»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:07 dayangkun joined 00:08 replore left 00:09 Exodist left 00:10 rulker left 00:25 zby_home__ left 00:27 zby_home__ joined 00:39 soh_cah_toa joined 00:40 frhodes joined 00:44 frhodes left 00:46 frhodes joined 00:48 jerome_ joined 00:55 alvis left 01:09 alvis joined 01:15 japhb_ left 01:25 jaldhar joined 01:26 ponbiki left 01:28 uasi joined
sorear mberends: ping 01:30
01:31 perlygatekeeper joined 01:37 thou left 01:40 Chillance left 01:46 djanatyn left 01:47 djanatyn joined 01:51 mtk left, mtk joined 01:54 uasi left 01:55 worr left 02:17 uasi joined 02:20 envi joined 02:21 cotto joined, athomason left, athomason joined 02:30 frhodes left 02:40 lue joined 03:03 shinobicl_ joined 03:04 cognominal_ joined 03:06 mkramer1 joined, y3llow_ joined, pothos_ joined 03:07 lue left, perlygatekeeper left, y3llow left, pothos left, cognominal left, Vlavv_ left, stepnem left, awwaiid left, kcwu left, yves_ left, dukeleto left, cosimo left, kshannon left, PacoLinux left, Woodi left, takesako left, cxreg left, flussence left, Lorn left, Helios left, y3llow_ is now known as y3llow 03:08 pothos_ is now known as pothos, kcwu joined 03:09 stepnem joined, Woodi joined, Lorn joined, Lorn left, Lorn joined 03:10 cxreg joined, cosimo joined, flussence joined, kshannon joined 03:13 frhodes joined, Helios joined, perlygatekeeper joined 03:16 frhodes left 03:20 Vlavv_ joined, lue joined, yves_ joined 03:23 dukeleto joined, takesako joined 03:25 PacoLinux joined 03:31 packetknife joined 03:33 awwaiid joined 03:38 bluescreen10 left 03:40 lumi_ left, lumi_ joined 03:43 packetknife left 03:57 f00li5h is now known as f00li5h-, f00li5h- is now known as f00li5h 04:20 orafu left 04:23 perlygatekeeper left 04:26 woosley joined, am0c joined 04:29 am0c left 04:30 am0c joined, am0c left 04:32 kaare_ joined 04:33 am0c joined 04:35 flatwhatson left, am0c left 05:05 REPLeffect left 05:06 ruoso left 05:13 ruoso joined 05:17 shinobicl_ left 05:18 REPLeffect joined 05:36 woosley left 05:37 woosley joined
mberends sorear: pong 05:42
05:44 ponbiki joined 05:48 ponbiki left 05:52 Exodist joined
sorear mberends: I am wondering if we should implement a stopgap solution to #56 for v10 05:54
05:55 soh_cah_toa left
mberends sorear: yes, any solution would be helpful. We could still track the issue as not fully resulved, and suggest applications do so too. 05:56
sorear: what do you think is the reason interface implementations cannot be called? They are still methods. Could it be only sometimes, for example when they are implemented in a parent class? 05:59
sorear mberends: are you familiar with the C#/CLR concept of "explicit interface implementations"? 06:00
mberends sorear: no, I'll read up on that. I understand explicit in some other contexts. 06:01
sorear in the CLR, methods that implement an interface slot don't have to be public
there are two ways to implement the IDisposable interface in C# 06:02
way 1: public void Dispose() { ... } - creates public method named Dispose 06:03
way 2: void IDisposable.Dispose() { ... } - creates private method named <gensym>
the CLR doesn't rely on names for interface mapping; classes have an explicit slot/method table
way 2 is very useful if two interfaces require different semantics for the same method name 06:04
06:04 REPLeffect left
mberends I see, and Gtk uses way 2. 06:04
sorear it's also nice for the API if you think "Dispose" is a bad method name for your class
the .net coding standards say any functionality should be available via some public method 06:05
the approach Niecza is currently using (reflect on the object's concrete type and bind public methods) is functionally the same as what DLR languages (IPy/IronRuby) do 06:06
mberends the freedom to rename .Dispose() seems like one of those dubious nice-to-haves that leads to confusion. 06:07
sorear mberends: you have a Gtk.Gc, correct?
no 06:08
couldn't be
mberends sorear: gtk-sharp probably wraps Gtk.Gc
06:09 dayangkun left
sorear it's a Cairo.Context 06:10
apparently, Cairo.Context has _no_ public method that calls Dispose functionality 06:11
sorear found the source
OTOH, the finalizer does call it. So the memory leak is not likely to be _destructively bad_ 06:12
mberends Cairo is provided through gdk-sharp instead of gtk-sharp, very similar to use.
yes, the finalizer is the source of the warning message that says "the programmer" forgot to call Dispose. 06:13
sorear: would your hack be along the lines of: "the reflection API denies that .Dispose exists, but let's call it anyway" ? 06:17
no, if it's private, it shouldn't be callable either. 06:18
06:18 REPLeffect joined
sorear mberends: my hack would be along the lines of adding a .dispose-hack method to CLR::System::Object which does the equivalent of ((IDisposable)obj).Dispose() 06:19
mberends: my real fix would be to implement $obj.Class::method() syntax in general for Perl 6 code, and reflect interfaces as roles, allowing you to do $obj.IDisposable::Dispose 06:20
the real fix, however, cannot be commuted with /serialize
mberends sorear: I understand. Strapping the fix on to Object is very global, thus brings the risk of .Dispose() being called inappropriately, but that's caveat utilitor. 06:23
I'd argue against the hack if it implies performance degradation. 06:24
sorear nah 06:25
there are already a handful of Object-strapons 06:26
of course I'd like to get rid of .dispose-hack ASAP :)
mberends sorear: the /serialize refactor is a much higher priority. Thus anything incompatible with /serialize should not even be considered.
dalek ecza/serialize: 6d3b702 | sorear++ | / (4 files):
Lookup variables in new model, can now handle some very simple variable uses
 06:27
ecza: f30bd36 | sorear++ | lib/NieczaCLR.cs:
Add stopgap .dispose-hack method for mberends++ GTK work
 06:34
06:35 cognominal_ left
mberends wow, such a quick turnaround! sorear++ 06:35
but: Unable to resolve method Dispose in class CLR::Cairo::Context :-( 06:47
sorear mberends: .dispose-hack 06:50
mberends sorear++: better name
sorear niecza: say { a => True }.any.perl 06:51
p6eval niecza v9-41-gbb5a7c3: OUTPUT«any("a", )␤»
sorear niecza: any("a").perl 06:52
p6eval niecza v9-41-gbb5a7c3: ( no output )
sorear niecza: say any("a").perl
p6eval niecza v9-41-gbb5a7c3: OUTPUT«any("a", )␤»
dalek ecza: ed11042 | sorear++ | test.pl:
Make test forgiving of Junction.perl format
 06:55
07:02 woosley left 07:08 woosley joined
mberends sorear: another CLR Interop problem is inability to convert Num parameters to System.Double. Shall I create a Github issue for this? 07:17
07:18 drbean joined
sorear what? 07:18
huh. 07:19
...oops 07:20
the code to marshal Num objects is under a when(Integral) condition :|
mberends in niecza/examples/gtk-clock.pl:32, passing Num (fractional) parameters in SetSourceRGB(0, 1, 1) fails. Thus we have 3-bit color.
sorear I knew I implemented it!
mberends you had said so, yes! 07:21
07:21 donri joined
mberends sorear: ok, I can fix a when() 07:22
sorear er, wait, not quite.
mberends waits
07:24 mikemol left, ruoso left
Woodi hallo #perl6.ppls :) 07:25
mberends hallo Woodi
07:34 flatwhatson joined
sorear hello Woodi 07:38
07:38 ruoso joined, mikemol joined 07:40 drbean left 07:41 drbean joined
Woodi what it can be: Can only use repr_get_attr_obj on a SixModelObject ? 07:44
i use field.hash.keys in for loop and this bumps 07:45
sorear That's an internal error and you should not be able to get it, unless you're doing improper things with pir:: 07:50
Somehow, an expression has evaluated to something other than a Rakudo object 07:51
07:56 [particle] joined
dalek ecza: 3896bb2 | sorear++ | lib/NieczaCLR.cs:
Implement marshalling for Num and Rat
 07:56
sorear I got carried away implementing exact Rat->decimal coercion
mberends ooh! sorear++ again :) 07:57
07:57 masak joined
masak goedemorgen, #perl6. 07:57
07:58 [particle]1 left
sorear niecza> 123456789123456789 / 10**20 07:58
0.00123456789123457
niecza> CLR::System::Decimal.marshal(FatRat.new(123456789123456789, 10**20))
0.00123456789123456789
o/ masak 07:59
mberends god morgon masak # thanks www.howtosayin.com 08:00
masak :) 08:02
masak conducts a structured attack on pack/unpack 08:03
08:07 xinming_ joined 08:10 xinming left
mberends Gtk clock looks less 1983 without its cyan background :) 08:11
sorear mm 1983 08:13
masak I was two. I didn't know the word "cyan" :) 08:14
sorear I was -7
masak :P
mberends the lo-res lo-color picture of the CGA display was so limited. It could barely resolve an 80x25 text screen with 16 colors. Consequently cyan and magenta represent "primitive color" in the memories of many oldies. 08:19
Woodi perl6: class A { has $!a; method !b() { $!a = 'internal' }; method c() { say self!b() } }; my $c = A.new; $c.c() 08:25
p6eval rakudo 09a598, niecza v9-43-ged11042: OUTPUT«internal␤» 08:26
..pugs: OUTPUT«*** ␤ Unexpected "{"␤ expecting operator, ":", "," or "}"␤ at /tmp/0HNIg2IhFX line 1, column 32␤»
Woodi mberends: why i must say self!method and no such restrictions for fields ?
i do not mean to add restriction for field access :) 08:27
mberends: definetly, 16 colors is enough for everyone :) and i realy like old, green 80x25 displays :) 08:28
mberends Woodi: as I understand it, fields are private but they can have public accessors with the . twigil or not public access with the ! twigil.
sorear Woodi: cga had 4 colors when in bitmapped mode 08:29
Woodi grean alone + bold, lighten and underline is enought :)
08:30 MayDaniel joined
Woodi mberends: i prefer to access methods with [.!]method inside class definition, like fields 08:30
/bold/dimish/ :) 08:31
mberends the only CGA colors in bitmapped mode were black, magenta, cyan and white. Red, green, blue, yellow etc were not even available.
sorear mberends: I thought you could at least change the palette (slightly)
Woodi sorear: it was hardware limitation probably 08:32
mberends sorear: I think VGA (not even EGA) added those hardware capabilities (palette). 08:33
moritz you could turn the display a bit brighter :-)
mberends moritz seems a bit brighter this morning :) 08:34
sorear mberends: my recollection was: CGA = 4 out of <=16; EGA = 16 out of 64; VGA = 256 out of 64^3
mberends: the earlier models had a small fixed pallet and an even smaller variable one 08:35
masak mberends: no, CGA allowed you 4 colors, but there were two palettes. this went even for non-EGA-capable hardware.
the unusual palette (the one you could switch to) was black, red, green, yellow. 08:36
08:36 cognominal joined
sorear Do I want to fix 14/163 failing test files before v10? Most of them probably just need refudging after nom work 08:40
mberends naah 08:41
sorear then I guess it is time to start writing announce.v10
sorear is currently struggling to figure out some kind of failure to initialize variables 08:42
Woodi Google will eat us one day 08:47
08:48 mj41 joined
Woodi maybe goverments should regulate moving user profiles betwin 'internet entities' in some way ? :) and how much it will cost us ? :) 08:48
dalek ecza/serialize: 7d57dc4 | sorear++ | lib/ (2 files):
WIP protopad generation
 08:49
sorear I give up for now
sorear sleep
Woodi gn sorear
mberends gn sorear
Woodi: thinking that way is a good start to prevent it from happening. I consciously use Google very little. 08:50
Woodi for now i prefer google over MS :) 08:53
and there no real alternatives probably 08:54
mberends I prefer whoever is not the apparent monopolist
monopolists become bullies. they can't help it. 08:55
Woodi wish bookmark synchronization plugin with option to store on server of choice not plugin-developer-company...
there was LDAP support in Netscape 4.x :)
Woodi can donate a bit for such LDAP plugin in FF :) 08:58
mberends can donate a byte
Woodi mberends: you understand mozilla internals ? :)
mberends Woodi: good gracious no! don't want to either! I was... joking ;) 08:59
Woodi patches counts :) becouse it should be opensource ofc :) 09:00
09:00 daniel-s left
mberends Woodi: if Niecza or Rakudo can do it, I'd rather write a custom web browser in Perl 6, just for the yak shave. I wonder if Niecza CLR Interop covers networking already? 09:02
Woodi thats remembered me something 09:05
moritz mberends: if I were to write a web browser, I'd base it on uzbl
that allows one to implement browser functionality by communicating to a process with pipes or sockets
Woodi it would be nice to have perl6 modules implementing misc standards, eg HTTP, HTML, CSS... just parsers of standard 09:06
moritz thinks that HTTP headers shouldn't be hard to parse 09:08
mberends moritz: thanks, more to read :) Sounds a bit like using netcat to make a webserver though, only a desperado would do that. Er.
Woodi i think monolitic CGI.pm was one of reasons ppls shitch to PHP so fast 09:09
09:09 daniel-s joined
dalek ecza: 9fccf74 | mberends++ | examples/gtk- (2 files):
[examples] update clock and sierpinski with CLR interop improvements
 09:12
huf Woodi: what do you mean? php *is* a monolithic CGI.pm
09:15 [particle] left
Woodi huf: probably. but development of web pages was from start different 09:15
and installation on web serwer :) 09:16
huf i think that was the real deal with php, the ease of deployment and config, especially for 10-minute hacks
Woodi so if CGI::Protocol module, CGI::HTML things could go another way 09:17
teknology for grandmas often wins :)
on the other hand it is Java replacing COBOL - becouse of business domain 'modules'... 09:18
business quality/warranty 09:20
COBOL *is not* dead...
so Java is new COBOL for 2080 :) 09:21
moritz thinks that the future of webserver interaction is with Plack and similar techniques 09:23
Woodi ok, its all about 'developers' or screensavers-like count and variety :) 09:24
right, midleware for big installations is livesaving :)
so 'a ? b : c' syntax is now 'a ?? b !! c' ? #half hour digresion :) 09:26
masak Woodi: aye. 09:28
Woodi thx :)
masak Woodi: anything we could've done to bring this information to you faster? :)
09:28 woosley left
moritz std: 1 ? 2 : 3 09:29
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Unsupported use of ?: for the conditional operator; in Perl 6 please use ??!! at /tmp/NqDFT2Y7eF line 1:␤------> 1 ?⏏ 2 : 3␤Parse failed␤FAILED 00:01 119m␤»
moritz rakudo: 1 ? 2 : 3
p6eval rakudo 09a598: OUTPUT«===SORRY!===␤Unsupported use of ?: for the conditional operator; in Perl 6 please use ??!! at line 1, near " 2 : 3"␤»
Woodi masak: yes ! you can :)
moritz well, rakudo even tells you :-)
09:29 packetknife joined
Woodi cannot find ??!! nor in S04 nor p5-to-p6 documentation :) 09:30
masak Woodi: S03
Woodi: why were you compelled to look in S04? 09:31
Woodi but i program in vim not irssi yet :)
becouse 'if' defined there
masak lol @ twitter.com/flyosity/status/117262301123461120
moritz Woodi: perlgeek.de/en/article/5-to-6#post_11 09:32
# ternary, conditional op $a == $b ?? 2 * $a !! $b - $a
Woodi ah, it's operator ! :> 09:33
moritz it even says "operator" in the error message :-) 09:34
Woodi ok, ok. but i was using only memories and vim... sory !!! 09:35
and domumentation about blocks :)
masak is it a coincidence that Best Practical has a logotype that is also a Perl 6 (meta)operator? bestpractical.com/images/bpslogo.png 09:37
also, I can't think of a real use for this particular operator.
when would I want to any() together two lists (or deeper structures) pairwise?
09:40 im2ee joined
moritz my $index = ('a'..'z' »|« 'A'..'Z').pairs.grep( *.value eq $char).key 09:41
just a wee bit contrieved :-)
dalek ecza: 16ef0c8 | (Paweł Murias)++ | lib/ (2 files):
cleanup the perl5 interpreter in Perl5Interpreter destructor
 09:43
moritz ronja tries to eat my round tuit
masak anyway, it's a nice symbol.
the »|« I mean, not the round tuit.
moritz aye :-) 09:44
masak moritz: sounds dangerous. a baby could choke on a thing like that. :/
moritz masak: it's too big for that 09:45
im2ee Hi! :) 09:47
moritz hello im2ee
masak im2ee! \o/
im2ee Today is perl6's day :) 09:48
masak im2ee: let's create something great today!
im2ee I going to do a lof of exercises :)
moritz wonders what we should do about rakudo release
im2ee I'm*
masak moritz: I'm simultaneously hopeful and dejected about recent Rakudo progress. 09:49
im2ee masak, i will! :)
masak moritz: we're making nice headway in a lot of areas. stalling in others.
