»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.
00:12 whiteknight joined 00:17 thou left 00:22 aloha left 00:24 replore joined 00:25 hugme left 00:26 Juerd left 00:27 [Coke] left, masak left, Util left, tadzik left, PerlJam left, thou joined 00:29 aloha joined 00:31 jaldhar left 00:32 Sarten-X left 00:39 abercrombie joined, Sarten-X joined 01:07 envi_ joined 01:09 jaldhar joined 01:16 Util joined, masak joined, [Coke] joined 01:17 PerlJam joined 01:20 Entonian left 01:29 Juerd joined 01:43 tadzik joined
japhb phenny ask moritz now that [Coke] has confirmed I have a CLA filed with TPF, what do I need to do next to apply for a Rakudo commitbit? 02:03
02:06 whiteknight left 02:40 packetknife joined 02:41 envi_ left 02:42 wolfman2000 joined 02:44 tokuhiro_ joined 03:25 cooper left 03:31 cooper joined
sorear good * #perl6 03:36
moritz good morning
sorear I need to find a way to /ignore comments directed to or from p6eval
03:36 japhb_ joined
sorear I'm doing the ?sorear|iecza search and it's mostly GlitchMr 03:36
japhb_ sorear, I'm at my Perl Mongers meeting, and someone just asked about bindings to native libraries from Perl 6, and I remembered someone had connected to GTK using niecza. Do you remember who it was, and (pushing my luck here) where that project is? 03:38
moritz japhb_: niecza repo, examples/gtk-* 03:39
03:39 Yappoko___ left, PZt left, apejens left, apejens joined
moritz japhb_: mostly mberends++ 03:39
japhb_ moritz++ # Thank you!
03:40 Yappoko___ joined, Trashlord left, domidumont left, lue left, domidumont1 joined
sorear needs to make time to drop by sandiego.pm some time 03:40
03:41 lue joined, Trashlord joined 03:42 thou left
abercrombie Does lines() read from $*ARGFILES or @*ARGS? 03:44
moritz $*ARGFILES 03:45
03:46 cooper left
moritz japhb_: as for the commit bit, you don't need to do anything. If pmichaud agrees, you'll get one. I've contacted him by email 03:46
abercrombie Thanks. But I found if I modify @*ARGS, then the subsequent lines() will read from what I set
03:46 cooper joined 03:47 packetknife left
moritz well, $*ARGFILES does take it list of filenames from @*ARGS (and if empty defaults to $*IN) 03:47
so the two are connected, and it's this connection you are seeing
japhb_ moritz, thank you again (re: commitbint) 03:48
*commit bit
abercrombie Gotcha. Thank you.
03:48 molaf joined 03:59 PZt joined 04:05 molaf left, alc joined 04:07 cooper left
PerlJam re commit bit for japhb_ -- +1 04:09
japhb_ PerlJam, thank you. :-) 04:10
04:11 soh_cah_toa_ left 04:27 abercrombie left 04:31 lue left, Yappoko___ left, apejens left, overrosy left, pochi left, japhb left 04:34 overrosy joined 04:37 lue joined, Yappoko___ joined, apejens joined, pochi joined, japhb joined 04:46 colomon left 04:49 colomon joined 05:06 wolfman2_ joined 05:08 wolfman2000 left 05:12 orafu left, orafu joined 05:13 uniejo joined 05:15 uniejo left 05:17 SHODAN joined 05:19 japhb_ left 05:38 daniel-s left 05:53 daniel-s joined 06:05 wtw joined
moritz japhb: welcome as a rakudo committer! 06:12
oops, accidentally sent the mail to rakudobug, not to pmichaud 06:13
sorear was wondering about that :D
How confused would perl6-ers be if niecza started producing files with names like Run.Test.dll instead of Test.dll? 06:14
moritz moritzfail
not more than that it generates .dll files at all :-)
.exe and .dll on linux take some time to get used to 06:17
japhb moritz, excellent, thanks! 06:18
06:20 koban joined
masak moritz: I forwarded Robert's reply about RT not finding anything on "given". 06:25
apparently, since "given" is a registered user, RT assumes that was what we were searching for and produces nothing.
moritz masak: thanks, seen it
masak also apparently, this is a "feature", but at least Robert puts it in quotes and says that it might change in later versions. 06:26
06:27 finanalyst joined
moritz I hope it does, it makes the search DOSable 06:27
masak how do you mean, "DOSable"? 06:28
finanalyst rakudo: class D {has $.a = 9; has $.b = $.a eq 0 or $.a gt 20 ?? $.a !! 20 }
p6eval rakudo 97aa73: OUTPUT«Use of uninitialized value in string context␤Use of uninitialized value in string context␤===SORRY!===␤error:imcc:syntax error, unexpected COMMA (',')␤ in file '(file unknown)' line 240␤␤»
masak finanalyst: 'eq' is for strings.
finanalyst rakudo: class E {has $.a = 9; has $.b = ($.a eq 0 or $.a gt 20) ?? $.a !! 20 }
p6eval rakudo 97aa73: ( no output )
masak finanalyst: 'or' has loose precedence, as opposed to '||' 06:29
moritz simply by registering the $n most common search terms as user names
finanalyst masak: i thought it was general and for numbers too
masak moritz: oh, I wasn't even thinking of it in terms of that; I just figgered it's in no way a feature and just Wrong.
finanalyst: what made you think that?
finanalyst any way, it works with brakets and when not in a class, but not in a class definition
masak: now you ask, I can't remember 06:30
masak rakudo: class E { has $.a = 9; has $.b = $.a }; say E.new.b
p6eval rakudo 97aa73: OUTPUT«9␤»
masak rakudo: class E { has $.a = 9; has $.b = $.a ?? $.a !! 20 }; say E.new.b
p6eval rakudo 97aa73: OUTPUT«9␤»
finanalyst masak: trying to keep up with changes is difficult
masak finanalyst: worksforme
finanalyst: yeah but in this case nothing changed :) 06:31
finanalyst masak: class E works, class D creates an error
masak finanalyst: 'eq' is for strings in perl 5 too.
