mentioned a moment ago
Because it's highly unlikely anyone will sensibly use with/without on a native type
TimToady okay, maybe that's enough to ignite my brane 21:28
it's got soggier these last few decades... :)
RabidGravy boo! 21:29
mind I've felt really stupid for the last week 21:30
I kind of woke up again today
jnthn has had one of his better weeks this one for productivity 21:32
TimToady fershure
21:34 bin_005 joined
jnthn kinda wants to show off .hyper/.race and also supply/whenever at this YAPC::Asia talk :) 21:35
*his
TimToady is probably gonna rewarm his fosdem talk, since that never got properly recorded 21:37
TimToady doesn't have the brane to write new talks right now anyway
jnthn Well, I get to re-use the YAPC::Asia one at Swiss Perl Workshop :) 21:39
RabidGravy jnthn, does the thing about not being able to bind a new CArray directly to an attribute in a CStruct and having to bind to a variable which is then bound to the attribute ring any bells? 21:40
jnthn RabidGravy: No, I was pretty confused by that... 21:42
RabidGravy well you can do it but it appears to mess the pointer up 21:43
ShimmerFairy TimToady: what are the "new andthen/orelse" mentioned in one of your commits yesterday? (That is, what's different from how they have been before?)
jnthn RabidGravy: FROGGS++ worked on that stuff much more recently than I have so may have an idea
RabidGravy all weird
the samplerate API is a bit special anyway, it gets a struct which contains both the input buffer and output buffer so quite niche 21:45
jnthn OK, time to wander afk to rest 21:48
Back tomorrow to juggle my condvars some more...
o/
TimToady \o
RabidGravy juggle your stuff in private matey ;-) 21:49
TimToady ShimmerFairy: they topicalize the for the RHS as specced
m: 42 andthen .say
camelia rakudo-moar 1a3f1a: OUTPUT«42␤»
TimToady and the internals are reworked to support with/without statement modifiers also 21:50
ShimmerFairy Ah, so you "just" implemented more of the spec on them then :)
TimToady well, I also tweaked the spec :)
RabidGravy if the spec don't fit tweak it 21:52
TimToady --> nap 22:05
22:05 TEttinger joined, telex left
dalek kudo-star-daily: 1b77bdd | coke++ | log/ (8 files):
today (automated commit)
 22:06
22:06 telex joined
RabidGravy boom! 22:12
dalek kudo-star-daily: 94d04df | coke++ | log/ (10 files):
archive parrot logs
 22:16
22:29 virtualsue left 22:36 Zoffix_ joined 22:38 llfourn joined
Sgeo fail reminds me a bit of Rust's try!(Err(...)) 22:40
22:41 Zoffix left, ][Sno][ joined, TimToady_ joined 22:43 llfourn left, mrf left 22:49 nys left, Grrrr left, [Sno] left, TimToady left, Grrrr joined 22:50 skids joined
RabidGravy is it possible to parameterise a subset? 22:53
lizmat RabidGravy: I have no idea 22:58
looking at blogs.perl.org/users/aaron_baugher/...erl-6.html
I wonder whether we shouldn't have a .contains, which would just return Bool ($i != -1) 22:59
so we can skip the .defined in that benchmark
23:00 nys joined
lizmat gets some sleep 23:06
RabidGravy to be fair index is always going to be way faster than a regex
b2gills I tested parsing with index and substr in Perl 5. It was faster for up to about 4 calls to index, after that the regex was faster. 23:13
23:33 RabidGravy left 23:36 TimToady_ is now known as TimToday, TimToday is now known as TimToady 23:37 mr_ron left 23:46 average joined
average gist.github.com/masak/66735ee4c40a...les-p6-L73 23:46
wow, I like how this looks
lines 85,86,93 much <3 23:47
oh, line 76 too
i'm wondering how the \in and \cap are implemented or if they can be overriden 23:48
I suppose one should be able to implement them for specific cases because they can have faster ones in particular cases 23:49
23:50 beastd left 23:54 captain-adequate left
average either way, much interesting stuff in the p6 blogosphere 23:55
10x
23:55 average left
ShimmerFairy I was about to say: there's "texas" versions of all those unicode operators. Rakudo's source suggests that you should override the texas versions, since the unicode versions are explicitly 'only' subs, but that doesn't actually stop anything. 23:56
m: class Foo { }; multi sub infix:<⊂>(Foo $a, Foo $b) { "OK!" }; say Foo ⊂ Foo; say Foo (<) Foo;
camelia rakudo-moar 1a3f1a: OUTPUT«OK!␤False␤»
ShimmerFairy supposedly, that infix:<⊂> is an 'only' sub and therefore shouldn't be overrideable. 23:57
However, I personally would prefer overriding the unicode operators instead, so I don't mind :) 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!