09:49 uasi left
moritz agreed 09:49
masak im2ee: ok, here's a deal: you publish the experiments you do with Perl 6, and I will review and comment on them :)
im2ee: I'm also a bit interested in porting GGE to nom and providing an API for the construct-all-possible-strings-from-regex thing. 09:50
moritz you can even run your experiments here with p6eval if you paste them to gist.github.com 09:51
masak moritz: if pmichaud doesn't show up to un-stuck us soon, we have something like three options: (1) slip, (2) release without pmichaud's bits, or (3) collectively emulate pmichaud and do his stuff. 09:52
(3) is either very very hard or impossible.
mberends votes (1) slip 09:53
masak (and (3) is exactly what Terry Brooks tells you not to attempt)
I'm fine with slipping. 09:54
moritz well, I'm willing to wait until tomorrow at least
masak +1
moritz in the medium term, we should try to increase the bus number on regex stuff though
but it's very hard to do that in the current stage, since pmichaud seems to have rather detailed plans on how to move forward 09:55
masak I'm extremely fine with him having that.
seems these things are best designed by ~1 designer.
moritz sure, my point is that we don't have that plan 09:56
masak and pmichaud has a fine grip on regular expressions.
moritz so it's hard to help
masak moritz: right. that's my point as well.
as far as I know, the intermediate result haven't been published anywhere.
mberends I'm interested in trying to port QRegex to 6model/c, for understanding too. I have no idea where or when to begin.
masak all I know is from private conversation during YAPC::EU. 09:57
09:59 wamba joined 10:07 whiteknight joined 10:26 cognominal left 10:27 MayDaniel left, whiteknight left 10:29 baest joined, whiteknight joined
tadzik good morning #perl6 10:34
mberends hi tadzik
im2ee masak, great! :) Thank You:) 10:35
10:37 packetknife left
masak tadzik! \o/ 10:37
jnthn *yawn*
tadzik oh, masak, I finally got my head around the blackboard vs whiteboard case 10:38
and I came to a conclusion, that what I really need, is a Giant Etch-A-Sketch :)
(after playing with it in my sisters' house)
erm, sister's
arnsholt tadzik: OOC why is an etch-a-sketch the superior choice? 10:45
jnthn masak, moritz: The NFA code that pmichaud was working on during YAPC::EU is at github.com/perl6/nqp/commits/nfa
tadzik arnsholt: because the pens are free, and tidying is easy 10:46
arnsholt Excellent points. Especialy the cleaning =) 10:47
tadzik yep :)
masak tadzik: *lol* 10:48
moritz likes whiteboards 10:49
masak likes blackboards
tadzik likes etchasketches 10:52
10:56 espadrine joined
masak I don't think we'll reach a consensus anytime soon. 10:57
moritz doesn't think consensus is necessary
masak to whom it may concern: gist.github.com/1239203 (A first approach to pack/unpack in Perl 6) 10:59
moritz feels concerned :-)
masak :P
moritz masak: for A I would allow Bufs, and space-pad them by counting bytes 11:02
masak moritz: ok, makes sense.
so the only difference between a and A is the space-padding? 11:03
moritz yes
uuencode encodes bytes iirc 11:04
so rather a Buf
masak oki.
fixed.
moritz didn't you do all that work for GSOC too? :-) 11:06
masak I started in on it. 11:07
I don't recall making this wide an approach. was more a little at a time. 11:08
also, I still have some vague dreams about auto-translating the 17k pack/unpack tests from Perl 5.
but it kind of depends on a solid understanding of the issues brought up in the gist. 11:09
moritz masak: the last three days of Buf recreated about two thirds of your gsoc work on buf, with lots of hindsight and better tools/foundations :-)
s/Buf/Buf work/ 11:10
most notably it wraps ByteBuffer, which didn't exist then (or was just being created, and not as powerful)
masak it was just being created back then. 11:11
a classical case of "HLL needs something, let's build it for them".
11:11 am0c joined, SHODAN joined
masak I'm very happy there's now a second iteration taking advantage of that. thanks moritz for spearheading that effort. 11:11
&
moritz masak: and I'm very happy the first iteration existed 11:12
11:14 cougnut left
mberends masak: some of the pack rules define precisely what the packed data should be, and implementing them will probably work fine. Other rules depend on particular Perl 6 implementations and should probably be dropped from the spec. For example j, J, f, d, F, D, p, P. They would be non-portable by definition. 11:17
jnthn mberends: I'm going to have to do the nqpclr porting of QRegex at some point in the not too distant future, so maybe that will help clarify things for your c one :) 11:18
mberends jnthn: aye :) 11:19
11:22 mj41 left
mberends masak: what a Perl 6 pack spec needs in the place of Perl 5's native types (meaning C types) is a lossless serialization format for each of Perl 6's data types such as Int, Rat and Num. 11:23
moritz not sure 11:24
afaict pack fills the niche of giving one a byte stream in a specific format 11:25
if you want to serialize, use Storable instead
mberends masak: those Perl 6 data serialization formats need to be canonical, so that Rakudo and Niecza can read/write each other's data.
moritz: ok, in that case those Perl 5 style rules should simply be dropped. 11:26
masak I'm all for dropping them. 11:35
that's the conservative thing to do.
error message could be "contact us if you have a sensible idea of what this should do" :P
moritz +2
tadzik :P 11:36
11:36 Psyche^ joined
tadzik oh, do not worry. They *will* contact us... 11:36
11:38 Patterner left, Psyche^ is now known as Patterner 11:50 Chillance joined
moritz nqp: say("\c[0]") 11:51
p6eval nqp: OUTPUT«␀␤»
moritz nqp: say("a" xx 2)
p6eval nqp: OUTPUT«Confused at line 1, near "say(\"a\" xx"␤current instr.: 'nqp;HLL;Grammar;panic' pc 23698 (src/stage2/gen/NQPHLL.pir:6757) (src/stage2/gen/NQPHLL.pm:329)␤»
11:54 mj41 joined
masak siggie DSL. 11:54
no, wait. that's a capture ;)
probably xx NI, then.
nqp: "a" xx 2
p6eval nqp: OUTPUT«Confused at line 1, near "\"a\" xx 2"␤current instr.: 'nqp;HLL;Grammar;panic' pc 23698 (src/stage2/gen/NQPHLL.pir:6757) (src/stage2/gen/NQPHLL.pm:329)␤»
masak yup.
jnthn NQP doesn't have xx, as far as I know.
colomon moritz: we probably should be supporting IEEE floating point format, in both little-endian and big-endian, in addition to "internal" format (which is just going to be IEEE for virtually every platform, I think)
moritz seems like :-)
tadzik I don't see why it should
moritz tadzik: because I just had an application where it would be convenient :-) 11:55
tadzik :0
:)
moritz parrot has the 'repeat' op after all 11:59
mberends colomon: "internal" risks creating a can of worms for future implementers to deal with
tadzik an op for _that_? Odd, imho
I looked at the string pmc
moritz tadzik: very useful for initializations 12:00
tadzik hmm, maybe
moritz: what's your usecase?
moritz tadzik: an initialization :-)
tadzik :)
turns out it's pretty commonly used in Parrot too 12:01
colomon mberends: I would think not -- at least, I'd interpret "internal" as "whatever the heck the system I'm on currently uses", ie explicitly not compatible. That should be the easiest thing for implementers to create. 12:03
12:07 wamba left
mberends colomon: if explicitly not compatible, I would prefer not to offer it at all, since adequate alternatives exist. That way no bad user can complain that Perl 6 caused data loss. 12:07
moritz nqp: say(8/3) 12:08
p6eval nqp: OUTPUT«2.66666666666667␤»
moritz nqp: say(nqp::div_i(8, 3)) 12:09
p6eval nqp: OUTPUT«2␤»
12:09 itz left
moritz nqp: say(nqp::ord(pir::shl__SSI("\c[1]", 3))) 12:10
p6eval nqp: OUTPUT«error:imcc:The opcode 'shl_s_s_i' (shl<3>) was not found. Check the type and number of the arguments␤ in file '(file unknown)' line 47977136␤»
moritz nqp: say(nqp::ord(pir::shl__PPI("\c[1]", 3)))
p6eval nqp: OUTPUT«error:imcc:The opcode 'shl_p_p_i' (shl<3>) was not found. Check the type and number of the arguments␤ in file '(file unknown)' line 45707951␤»
moritz nqp: say(nqp::ord(pir::sh_l__PPI("\c[1]", 3)))
p6eval nqp: OUTPUT«error:imcc:syntax error, unexpected PREG, expecting '(' ('$P102')␤ in file '(file unknown)' line 40452783␤␤»
moritz nqp: say(nqp::ord(pir::shl__PPI("\c[1]", 3)))
p6eval nqp: OUTPUT«error:imcc:The opcode 'shl_p_p_i' (shl<3>) was not found. Check the type and number of the arguments␤ in file '(file unknown)' line 55906991␤»
moritz nqp: say(pir::bors__SSS("a", "c")) 12:11
p6eval nqp: OUTPUT«error:imcc:syntax error, unexpected SREG, expecting '(' ('$S102')␤ in file '(file unknown)' line 40␤␤»
moritz seems that nqp doesn't load str_ops 12:12
erm, bit_ops
nom: say 1 +< 3 12:15
p6eval nom 09a598: OUTPUT«8␤»
moritz nom: say(0b11) 12:16
p6eval nom 09a598: OUTPUT«3␤»
12:20 [Coke] left 12:21 pmurias joined
moritz nom: say(pir::repeat("x", -1)) 12:21
pmurias hi
p6eval nom 09a598: OUTPUT«Cannot repeat with negative arg␤ in <anon> at /tmp/dLwjat2TiA:1␤ in <anon> at /tmp/dLwjat2TiA:1␤␤»
moritz nom: say(pir::repeat("x", 0)) 12:23
p6eval nom 09a598: OUTPUT«Method 'gist' not found for invocant of class 'String'␤ in sub say at src/gen/CORE.setting:4859␤ in <anon> at /tmp/0umtdAuoYj:1␤ in <anon> at /tmp/0umtdAuoYj:1␤␤»
moritz nom: pir::say(pir::repeat("x", 0))
p6eval nom 09a598: OUTPUT«␤»
pmurias how do i rethrow the exception from eval? 12:24
moritz pmurias: eval is supposed not to catch the exception in the first place 12:25
(rakudo gets that wrong)
colomon mberends: not sure why you'd want to drop "internal" floating point format -- as I said, it's the easiest to implement and it has several very good uses, being the quickest format for handling data which isn't supposed to be portable, and if you're on a platform where the floating point format is actually weird, it's needed for easy compatibility with other programs on that platform. 12:26
colomon just discovered on wikipedia that IEEE has now expanded their floating point spec to handle several new types.... 12:27
mberends colomon: I see your point, yet remain unconvinced of its value in practice 12:30
colomon mberends: note that (if moritz++'s page is correct, the "internal" floating point format is the only one Perl 5 currently supports. Adding support for IEEE 754 floating point format was the point of my proposal. 12:33
mberends colomon: +1 to IEEE 754, and too bad for implementations that choose something else for their "internal" floating point. 12:36
12:38 MayDaniel joined
colomon mberends: it's not implementations, it's platforms 12:39
IBM mainframes, for instance.
ingy hi moritz
pmurias moritz: good
colomon cray supercomputers... 12:40
12:41 pmurias left
ingy is there an idiom in p6 regex to say [lcurly|lsquare] ... [rcurly|rsquare] ? 12:41
colomon has actually done significant programming on more than one platform that didn't support IEEE floating point, though admittedly, it's been over a decade now since those days...
12:41 [Coke] joined
ingy such that you only use curly or square? 12:41
mberends colomon: true, but then as moritz++ pointed out, pack should be about bytes in files, and Storable about serialization. 12:42
12:42 woosley joined
ingy also I was considering renaming pegex.receiver to pegex.action. Is there any other PEGs using the terminology "action" besides p6? 12:42
I just got the python parser port working... 12:43
ribasushi: o/ I got your rt tickets. care to fix on github? if so I give you teh commitz... 12:45
moritz wonders if there should be a built-in mirror rule 12:47
so that you can say (<[({]>) ~ <.mirror($0)> <thing>
snarkyboojum ingy: yep piumarta.com/software/peg/peg.1.html 12:53
12:57 uasi joined 13:06 uasi left 13:08 mj41 left 13:09 PacoLinux left, PacoLinux_ joined 13:15 Limbic_Region joined, pmurias joined
ingy moritz: what does that ~ do again? 13:16
snarkyboojum: ta
moritz ingy: that's the parser generator for matching delimited stuff
13:22 packetknife joined
pmichaud good morning, #perl6 13:23
masak pmichaud! \o/
moritz: I'm not sure there needs to be a built-in rule for mirroring. just possible to supply.
moritz good am, pm
jnthn pmichaud! \o/
pmichaud sorry about missing the release yesterday. some of Paula's colleagues on her projects dropped a few balls, which meant that lots of stuff got pushed to me unexpectedly :-/ 13:25
13:25 packetknife left
masak that happens. 13:25
moritz what's the updated plan?
ingy snarkyboojum: that uses the term 'action' as a code block that gets executed on a match. I'm thinking of the name of the 'action containing class', which I call a receiver. A receiver contains actions. Still undecided. 13:26
moritz: I get ~ now. thanks. I like the idea of a mirror token. :)
pmichaud I'm thinking I'll have time this afternoon to work a bit on regexes and release prep, then will do the release tomorrow morning (about 26 hrs from now), unless anyone has objections or other suggestions 13:27
masak perl6: my $a = "This value will never be set" if 0; say $a.perl
p6eval rakudo 09a598, niecza v9-46-g16ef0c8: OUTPUT«Any␤»
..pugs: OUTPUT«\undef␤»
pmichaud (I'm very open to other suggestions :)
masak haha, '\undef'. Pugs, you so funny.
moritz pmichaud: sounds good 13:28
masak pmichaud: +1
jnthn pmichaud: +1
13:28 MayDaniel left
masak by the way, my offer to combine resurrecting Pugs with a weekly online Haskell study group still stands. just need one or two interested people to get going. :) 13:29
ingy I use /~<thing>~/ as short for /<ws>*<thing><ws>*/. I wonder if I should change that sigil. I don't *need* to because I can still use ~ outside //.
ingy hmms 13:30
13:30 packetknife joined
arnsholt I assume that means you use hidden Markov models to classify the tokens in your grammar? =) 13:30
ingy masak: are you resurging Pugs??!!
masak ingy: no, not yet.
but I want to. 13:31
ingy :D
13:31 Limbic_Region left
masak it's the kind of project that needs at least two people, just to establish some kind of mutual moral support. 13:31
ingy yeah. I got so much acmeism stuff on the fire. 13:32
still I support you
masak feels supported o/ 13:33
ingy are there other infix operators besides ** and ~?
13:33 am0c left
moritz now there's the new % 13:33
masak thought ** was a postfix 13:34
ingy yeah?
probably is
masak postfix with extras :P
ingy I think moritz regrets it though ;)
arnsholt And I've sort of classified ~ as a ternary
Only the second half of it is invisible
masak yeah. but infixes can be ternary.
moritz je ne regret rien 13:35
masak arnsholt: ?? !! is an infix.
arnsholt Tru dat
masak ...as I just found out the other day :)
13:35 packetknife left
ingy moritz: what is %? 13:35
masak ingy: it's what ** used to be for separators.
ingy ?! 13:36
pmichaud \w+ % ',' / # if I read correctly
/ \w+ % ',' / # if I read correctly
masak ingy: see github.com/perl6/specs/commit/a455...0fa3628e3d
13:38 MayDaniel joined
ingy seriously?! 13:38
did pegex talk spawn that?
masak yeah!
ingy thanks for clueing me in :\
now I have to change 13:39
masak stop by more often! :P
pmichaud Perl happily steals from many sources :)
ingy bleh
jnthn This is the risk of brining things up on #perl6. It may buy you some re-implementation work. ;-)
ingy moritz: is there any analogs to the ?** stuffs I wanted?
pmichaud yeah, I bet if we stopped discussing Perl 6 we'd all have a lot less to do :)
masak ingy: why you keep addressing moritz? :) 13:40
I'm over here...
ingy masak: moritz has been my guy for this stuff 13:41
masak oh, ok. so it wasn't a mistab.
ingy you wanna take over ingysitting?
masak ingy: anyway, I wouldn't be surprised if most/all of your use cases can be done with %
heh, I don't mind answering questions. I'm sure moritz does a good job, though.
pmichaud afk again -- bbl 13:42
ingy 1) one or more with commas 2) zero or more 3) 2 or more 4) trailing comma allowed
masak: ^
that was basically it 13:43
masak those can be done with a bit of repetition around the % construct.