finanalyst: that's because you used 'or'.
finanalyst: which has loose precedence.
that is, looser than '='
finanalyst masak: when the expression is used in a normal expression, I got what i expected 06:32
masak so it parses as '(has $.b = $.a eq 0) or $.a gt 20 ?? $.a !! 20'
finanalyst when in a class definition, i got a funny IMCC error
ah, now I see
masak finanalyst: if you get what you wanted with '=' and 'or', it's by luck and you're not testing all cases. 06:33
sorear o/ masak!
finanalyst masak: i'm not trying to test all cases. I got an error that was strange, so I tried to reproduce it
masak sorear: \o
finanalyst: as far as that goes, it's great. keep it up :) 06:34
finanalyst: I should be explicit about separating my thanks for your bug reporting from my critique of your choice of operators. :)
finanalyst masak: although my syntax was wrong, it seems to me that somehow the error should be trapped better 06:35
masak definitely.
sorear masak: how foolish would I be to merge /serialize in the weekend before the release?
masak: do you think I have enough users to care about waiting for it to stablize? :)
finanalyst masak: is == still the test operator for numbers, as in perl5?
masak sorear: not sure, but I'm willing to test it against my mid-sized application before you merge. I'm sure mberends et all will do the same. 06:36
finanalyst: yes.
finanalyst masak: thanx. for some reason I thought that == had been eliminated in favour of a more general operator
sorear masak: maybe we need something like plumage-smoke for Perl6land 06:37
masak finanalyst: '== < > <= >=' is for numbers, 'eq lt gt le gt' is for strings.
sorear: yes.
finanalyst: there are these interesting "phantom improvements" that people think have happened to Perl 6 but that haven't. yours is a bit unusual.
sorear finanalyst: it's sort of the other way around. Str/Num allomorphism means you have to be clear what comparison type you're using 06:38
masak the most common phantom improvement is where people believe that bareword keys in hashes are outlawed. we tell them "no, that's still fine" and they go "O RLY?"
sorear finanalyst: Perl 6's version of the allomorphism is a bit weaker and we have an === operator that can compare anything
masak Perl 6's allomorphism is cool. :) 06:39
finanalyst i'm at a stage where i understand perhaps over 50% of perl6, but nowhere near everything. I'm also anxious to use perl6 for all the cool things. But I keep getting tripped up by changes
sorear TimToady++ # introducing me to the concept of allomorphism. Took longer to "get" than, say, monads 06:40
I'm an old Perl5 hand, and I like to think of 42 and "42" as fundamentally the same, so I don't use === much
finanalyst so when i've misunderstood something, i'm not sure whether its because i never did understand it, or i understood it once, and the language changed
sorear I know my code isn't careful to keep them straight
It's possible Perl 6 is ultimately a simpler language than Perl 5. It's certainly much easier to parse. 06:41
finanalyst allomorphism????? wooo what?? where can i read up on it
sorear I've long since forgotten, if I got you a link it would be from Google 06:42
finanalyst ok
masak finanalyst: you'll find that it's a fancy term for something that you already know. :) "scalar" is the allomorphic thing in Perl 5 for strings, ints, floats. 06:46
sorear well yeah 06:48
all good ideas are the same as things I already know ;) but knowing the terms helps a lot just the same. 06:49
masak "allomorphic, n, Linguistics. one of the alternate contextually determined phonological shapes of a morpheme, as I<en> in I<oxen>, which is an allomorph of the English plural morpheme."
sorear: agreed.
sorear oh hey 06:50
masak oh, sorry. that was the entry for "allomorph".
sorear plunks on'yomi/kun'yomi into a more general mental bucket.
masak I wonder what the exact relation between allomorphy and type punning is. 06:57
07:26 xinming joined
masak uses @AoA»[0] in real code and feels pretty great about it 07:36
07:37 wk left
moritz www.perlmonks.org/?node_id=933807 that rakudo dog would be called "Null PMC Access", no? :-) 08:00
08:03 agentzh joined
sorear InvalidCastException here I'm sure :-) 08:06
moritz aye :-)
niecza: +'a'
p6eval niecza v10-61-gbacff10: OUTPUT«␤Unhandled Exception: System.FormatException: Unknown char: a␤ at System.Double.Parse (System.String s, NumberStyles style, IFormatProvider provider) [0x00000] in <filename unknown>:0 ␤ at System.Double.Parse (System.String s, IFormatProvider provider) […
moritz or this one as second candidate :-)
08:08 mj41_nb joined, mj41 joined 08:16 koban left
dalek ast: be0ff5f | moritz++ | S12-construction/new.t:
test nextwith() in .new method
 08:22
moritz rakudo: try { die 1; CATCH { die 2 } } 08:23
p6eval rakudo 97aa73: OUTPUT«2␤ in block <anon> at /tmp/ke9dCRSykE:1␤ in block <anon> at /tmp/ke9dCRSykE:1␤ in block <anon> at /tmp/ke9dCRSykE:1␤ in <anon> at /tmp/ke9dCRSykE:1␤»
08:26 dakkar joined
dalek ast: c3c9fb7 | moritz++ | S04-exception-handlers/catch.t:
test throwing exceptions in CATCH blocks
 08:27
moritz closes another 2-year old bug
nom: ...('foo') 08:28
p6eval nom 97aa73: ( no output )
moritz nom: !!!('foo')
p6eval nom 97aa73: OUTPUT«Stub code executed␤ in block <anon> at /tmp/kKGKvghO_P:1␤ in <anon> at /tmp/kKGKvghO_P:1␤»
moritz rakudo: try { die 'omg!'; CATCH { say '$!'; } } 08:38
p6eval rakudo 97aa73: OUTPUT«$!␤omg!␤ in block <anon> at /tmp/oCb_62sHiz:1␤ in block <anon> at /tmp/oCb_62sHiz:1␤ in <anon> at /tmp/oCb_62sHiz:1␤»
dalek ecza/serialize: 9083142 | sorear++ | / (7 files):
Switch to single-AppDomain execution

As I thought, remote method invocation seems to have been the major bottleneck in /serialize. Using a lighter-weight isolation protocol has improved performance:
  (time mono-sgen run/Niecza.exe -C CORE)
master: user 0m30.898s before: user 1m26.453s now: user 0m29.246s
The new procedure uses only Assembly-level isolation: runtime assemblies are renamed to have names starting with Run.; so CORE.dll is the compiler's CORE and Run.CORE.dll is the user's. This seemed preferable over the reverse to avoid name clashes with perversely named user modules.		 08:45
sorear this is the "one big optimization" I was talking about yesterday. glad to see numbers confirming my hunches
moritz \o/ 08:46
rakudo: my %a; %a{ CATCH{} }
p6eval rakudo 97aa73: ( no output )
sorear interestingly, the time required for 'mono-sgen run/Niecza.exe -e "say 2 + 2"' is statistically significant faster (3.8s vs 6.0s) 08:47
diakopter: ^^^
moritz what about niecza startup time?
dalek ast: af82544 | moritz++ | S04-exception-handlers/catch.t:
RT #80864
 08:48
sorear moritz: won't be affected until I redo the bootstrap at the end of the month
moritz sorear: ok
perl6: do { CATCH { }; CATCH { } }
p6eval niecza v10-61-gbacff10: OUTPUT«===SORRY!===␤␤Limit one CATCH phaser per block, please. at /tmp/udH7y94R5a line 1:␤------> do { CATCH { }; CATCH { } ⏏}␤␤␤Unhandled Exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 766 (CORE …
..rakudo 97aa73: OUTPUT«===SORRY!===␤only one CATCH block allowed at line 1, near "}"␤»
..pugs b927740: ( no output )
sorear I now have the challenge of getting "make reboot" working again before Oct31
moritz sorear: that error message doesn't match my understanding of English 08:49
sorear moritz: it's a pun off a phrase commonly seen on advertisements in the USA - Great deal!!! Limit one $ITEM per customer. 08:50
moritz ie I'm missing either a verb (if Limit is a noun), or a subject (if 'Limit' is a verb)
sorear the implied verb is "is"
moritz sorear: ah
sorear aka "sorear being too cute"
dalek ast: 7bf7582 | moritz++ | S04-exception-handlers/catch.t:
only one CATCH per block, please
 08:52
moritz is pleased to announce that all but one rakudobugs related to 'try' or 'CATCH' have been closed. mls_++ 08:54
sorear -> sleeeep 08:55
jnthn moritz: \o/ 08:56
mls_++ indeed!
moritz: Which one remains? 08:57
moritz jnthn: one related to warn 08:58
jnthn: which currently just cheats, and writes to $*ERR
jnthn ah, that may be resolved by an mls patch I didn't get to yet.
moritz nom: -> { return 1 } 09:00
p6eval nom 97aa73: ( no output )
moritz shouldn't we be able to catch that at compile time, since return is supposed to be lexotic? 09:01
nom: -> { return 1 }.()
p6eval nom 97aa73: OUTPUT«Attempt to return outside of any Routine␤ in block <anon> at /tmp/buE64W0CCC:1␤ in block <anon> at /tmp/buE64W0CCC:1␤ in <anon> at /tmp/buE64W0CCC:1␤»
jnthn moritz: We could 09:04
moritz otoh ... return isn't special syntax, so it's just a call 09:05
moritz did want to override return occasionally
erm, on exactly one occasion so far
nom: sub f { my sub return($x) { nextwith(2*x) }; return 5 }; say f 09:06
p6eval nom 97aa73: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&x' called (line 1)␤»
moritz nom: sub f { my sub return($x) { nextwith(2*$x) }; return 5 }; say f
p6eval nom 97aa73: OUTPUT«No dispatcher in scope␤ in sub return at /tmp/r_CL9MbfVW:1␤ in sub f at /tmp/r_CL9MbfVW:1␤ in block <anon> at /tmp/r_CL9MbfVW:1␤ in <anon> at /tmp/r_CL9MbfVW:1␤»
moritz ah, because &return isn't a proper sub, but rather defined as my &return := BLOCK 09:07
jnthn Well, also because...where are you going to nextwith to in a sub? :)
(an only sub, that is)
moritz right :-)
should be more like &OUTER::return() 09:08
jnthn oh, I see what you were trying to do 09:10
CORE::return perhaps :)
moritz is that implemented?