<stuff>* % <comma> <comma>? 13:44
or, hm, <stuff>+ % <comma> <comma>? 13:45
ingy well to me, a % b and a [b a]* are not the same thing
13:45 packetknife joined
masak no, it's written a+ % b 13:45
% is a quantifier modifier.
pmurias where can i lookup what features niecza implements? 13:46
ingy one produces data of the form [a+] and the other [a, [a+]*]
masak pmurias: perl6.org/compilers/features ?
ingy: oh, that's true.
ingy: but % doesn't have that problem, seemsse. 13:47
seemsme*
pmurias masak: is the matrix automatically update or manual built?
ingy I want all my use cases of % creating [a]
masak pmurias: the latter.
pmurias: well, both ;)
ingy masak: how do I get trailing b? 13:48
masak ingy: <stuff>+ % <comma> <comma>?
ingy and how do I get 2 or more?
masak ingy: (<stuff>)+ % <comma> { $0.elems >= 2 } <comma>? 13:49
ingy: (<stuff>)+ % <comma> { $0.elems >= 2 }
not sure the .elems is needed.
ingy 2 or more is a real use case. like for a | b
masak wait, yes, I think it is. Match numification goes through Match stringification.
ingy um, I thought {} is an action, and actions don't affect parsing 13:50
masak oops
ingy: (<stuff>)+ % <comma> <?{ $0.elems >= 2 }>
ingy anyway I have no inline actions so I'd like a syntax
masak we have inline actions, so we don't need extra syntax ;P 13:51
pmurias masak: it would be cool if the given spot in the table was filled in depending if a set of tests passes ;)
masak pmurias: :)
ingy I will do a+2 % b
<stuff>+2 % <PIPE> 13:52
masak is happily surprised to see how many gray boxes have been colored in on perl6.org/compilers/features
ingy inline actions are so unacmeist :P 13:53
which is why p6 rules dies with p6 13:54
13:54 MayDaniel left
ingy anymo... 13:54
thanks for the trix
masak from our perspective "dies with p6" sounds more like "work really well with p6". 13:55
but yes, it's a bit of a monopolistic practice. 13:56
ingy yes
jnthn masak: japhb++ encouraged me to fill in the missing Rakudo ones the other day, and I managed to do so...after busting the json file ;) 13:57
ingy masak: maybe you have a good answer to this hand grenade...
[Coke] rakudo: rakudo: enum A <a>; say a; eval 'sub a { "O HI" }'; say a 13:58
p6eval rakudo 09a598: OUTPUT«===SORRY!===␤Confused at line 1, near "rakudo: en"␤»
ingy masak: <hand-grenade> will there ever be a day when one does this command: 'sudo apt-get install perl6.0.0'
[Coke] masak: haskell study group? I am... intrigued. 13:59
masak jnthn: yes, I remember. I pointed out the error to you. :P
mberends [Coke]: [Coke]: don't say rakudo: rakudo:
masak [Coke]: ooh!
ingy: don't ask me, I'm just a happy end user! :P
ingy: why would I care about this silly "when will it be released" circus? 14:00
[Coke] mberends: yes, yes. typo.
rakudo: enum A <a>; say a; eval 'sub a { "O HI" }'; say a
p6eval rakudo 09a598: OUTPUT«A::a␤A::a␤»
ingy I was asked that question at the flemish perl mongers, and I said "That doesn't make sense to me. Perl6 is a spec. THere will never be an installed perl6.0.0 binary."
14:01 mj41 joined
[Coke] adds "cmasak" to his skip rule when trying to find updatable perl6 bugs. 14:01
ingy and the people said, "If that is true I will never give another dime to perl 6 dev"
and also, "If that is true, my company will stop using perl" 14:02
pmurias ingy: no inline rules, you mean pegex won't allow me to write actions in my favourite acmeish dilect? ;)
masak ingy: of course the spec could be declared 6.0.0. but I don't see why it'd matter much for end users.
[Coke] masak: wow. I had already excluded some stuff, and was down to about 700 tickets to look at. exclude you, I'm down under 300. ;) 14:03
ingy masak: I didn't think so either. but it is HUGE to rela people
masak ingy: also, I'm not too impressed by some flemish perl mongers who have probably never contributed to Perl 6 making random threats about withholing support.
ingy they all are waiting for something that will NEVER happen
masak ingy: yes.
basically, yes.
to a first approximation.
benabik If the spec is declared 6.0.0, then any compiler that follows it could provide perl6.0.0. Then trying to apt-get it will attempt to select one.
masak ingy: consider it a honeypot for trolls and similar-minded people.
ingy and when that becomes know, they are gonna be PISSED
[Coke] rakudo: my @a=1..5; @a >>+>> 1; @a >>->> 2
p6eval rakudo 09a598: ( no output )
masak [Coke]: that sounds about right to me. :) 14:04
ingy people are giving you money under false assumptions
[Coke] rakudo: my @a=1..5; @a >>+>> 1; @a >>->> 2; say @a;
p6eval rakudo 09a598: OUTPUT«1 2 3 4 5␤»
mberends masak: I'm moderately interested in joining a pugs and haskell group, if it could be scheduled semi-formally during example transatlantic hours on Sundays.
ingy this needs to be put out in the open clearly, and soon
masak ingy: who is giving us money? and why do they have these assumptions?
[Coke] wait, there will never be a perl6 release? what? 14:05
I think we're talking past each other here.
masak mberends, [Coke]: ok, cool. let's schedule something.
[Coke]: I didn't say that, no.
pmurias why would a rakudo 6.0.0 or niecza 6.0.0 be any different from a perl 6.0.0 release
?
masak pmurias: it could coincide, but definitely wouldn't have to. 14:06
[Coke] masak: then I would be a lot more careful about answering that sort of question.
ingy people just don't understand it at all. I was surprised but they don't
and they were really upset
Liz and Wendy gave $5k
[Coke] phenny: ask jnthn if we can close RT#76818 with tests. 14:07
phenny [Coke]: I'll pass that on when jnthn is around.
masak ingy: maybe we need a document outlining all this.
ingy and if you told them that perl6 isn't coming, they would ask it back
masak: YES!
masak ingy: I really don't think people would be upset if they knew how things actually are.
ingy masak: me neither
masak and not all of us here are as disillusioned about a "Perl 6 release" as I am. :)
I just don't consider it that important, is all. 14:08
ingy another guy told me his work is waiting for perl 6 to ship on debian
masak now, a stable spec, stable/fast/complete/robust implementations, *that's* important.
ingy: I thought there were already debian packages for Rakudo.
ingy and there is no way to sell the powers that be the rakudo story 14:09
they really are looking for 'perl6.deb'
benabik There's no C98.deb
masak I too have grown accustomed to the fact that outsiders will never really appreciate (or care about) the Perl6-Rakudo distinction. or even the Perl6-Parrot distinction. :/
ingy benabik: right, but nobody outside this irc channel gets that perl6 is a spec 14:10
[Coke] moritz: can you respond to my question from July on 63650 ?
jnthn I'm not sure they should have to.
phenny jnthn: 14:07Z <[Coke]> ask jnthn if we can close RT#76818 with tests.
[Coke] ingy: well, that's crazy.
"we want perl6" "here it is" "no it isn't" "..."
masak ingy: it seems to me most of these arguments would be moot if we got to the point where we could just point to our implementations and say: here you are. Perl 6 can do this. it works great. enjoy.
[Coke]: I have that conversation all the time. :/ 14:11
jnthn nom: my @a=1..5; @a >>+>> 1; @a >>->> 2
ingy no it's not, because you guys don't have any clear statement that that's the case
p6eval nom 09a598: ( no output )
jnthn nom: my @a=1..5; @a >>+>> 1; @a >>->> 2; say @a
p6eval nom 09a598: OUTPUT«1 2 3 4 5␤»
masak [Coke]: "when is Perl 6 going to be released?" "we release every month." "no, here's what I mean by release..." :/
jnthn nom: my @a=1..5; @a >>+>> 1; @a >>->> 2; say @
p6eval nom 09a598: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name at line 1, near "@"␤»
[Coke] I certainly think a doc explaining that should exist on perl6.org... but I'd be surprised if it didn't already.
jnthn nom: my @a=1..5; say @a >>+>> 1; say @a >>->> 2;
p6eval nom 09a598: OUTPUT«2 3 4 5 6␤-1 0 1 2 3␤»
jnthn [Coke]: Closable with tests, looks like.
ingy it should be the fucking HEADLINE
[Coke] jnthn: danke. 14:12
masak hugs ingy
ingy I was going to write a blog post "perl6 will never ship" and explain the whole thing in detail 14:13
jnthn ingy: We don't need more drama. We need to quietly get on with making awesome Perl 6 implementations that people can get their work done with. Because I do plenty of conferences and that's what excites people: seeing stuff working.
masak ingy: please consider carefully what people who don't read the blog post will take away from that title. :/
also, full ACK to what jnthn++ just said.
ingy jnthn: if you don't just explain it now and deal with the drama, you are shooting yourself in the foo\whead 14:14
masak releasing Rakudo Star was a lot of fun. we pulled all the stops. people still just wanted more and faster. so that's what we're working on now. :)
ingy I'm just reporting what real people told me when I tried to sell them your story 14:15
[Coke] I would prefer a title of "perl6 is already shipping", but I see what ingy's shooting for.
masak I also see what ingy is shooting for.
ingy I was advocating p6, and this is what I got
masak I've wanted to write that blog post myself sometimes.
ingy masak: please do 14:16
choose a better title
masak I just might.
ingy but please start the discussion
[Coke] rakudo: my $foo; say $foo.WHAT
p6eval rakudo 09a598: OUTPUT«Any()␤»
pmurias ingy: *a* perl 6.0.0 will ship, *the* perl 6.0.0 won't
[Coke] rakudo: my $foo; say $foo.WHAT.Str;
p6eval rakudo 09a598: OUTPUT«Use of uninitialized value in string context␤␤»
[Coke] jnthn: that's another one of yours: RT#76482 14:17
"Haven't you heard? They come in six packs."
masak why do people love version numbers so much? I don't give a Rat's denominator about 6.0.0! I want an implementation that I can use in my daily life and work!
ingy I personally don't give a rats about all this. but I saw clearly that you guys were heading for a shitstorm
masak call it 0.93, I don't care!
ingy masak: if you guys don't want any money to get p6 done, then do whatever you want 14:18
masak *sigh*
[Coke] masak: I'm trying to figure out why you're upset about this feedback. 14:19
masak maybe we need to clearly distinguish PR and actual internal work here.
[Coke] where on perl6.org is this documented?
I think it's pretty clear ingy is talking about PR here.
masak [Coke]: I'm not, really. I'm happy that ingy is stepping forward reporting a real-life PR experience.
jnthn [Coke]: Right on the top of www.perl6.org/compilers/
masak [Coke]: I'm not sure it's grounds for panicking.
[Coke] jnthn: I think a little more text there is a good idea. 14:20
masak specifically, the message "if sponsors only knew that a 6.0.0 is not forthcoming, they'd ask their money back" sounds extremely alarmist.
I'm in this not to get a 6.0.0 release of anything, just to bring Perl 6 to fruition. 14:21
ingy it's what I heard
masak yes. and that's interesting.
ingy </hand-grenade> 14:22
Woodi I personaly think that Perl 6 is implemented and not optimizet at the moment. and that more exotic features are still in scientific research. maybe there should be feature matrix partitionet like: 1. 'ALL features from C/C++' (implemented) :) 2. features from Haskel (show what work); 3) features with spec under development; 4) pure research :)
to late one line :) 14:23
ingy Woodi: you are right. but the thing is that not only does the world not get it, the majority of the perl community does not get it 14:25
[Coke] ingy: welcome to 10 years ago. ;)
ingy it's a pretty easy thing to set straight, in no uncertain terms.
[Coke] ingy: ORLY? how? 14:26
masak ingy: I think that I've somehow gotten used to that no matter what we do here in the echo chamber, people will still be tapping their foot, arms akimbo outside of it. what gives me hope is that there are plenty of people who *want* Perl 6 to succeed. but there's no easy way to succeed by outsiders' criteria, except to keep implementing a huge amount of stuff well.
[Coke] masak++; but if there's something we can add to the website to help out, by all means. 14:27
masak agreed.
Woodi ingy: i was thinking same thgings sometimes and then Perl5 development restarted :) so not so much presure on v6
and i am on debian now with two sets of Perl6 packages :) 14:29
one was apt-get'ed :)
ingy [Coke]: On the top of your main web page say this: "The term 'Perl 6' refers to a language specification. There are many implmentations of this spec, but none of them will ever be called 'Perl 6'. The current implmentations are: ..."
[Coke] Note to bug reporters: when opening tickets and saying "this is weird", please include your expected output. Your bugadmins thank you.
ingy: but they are called Perl 6.
do you mean "they will not be called "perl6" " ? 14:30
ingy just do that, and go back to work
yes, I mean perl6
[Coke] ok. rakudo's executable is called perl6.
so this is not an easy thing to sell.
Woodi ingy: on contrary you are right :) ppls want something declared beta at least 14:31
ingy any implementation can install /usr/bin/perl6
but nobody can ship perl6.deb
Woodi -rwxr-xr-x 1 root root 15252656 Aug 7 10:07 /usr/bin/perl6 14:32
[Coke] how about "... will ever be *the* Perl 6. The current imp...."
jnthn [Coke]: #76482 loosk like correct by current spec.
14:33 packetknife left
ingy there will never be a Perl 6 implementation know simply as "Perl 6". All implementations can use the generic term a "Perl 6" implementation, but the specific name will not be "Perl 6" 14:34
golf that into a concise statement 14:35
TimToady should probably craft it
then put it out front and center
and let other people deal with drama while you hack :) 14:36
masak it should be noted that Perl 6 has been a language implementation since day 1. 14:37
14:37 masonkramer joined
masak this really shouldn't be a surprise to people. 14:37
14:37 replore_ joined
masak that, in essence is why I don't think there's really an oncoming "shitstorm". 14:38
ingy masak: keep your head in the sand
masak *sigh*
14:38 masonkramer left
masak I'll allow for the fact that we haven't been utterly clear PR-wise. we suck at PR. 14:39
but really, this is the way it's been, always. Perl 6 is a spec.
release of Perl-the-spec 6.0.0 will have close to zero significance for people who want to use Perl 6.
other things matter a lot more.
ingy masak: the world was always round
masak ingy: I agree with that but fail to close the analogy circuit. :/ 14:40
ingy masak: you are wrong about the zero significance
masak aye, maybe I am.
maybe I've been here too long.
I should say, it really won't matter one ounce *to me*. 14:41
ingy there are real people trying to keep their companies on board with perl, and waiting for a perl6 release etc.
at least just tell them the truth loudly
then you are off the hook 14:42
benabik Point them at 5.14… Lots of good improvements happening. Perl is still moving.
masak so... the companies are waiting for a Perl 6 release... but they're unaware of the distinction between spec and implementations... and they won't consider Rakudo to be "Perl 6"...? :/
ingy masak: correct
masak ingy: I'm not sure what to do about that.
14:43 espadrine left
masak ingy: clearly, it sucks, but it's also the way it is. 14:43
flussence IMO we could learn a few things from the WHATWG - just let some slow-moving W3C-like group have "6.0.*" so the people who prefer getting stuff done can carry on doing so :P
ingy masak: when I am at a computer conference i meet a lot of people who never heard of yaml
which seems unrela to me
unreal
but it is real
masak: I told you a way to start solving it immediately 14:44
benabik masak: For perl people, that doesn't surprise me too much. There was no difference between p5 spec and implementation. I would think that C/C++ people would get it, but Java people probably not.
ingy people are great about saying what $perl6 will be, but suck at shouting what it won't be
masak frankly, I feel neither capable of nor very interested in addressing this issue. 14:45
at least not right now.
helping with nom feels more important.
I'm not saying ingy is wrong in his analysis... just that there will always be people who don't see all the distinctions and shades of gray we do inside the Perl 6 community. 14:46
flussence speaking of nom, that Buf stuff helped a lot: github.com/flussence/specgraphs/ra...-tests.png
masak sure, put something up on the website to be even more clear. that's perfectly alright. 14:47
but let's not overreact.
ingy I just asked a geek who sat down next to me at the OSDC conference if he knew yaml
nope
for real just now
I asked him if he knows perl 6
masak "knows" or "knows about"? 14:48
14:48 mj41 left
ingy he says, joking, yeah sure coming out any day now... 14:48
*laughs*
rakudo?
blank
flussence I bet he knows PHP and/or Java though...
benabik yaml was a bit of a flash in the pan… json has become far more common
masak ingy: you're not asking us to change the "any day now" label? that's been with us since, like, 2001...
ingy oh so yaml is dead?
really
yaml is more popular than ever 14:49
masak ingy: the "any day now" meme will *survive* any kind of release we ever make by perhaps a decade.
ingy anyway, you guys are soaking in p6
but nobody else has a clue
masak right.
ingy even your friends at yapc
masak ingy: if you believe that we are withholding information and PR on purpose... 14:50
reaching out is *hard*.
ingy again, just real feedback from real human
masak hey, he knew about Perl 6!
that's about what I'd expect from someone who isn't involved. 14:51
ingy he knows it as the next version of perl
masak Perl 6 has a real brand going for it. and people want us to change name. :P
ingy I don't
masak me neither.