nom: say CORE 09:11
p6eval nom 97aa73: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&CORE' called (line 1)␤»
moritz nom: say CORE::return
p6eval nom 97aa73: OUTPUT«Could not find symbol 'CORE::&return'␤ in block <anon> at /tmp/wVyNz17DDv:1␤ in <anon> at /tmp/wVyNz17DDv:1␤»
moritz seems not :-)
jnthn ah, no, didn't do that yet
moritz nom: sub f { my sub return($x) { pir::find_outer_lex__PS('&return').(2*$x) }; return 5 }; say f 09:12
p6eval nom 97aa73: OUTPUT«Use of uninitialized value in string context␤Use of uninitialized value in string context␤===SORRY!===␤error:imcc:syntax error, unexpected PREG, expecting '(' ('$P100')␤ in file '(file unknown)' line 77177405␤␤»
jnthn That won't work because return in outer is...the return you're currently defining :)
You'd have to look 2 frames out. :)
moritz nom: sub f { my sub return($x) { pir::find_caller_lex__PS('&return').(2*$x) }; return 5 }; say f
p6eval nom 97aa73: OUTPUT«maximum recursion depth exceeded␤ in sub infix:<*> at src/gen/CORE.setting:2306␤ in sub return at /tmp/b7ZCXDD4yx:1␤ in sub return at /tmp/b7ZCXDD4yx:1␤ in sub return at /tmp/b7ZCXDD4yx:1␤ in sub return at /tmp/b7ZCXDD4yx:1␤ in sub return at /tmp/b7ZCXDD4yx:1␤ … 09:13
moritz nom: sub f { my sub return($x) { pir::find_caller_lex__PS('&return').(10) }; return 5 }; say f
p6eval nom 97aa73: OUTPUT«(signal SEGV)maximum recursion depth exceeded␤»
moritz nom: sub f { my sub return($x) { callframe(1).my<&return>.(10) }; return 5 }; say f
p6eval nom 97aa73: OUTPUT«Method 'postcircumfix:<( )>' not found for invocant of class 'Any'␤ in <anon> at src/gen/Metamodel.pm:3177␤ in sub return at /tmp/NJ0ewLo9ph:1␤ in sub return at /tmp/NJ0ewLo9ph:1␤ in sub f at /tmp/NJ0ewLo9ph:1␤ in block <anon> at /tmp/NJ0ewLo9ph:1␤ in <anon> at …
moritz nom: sub f { my sub return($x) { callframe(2).my<&return>.(10) }; return 5 }; say f 09:14
p6eval nom 97aa73: OUTPUT«Method 'postcircumfix:<( )>' not found for invocant of class 'Any'␤ in <anon> at src/gen/Metamodel.pm:3177␤ in sub return at /tmp/WCTRCwHqoW:1␤ in sub f at /tmp/WCTRCwHqoW:1␤ in block <anon> at /tmp/WCTRCwHqoW:1␤ in <anon> at /tmp/WCTRCwHqoW:1␤»
moritz meh
jnthn one more :)
moritz nom: sub f { my sub return($x) { callframe(3).my<&return>.(10) }; return 5 }; say f
p6eval nom 97aa73: OUTPUT«Method 'postcircumfix:<( )>' not found for invocant of class 'Any'␤ in <anon> at src/gen/Metamodel.pm:3177␤ in sub return at /tmp/RxihKxF1K9:1␤ in sub f at /tmp/RxihKxF1K9:1␤ in block <anon> at /tmp/RxihKxF1K9:1␤ in <anon> at /tmp/RxihKxF1K9:1␤»
jnthn hmm.
oh, wait...
is callframe lexical or caller chain?
moritz caller chain
jnthn Well, the setting isn't in the caller chain :) 09:15
It's in the lexical one.
moritz right
jnthn should implement CORE, OUTER etc.
mls_ nom: sub f { my sub return($x) { pir::find_lex_skip_current__PS('&return').(10) }; return 5 }; 09:22
p6eval nom 97aa73: ( no output )
mls_ (morning!)
09:23 envi_ joined, Vlavv_ joined
moritz nom: sub f { my sub return($x) { pir::find_lex_skip_current__PS('&return').(10) }; return 5 }; say f 09:23
p6eval nom 97aa73: OUTPUT«(signal SEGV)maximum recursion depth exceeded␤»
moritz still finds my return, not the outer
nom: sub f { my $r := pir::find_lex_skip_current__PS('&return'); my sub return($x) {$r(2 * $x) }; return 5 }; say f 09:24
p6eval nom 97aa73: OUTPUT«10␤»
moritz \o/
09:25 Vlavv` left
moritz nom: class A is int { has $.x } 09:28
p6eval nom 97aa73: OUTPUT«===SORRY!===␤Method 'parrot_vtable_mappings' not found for invocant of class 'Perl6::Metamodel::NativeHOW'␤»
jnthn Well, that's LTA... 09:29
jnthn wonders if you should be allowed to do that.
moritz no 09:30
jnthn nom: native A is int { }
p6eval nom 97aa73: ( no output )
jnthn nom: native A is int { has $.x }
p6eval nom 97aa73: OUTPUT«===SORRY!===␤A native cannot have attributes at line 1, near " }"␤»
moritz I remember that deriving from native types may not change storage
jnthn right.
It should fail but currently it fails for the wrong reasons.
09:31 alc left 09:32 kfo_ joined
moritz akshually I found a small fail in the error message that is supposed to show up, and wanted to demonstrate that :-) 09:33
I guess I'll just commit it (after testing) 09:34
nqp::bindattr(1, '$!foo', int, 3)
nom: nqp::bindattr(1, '$!foo', int, 3) 09:35
p6eval nom 97aa73: OUTPUT«get_string() not implemented in class 'int'␤ in block <anon> at /tmp/4cCrXBqNSA:1␤ in <anon> at /tmp/4cCrXBqNSA:1␤»
09:35 kfo left
moritz nom: nqp::bindattr(1, int, '$!foo', 3) 09:35
p6eval nom 97aa73: OUTPUT«Can not bind non-existant attribute '$!foo' on class 'int'␤ in block <anon> at /tmp/lPj9ip4WsN:1␤ in <anon> at /tmp/lPj9ip4WsN:1␤»
dalek p: fd63797 | moritz++ | src/6model/reprs/P6 (2 files):
fix copy&pastos
 09:36
moritz nom: int.^add_attribute(int, 1) 09:37
p6eval nom 97aa73: OUTPUT«Method 'dispatch:<.^>' not found for invocant of class 'int'␤ in block <anon> at /tmp/bx5Gs5aHcn:1␤ in <anon> at /tmp/bx5Gs5aHcn:1␤»
moritz fails to show the failing fail :-)
masak I had a "caught exceptions too happily" situation today. one day it will sink in that I can't just check for failure, period. 09:39
moritz: the sooner your exception work lands, the better :)
moritz masak: agreed 09:40
but it must be done before it can land :/
nom: say X::Base.new
p6eval nom 97aa73: OUTPUT«Base.new(message => Mu)␤»
moritz has been plowing his way through nqp/src/6model/reprs/ 09:46
and all the .h files and most .c files look pretty understandable
but P6opaque.c... wow.
jnthn :) 09:48
moritz: Well, it is basically computing C structure layouts along its way :)
Then making sure we can efficiently GC-mark the appropriate parts of them. :)
mls_ it creates a entry->offset table
jnthn It's more fiddly than hard, really. 09:49
moritz jnthn: I noticed the struct building. Which is why it does the crazy casting to char and then back to a void pointer and then back to a PMC pointer
jnthn crazy code is crazy 09:50
mls_ (btw, why is the "hint" not simply called "slot"?)
jnthn mls_: We may not always be able to use it. 09:51
moritz hysterical raisins, it seems
jnthn No, I still find hint fairly sane
mls_ yes, but in that case UNKNOWN_SLOT (aka -1) 09:52
jnthn It conveys "here's a clue that you may be able to use to lookup faster"
mls_: No, it's more than that.
mls_: We can only rely on those to work out in SI situations.
moritz jnthn: so far it complains that the attribute doesn't exist if hint == 0
does that mean that some MI situations are still NYI?
jnthn No 09:53
MI works
It just means we can't do that particular optimization in the case of MI.
mls_ see: slot = hint >= 0 && !(repr_data->mi) ? hint : ...
jnthn MI is still way better now than in the original way I was going to do P6opaque though :) 09:54
moritz ah, I see(-ish) 09:55
in MI we have the additional try_get_slot call
jnthn nod
tadzik jnthn: btw, did you notice the time jump in settings compilation/spectests after the optimizer merge?
jnthn tadzik: Yeah 09:56
tadzik: The spectests are about the worst case for the optimizer. 09:57
tadzik: No real loops or anything.
tadzik: Setting compilation - it's worth the time spent there.
tadzik oh, so that's the reason for what I noticed recently
sure
jnthn yeah but...for me it's not actually a huge jump.
tadzik :)
I can imagine
jnthn Well but I mean as a percentage. 09:58
tadzik hmm
jnthn It wasn't like "add optimizer, now it's twice as slow"
tadzik for me, it was
moritz eagerly awaits his 12-core $work machine
jnthn OK, that's...odd. 09:59
jnthn wonders why there are such different relative differences.
09:59 que joined
tadzik irclog.perlgeek.de/perl6/2011-10-19#i_4590435 09:59
que hi perl6 10:00
10:00 replore left, envi_ left
tadzik there is a possibility that there were many, many more tests though :) 10:00
hello que
10:00 envi_ joined
tadzik I should check out flussence++'s graphs from those dates 10:00
que I have a small question about mutable immutable
tadzik that's like a month
que my Int $i =5;$i =6; $i.say ; #6 10:01
moritz hopes that immutables are not mutable :-)
tadzik well, it's not that dramatic it seems (number of tests)
jnthn tadzik: S05-rx/mass.t can be time consuming
lunch - bbs
tadzik . o O ( launch ) 10:02
que Int is a immutables type ,but it mutable
moritz que: the variable $i is mutable
nom: 5 = 6
p6eval nom 97aa73: OUTPUT«===SORRY!===␤Method 'lvalue' not found for invocant of class 'Float'␤»
moritz wow, didn't we have better error messages for that?
que $i is Int object?
tadzik yes 10:03
moritz que: $i is a variable which holds an Int object
nom: my $i := 5; $i = 6
que Int is immutables?
tadzik hmm
p6eval nom 97aa73: OUTPUT«Cannot assign to a non-container␤ in block <anon> at /tmp/G3JOnQW7KY:1␤ in <anon> at /tmp/G3JOnQW7KY:1␤»
tadzik true. Int is immutable, $i can hold one immutable object, and then another
the object doesn't get changed, the value of $i does
it's not like it's a constant 10:04
que can you show me a example ? immutable's example
moritz num: my $i := 3; $i = 6 10:05
nom: my $i := 3; $i = 6
tadzik num num num :)
p6eval nom 97aa73: OUTPUT«Cannot assign to a non-container␤ in block <anon> at /tmp/QPGSxc4U7M:1␤ in <anon> at /tmp/QPGSxc4U7M:1␤»
moritz que: there you go
que: $i is bound to an Int, so you can't assign a new value to it
que thanks moritz
moritz que: does that make sense to you?
nom: 5++
p6eval nom 97aa73: OUTPUT«Cannot assign to a non-container␤ in sub postfix:<++> at src/gen/CORE.setting:2270␤ in block <anon> at /tmp/mWTnM2GQzP:1␤ in <anon> at /tmp/mWTnM2GQzP:1␤»
moritz another example
que not ?
moritz can't parse that line 10:06
que: the other important thing is that there are no methods in class Int which modify an Int instance
tadzik Date is immutable, you can't change the month. But keeping a Date in $a, you can assign a different Date to $a, and Date does not get changed, $a does. It's the same with Int, iiuc
moritz compare 10:07
nom: my @a = 3, 4; @a.push(5); say @a # the push mutates @a
p6eval nom 97aa73: OUTPUT«3 4 5␤»
moritz there aren't any comparable methods in Int
10:08 lue left 10:09 saaki left 10:10 MayDaniel joined
que moritz:my $i := 5 ,$i must be 5 can't assign it 6,but can the example show Int is immutable? 10:13
moritz que: no. The absense of a feature (mutability) can never be proven by examples 10:14
que: I can just show you examples how other objects mutate, and show that they don't work for Int. Maybe that convinces you, maybe not
que show me 10:15
moritz I did 10:16
10:16 envi_ left
que Oh?? thanks 10:16
10:17 [Sec] joined
moritz masak: re gist.github.com/1293853, in role R[ {{{$foo}}}] { }, what does $foo contain? the AST of a parameter list? 10:17
10:18 saaki joined 10:19 envi_ joined 10:21 lue joined 10:24 grondilu joined
grondilu rakudo: my int $n; say $n = 42; 10:24
p6eval rakudo 97aa73: OUTPUT«Method 'gist' not found for invocant of class 'Integer'␤ in sub say at src/gen/CORE.setting:5281␤ in block <anon> at /tmp/0oeqA8MH_x:1␤ in <anon> at /tmp/0oeqA8MH_x:1␤»
moritz and if yes, does that mean that arbitrary parts of syntax can be replaced with a {{{ }}} block? 10:25
std: my $name; quasi { class {{{$name}}} { } }
grondilu I thought native integer were implemented?