14:52 uasi joined
ingy I want you to tell everyone precisely what to expect 14:52
then they will say
masak the most constructive thing I've come up with to counteract the "Perl 6 will never be released" meme is to have people start saying "haha, yes. but seriously, I used it yesterday, and it rocks." for that, we need to implement it as much as possible. preferably soon.
ingy oh! that makes sense. ok
done 14:53
masak ingy: people come in here a lot, asking a lot of questions. we tell them what to expect. they don't listen and then they go away. and they're the *interested* ones.
ingy I'm not talking about them
masak good luck informing the mildly interested or disinterested ones.
ingy I'm talking about the world that trusts the perl community to DTRT and is patiently waiting.... 14:54
for something that will NEVER come
by definition
masak I think you're exaggerating.
ingy I'm not
I run into this every day
I travel all over the world pimping perl 14:55
and perl 6 brings up anger
jnthn *sigh*
masak sorry to hear that.
ingy and it is almost always off track
jnthn afk, for quite a while
masak ingy: I hear a lot of misconceptions about Perl 6 too. I travel places too, although perhaps not as much as you. 14:56
ingy they are wrong, but there's no reason they shouldn't be
masak ingy: to the best of my knowledge, I've managed to answer similar questions to people's satisfaction.
maybe you just suck at answering those questions :P
ingy masak: yes
masak ok, so it's a very local shitstorm we're talking about... 14:57
ingy masak: why are you not trying to solve this?
masak because I'm, you know, doing real stuff.
ingy it's great that you can calm a person down and set them stright
straight
just do it once in writing
then you can just point them a a url 14:58
masak I hear you.
will think about it a bit.
ingy and get MORE hacking done
masak that doesn't follow, but sure :P
ingy so is % an infix? 14:59
in rules
masak I think so. primarily, it's a quantifier modifier. 15:00
which could be seen as an infix with some restrictions.
ingy it's not a cast modulus?
I thought someone (moritz?) said that
masak in STD, it's a token. github.com/perl6/std/commit/c22b2e...995bba50d3
15:01 cognominal joined
sorear good * #perl6 15:01
colomon o/ 15:02
sorear: got asinh working last night, but seem to have broken it since.
ingy The C<%> because it can be pronounced
'modulo' and because it looks like the relationship of two things.
masak I don't know what you mean by 'cast modulus'. 15:03
sorear: *
sorear Woodi: there are two ls in Haskell. 15:04
mberends: ping 15:05
ingy masak: 5 % 3 15:07
masak ingy: yes. that is en.wikipedia.org/wiki/Modulo_operation -- I'm familiar with it. 15:08
ingy % is modulus
masak OK.
Woodi sorear: i was aware i can mispell that name. i was active betting to use 'a' not 'e' after 'h' :) will try memoize :)
ingy and you know what a cast is
masak a type cast?
sure, but what does that have to do with parsing?
sorear masak: do you think niecza deserves a +- or a + for printf? 15:09
15:09 [particle] joined
masak sorear: haven't used it enough. #58 doesn't seem like grounds for giving it a +- 15:09
sorear: but other things might, I dunno.
mberends sorear: pong
ingy I'm asking if % is cast to a new behaviour when given rule args
masak sorear: how much of the spectest suite does it pass?
sorear mberends: what I just asked masak.
I'll give it a +- because there's no .fmt 15:10
masak ingy: oh! I just consider it two completely unrelated languages.
ingy: the fact that it's the same character for the two operators is more of a nice symmetry/rhyme.
15:10 mj41 joined
ingy well rules are just methods in grammars 15:11
dalek atures: f42e110 | sorear++ | features.json:
Update niecza features
masak ingy: sure, but it's still two different grammars parsing the two kinds of %
ingy which is what I was asking :) 15:12
masak and the answer is: they're not at all the same operator. they just look alike, and they can both be called "modulo". 15:13
ingy thx
sorear like the relationship between infix:<~> and prefix:<~> ?
mberends sorear: I agree +- because there quite a few formatting options are NYI. The coverage by tests in roast is insufficient to consider sprintf or printf to be complete. 15:16
yeah, grammar fail :/
masak sorear: almost; but those are two different operators of different category in the same sublanguage. 15:17
sorear: the two %s are two different operators of the same category (provided it is an infix in P6Regex) in different sublanguages. 15:18
15:18 cognominal left
Woodi how to say return undefined in perl6 ? 15:20
15:20 cognominal joined
masak Woodi: 'return' 15:23
you could say 'return Nil', but that's just a longer way of saying it.
also, consider whether it's 'fail' you want.
Woodi hmm, compile error message is quite descriptive 15:24
perl5 is more scriptable :) 15:25
masak how do you mean?
Woodi probably one syntax just work there
masak I still don't understand what you mean. 15:26
Perl has never been about "one and only one way".
Woodi maybe here will to with some automatic conversions
masak and if the latter two ways bother you, just forget I mentioned them :P
Woodi masak: there was no types in perl so far :)
benabik Didn't even get into typed undefineds.
masak fwiw, 'return' is how I would do it in Perl 5, too. 15:27
not 'return undef'.
mberends heh, in Gtk a group of tabbed windows is called a notebook, so in Perl 6 you create one with 'sub notebook' :)
colomon sorear: okay, I've got all the hyperbolic trig functions working too, except the values at Inf and -Inf are wrong for a lot of them. 15:29
masak heh -- twitter.com/audreyt/status/117621081526894592 15:36
by this I have to assume that au++ is also aware of our loose Haskell study group + Pugs revival plans... 15:37
[Coke], mberends: so, is tomorrow fine with you guys? :)
what UTC?
mberends masak: yes, tomorrow. vaguely from 8:00 to 21:00 UTC, not continuously 15:39
masak nod. agreed. 15:41
let's start just trying to get Pugs to run from source. 15:42
mberends +1
masak cloning the repository and making lists of LHF can come later :)
mberends we'll need commitbits 15:43
or each hack on an own clone I spose 15:44
sorear haskell studygroup? 15:45
mberends sorear: yes, will you join? this will be a beginner group. 15:47
masak mberends: either of us forks au's repo on github, and gives the others commitbits.
sorear is not a Haskell beginner ;)
15:47 cognominal left
mberends masak: yes, it was more of a warning ;) 15:48
masak he's still free to join the group, of course, and definitely welcome to hack on Pugs with us...
but something tells me he's having enough fun with Niecza.
wow! just imagine, putting Pugs on 6model!
masak boggles
mberends masak: could we make a perl6/pugs repo, and so give commitbits via hugme instead? 15:52
masak fine with me. 15:53
moritz should I do that?
masak let's wait until tomorrow ;) 15:54
moritz has enough permission to fork into the perl6 organization
masak it's all contingent on whether we can get Pugs to compile and run.
sorear masak: pugs on 6model? 15:55
mberends: I just want to make it known, that I am willing to help
masak sorear++ # cool! 15:56
mberends sorear: that's great :)
masak discovers defunkt.io/hub/
sorear: it was just a thought. apparently it's easy to let one's ideas get ahead of oneself when planning to wake up Pugs from her beauty sleep ;)
16:13 ggoebel left
benabik can provide some assistance with Haskell. 16:14
I don't know the pugs code, but I'm reasonably proficient with lambdastuff
TimToady 'course, by the time you replace the parser and the MOP, you mostly just have the runtime left
16:16 daniel-s left 16:17 whiteknight left, daniel-s joined 16:18 envi left 16:19 ggoebel joined 16:21 envi joined
dalek ecza/serialize: 0b7a754 | sorear++ | lib/ (2 files):
Fix ordering bug in protopads
 16:25
diakopter niecza: "
p6eval niecza v9-46-g16ef0c8: OUTPUT«Unhandled exception: System.IndexOutOfRangeException: Array index is out of range.␤ at Lexer.Run (System.String from, Int32 pos) [0x00000] in <filename unknown>:0 ␤ at Lexer.RunDispatch (Niecza.Frame fromf, Niecza.P6any cursor) [0x00000] in <filename unkn…
16:29 gbacon joined
diakopter std: :16{A705E} # should parse as a Blob literal 16:30
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Malformed radix number at /tmp/oTG4WKBGmr line 1:␤------> :16⏏{A705E} # should parse as a Blob literal␤Parse failed␤FAILED 00:01 117m␤»
moritz someone please remind me what a Blob is 16:33
diakopter perlcabal.org/syn/S02.html#Blob_literals
I don't know any more than that ;)
pugs: say :16{A705E} 16:34
p6eval pugs: OUTPUT«526375770699927␤»
diakopter pugs: say :16{A705E}.WHAT 16:36
p6eval pugs: OUTPUT«Int␤»
moritz ah, a blob is an immutable buf
flussence Surprise! github.com/flussence/specgraphs/ra...output.png
(this one is *much* faster to fill in than the nom one :) 16:39
TimToady well, nothing makes progress if you look at it close enough :) 16:40
TimToady is going to chew up his entire Saturday just backlogging #perl6... 16:41
std: my $a = "xxx" ; { say $a ; my $a; } 16:43
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Lexical symbol '$a' is already bound to an outer symbol (see line 1);␤ the implicit outer binding at line 1 must be rewritten as OUTER::<$a>␤ before you can unambiguously declare a new '$a' in this scope at /tmp/IpnGmWphdn line 1:␤---…
TimToady niecza: my $a = "xxx" ; { say $a ; my $a; }
p6eval niecza v9-46-g16ef0c8: OUTPUT«===SORRY!===␤␤Lexical symbol '$a' is already bound to an outer symbol (see line 1);␤ the implicit outer binding at line 1 must be rewritten as $a␤ before you can unambiguously declare a new '$a' in this scope at /tmp/wBJZi682Mg line 1:␤-…
TimToady here's a bug . . . . . . . . . . . . . . . . . . . . . . . . . . ^^ 16:44
sorear: ^^
niecza: my $a = "xxx" ; {{{{ say $a ; my $a; }}}} 16:45
p6eval niecza v9-46-g16ef0c8: OUTPUT«===SORRY!===␤␤Lexical symbol '$a' is already bound to an outer symbol (see line 1);␤ the implicit outer binding at line 1 must be rewritten as $a␤ before you can unambiguously declare a new '$a' in this scope at /tmp/l_Hlvfiary line 1:␤-…
diakopter niecza startup time is so much faster on linux than on windows mono; very noticeable on p6eval vs any other machine I've tried; I dunno why
TimToady std: my $a = "xxx" ; {{{{ say $a ; my $a; }}}}
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Lexical symbol '$a' is already bound to an outer symbol (see line 1);␤ the implicit outer binding at line 1 must be rewritten as OUTER::OUTER::OUTER::OUTER::<$a>␤ before you can unambiguously declare a new '$a' in this scope at /tmp/Z…
diakopter benabik: there are plenty of jvm implementations, re: your point about C/C++ vs java 16:46
TimToady std: my $a = "xxx" ; {{{{ say OUTER::<$a> ; my $a; }}}} 16:47
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Variable OUTER::<$a> is not predeclared at /tmp/O7RBAWI_R9 line 1:␤------> my $a = "xxx" ; {{{{ say OUTER::<$a>⏏ ; my $a; }}}}␤Potential difficulties:␤ $a is declared but not used at /tmp/O7RBAWI_R9 line 1:␤------> …
TimToady niecza: my $a = "xxx" ; {{{{ say OUTER::<$a> ; my $a; }}}}
p6eval niecza v9-46-g16ef0c8: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/Q94wuMwysW line 1:␤------> $a = "xxx" ; {{{{ say OUTER::<$a> ; my ⏏$a; }}}}␤␤xxx␤»
TimToady arguably niecza is correcter, and one OUTER will be sufficient 16:48
16:48 woosley left
TimToady except for the fact that the error message doesn't mention OUTER 16:48
16:48 im2ee_ joined 16:49 im2ee left 16:51 packetknife joined
sorear cute bug 16:51
TimToady but I don't think you need to reproduce OUTER::OUTER::OUTER::OUTER
since one OUTER says where to start looking
sorear at one point niecza was generating OUTER chains like that 16:52
TimToady well, it does have the advantage of being explicit, in case someone inserts another my $a somewhere in between 16:53
diakopter niecza: my $a = "xxx" ; { my $a = "yyy"; {{{ say OUTER::OUTER::<$a> ; }}}} 16:54
p6eval niecza v9-46-g16ef0c8: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/Y0C_hskFvT line 1:␤------> my $a = "xxx" ; { my ⏏$a = "yyy"; {{{ say OUTER::OUTER::<$a> ;␤␤yyy␤»
diakopter why did it say yyy
TimToady because
diakopter when OUTER::<$a> also gives yyy
TimToady yyy is correct
try more OUTERs 16:55
or fewer braces 16:56
masak or just name the variables differently :P
sheesh
diakopter niecza: my $a = "xxx" ; $a; { my $a = "yyy"; {{{ say OUTER::<$a>; say OUTER::OUTER::<$a>; say OUTER::OUTER::OUTER::<$a>; say OUTER::OUTER::OUTER::OUTER::<$a>; $a }}}} 16:57
p6eval niecza v9-46-g16ef0c8: OUTPUT«yyy␤yyy␤yyy␤xxx␤»
Woodi if role implements method then it should be available in class doing that role ? 16:58
16:58 [Coke] left
TimToady Woodi: yes, assuming it doesn't collide with something 16:59
Woodi so i do not understand why IO::Socket::INET.read is absent... 17:00
masak Woodi: if the program compiles and the class doesn't override, then yes :)
17:00 [Coke] joined
benabik diakopter: Yes, but "Java" means "Oracle Java" to 90% of people. Really changing due to dalvik, I suppose. 17:01
TimToady nom: say IO::Socket::INET.^methods
p6eval nom 09a598: OUTPUT«Could not find symbol 'IO::Socket::&INET'␤ in <anon> at /tmp/FLqCmwIDKH:1␤ in <anon> at /tmp/FLqCmwIDKH:1␤␤»
Woodi roles are not abstract ?
TimToady they're immutable, but of varying abstraction
they may suggest an implementation 17:02
masak TODO passed tests 2-3 in t/spec/S19-command-line/dash-e.rakudo
Woodi: roles may definitely implement methods.
even if that implementation is just 'method foo { ... }' 17:03
(three dots meant literally)
Woodi ok
17:04 im2ee_ left 17:05 im2ee joined
Woodi nom shows 'read' on ^methods but explodes with no STORE in ANY() 17:05
diakopter benabik: what % of people would you say "C/C++" means each of these three: msvc, gcc, clang
TimToady well, it looks to me like IO::Socket::INET isn't there at all
moritz it's disabled in evalbot
for security reasons 17:06
TimToady it's not there locally either
oh wait, it is
just a 'use' doesn't work :)
sorear diakopter: what's a good jvm that will run comfortably in 384mb
benabik diakopter: Yes, it's fairly well known that the compiler isn't the definition of the language. But most people would say Oracle Java is the definition of Java. It's far more grey, but they're far more intertwined. 17:07
moritz TimToady: how up-to-date is your copy of nom?
17:07 REPLeffect left
diakopter sorear: I dunno 17:07
TimToady couple days
diakopter benabik: that wasn't my point; I was just asking what you imagined the distribution was
moritz I'm pretty sure it exists :-)
TimToady it finds the methods, it just fails a 'use IO::Socket::INET' if I do that first
which bamboozled me 17:08
moritz well, it's in core
no need to 'use' it
TimToady seems like we should allow a 'use' on something that's already there...
moritz use Int; #?
Woodi seems use 'use' on core modules is user anti-case :) 17:09
TimToady what if we decide to kick Int out of the core someday? :P
17:09 simcop2387 left
benabik diakopter: A majority of people I deal with know that gcc is _a_ C compiler, not _the_ C compiler. From what I'm told, msvc people may be fuzzier on the distinction, but eh. 17:09
diakopter heh 17:10
TimToady I'd think they'd be clearer on that distinction :)
sorear heehe
benabik: do you know of any alternate vms that would work for me?
TimToady VMS?!?!
TimToady goes back into ptsd mode...