p6eval std 580b69a: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/kVz_Nl_DW5 line 1:␤------> my $name; quasi { class {{{$name}}} ⏏{ } }␤ expecting infix or meta-infix␤Parse f…
moritz grondilu: yes, just probably not every case
nom: my int $x = 3; say $x
p6eval nom 97aa73: OUTPUT«3␤»
moritz nom: my int $x = 3; say $x = 4
p6eval nom 97aa73: OUTPUT«Method 'gist' not found for invocant of class 'Integer'␤ in sub say at src/gen/CORE.setting:5281␤ in block <anon> at /tmp/kjpskFwLvE:1␤ in <anon> at /tmp/kjpskFwLvE:1␤»
grondilu oh, ok 10:26
moritz that's a parrot Integer leaking through
jnthn: ^^
nom: sub f($x) { $x * 2 }; say f(my int $a = 3) 10:27
p6eval nom 97aa73: OUTPUT«6␤»
moritz nom: sub f($x) { $x * 2 }; my int $a; say f($a = 3)
p6eval nom 97aa73: OUTPUT«6␤»
moritz nom: sub f($x) { $x * 2 }; my int $a; say f(($a = 3).gist)
p6eval nom 97aa73: OUTPUT«Method 'gist' not found for invocant of class 'Integer'␤ in block <anon> at /tmp/S3WY55l0dR:1␤ in <anon> at /tmp/S3WY55l0dR:1␤»
moritz nom: my int $x = 2; $x.gist
p6eval nom 97aa73: ( no output )
moritz nom: my int $x = 2; say $x.gist
p6eval nom 97aa73: OUTPUT«2␤»
moritz sighs. We had such a nice insulation between parrot and rakudo types, and now they leak through when using native types :/ 10:30
10:35 alc joined
dalek ast: 4cafc7d | moritz++ | S02-types/native.t:
RT #102244, native ints leaking through in assginment
 10:35
10:39 Trashlord left
jnthn moritz: It's just PAST::Compiler trying to be helpful and failing. 10:48
moritz: File ticket (if you didn't already); it's probably something in assign_op 10:49
(Like it doesn't call box_native_if_needed on the PAST tree it produces)
Another thing to deal with in QAST - let the language configure its box/unbox regime. 10:51
10:56 grondilu left
masak would like a guide for pronouncing "QAST" 10:57
jnthn
.oO( d e qast )
 11:01
masak :)
11:02 que left 11:09 ab5tract joined
masak moritz: I think a lot of people who have the wrong expectations on junctions really want the amb special form from Lisp: community.schemewiki.org/?amb -- and IIRC a year or so back nothingmuch stopped by, had those expectations, and then implemented amb as a CPAN module. 11:09
maybe that should go in the FAQ :)
ah, here's his blog post about it: blog.woobling.org/2009/08/abstracti...guity.html 11:14
but I seem to misremember about the CPAN module.
perl6: sub foo { our $a = 5 }; sub bar { say our $a }; bar; foo; bar 11:17
p6eval rakudo 97aa73, niecza v10-61-gbacff10: OUTPUT«Any()␤5␤»
..pugs b927740: OUTPUT«␤␤»
masak \o/
...wonder what model Pugs has for 'our' variables...
11:34 finanalyst left 11:36 DataLinkDKT joined 11:41 Patterner left
moritz jnthn: ticket filed 11:42
11:43 Psyche^ joined, Psyche^ is now known as Patterner 11:56 JimmyZ joined
moritz masak: fix it :-) 11:58
masak Pugs? yeah, in time... :) 11:59
I still haven't dived into setting up a Pugs development environment, but thanks to ingys explorations and yours, it feels quite approachable now.
12:03 que joined, replore_ joined
que moritz: can you give me an example about mutable 12:04
moritz que: did you like my example with .push on the Array?
que I need to read it again 12:05
moritz wonders why he does all the typing when his examples are ignored anyway
masak hugs moritz
moritz thanks 12:06
masak moritz: just saw your question at irclog.perlgeek.de/perl6/2011-10-26#i_4616492 12:07
moritz: when I wrote it I pictured it as one single parameter, not a parameter list. but yes, AST.
moritz: I think it makes more sense to pass in lists of ASTs, not ASTs of lists, into macros.
moritz masak: now I'm mightily confused
masak let's hear it. 12:08
moritz masak: because I thought that you can only substitute terms with {{{ $var }}} holes
12:08 colomon left
masak aye. (btw, I prefer "placeholder" to "hole" nowadays) 12:08
moritz std: my $op; quasi { 1 {{{ $op }}} 2 }
p6eval std 580b69a: OUTPUT«===SORRY!===␤Unexpected block in infix position (two terms in a row, or previous statement missing semicolon?) at /tmp/g5MaPfOhaW line 1:␤------> my $op; quasi { 1 ⏏{{{ $op }}} 2 }␤ expecting infix or meta-infix␤Parse failed␤FA…
masak right.
that will never work.
moritz masak: but the inside of role R[...] is parsed as a signature, not as a list of terms 12:09
masak almost right.
12:09 Trashlord joined
masak it *is* parsed as a signature. 12:09
but terms are part of a signature, in the form of type names.
so it's OK there.
moritz std: my $thing; quasi { role R[{{{$thing}}}] { } } 12:10
p6eval std 580b69a: OUTPUT«===SORRY!===␤Unable to parse generic role at /tmp/FDHAuBs39A line 1:␤------> my $thing; quasi { role R⏏[{{{$thing}}}] { } }␤Couldn't find final ']'; gave up at /tmp/FDHAuBs39A line 1:␤------> my $thing; quasi { role R[⏏…
moritz seems not very OK there
que moritz:another small question
> my int $x =5 Use of uninitialized value in string context Use of uninitialized value in string context Use of uninitialized value in string context Use of uninitialized value in string context ===SORRY!=== Lexical '$x' is of wrong register type in lexical lookup
masak moritz: I don't think STD's quasi parsing handles that.
moritz masak: even if you could, that would only allow injection of constraints, not of parameters, right? 12:11
masak moritz: also, I might simply be wrong about terms being OK there.
moritz waits for the question
masak moritz: right, but the rest of the signature could be put outside of the placeholder as usual.
moritz oh, it seems that the REPL doesn't like native lexicals yet 12:12
masak: well, if terms are allowed, than only in the form of multi fib(0) { 1 }
masak: so as a constraint, not as a declaration 12:13
12:13 pernatiy joined
moritz so I don't see a big problem here 12:13
(hey, that's new, isn't it? *I* don't see a problem with something macro-related where masak sees a problem :-) 12:14
masak :-P 12:15
I'm not sure what arbitrary distinction you draw between constraints-as-types and constraints-as-values here... 12:16
both are terms.
12:16 molaf joined
que thanks 12:16
12:16 que left
moritz que: I've sent a bug report 12:16
masak and in both cases, we're prevented from actually creating the role until macro-call o'clock. 12:17
12:17 alc left
moritz masak: constraints-as-types are optionally followed by *another* term-like thing (the parameter), so they need special parsing 12:17
hm, are we? do we need the constraints on an anonymous formal parameter to build the role? 12:18
in analogy: sub f(0) { the body here doesn't need to know about the 0 }
12:19 alc joined
masak true, but there are conceptual problems with building the signature (and therefore the sub) before we've filled in the placeholders. 12:20
these are immutable things, so it's hard to fake it and go back and fill in afterwards.
moritz right
jnthn But we don't parse term in the sub f(0) { case 12:21
We parse value.
masak clearly more research is needed in this area.
right, it can't be term because that would get '$x' wrong.
moritz so by current understanding that means that no quasi placeholder can appear in the signature, except in where BLOCK clauses maybe 12:22
jnthn Not quite true 12:23
Defaults are parsed as term.
moritz std: my $h; quasi { sub f($x = {{{$h}}}) { say $x } } 12:24
p6eval std 580b69a: OUTPUT«ok 00:01 133m␤»
JimmyZ nom: my num $num; say $num++; 12:25
p6eval nom 97aa73: OUTPUT«Cannot assign to a non-container␤ in sub postfix:<++> at src/gen/CORE.setting:2697␤ in block <anon> at /tmp/snsnzj7U8Z:1␤ in <anon> at /tmp/snsnzj7U8Z:1␤»
jnthn std: my $h; quasi { sub f({{{$h}}}) { say $x } }
p6eval std 580b69a: OUTPUT«===SORRY!===␤Unable to parse signature at /tmp/n3DtpymxwY line 1:␤------> my $h; quasi { sub f⏏({{{$h}}}) { say $x } }␤Couldn't find final ')'; gave up at /tmp/n3DtpymxwY line 1:␤------> my $h; quasi { sub f(⏏{{{$h}…
jnthn Right, that's what I was expecting.
(sorry if you already did it - I kinda skim-backlogged after a long meeting :))
JimmyZ: rw native parameters nyi 12:26
(so we can't have native ++ yet)
moritz jnthn: are they even specced?
jnthn moritz: Well, they aren't outlawed by the spec... :)
moritz isn't sure how much sense they make 12:27
jnthn Well, I'm not sure saying "you can't $i++ if $i is a native int" is going to fly
moritz neither, but OTOH "we want that feature, so make it happen" makes sense and is possible in all cases 12:28
the very point of int (vs. Int) is that it does *not* pass a reference, but the value 12:29
jnthn yeah
Maybe it only works in the static case.
moritz static case?
jnthn e.g. we have to statically know it's going to bind to an rw parameter
As in
moritz ah
jnthn If you don't statically know what you're calling, it won't work out. 12:30
JimmyZ $num is not NaN, so what will $num++ be？
1？
jnthn heh
moritz JimmyZ: the same as NaN + 1
jnthn NaN ;)
moritz perl6: say NaN + 1
p6eval pugs b927740, rakudo 97aa73, niecza v10-61-gbacff10: OUTPUT«NaN␤»
JimmyZ I'm curious, why my int $int default is 0, and my num $n default is NaN 12:32
moritz JimmyZ: the idea is to mimic "undefined" as good as possibles
JimmyZ: but 'int' doesn't have anything close to "undefined", so something else must be used. 0 is an obvious choice 12:33
JimmyZ maybe NaI
moritz JimmyZ: but floating point numbers support NaN, so it makes sense to use them
JimmyZ: how would you store that in an integer register of your CPU?