17:11 simcop2387 joined
diakopter sorear: I was reading about ildjit 17:11
17:11 replore_ left
benabik sorear: JVMs? 17:11
sorear has been thinking about Niecza/JVM a little bit lately
benabik: ye
masak perl6: =END␤say "OH HAI" 17:12
p6eval pugs, rakudo 09a598: ( no output )
..niecza v9-46-g16ef0c8: OUTPUT«OH HAI␤»
17:12 Sarten-X left
masak perl6: say "alive";␤=END␤say "OH HAI" 17:12
p6eval niecza v9-46-g16ef0c8: OUTPUT«alive␤OH HAI␤»
..pugs, rakudo 09a598: OUTPUT«alive␤»
masak sorear: nieczabug.
benabik sorear: I've tended to just use Sun/Oracle's, honestly.
im2ee rakudo: multi combs(@, 0) { "" }; multi combs { combs(@^dict, $^n - 1) X~ @dict }; (.say for combs(<a b c>, $_)) for 1..4; 17:15
p6eval rakudo 09a598: OUTPUT«a␤b␤c␤aa␤ab␤ac␤ba␤bb␤bc␤ca␤cb␤cc␤aaa␤aab␤aac␤aba␤abb␤abc␤aca␤acb␤acc␤baa␤bab␤bac␤bba␤bbb␤bbc␤bca␤bcb␤bcc␤caa␤cab␤cac␤cba␤cbb␤cbc␤cca␤ccb␤ccc␤aaaa␤aaab␤aaac␤aaba␤aabb␤aabc␤aaca␤aacb␤aacc␤abaa␤abab␤abac␤abba␤abbb␤abbc␤abca␤abcb␤abcc␤acaa␤acab␤acac␤acba␤acbb␤acbc␤acca…
im2ee masak, works here.
masak b: multi combs(@, 0) { "" }; multi combs { combs(@^dict, $^n - 1) X~ @dict }; (.say for combs(<a b c>, $_)) for 1..4; 17:16
p6eval b 1b7dd1: OUTPUT«No applicable candidates found to dispatch to for 'combs'. Available candidates are:␤:(Positional , Int where (0))␤:(;; Any $n, @dict)␤␤ in <anon> at line 22:/tmp/cdcc6w5zVv␤ in main program body at line 1:src/metamodel/RoleToInstanceApplier.nqp␤»
masak that's the error.
im2ee b is the latest version?
of rakudo?
masak latest version of the last development branch.
problem is, the current one hasn't had a release yet.
when you go 'b:' you get the last one, the one you have locally. 17:17
when you go 'rakudo:' or 'nom:', you get the new one.
the one that isn't all done yet. :)
but it sure seems to be handling this program better...
moritz and be was "all done"? :-)
masak moritz: sssh, I'm doing PR here :P 17:18
17:18 frhodes joined
TimToady if we ever do kick a core module out someday (and we probably will), it seems to me we'll end up with a class of auto-used names; and if we have that, we should at least think about making them all autousable, rather than avoiding Occam. 17:18
im2ee masak, so.. could You help me to write this program to the latest version?
masak im2ee: I thought that's what I did.
im2ee: it works on nom but not on Rakudo Star. 17:19
oh, or you meant the latest release?
ok, sure.
im2ee Yes, i mean the Rakudo Start :)
17:19 daniel-s left
masak "star", not "start". 17:19
im2ee Star, yes - sorry, im bit in hurry :) 17:20
17:20 REPLeffect joined
TimToady
.oO(Warning: had to assume a 'use IO::Socket::INET' before foo line 42)
 17:20
Woodi btw, TimToady s/core/standard/ modules|distribution :) Linux-way :)
im2ee masak, do You have a time to do this? :) It's nothing important
masak b: multi combs(@d, 0) { say "base case" }; multi combs(@d, Int $n) { say "recursive case" }; combs([], 0); combs([], 5)
p6eval b 1b7dd1: OUTPUT«base case␤recursive case␤»
masak im2ee: I think I know what's wrong. hold on.
TimToady
.oO(Warning: had to 'cpanm IO::Socket::INET' before foo line 42 :)
 17:21
im2ee masak, ok. Great :)
masak b: multi combs(@, 0) { "" }; multi combs(@dict, Int $n) { combs(@dict, $n - 1) X~ @dict }; (.say for combs(<a b c>, $_)) for 1..4;
p6eval b 1b7dd1: OUTPUT«a␤b␤c␤aa␤ab␤ac␤ba␤bb␤bc␤ca␤cb␤cc␤aaa␤aab␤aac␤aba␤abb␤abc␤aca␤acb␤acc␤baa␤bab␤bac␤bba␤bbb␤bbc␤bca␤bcb␤bcc␤caa␤cab␤cac␤cba␤cbb␤cbc␤cca␤ccb␤ccc␤aaaa␤aaab␤aaac␤aaba␤aabb␤aabc␤aaca␤aacb␤aacc␤abaa␤abab␤abac␤abba␤abbb␤abbc␤abca␤abcb␤abcc␤acaa␤acab␤acac␤acba␤acbb␤acbc␤acca␤accb…
masak im2ee: there you go ;)
nom: multi combs(@, 0) { "" }; multi combs(@dict, Int $n) { combs(@dict, $n - 1) X~ @dict }; (.say for combs(<a b c>, $_)) for 1..4;
p6eval nom 09a598: OUTPUT«a␤b␤c␤aa␤ab␤ac␤ba␤bb␤bc␤ca␤cb␤cc␤aaa␤aab␤aac␤aba␤abb␤abc␤aca␤acb␤acc␤baa␤bab␤bac␤bba␤bbb␤bbc␤bca␤bcb␤bcc␤caa␤cab␤cac␤cba␤cbb␤cbc␤cca␤ccb␤ccc␤aaaa␤aaab␤aaac␤aaba␤aabb␤aabc␤aaca␤aacb␤aacc␤abaa␤abab␤abac␤abba␤abbb␤abbc␤abca␤abcb␤abcc␤acaa␤acab␤acac␤acba␤acbb␤acbc␤acca␤ac…
diakopter std: cpanm IO::Socket::INET
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Undeclared name:␤ 'IO::Socket::INET' used at line 1␤Undeclared routine:␤ 'cpanm' used at line 1␤Check failed␤FAILED 00:01 118m␤»
masak im2ee: works fine on both branches ;)
17:22 daniel-s joined
im2ee You changed only Int $n instead of $n? :) 17:22
17:22 Sarten-X joined
im2ee Hm, i mean you add () to second multi combs? 17:23
17:23 masak left
arnsholt joshuablankenship.com/blog/2011/09/...e-purists/ made me think of Perl 6, in a way 17:23
17:24 masak joined
masak im2ee: yeah. I had a feeling the problem was related to signature type narrowness. 17:24
im2ee masak, great. It's nice lesson. :) Thanks.
masak im2ee: probably the $n parameter is too loose if it's not typed as Int. 17:25
for some reason, that's not a problem in nom.
pmurias sorear: hi 17:27
diakopter: what'ls ildjit? 17:28
* what's
TimToady arnsholt: the question is, which way did it make you think about Perl 6? :)
arnsholt Mostly the "What's wrong with Perl 5/Why aren't you working on Perl 5 instead?" crowd, I think 17:29
pmurias sorear: re niecza/jvm what would the benefits of that be? 17:30
jlaire sb goto -50 17:31
oops
TimToady just uses pageup/pagedown...
jlaire is on a laptop with a broken keyboard :( 17:32
the most painful bug is that up-arrow doesn't work
TimToady I've always wished that irssi would respond to spacebar or backspace at the first position and do scrolling
jlaire interesting 17:33
left/right might be nice for that, too
pmurias masak: what's the point of reviving pugs? purely fun? 17:34
TimToady learning Haskell...again...
17:34 kaare_ left
sorear pmurias: niecza/jvm would be useful for all the people who beleive the anti-clr patent fud :) 17:40
moritz what about those who believe the anti-java patent fud? :-) 17:41
pmurias they have niecza/clr ;) 17:43
diakopter moritz: good point 17:44
17:44 bbkr1 left
pmurias sorear: is the anti-clr patent fud stopping people from contributing to niecza? 17:45
diakopter pmurias: lmgtfy.com/?q=ildjit
17:46 bbkr joined
TimToady with respect to the 40-match puzzle, haven't thought about it analytically, but it seems equivalent to me to a Galton's Quincunx that is diamond shaped rather than triangular 17:46
or maybe that *is* thinking about it analytically :)
masak: ^^ 17:47
17:48 envi left
TimToady has a geometric brane, not an algebraic brane... 17:49
to the extent that, when I view the Galton's Quincunx in my head, I "know" I'm facing north! 17:50
I guess you'd call that a geographic brane 17:51
"The spider is crawling up your west arm." 17:52
masak pmurias: the point, I guess, is to have a working Pugs. 17:53
masak googles "Quincunx"
lucs What's the "40-match puzzle"?
TimToady irclog.perlgeek.de/perl6/2011-09-23#i_4469618 17:54
masak ...and subsequent discussion.
TimToady now you know how far back I am in the backlog... :/
lucs TimToady: Thanks.
17:55 uasi left
im2ee How to create TCP server in p6? :) 17:55
Just open a port. 17:56
masak TimToady: yes. Galton's Quincunx, along with folding up in half, seems to be a way to analyze -- er, geometrize -- the problem, too.
im2ee: have a look at IO::Socket::INET 17:57
im2ee masak, hmm, i'm trying it, and i have some problems. :)
masak, but i have some ideas :)
TimToady well, I was thinking more of folding down in double
moritz my $s = IO::Socket::INET.new(:localhost<localhost>, :localport(1234), :listen); 17:58
my $connection = $s.accept
TimToady or of a quincunx that has the bottom two angles trimmed off
masak TimToady: I'm not a native English speaker, all appearances to the contrary nonwithstanding :P
moritz say $connection.read(20).decode
TimToady to constrain the distribution to 40 choices left or right
masak right. 17:59
17:59 pmurias left
TimToady the constraints being a form of "gambler's ruin" 17:59
im2ee moritz, thanks :)
I wrote LocalHost and LocalPort, like in p5. :)
TimToady in fact, one could pose it exactly as a symmetrical gambler's ruin problem 18:00
moritz we CamelCase only class names in p6
masak .oO( CameliaCase ) 18:01
TimToady $She-Who-Must-Be-Obeyed is a possibility though :)
im2ee moritz, thanks :)
TimToady but yeah, we like our hyphens in preference to caps now 18:03
that leaves the letters free to be caps when it's natural, and not when it's not 18:04
Mutiny-on-the-Bounty
18:04 packetknife left
diakopter masak: golf: produce a list of random ascii words of random length < N 18:04
TimToady much easier to read than MutinyOnTheBounty
im2ee is afk 18:05
moritz nom: ('a'..'z').roll((1..6).pick)
p6eval nom 09a598: ( no output )
lucs masak: "[...] as one increases N to the millions it becomes 1." There tends to be a single match left in the other box when N is very high? I find that hard to believe.
moritz nom: say ('a'..'z').roll((1..6).pick) 18:06
p6eval nom 09a598: OUTPUT«l j␤»
moritz nom: say ('a'..'z').roll((2..7).pick).join for 1..8
p6eval nom 09a598: OUTPUT«gelctgv␤ojisg␤jqg␤jwtmbid␤kt␤ixx␤sfxnu␤bud␤»
diakopter cool
masak lucs: N being the number of simulations run, not the number of matches.
lucs Oh! Not what I'd understood :)
18:07 snikkers joined
moritz nom: say (1..8).map: {('a'..'z').roll((2..7).pick).join} 18:07
p6eval nom 09a598: OUTPUT«cgbocu xtoo fmlg zmgcqot ud xvdipo wwz ko␤»
masak lucs: I've already apologised about the confusion in the backlog :P
lucs masak: Sorry, only just skimming :)
TimToady
.oO(lucs makes his sanity roll)
 18:08
TimToady doesn't...
diakopter mmm rolls
18:08 snikkers left
masak nom: say ~('a'..'z').roll(2..7) for ^8 18:09
p6eval nom 09a598: OUTPUT«Method 'pred' not found for invocant of class 'Range'␤ in sub postfix:<--> at src/gen/CORE.setting:1054␤ in sub coro at src/gen/CORE.setting:3855␤ in method reify at src/gen/CORE.setting:3825␤ in method reify at src/gen/CORE.setting:3629␤ in method reify at src/g…
diakopter eh
TimToady pred?!?
masak huh?
this proves it. everything I touch exposes a bug. :/
nom: say ~('a'..'z').roll((2..7).pick) for ^8 18:10
p6eval nom 09a598: OUTPUT«m k c r c z␤d f u e r m l␤c l h h h␤s k t␤h o h c k␤h p c␤s l h f v␤p f r r b e␤»
masak hm.
oh, right.
18:10 pmurias joined
moritz I guess it just copies the argument, and -- it until it's 0 18:10
masak nom: say ('a'..'z').roll((2..7).pick).fmt for ^8
p6eval nom 09a598: OUTPUT«d l l p␤e a f l a k x␤c z g k s j␤q i q k p g s␤m y w b m l a␤f i w␤v c x␤z n p␤»
TimToady oah
hmm, nice word
moritz nom: say ('a'..'z').roll((2..7).pick).fmt('') for ^8 18:11
p6eval nom 09a598: OUTPUT« ␤ ␤ ␤ ␤ ␤ ␤ ␤ ␤»
diakopter heh
moritz nom: say ('a'..'z').roll((2..7).pick).fmt('%s','') for ^8
p6eval nom 09a598: OUTPUT«nmkg␤rqheyfv␤vlp␤ho␤qxil␤xeqi␤veydh␤fivqtfy␤»
masak TimToady: any reason not to make .roll(2..7) mean .roll( (2..7).pick )? :)
moritz .join is still shorter
masak moritz: thanks.
nom: say ('a'..'z').roll((2..7).pick).join for ^8
p6eval nom 09a598: OUTPUT«imgqtfm␤dcnec␤hxww␤zkk␤dmq␤xtaia␤hxizfve␤azx␤»
masak moritz: and somehow more descriptive.
moritz masak: oh come on, this is about golfing :-) 18:12
masak troo
diakopter .roll(Range) sounds cool
TimToady masak: but maybe they meant .roll(any 2..7)
masak TimToady: that's an odd use of junction... :)
junctions*
TimToady nevertheless, someone might think it 18:13
I think the huffman value of it is minimal
it's a pretty low-wattage dwim
druther save my dwims for high-wattage locations 18:14
diakopter LOL .rroll for range roll
TimToady pick-and-roll(2..7) :)
diakopter I was thinking rick-roll
TimToady speaking of low-wattage... 18:15
diakopter ... or desert you
TimToady goes back to lackbogging
lucs masak: Did you come up with a reasonable answer to the matches problem? 18:16
lucs would instead ask "How many matches at most would remain in the other box, 50% (for example) of the time?" 18:17
masak lucs: sorear and moritz solved it to my satisfaction.
lucs masak: Will I find their solution if I keep reading the backlog? 18:18
masak depends if you skim or not :P
lucs :) 18:19
I think I misphrased my question. What I'm trying to get at is something like: "In 100% of test runs, there will be at most 40 matches in the other box. In 98.3% (wild guess), there will be at most 39 matches left in the other box. ... . In ?% of test runs, there will be at most X matches left in the other box." 18:24
masak lucs: that's what a distribution describes. 18:25
lucs Oh. Isn't it known how to solve that? Can't be that hard, can it? 18:26
masak sorear and moritz already did, so it's certainly possible :) 18:28
Woodi was again catched by backlog :) static.themetapicture.com/media/fun...eck-me.jpg
lucs masak: Aha. Got it. Thanks. 18:30
18:37 orafu joined, soh_cah_toa joined 18:41 buubot_backup left
jlaire it's very basic dynamic programming 18:42
for relatively small values of 40 :)
masak :) 18:44
I've always felt that "dynamic programming" is way too bombastic a term for the actual concept.
jlaire me too 18:45
but it's standard
what confuses me even more is that many people find it difficult to understand 18:46
and some algorithm texts treat it as an advanced concept
masak "subproblem caching" would've been fine.
jlaire that's descriptive 18:47
masak "dynamic programming" is like deciding to call closures "extreme code objects". 18:51
jlaire
.oO( dynamic code object )
 18:53
masak or choosing to call constants "public static final"... 18:55
TimToady some of us are catty, and some of us are just dogged... 19:01
diakopter some are fishy
masak or sheepish. 19:02
TimToady is bullish
but occasionally just bullheaded 19:03
moritz is CamliaCased
masak 不是 :)
jlaire niecza: sub f($n) { my @p; @p[0][$n] = @p[$n][0] = 1; for 0..40 -> $x { for 0..40 -> $y { @p[$x][$y] //= 0; @p[$x+1][$y] += @p[$x][$y]/2 if $y > 0; @p[$x][$y+1] += @p[$x][$y]/2 if $x > 0; } }; @p[40][40] }; my @dist = (0..40).map: { f($_) }; say [+] @dist; say ~@dist 19:04
p6eval niecza v9-46-g16ef0c8: OUTPUT«1␤0/1 0.0900535481254757 0.0900535481254757 0.0888840215264435 0.0865449683283792 0.083083169595244 0.0785921874549606 0.0732091609169496 0.0671083975072038 0.0604920766262119 0.0535786964403591 214859239356147155/4611686018427387904 0.0397386751092177 3829…
Woodi 'Whatever Wildcard (like Any, but subject to do-what-I-mean via MMD)' what is MMD ? and is it 'just DWIM' ? 19:05
moritz MMD = Multi Method Dispatch
TimToady using the term "method" loosely 19:06
jlaire niecza: sub f($n) { my @p; @p[0][$n] = @p[$n][0] = 1; for 0..40 -> $x { for 0..40 -> $y { @p[$x][$y] //= 0; @p[$x+1][$y] += @p[$x][$y]/2 if $y > 0; @p[$x][$y+1] += @p[$x][$y]/2 if $x > 0; } }; @p[40][40] }; say f(40) == 1/2**39
p6eval niecza v9-46-g16ef0c8: OUTPUT«Bool::True␤»
moritz maybe "MD" would have been more appropriate, but Mr. Mark Dominus might have objected :-)
TimToady or thinking of function calls as method calls on a lexical scope
moritz oh wait, that'd be MJD 19:07
gbacon you can't ... oh, nevermind
TimToady anyhoo, we aren't picky about that
moritz jlaire++
TimToady gbacon: lemme guess, the next word is "just"
masak moritz: Multi-Junction Dispatch :P 19:08
Woodi but should i use Whatever to undefine field containing possible error text or Any or Mu ? just '' is defined ?
gbacon TimToady: we are the dreamers of dreams
Woodi EtoManyChoises 19:09
moritz Woodi: probably Any
nom: my $x = enum A <b c d>; say $x.WHAT
TimToady probably Nil, if you're undefining something
p6eval nom 09a598: OUTPUT«A()␤»
moritz nom: my $x = enum A <b c d>; say $x.HOW
p6eval nom 09a598: OUTPUT«Method 'gist' not found for invocant of class 'Perl6::Metamodel::EnumHOW'␤ in sub say at src/gen/CORE.setting:4859␤ in <anon> at /tmp/hKo9p7zF2k:1␤ in <anon> at /tmp/hKo9p7zF2k:1␤␤» 19:10
TimToady or just undefine() it
Woodi ok, i undefine and check :)
TimToady perl6: my Int $x = Any;
p6eval niecza v9-46-g16ef0c8: OUTPUT«Potential difficulties:␤ $x is declared but not used at /tmp/SR9l8wFV70 line 1:␤------> my Int ⏏$x = Any;␤␤Unhandled exception: Nominal type check failed for scalar store; got Any, needed Int or subtype␤ at /tmp/SR9l8wFV70 line 0 (MAIN …
..rakudo 09a598: OUTPUT«Type check failed in assignment to '$x'; expected 'Int' but got 'Any'␤ in <anon> at /tmp/XfK0aa5gTm:1␤ in <anon> at /tmp/XfK0aa5gTm:1␤␤»
..pugs: ( no output )
19:10 buubot_backup joined
TimToady perl6: my Int $x = Nil; 19:10
p6eval rakudo 09a598: OUTPUT«Type check failed in assignment to '$x'; expected 'Int' but got 'Nil'␤ in <anon> at /tmp/9KQJ39klXi:1␤ in <anon> at /tmp/9KQJ39klXi:1␤␤»
..niecza v9-46-g16ef0c8: OUTPUT«Potential difficulties:␤ $x is declared but not used at /tmp/sGUCQ43ITM line 1:␤------> my Int ⏏$x = Nil;␤␤»
..pugs: OUTPUT«*** No such subroutine: "&Nil"␤ at /tmp/QLOVkvee3y line 1, column 8-16␤»
masak moritz: I'll just submit that EnumHOW.gist thing as a rakudobug 19:11
moritz nom: my $x = (enum A <b c d>).enums; say $x.WHAT
p6eval nom 09a598: OUTPUT«Hash()␤»
TimToady well, niecza gets it right
moritz nom: my $x = (enum A <b c d>).enums; say $x.perl
p6eval nom 09a598: OUTPUT«{"b" => 0, "c" => 1, "d" => 2}␤»
Woodi perl6: return undef; 19:13
p6eval pugs: OUTPUT«*** Cannot use this control structure outside a 'routine' structure␤ at /tmp/T2LTUM4cmV line 1, column 1-13␤»
..rakudo 09a598: OUTPUT«===SORRY!===␤Unsupported use of undef as a value; in Perl 6 please use something more specific:␤ Mu (the "most undefined" type object),␤ an undefined type object such as Int,␤ Nil as an empty list,␤ !*.defined as a matcher or method,␤ Any:U as a type constraint␤
..or…
..niecza v9-46-g16ef0c8: OUTPUT«===SORRY!===␤␤Unsupported use of undef as a value; in Perl 6 please use something more specific:␤ Mu (the "most undefined" type object),␤ an undefined type object such as Int,␤ Nil as an empty list,␤ :!defined as a matcher,␤ Any:U as
19:13 Sarten-X left
p6eval ..a typ… 19:13
19:15 mkramer joined, buubot_backup left
moritz quick poll: should IO::Socket.recv return a Buf or a Str? 19:16
dalek kudo/nom: a352f1b | moritz++ | src/core/Buf.pm:
Buf.bytes and .chars
kudo/nom: e305cff | moritz++ | src/core/IO/Socket.pm:
in IO::Socket.recv, interpret $bufsize as a maximal length, not a minimal one
moritz are enums 'our' by default? 19:17
Woodi perl6: class A { has $!er = 'Err'; method b{ $!er.undefine } }; my $a = A.new; $a.b();
p6eval pugs: ( no output )
..rakudo 09a598: OUTPUT«Method 'undefine' not found for invocant of class 'Str'␤ in method b at /tmp/CrJyWSTlEl:1␤ in <anon> at /tmp/CrJyWSTlEl:1␤ in <anon> at /tmp/CrJyWSTlEl:1␤␤»
..niecza v9-46-g16ef0c8: OUTPUT«Unhandled exception: Unable to resolve method undefine in class Str␤ at /tmp/7S1lk4E5Ej line 1 (MAIN A.b @ 4) ␤ at /tmp/7S1lk4E5Ej line 1 (MAIN mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2066 (CORE C971_ANON @ 2) ␤ at /home/p6eval/niec…
19:18 mkramer1 left
TimToady moritz: yes, as are all constants 19:18
since constants are just very small types :)
moritz nom: class A { enum B <a c> }; say A::B; say A::B::c
p6eval nom 09a598: OUTPUT«B()␤B::c␤»
moritz \o/
TimToady and that's one reason for it
Woodi $v.undef[ine]()should just work...
TimToady the word "undef" is banned in p6 19:19
benabik Mu
TimToady std: undef
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Unsupported use of undef as a value; in Perl 6 please use something more specific:␤ Mu (the "most undefined" type object),␤ an undefined type object such as Int,␤ Nil as an empty list,␤ :!defined as a matcher,␤ Any:U as a type constrain…
moritz Woodi: iirc undefine() is only specced as a function, not as a method 19:20
Woodi so perl6 is less scripting language intentionally :)
jlaire T9Yur9puxN4, sErDJOP8PHc
ww. 19:21
TimToady we can't be dwimmy when we don't know what the user means
it's like asking the universe to create a quark without color 19:22
Woodi perl6: class A { has $!er = 'Err'; method b{ undefine($!er); say $!er } }; my $a = A.new; $a.b();
p6eval niecza v9-46-g16ef0c8: OUTPUT«===SORRY!===␤␤Undeclared routine:␤ 'undefine' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 685 (CORE die @ 2) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1136 (STD P6.comp_unit @ 36)…
..rakudo 09a598: OUTPUT«Any()␤»
..pugs: OUTPUT«␤»
Woodi that work for me
TimToady P6 is actually more strongly typed than languages that mix up differently typed values as nil 19:23
undefs are "type charge" carriers in P6, like gluons
Woodi learning perl6, v5 was handy :) 19:24
19:24 Sarten-X joined
TimToady we still try to be handy most of the time, and even allow the user to have mushy ideas when we can figure out what they mean 19:24
but we can't always do that
dalek kudo/nom: aed9fe7 | moritz++ | t/spectest.data:
track test rename
 19:25
TimToady and at that point we try to give sufficient feedback as to what kind of distinction they need to learn to make
Woodi TimToady: backward compatybility is important when it is possible...
TimToady sure, but that's the one thing we were intentionally breaking in P6 from the start
moritz TimToady: I find it odd that anonymous enums return a hash, while named enums return something where you have to call .enums to get to the hash 19:28
Woodi understand that, but about api... some handy conventions can be unchanged (when it is not necesary
moritz that's slightly at odds with the normal convention that omitting the name just... well, means that the result doesn't have a name anymore
Woodi stops complains :) 19:29
TimToady undef was also a syntactic problem in that it confuses the term/operator expectations, when P6 is trying to keep that as unambiguous as possible. 19:32
std: Mu $_
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Two terms in a row at /tmp/HlWyXUiJ7O line 1:␤------> Mu ⏏$_␤ expecting any of:␤ bracketed infix␤ infix or meta-infix␤ statement modifier loop␤Parse failed␤FAILED 00:01 120m␤»
masak TimToady: why is that a problem for 'undef', though, and not for '*'?
TimToady p6 knows that Mu is a term in a way that p5 can't know about undef
* is simply a term, and never takes an argument 19:33
masak oh, troo
TimToady (undef) and (undef $x) are two completely different things in p5
p5 only gets away with it most of the time because few infix operators start with a sigil, and none of the ones that do make sense with undef on the left 19:34
19:35 ZapZ_ joined
TimToady but we want to be more extensible, and that means we can't predict the future contents of the infix namespace, or the sigil namespace for that matter 19:35
or if we'll even have sigils in some dialects 19:36
masak TimToady: quietfanatic called you a "language designer designer" on Twitter. I'm starting to see how that's very accurate. :)
[Coke] masak: My sundays are ... I have no idea if I'm free. 19:37
19:37 gbacon left
TimToady well, that probably makes Paul Graham a language designer designer designer, at least in terms of thinking about a 100-year language 19:37
[Coke] I just got back from a slightly unexpected 3 hour volunteering stint at the local elementary school, and am too zoned to *SQUIRREL*!
masak [Coke]: ok. sounded like mberends might have trouble any other day. but I'll field the question to him Justin Case.
19:38 buubot_backup joined
TimToady the syntactic categories can change however they want, but TTIAR is how we keep infixes and prefixes in different namespaces, and nospace is how we keep infixes and postfixes in different namespaces. So TTIAR and nospace are more fundamental. 19:40
s/nospace/unspace/ if you prefer
[Coke] masak: I really really don't have time to learn haskell. ;) 19:41
TimToady we probably need a separate name for that concept
you can learn it lazily
19:42 gbacon joined
Woodi i hear monads are obsolote :) 19:42
masak [Coke]: so... you want to help with Pugs, but you'd prefer not to have to learn Haskell? :)
Woodi: Monads are outside of time, and incapable of becoming obsolete :P 19:43
TimToady just write an emitter for STD that emits Haskell as you understand it, and then let someone else fix the bugs. :)
dalek ast: 44efcc5 | moritz++ | S (2 files):
rename S02-types/enum.t to S12-enums/misc.t and delete lots of crufty tests
 19:45
ast: 62d5ffd | moritz++ | S12-enums/misc.t:
tests for conflicts between enums and subs (RT #63650)
TimToady wonders if the Arrow of time is necessarily outside of time...
masak I should hope so!
TimToady then why don't you? 19:46
19:46 DarthGandalf left, buubot_backup left
moritz [Coke]++ # ticket triaging 19:46
masak I accidentally dropped the remote control over the Railing of Time...
TimToady
.oO(Remotes fly like an Arrow)
masak *lol* 19:48
TimToady
.oO(I shot my Remote in the air, and where it lands, I know not where...)
 19:49
Woodi anyone else gets Method 'STORE' not found for invocant of class 'Any' all today day ?
moritz nope 19:50
TimToady nom: Any = 42;
p6eval nom 09a598: OUTPUT«Cannot assign to a non-container␤ in <anon> at /tmp/wP8sQVjuzJ:1␤ in <anon> at /tmp/wP8sQVjuzJ:1␤␤»
TimToady aww
moritz nom: my $x := Any; $x = 3
p6eval nom 09a598: OUTPUT«Cannot assign to a non-container␤ in <anon> at /tmp/r0nOAQQMd6:1␤ in <anon> at /tmp/r0nOAQQMd6:1␤␤»
moritz nom: Any.STORE
p6eval nom 09a598: OUTPUT«Method 'STORE' not found for invocant of class 'Any'␤ in <anon> at /tmp/pNE1fZetI2:1␤ in <anon> at /tmp/pNE1fZetI2:1␤␤»
moritz :-)
TimToady \o/
Woodi so probably it's my modification responsible :)
TimToady this wouldn't happen in Haskell :) 19:51
19:52 DarthGandalf joined
moritz nom: say Mu.^can('gist') 19:52
p6eval nom 09a598: OUTPUT«gist␤»
moritz nom: say ClassHOW.^can('gist')
p6eval nom 09a598: OUTPUT«Could not find sub &ClassHOW␤ in <anon> at /tmp/qoP1b_RgWo:1␤ in <anon> at /tmp/qoP1b_RgWo:1␤␤» 19:53
19:57 daniel-s left
masak wow. so, I've started to translate the pack.t test file from Perl 5 to Perl 6. this current test *opens the perl executable*, reads 8192 bytes, then runs unpack in two different ways and checks the results against each other. 19:57
I'm... I'm not sure I want to have a spectest that does that. 19:58
:)
19:58 daniel-s joined
masak decides to hardcode the first 8192 bytes of his perl executable into the test file instead 19:59
moritz nom: say ([+] *.list).([1, 2, 3]) 20:02
p6eval nom aed9fe: OUTPUT«No applicable candidates found to dispatch to for 'Numeric'. Available candidates are:␤:(Mu, Mu %_)␤␤ in method Numeric at src/gen/CORE.setting:602␤ in sub infix:<+> at src/gen/CORE.setting:1924␤ in sub <anon> at src/gen/CORE.setting:7356␤ in <anon> at /tmp/TUbeM6…
moritz should that work?
TimToady Truman beat Dewey when I was -7, sigh
er, no, that was when I was -6 20:03
I forget...
moritz: no, whatevers don't do anything in lists 20:04
dalek ecza: 4266b44 | (Paweł Murias)++ | / (4 files):
Argumentless method call on perl5 objects.
 20:05
pmurias sorear: how can i create an STable which inherits from Any?
sorear pmurias: look at NewWrapper in nieczaCLR.cs 20:06
moritz www.perlmonks.org/?node_id=927675 # any neat Perl 6 solutions? 20:07
sorear pmurias: chromatic has said in so many words that he ignores niecza because Microsoft. He can't be alone.
TimToady nom: (class {}).HOW
p6eval nom aed9fe: ( no output )
TimToady nom: (class {}).HOW.say 20:08
moritz sorear: I think chromatic ignores niecza because otherwise he'd have to admit that you don't need parrot for running Perl 6
p6eval nom aed9fe: OUTPUT«Method 'say' not found for invocant of class 'Perl6::Metamodel::ClassHOW'␤ in <anon> at /tmp/uG7ta6Zl9d:1␤ in <anon> at /tmp/uG7ta6Zl9d:1␤␤»
masak moritz: that does sound like chromatic.
TimToady nom: (class {}).HOW.Str
p6eval nom aed9fe: OUTPUT«No applicable candidates found to dispatch to for 'Str'. Available candidates are:␤␤ in hash Str at src/stage2/NQPCORE.setting:57␤ in <anon> at /tmp/95FwJHSrhf:1␤ in <anon> at /tmp/95FwJHSrhf:1␤␤»
TimToady nom: (class {}).HOW.gist.say
p6eval nom aed9fe: OUTPUT«Method 'gist' not found for invocant of class 'Perl6::Metamodel::ClassHOW'␤ in <anon> at /tmp/rQEr_D4akr:1␤ in <anon> at /tmp/rQEr_D4akr:1␤␤»
masak moritz: I haven't been picking up much resistance to the idea that Rakudo is free to explore other VMs from any Parrot developers aside from chromatic.
TimToady nom: say (class {}).^methods 20:09
p6eval nom aed9fe: OUTPUT«eager elems end classify infinite flat hash list pick roll reverse sort values keys kv pairs Array grep first join map min max minmax postcircumfix:<[ ]> at_pos all any one none postcircumfix:<{ }> reduce ACCEPTS WHERE WHICH WHY Bool so defined new CREATE bless BUILDA…
TimToady nom: say (class {}).^methods(:local) # still want default to not be :all 20:10
donri does temp $*CWD make sense?
p6eval nom aed9fe: OUTPUT«␤»
diakopter moritz: I wouldn't say chromatic totally ignores niecza; he mentions it in a blog post and slightly explains why he can't consider using it
flussence the only realistic reason I can come up with to avoid mono is "the people developing it sound nasty", but by that logic I may as well avoid parrot too.
diakopter mono folks are the kindest around 20:11
flussence I've seen them say pretty nasty things to distro package maintainers; they don't look like the sort of people I'd want to deal with...
masak moritz: that meditation is a very interesting read. thank you. 20:12
diakopter flussence: oh
masak diakopter: I think moritz meant "ignores" as in "won't use".
for his own projects and so.
pmurias diakopter: where can i find chromatics blog?
diakopter modernperlbooks dot something 20:13
er
sorear 0 i.~ +./"1 arr -- works :>
diakopter modernperlbook
flussence diakopter: this sort of thing specifically - blog.flameeyes.eu/2009/07/04/the-ne...a-debugger
diakopter pmurias: www.modernperlbooks.com/mt/index.html
pmurias found it, it seems chromatic generally dislikes mono instead of the patents being the main reason 20:15
im2ee Can i add something to new method? Without changing default action? And how if yes :) 20:16
masak nom: my @array = [1, 1, 1, 1, 1], [0, 1, 1, 0, 0], [0, 0, 0, 0, 1], [0, 0, 0, 0, 0], [0, 1, 0, 1, 0]; say @array.pairs.first({ all(@(.value)) == 0 }).key
p6eval nom aed9fe: OUTPUT«3␤»
masak moritz: ^
moritz masak: that was my first solutioin too :-) 20:17
I also have
say @array.pairs.first({ 0 == [+] .value.list}).key;
20:17 gbacon left
im2ee masak, what means "{ all(@(.value)) == 0 }" 20:18
? :)
flussence (all values of (.value in list context)) == 0
masak moritz: well, as long as you know you don't have negative values in the array... 20:19
im2ee: what flussence said.
moritz masak: correct
I also have
say @array.pairs.first({ not [||] .value.list}).key;
im2ee .value == $_.value, right? :)
masak aye.
TimToady um, 0 == [+] is wrongish
-1 + 1 == 0
masak TimToady: I pointed that out above :P
moritz you can >>.abs if you allow negative numbers 20:20
TimToady oic
moritz masak: oh, and all(..) == 0 can be simplified to none()
masak oh!
of course!
moritz www.perlmonks.org/?node_id=927681
masak I even thought that before starting, but subsequently forgot...
sorear flussence: Having read that blog post, I side with miguel and directhex. They're two of the nicest people I know outside here and I don't think their responsese were out of line at all
masak moritz++ 20:21
donri what is the purpose of "fail", given that it seems to only/mainly do string errors?
TimToady seems to me that a lazy .[0] doesn't need .first
masak sometimes I want to write @.pairs.first(...).key as @.first(...) :key
flussence sorear: probably. It's not like gentoo's devs are saints themselves... 20:22
20:22 mishin_ joined
masak TimToady: .first(...) is a nice Huffmanization of .grep(...)[0] 20:22
donri also fail seems to be fatal for me (old rakudo though) but specs say default is no fatal
20:23 DarthGandalf left
TimToady fail can contain any unthrown exception 20:24
and it's only supposed to throw it if it isn't checked at some point
or if the user has 'use fatal' in effect
20:25 carlin joined
TimToady but most of the time, as soon as the failure gets into void context, it should throw 20:25
donri ah, i see
20:25 DarthGandalf joined
moritz not that the implementation of fail() in the old rakudo was quite bad 20:25
nom's is still bad, though not quite as bad
donri also, why is 'die' *not* fatal? :) 20:26
moritz it is not?
where?
when?
donri well it's not exiting
moritz example?
donri or is that just the interactive shell?
moritz could be
there it's a feature :-) 20:27
TimToady it's a game where you get multiple lives
donri but then, why is "fail" fatal in the shell?
TimToady nom: say Failure.gist
p6eval nom aed9fe: OUTPUT«Failure()␤»
20:28 buubot_backup joined
moritz donri: bug 20:28
TimToady nom: say Failure
p6eval nom aed9fe: OUTPUT«Failure()␤»
TimToady nom: print Failure
p6eval nom aed9fe: OUTPUT«Use of uninitialized value in string context␤»
donri is "nom" back, i thought that was "rakudo" now
TimToady rakudo is the generic name
moritz 'nom' and 'rakudo' are synonyms, as far as p6eval is concerned
TimToady nom is the version name
donri my local install: This is Rakudo Perl 6, version 2011.04 built on parrot 3.3.0
TimToady the next major version will be rakudo but not nom 20:29
donri le old
masak perl6: say "lots of different vowels".comb("o").elems
p6eval pugs: OUTPUT«3␤»
..rakudo aed9fe: OUTPUT«No applicable candidates found to dispatch to for 'comb'. Available candidates are:␤:(Str, Mu %_)␤:(Str, Regex $pat, Any $limit, Any $match, Mu %_)␤:(Cool, Mu %_)␤:(Cool, Regex $matcher, Any $limit, Mu %_)␤␤ in method comb at src/gen/CORE.setting:1559␤ in <anon> …
..niecza v9-47-g4266b44: OUTPUT«Unhandled exception: Unable to resolve method postcircumfix:<( )> in class Str␤ at /home/p6eval/niecza/lib/CORE.setting line 296 (CORE Cool.comb @ 8) ␤ at /tmp/FO7g6fm77X line 1 (MAIN mainline @ 1) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2066 (COR…
masak b: say "lots of different vowels".comb("o").elems
p6eval b 1b7dd1: OUTPUT«No applicable candidates found to dispatch to for 'comb'. Available candidates are:␤:(Mu : Any :limit($limit);; *%_)␤:(Mu : Regex $matcher, Any $limit = { ... }, Any :match($match);; *%_)␤␤ in main program body at line 22:/tmp/vplMQUpZml␤»
masak oh, well.
donri so what is "b"? :)
TimToady the one b-fore 20:30
moritz the old rakudo
donri "buggy"
TimToady beta, but don't say that!
masak "B-star"
moritz beijing # codename of last release
TimToady so this one should also be c: 20:31
masak c as in "gamma" :)
TimToady which would confuse the heck out of DOS programmers...
masak C:\>
those were the newer DOSes.
the older ones were just A> 20:32
moritz wants a new HPMoR chapter, but EY seems to be quite busy otherwise :(
TimToady spivaking won't rescue ambiguous pronouns :)
masak moritz: I've been reading a bit of his other stuff. it's generally very high quality. 20:33
20:33 buubot_backup left
masak TimToady: it wasn't a spivak, it's his initials :P 20:33
moritz masak: right, but most of it isn't quite as entertaining as hpmor
masak moritz: right.
TimToady you could also use [==] 0, @list 20:35
and that is supposed to short circuit as well 20:36
masak ooh
what's the colloquial way to compare two Buf variables for equality? 20:37
moritz that one is nice
TimToady that's after a DOH! moment when I wrote 0 == [==] @list
moritz masak: infix:<eqv>
masak thanks.
moritz (works both in b and in nom)
masak moritz++
moritz nom: say $*IN.read(2) eqv $*IN.read(2) 20:38
p6eval nom aed9fe: OUTPUT«Bool::False␤»
TimToady [==] also has the benefit of coercing to numeric
moritz nom: say Buf.new(1, 2) eqv Buf.new(1, 2)
p6eval nom aed9fe: OUTPUT«Bool::True␤»
masak I'm making good headway into the pack.t file, actually.
moritz TimToady: just how stringy is Buf?
TimToady: should infix:<eq> work on bufs? what aboug gt/ge/lt...
masak automatic translation of pack.t is out of the question. at least in 2011. 20:39
TimToady I think all those can usefully work if defined as integer comparisons, even without assuming ASCII
moritz thought so
masak .oO( oh right, *that's* what prefix:<~> means in Perl 5... oh dear... ) 20:41
people, what would you say a decent translation of Perl 5's C< ~0 > would be? 20:42
TimToady -1
masak ok :) 20:43
TimToady perl6: say +^0
p6eval pugs, rakudo aed9fe, niecza v9-47-g4266b44: OUTPUT«-1␤»
masak my local Perl 5 says 18446744073709551615
TimToady wowee
moritz Inf?
masak moritz: minus one :) 20:44
TimToady 2 ** 64 - 1
masak nodnod
20:44 bazqux joined
masak 2 ** 64 is the number I never memorized, remember? :/ 20:44
a girl in my class did, and recited it to me, triumphantly. 20:45
TimToady me either, but if it's about that long, and starts with 18, it probably is :)
masak yeah :)
that's my rule of thumb too.
2 ** 32 is one of my favorite numbers. 20:46
perl6: say 2 ** 32
p6eval pugs, rakudo aed9fe, niecza v9-47-g4266b44: OUTPUT«4294967296␤»
TimToady in anime you can tell all the autistic girls because they have blue-ish hair
masak TimToady: oh! I thought that was just in Soranowoto
donri rakudo website is wrong: nom is the default branch of git, so the steps are the opposite - need to git checkout to get non-nom, not to get nom 20:50
masak ソ・ラ・ノ・ヲ・ト
TimToady ヲ, not を? 20:51
innersting 20:52
well, consistent, anyway
tadzik one-eyed smile, :D, needle, antenna, some futhark character?
20:52 pmurias left 20:53 pmurias joined
masak TimToady: straight from Wikipedia. 20:53
tadzik: no, no, no, no, and no :) 20:54
tadzik dang
TimToady it's just ヲ is usually considered somewhat archaic
tadzik ト likes a piece of futhark
masak characters have no likes or dislikes.
20:55 buubot_backup joined
tadzik arm, right 20:56
well, nope
for example, Arya doesn't like sewing
she's a character with a dislike :)
20:57 DarthGandalf left, worr joined
masak :P 20:57
I stand corrected.
or at least confused.
tadzik fair enough
masak she likes needles, though :)
im2ee how are You tadzik ? :)
tadzik some of them, yes
im2ee: pretty fine, thanks :) How are things?
TimToady ウオ is the usual way to write ヲ these days 20:58
that's how my last name starts out, in fact
im2ee tadzik, good enough :)
tadzik Toady? :)
TimToady no, that's my first name 20:59
20:59 mj41 left
sorear thinks vowels are funny 20:59
TimToady thinks vowels is funny 21:00
tadzik never knows which are which
sorear suprissed it's not ワ...
im2ee Hm, can i add to default "new" method of class actions some of own action? I want to save "auto-assigning" and add some own functions :)
TimToady that's wa, not uo
masak tadzik: vowels are the ones you say vocally. consonants are the ones you sound with. :P
21:01 DarthGandalf joined, bluescreen10 joined
masak im2ee: sounds like you want a BUILD submethod. 21:01
TimToady vowels are the continuants in the middle of a syllable
im2ee masak, probably :) in BUILD submethod i don't have to writing all assigning stuff? :)
TimToady so anything that's not a stop or affricate can function as a vowel
my favorite is the one in psst 21:02
sorear there seems to be a lot more consistency between languages in the consonants than in the vowels
masak im2ee: there's a way to write them in the siggie of the BUILD.
TimToady and what is the vowel in the second syllable of "didn't"? 21:03
tadzik y 21:04
di-dynt
sorear I hear a pure consonant. d-i-dn, t
tadzik how do you say 'dn'? For me it's quite like 'dyn', in my pronounciation of 'dyn' 21:05
TimToady I'm referring to the dialect in which your tongue never leaves your alveolar ridge between the middle d and the final t
all we really do in my dialect is open up our nose to let the n out
im2ee masak, thanks. :)
TimToady and that's the vowel
benabik The vowel is ', obviously.
masak didənt 21:06
TimToady well, to be precise, [n] is the vocoid, and /'/ is (perhaps) the vowel :)
masak the schwa. it's always the schwa. 21:07
TimToady only there is no ə in my dialect that is not already nasalized
tadzik e sounds funny when you stand on your head I suppose
masak məh.
TimToady my tongue touches twice, not three times
masak mine once.
oh wait, twice. 21:08
TimToady you do you pronouce the did then
right
masak things get complicated once you start thinking about them :P
sorear has to look up what an alveolar ridge is
TimToady indeed, lots of etic vs emic distinctions here, which is why it's the prototype for all etic/emic distinctions
sorear knew ey didn't inhale eir tounge :) 21:09
21:09 buubot_backup left
TimToady well, you do with clicks 21:09
kinda
I'd say, don't get me starting on phonetics, but it's kinda too late... 21:10
masak hah, just got callbinding errors because of whitespace between function name and parentheses. the perils of translating from p5 to p6... I never make that error when writing Perl 6 from scratch.
japhb rakudo: given "a" { if /^(a)$/ { say $0 } } 21:13
p6eval rakudo aed9fe: OUTPUT«a␤»
japhb rakudo: given "a" { when /^(a)$/ { say $0 } }
p6eval rakudo aed9fe: OUTPUT«Nil␤»
japhb ^^ known?
21:14 packetknife joined
masak b: given "a" { when /^(a)$/ { say $0 } } 21:15
p6eval b 1b7dd1: OUTPUT«a␤»
masak japhb: probably part of the whole regexes not being quite done in nom yet thing.
japhb masak, nodnod. Strange effect though. I wouldn't expect a difference between those two. 21:16
masak indeed.
masak submits rakudobug
japhb masak, thx
flussence at last, I have something interesting to show for all those weeks of CPU time... github.com/flussence/specgraphs/ra.../impls.png 21:17
masak flussence: wow, cool! flussence++ 21:18
I swear Niecza *feels* more implemented than just 25% of Rakudo's features...
tadzik oo, that's cool 21:19
masak but maybe there's a bias in there, given that most of the spectest suite has been built for/adaptet to Rakudo.
flussence just because the lines overlap that way doesn't mean the features do :)
masak adapted*
flussence (there's a huge grey area that nom doesn't do yet...)
dalek ecza: a20956d | (Paweł Murias)++ | / (3 files):
Arguments are passed to perl 5 methods.
 21:20
colomon masak: I think a large-scale effort to get the spectest fudged for Niecza would get it to more like 75%
TimToady features probably gives a better comparison 21:21
colomon has 90% of trig working on his local copy of Niecza, just need to figure out how to elegantly integrate it to the basic build.
flussence also, the niecza numbers on my graphs are slightly more correct since it checks out t/spec at a matching date, whereas I just use make spectest on rakudo which does a git pull 21:24
(if someone wants to copy-paste the code I'm using for that to make rakudo's makefile DTRT, go ahead) 21:25
masak colomon: that feels reassuring. the 75% figure, I mean. 21:26
pmurias colomon: what's the tricky part with the integration? 21:27
im2ee b: class abc { has $.lol = 'abc'; submethod BUILD { say $.lol; }}; my $abc = abc.new(); 21:31
p6eval b 1b7dd1: OUTPUT«Any()␤»
im2ee why? :)
pmurias sorear: what do the STable entries do?
tadzik im2ee: rakudobug 21:32
im2ee: thought it's fixed in nom
im2ee nom: class abc { has $.lol = 'abc'; submethod BUILD { say $.lol; }}; my $abc = abc.new();
tadzik nom: class abc { has $.lol = 'abc'; submethod BUILD { say $.lol; }}; my $abc = abc.new();
p6eval nom aed9fe: OUTPUT«Mu()␤» 21:33
im2ee like in:
rakudo: class abc { has $.lol = 'abc'; submethod BUILD { say $.lol; }}; my $abc = abc.new();
p6eval rakudo aed9fe: OUTPUT«Mu()␤»
tadzik someone will now tell me it's a feature, and I'm going to whine again, and masak will join again, and we'll both be dissatisfied again :/
nom: class abc { has $.lol = 'abc'; submethod BUILD { callsame; say $.lol; }}; my $abc = abc.new();
p6eval nom aed9fe: OUTPUT«Mu()␤»
tadzik b: class abc { has $.lol = 'abc'; submethod BUILD { callsame; say $.lol; }}; my $abc = abc.new();
p6eval b 1b7dd1: OUTPUT«abc␤»
tadzik great, nothing works in nom now
masak tadzik: no, I actually understood last time. 21:34
im2ee callsame?
tadzik really?
im2ee tadzik, what is callsame? :)
masak tadzik: AFAIR, what I was dissatisfied about was just a bug in Rakudo.
tadzik so what's then new easy-like-in-Moose way?
masak: well, look above
masak tadzik: not sure I want to :)
I might get upset ;) 21:35
tadzik masak: tl;dr: now it's broken even with callsame :)
masak ooh!
tadzik im2ee: call the next method with this name in the MRO chain
flussence rakudo: class A { multi method foo(Int $a) { } }; A.new.?foo('str'); say 'alive'
p6eval rakudo aed9fe: OUTPUT«No applicable candidates found to dispatch to for 'foo'. Available candidates are:␤:(A, Int $a, Mu %_)␤␤ in method foo at src/gen/CORE.setting:250␤ in method dispatch:<.?> at src/gen/CORE.setting:711␤ in <anon> at /tmp/dxrWCth0YM:1␤ in <anon> at /tmp/dxrWCth0YM…
masak well, that's just reality not living up to Platonic reality. I can live with that :)
tadzik im2ee: so in this case, we default to the default BUILD
flussence should that work?
tadzik flussence: I don't think so
'str' in not an Int
flussence but I'm using .? to call it... 21:36
tadzik oh, I see what you mean I think
and I don't know how exactly .? should work too ;)
im2ee What is .? for? :)
tadzik I read it as $a.can('foo') && $a.foo
flussence rakudo: class A { }; A.new.?foo('str'); say 'alive'
p6eval rakudo aed9fe: OUTPUT«alive␤»
tadzik in which case that'd be right
because A can foo()
masak what's the colloquial way in Perl 6 to leave a do-once block?
flussence im2ee: it works the same as in p5 regex
im2ee flussence, ok, thanks. :) 21:37
flussence (?, + and * work)
im2ee Great. :)
TimToady <ingy> and how do I get 2 or more? <masak> we have inline actions, so we don't need extra syntax ;P <--- neither of you understands % yet
<stuff> ** 2..* % <comma>
tadzik masak: my first thought is do { } while 0; and break inside
im2ee tadzik, what means "int the MRO chain"? :)
tadzik or however is 'break' in Perl 6 21:38
masak TimToady: oh! of course!
TimToady: it totally subsumes all of ingy's use cases.
donri "succeed"?
21:38 benabik left
masak TimToady: I understand it, I just don't grok it yet :) 21:38
tadzik im2ee: MRO is Method Resolution Order. It's your class, then the parent classes, then their parents and so on, til the beginning of time
masak tadzik: that would be 'repeat { } while 0'.
tadzik masak: mebbe, yes
colomon pmurias: the trig functions are autogenerated. I mostly just need to figure out how to properly add that to the niecza build toolchain. 21:39
masak tadzik: 'do' does something differently in Perl 6.
tadzik there are things that, for some reason, I don't use outside of C
masak donri: "succeed" is actually a nice suggestion.
21:40 zby_home__ left
masak donri: I could do that from within a 'given' block, which "loops" once :) 21:40
pmurias colomon: what is the autogenerator written in?
im2ee tadzik, so callsame calls "new" method?
colomon pmurias: p5
tadzik im2ee: why "new"?
im2ee: I'll write an example
21:40 SHODAN left
pmurias colomon: is p5 already a niecza dependency? 21:40
tadzik nom: class A { method a { say 'first!' } }; class B is A { method a { callsame; say 'second!' } }; B.new.a 21:41
p6eval nom aed9fe: OUTPUT«first!␤second!␤»
sorear colomon: I recommend checking the autogenerated code into git; it's unlikely it will need to be run often
pmurias: no
colomon pmurias: no idea
sorear (also I made a point of purging the p5 dependency back in Jan-Feb)
im2ee tadzik, so in my example what is called by callsame? :)
colomon sorear: ah. just paste it into Builtins.cs? That's how I've got it locally now
sorear colomon: I'd be more inclined to put it in lib/GeneratedTrigFunctions.cs 21:42
colomon sorear: ah! well, then, I need to figure out how to add that file to the build process. :)
tadzik im2ee: the default BUILD that every class gets 21:43
sorear colomon: just add it to cskernel in Makefile
21:44 Exodist left
im2ee tadzik, ok, i had it in mind. :) 21:44
tadzik, thanks.
tadzik you're welcome 21:45
colomon sorear: is public partial class Builtins { what I need to declare at the top?
sorear colomon: yes, also using Niecza; using System; 21:46
colomon sorear: yes, I figured I'd just copy the usings until it worked. :)
sorear: okay, it's dinnertime here, but I'll try to tackle the conversion before bedtime
jnthn back 21:50
masak std: 'x'x16 21:54
p6eval std c22b2ee: OUTPUT«===SORRY!===␤Whitespace is required between alphanumeric tokens at /tmp/d3u6mj477c line 1:␤------> 'x'x⏏16␤Check failed␤FAILED 00:01 120m␤»
masak thought so
21:56 diakopte1 joined, diakopte1 left
diakopter eh 21:56
jnthn tadzik: Why do you want to callsame to BUILD? 21:57
tadzik jnthn: that worked in b 21:58
TimToady it's "unwarranted chumminess with the implementation"
jnthn tadzik: That doesn't answer my question at all.
tadzik jnthn: and I want to have attributes set in in BUILD
TimToady we need some other way of saying, "apply the defaults now"
jnthn tadzik: BUILD is for setting attributes
Anything you don't touch in BUILD is set to the default value.
tadzik jnthn: Okay, then I want the Moose's BUILD feelalike
TimToady LISTEN TO ME!!!! 21:59
thank you
I'm tired of this discussion
jnthn also
TimToady we need a verb
do-it-now()
whatever
BUILD is highly magical, and can be magicaller 22:00
jnthn Or people could just write submethod BUILD($!x = oh-look-my-default()) { }
TimToady whatever operation is assumed at the end of BUILD, we should be able to call sooner explicitly
pmurias what would be a good niecza p5 interop example? (i need something to determine the features i need to implent first)
japhb TimToady, I have an unrelated question re: your intent with a spec item, when you have a chance.
TimToady but the point of BUILD is to build the object, not to be a hook after it gets prebuilt 22:01
I don't care how Moose misuses the term
im2ee jnthn, i wanted to make connection via IO:Socket using .new method.
TimToady japhb: go ahead
masak TimToady: +1
im2ee (...) using .new method in my class 22:02
tadzik okay, I forget about Moose and BUILD then. What is the best way to have something done just after the object is good to go?
jnthn Well, back in b we had the default BUILD be the thing that applied defaults. This was terribly broken. Now BUILDALL does all those things. There's no default BUILD in nom; we only call the user-defined ones. And defaults, applying things from the init hash, etc are all done in BUILDALL - after BUILD.
TimToady use my verb
pmurias TimToady: +1 22:03
TimToady init-now() call-it-good() set-defaults() # I don't care
write a BUILD, call the verb, and do your own stuff
but it had better not depend on anything any subclass does
japhb TimToady, the spec says that command-line arguments will be passed through the val() function to attempt to convert them to non-Str classes as appropriate (integers to Int, etc.) before passing to Main. But it's not clear to me how *hard* val() tries to do this. In other words, does it try to convert strings that look like integers in radix notation? Does it change 'NaN' into a real NaN? etc.
jnthn TimToady: I can implement stuff like that. 22:04
japhb *MAIN
pmurias TimToady: so every BUILD will have to contain the special verb?
tadzik I once implemented a method AFTERPARTY which was called just after the object was fully initialized
TimToady every BUILD does already imply it 22:05
just at the end
tadzik pmurias: I read that as "it's not BUILD we're looking for"
jnthn TimToady: Are we expect to not re-do it if it was already done?
tadzik but seems like I read it wrong
jnthn TimToady: Tracking that could be annoying.
Though possible.
TimToady jnthn: yes, well, it touchs the attributes, so should just work :)
sorear TimToady: in my opinion, a method that is called after setting attributes is far more useful in most cases than what BUILD is currently specced to do
TimToady: what should I call it when I implement it? 22:06
pmurias can the BUILD choose not to set attributes?
jnthn pmurias: Under current spec, any attribute that BUILD doesn't touch gets its defaults applied, or takes a value from those passed to bless.
diakopter done() 22:07
tadzik sorear++
TimToady sorear: well, I think a special verb would be more optimizable than an extra dispatch
sorear is frustrated by the resistance the Perl 6 community has to back-stealing good ideas from Moose
masak sorear: operative word "good" :) 22:08
jnthn: it should be possible to tell from an attribute whether it's been initialized.
TimToady I only saw a valid use case for before and after this last summer for the first time
tadzik hell, people already workaround a lack of that
even in Rakudo's settings
masak jnthn: no tracking necessary.
TimToady: ooc, what was that use case?
sorear before and after: like callsame/nextsame, but better Huffmanization 22:09
jnthn masak: Well, it's more about avoding the work of re-checking.
TimToady knowing whether a subclass hid something on accident or on purpose
jnthn masak: But come of think of it, that's quite cheap these days anyway.
masak TimToady: ah. yes, makes sense. 22:10
TimToady japhb: yes, yes, and yes
masak japhb: think of val() as a side-effect-free eval() 22:11
22:11 buubot_backup joined
diakopter eval use Moose 22:12
buubot_backup diakopter: ERROR: Can't locate Moose.pm in @INC (@INC contains:) at (eval 20) line 2. BEGIN failed--compilation aborted at (eval 20) line 2.
japhb TimToady, wow, really? That's simultaneously cool and makes me worry that people will try to write file system utilities and find out that their filenames have become Num's. I guess as long as they perfectly roundtrip as strings, that's OK ... but radix-notated integers wouldn't roundtrip, neh?
TimToady they will if you use them as strings, that's why val() is trying hard to represent allomorphic literals 22:13
sorear japhb: values returned from &val are annotated with their original string form
in a 0 but True kind of way
japhb sorear, ah, interesting. 22:14
... I had a long question, but then I think I may have figured it out, lemme see: 22:16
So val() returns things that isa Num or Rat or what have you, but mix in roles that override stringification 22:17
sorear That is the idea, IIUC.
japhb interesting.
Is there already a decided name for that stringification-override role? 22:18
sorear figures it would be anonymous 22:19
jnthn Could just be an anonymous one
masak rakudo: my $val = 42 but role { method gist { "apfelstrudel" } }; say $val
p6eval rakudo aed9fe: OUTPUT«apfelstrudel␤»
masak \o/
japhb oooh
masak rakudo: my $val = 42 but role { method gist { "apfelstrudel" } }; say +$val
p6eval rakudo aed9fe: OUTPUT«apfelstrudel␤»
japhb OK, which methods do I need to override?
masak eh.
japhb darn
tadzik rakudo: my $val = 42 but role { method gist { "apfelstrudel" } }; print $val
p6eval rakudo aed9fe: OUTPUT«42»
jnthn masak: hehe
tadzik hah
jnthn masak: Numeric.Numeric returns self.
22:19 buubot_backup left
jnthn masak: It then .gists :) 22:20
Not so sure what we can do about that. It's not wrong, it is however a bit confusing. 22:21
TimToady rakudo: my $val = 42 but role { method Str { "apfelstrudel" } }; print $val
p6eval rakudo aed9fe: OUTPUT«apfelstrudel»
TimToady rakudo: my $val = 42 but role { method Str { "apfelstrudel" } }; print +$val
donri rakudo: my $val = 42 but "fourty two"; say $val
p6eval rakudo aed9fe: OUTPUT«apfelstrudel»
rakudo aed9fe: OUTPUT«No applicable candidates found to dispatch to for 'infix:<but>'. Available candidates are:␤:(Mu $obj, Positional @roles)␤:(Mu $obj, Mu $role)␤␤ in sub infix:<but> at src/gen/CORE.setting:7153␤ in <anon> at /tmp/ZC7lGe13C9:1␤ in <anon> at /tmp/ZC7lGe13C9:1␤␤»…
TimToady rakudo: my $val = 42 but role { method Str { "apfelstrudel" } }; print 0+$val 22:22
p6eval rakudo aed9fe: OUTPUT«42»
japhb Ewww. Having +$val and 0+$val be different is surprising.
donri why doesn't mine work
TimToady I thought recent spec did it by subclassing a Num on top of a Str, but I could be hallucinating 22:23
jnthn donri: Because it relies on a curious bit of S14 that didn't make it back into nom (yet).
donri oh :)
b: my $val = 42 but "fourty two"; say $val
p6eval b 1b7dd1: OUTPUT«42␤»
jnthn TimToady: I can't understand "subclassing a num on top of a Str"
s/can't/don't
donri b: my $val = 42 but "fourty two"; say ~$val
p6eval b 1b7dd1: OUTPUT«42␤»
jnthn ...I'm sure I'm capable of understanding it ;) 22:24
sorear japhb: gist and Str are probably the minimal set
jnthn donri: I fear that even *if* I implement the same thing in nom as master has, it'll only give you a .Str in the role, and say looks at .gist, so it may still not work out.
sorear japhb: 0+$val eqv $val has never been true, suppose $val = [1,2] 22:25
japhb: I'd find $val+1-1 !eqv $val+0 to be a more worrisome state of affai
wait I misundersood you nvm
japhb I didn't say '$val' and '0+$val'. That first one was '+$val'. That's different. 22:26
right
japhb slow and distracted
TimToady jnthn: I'm just groping for a way to make Str consider itself to be epsilon looser than Num in a multi dispatch
jnthn TimToady: Hmm. 22:27
TimToady so it'll pick the numeric methods over the string methods
jnthn TimToady: I don't have any ideas right off...I'll ponder it some.
TimToady but for anything that doesn't want numberic params, hits the Str type
this might also work out well for allomorphic scalars coming in from Perl 5 22:29
22:32 buubot_backup joined
TimToady I mean, we could fake it by declaring all our Str params as Cool right now, but that's overkill, I think 22:32
jnthn Yeah, it feels like there should be a better way 22:34
Does this work in the slightest if we flip it around somehow, and have it be a Str with something else mixed in?
"42" but role :: does Numeric { method Numeric() { 42 } }
TimToady doesn't necessarily help with dispatching to an Int param 22:35
jnthn oh wait, that may go terribly wrong...
Yeah.
duh. :)
There's always multiple inheritance :P 22:36
You encode the epsilon narrowness in the mro :)
But...I really don't like MI solutions.
Plus you can't add extra parents by mixing in.
Well, you could if you wrote a repr that allowed it I guess. 22:37
22:37 bazqux left
jnthn oh, there maybe is a way 22:37
TimToady there's a sense in which Cool *means* Stringy 22:38
in the TCL everything is a string sense 22:39
and the numeric types just optimize away the string representation when it's redundant
im2ee Good night! :) 22:40
TimToady o/
22:40 im2ee left
sorear ponders modifying the multiple dispatch system by the addition of cross-coercions between types 22:42
jnthn dies a bit inside every time he sees "modifying the multiple dispatch system" :/ 22:44
Though I wonder if something could just naturally fall out of coercion types here...somehow...
TimToady I always thought they'd just install two different entries into the list 22:45
one with the exact types, and one with the looser types
jnthn Would there be an ordering of any kind?
TimToady just according the ordering that already exists
jnthn OK 22:46
TimToady so you could think of multi foo (Int(Cool) $x) as either Int or Cool
sorear what do coercion types look like?
ah
TimToady and Int() defaults to something 22:47
jnthn TimToady: To say we introduced coercion types because the "as" was in the wrong place, I still have fun remembering which way around these go.
TimToady maybe Cool, maybe Any
jnthn :)
TimToady same as the coercion, Int($cool) ends up with an Int
jnthn oh! 22:48
Good way to remember it.
TimToady but I'm not sure this is related to allomorphic preferences, since those aren't necessarily about coercion at all 22:49
jnthn Ah. Dang.
TimToady except insofar as if you use a number as a string, it has to figure something out
so maybe the Str params should be Str(), and that loosens them a bit 22:50
but not under my two-different-entries approach, alas
22:51 geekosaur left
TimToady the other way to view Foo(Bar) types is that Foo is the only type visible inside the routine, and Bar is the only type that is visible to MMD 22:51
sorear masak: how much would we lose by allowing multi foo(Str $x) to accept a Num value? 22:52
TimToady basically, making Str a synonym for Cool 22:53
when used in a sig
seems like a hack
jnthn iiuc, it'd have to be Str() though?
TimToady well, Str() would make it official
jnthn Well, that's better than a hack. :) 22:54
Trouble is that if the default thing in the parens is any then
multi x(Str() $x) {
}
multi x(Cool $x) { } 22:55
The second is narrower.
*Any
TimToady so I'm thinking Str defaults to Cool or narrower 22:56
Str() rather
or maybe there's just some happy way to mark our coercions so they work this all out among themselves :)
but I think I like the explicit coercion-allowed-here sytnax 22:57
*nt
masak sorear: it sounds rather like Moose's type system.
TimToady otherwise you get stuck in the C++ trap of having all your semantics defined somewhere else...
well, Moose has been having to deal with allomorphism for a long time now :) 22:58
and if they've made strings a little looser than numeric types, I think that's the right direction 22:59
for when we're stuck in such a situation
which I'd only been envisioning for <1 2.3 foo> and MAIN situations, as well as input, but we can get allomorphic objects from other language, not the least of which is Perl 5 23:00
jnthn It's almost like they're a conjunction type
Well, or disjunction depending on how you see it.
TimToady well, it's one thing pretending to be different things according to context 23:01
jnthn True :)
TimToady a rather tagmemic thing, actually
Dogging a dog during the dog days of summer. 23:02
or maybe we just have a way for Str to say "I defer to Numeric when tied" 23:03
then siblings don't have to be forced into unnatural relationships 23:04
er...
23:04 geekosaur joined
sorear I propose to put that declaration in Numeric and spell it method coerce:<Str>() { self.Str } 23:04
Treat any coerce: method as defining a virtual superclass
23:05 mishin_ left
TimToady which gets back into the problems we had at the beginning with derivation, it seems like 23:06
and it just feels more like a me-first solution than a you-first solution 23:07
23:08 newsbad_com joined
TimToady not claiming to have thought it through, of course... 23:09
in fact, I'm gonna nap on it &
masak 'night, #perl6 23:15
23:16 masak left 23:17 Woodi left 23:18 Exodist joined, [Coke] left 23:20 [Coke] joined 23:24 pmurias left
jnthn sleep & 23:26
23:31 worr left 23:40 newsbad_com left
japhb nom: 42 but role {} 23:41
p6eval nom aed9fe: ( no output )
japhb nom: 1/2 but role {}
p6eval nom aed9fe: OUTPUT«Not enough positional parameters passed; got 1 but expected 3␤ in method BUILD at src/gen/CORE.setting:5304␤ in method BUILD_LEAST_DERIVED at src/gen/CORE.setting:573␤ in sub infix:<but> at src/gen/CORE.setting:7155␤ in <anon> at /tmp/MBijIIXbzu:1␤ in <anon> at /…
japhb ^^ Wheee, another bug
23:42 newsbad_com joined 23:45 wolfman2000 left
diakopter nom: 1/(2 but role {}) 23:53
p6eval nom aed9fe: ( no output )
23:54 benabik joined
dalek ecza/serialize: f1cb241 | sorear++ | / (4 files):
Implement sub definitions
 23:54
sorear o/ benabik
benabik o/ sorear
japhb diakopter, I had golfed that down from something that had a variable on the left of 'but' and a real role on the right; the same error pattern occurs, based on the type of the left argument.
colomon sorear: back and ready to hack. 23:58
sorear o/
benabik sorear: Anything interesting happening? :-)
sorear v10 release on Monday 23:59
and we just got out of a slightly heated discussion of MAIN coercisons

Powered by Cro and the Raku® Programming Language.

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

Thank you!