JimmyZ perl6: say NaN ~ 1 + 1; 12:38
p6eval pugs b927740, rakudo 97aa73, niecza v10-61-gbacff10: OUTPUT«NaN2␤»
12:38 benabik left
dalek kudo/nom: cbc652c | moritz++ | docs/ChangeLog:
add some ChangeLog entries
 12:38
flussence moritz: ever seen how javascript does it? it's scary :)
moritz flussence: javascript, or v8? 12:39
flussence the one I read was about mozilla's implementation - they put the type data and the value in a single register...
moritz iirc using the highest bit to indicate "not a real int, interpret as a reference" or something
masak yuck. 12:40
JimmyZ nom: my Num $num; say $num++ 12:41
p6eval nom 97aa73: OUTPUT«0␤»
mux GHC (haskell) does something similar for optimisation purposes, albeint in a totally different context obviously
(taking one or two bits our of an int) 12:42
JimmyZ my Num $num default is 0?
moritz JimmyZ: no
jnthn Default is the type object there.
moritz nom: my Num $num; say $num.perl
p6eval nom 97aa73: OUTPUT«Num␤»
jnthn If you succ on a Num, it becomes a 1
.oO( that sounded weird out loud )
 12:43
JimmyZ num $num ++ is different from Num $num ++
moritz if you left out that 'on', it would've been much less weird :-)
JimmyZ: num is different from Num in that num-typed variables can't hold a type object
not sure if NaN is the best default though
JimmyZ nom: my Num $num = NaN; say $num++ 12:44
p6eval nom 97aa73: OUTPUT«NaN␤»
moritz nom: say num ~~ num
p6eval nom 97aa73: OUTPUT«Method 'ACCEPTS' not found for invocant of class 'num'␤ in block <anon> at /tmp/X4fqoPFeNW:1␤ in <anon> at /tmp/X4fqoPFeNW:1␤»
moritz nom: say num ~~ Num
p6eval nom 97aa73: OUTPUT«Bool::False␤»
moritz nom: say num ~~ Mu
p6eval nom 97aa73: OUTPUT«Bool::False␤»
moritz woah.
JimmyZ nom: my Num $num = NaN; say $num++; say $num++; say $num++
p6eval nom 97aa73: OUTPUT«NaN␤NaN␤NaN␤»
JimmyZ nom: my Num $num; say $num++; say $num++; say $num++ 12:45
p6eval nom 97aa73: OUTPUT«0␤1␤2␤»
JimmyZ nom: my num $num; say $num + 1； say $num + 2; 12:46
p6eval nom 97aa73: OUTPUT«===SORRY!===␤Confused at line 1, near "say $num +"␤»
jnthn nom: say int ~~ Mu
p6eval nom 97aa73: OUTPUT«Bool::False␤»
jnthn hm :)
JimmyZ nom: my num $num; say $num+1； say $num+2;
p6eval nom 97aa73: OUTPUT«===SORRY!===␤Confused at line 1, near "say $num+1"␤»
jnthn Well, it's a native, not an object but... :)
moritz nom: say num.defined
p6eval nom 97aa73: OUTPUT«Method 'defined' not found for invocant of class 'num'␤ in block <anon> at /tmp/IlMbHkpIGO:1␤ in <anon> at /tmp/IlMbHkpIGO:1␤»
jnthn Yeah, that's wrong
nom: say num
p6eval nom 97aa73: OUTPUT«Method 'gist' not found for invocant of class 'num'␤ in sub say at src/gen/CORE.setting:5281␤ in block <anon> at /tmp/VOY_dFDZTm:1␤ in <anon> at /tmp/VOY_dFDZTm:1␤»
jnthn huh 12:47
I'm sure that used to work.
nom: say int
p6eval nom 97aa73: OUTPUT«Method 'gist' not found for invocant of class 'int'␤ in sub say at src/gen/CORE.setting:5281␤ in block <anon> at /tmp/Q89tPGJ8Lj:1␤ in <anon> at /tmp/Q89tPGJ8Lj:1␤»
moritz jnthn: it even says 'class' in the error message, so if it's a class, it must be ~~ Mu :-)
jnthn I'm really sure that one did.
nom: say int.^methods
p6eval nom 97aa73: OUTPUT«Method 'dispatch:<.^>' not found for invocant of class 'int'␤ in block <anon> at /tmp/9kTKsV5nJE:1␤ in <anon> at /tmp/9kTKsV5nJE:1␤»
moritz writes more tests
JimmyZ nom: my Num $num = NaN; say $num.WHAT
jnthn nom: say int.HOW.mro(int)
p6eval nom 97aa73: OUTPUT«Num()␤»
nom 97aa73: OUTPUT«Method 'gist' not found for invocant of class 'int'␤ in method gist at src/gen/CORE.setting:3680␤ in sub say at src/gen/CORE.setting:5281␤ in block <anon> at /tmp/Bi4RGUq83a:1␤ in <anon> at /tmp/Bi4RGUq83a:1␤»
jnthn oh
that'll do it. 12:48
masak bug?
jnthn yeah, looks like
masak submits
jnthn somehow it's not building the MRO properly.
And since method dispatch and type checking are all done by MRO...yeah, not surprising.
ENEEDMOARTESTS ;)
moritz really? I thought the problem is that .mro returns a Parcel, and Parcel.gist recurses into the elements
and hits int.gist, and dies 12:49
at least that's what the p6eval output tells me
JimmyZ anyway, (my Num $num)++ is deferent from (my num $num)++ is a little strange to me 12:50
moritz nom: my int $x = 3; say $x.gist
p6eval nom 97aa73: OUTPUT«3␤»
JimmyZ s/deferent/different/ 12:51
12:54 im2ee joined
moritz masak: got a ticket number for the bug you just submitted? 12:54
im2ee Hi. :) 12:55
moritz hi im2ee
masak hi im2ee
moritz: not yet; still looking for the start of the bug in the logs :) 12:56
...and I got derailed into finding out whether irssi has search functionality... :/
dalek ast: 29a84dc | moritz++ | S02-types/native.t:
basic tests for native type objects
moritz masak: there's /lastlog, but it might not be flexible/nice enough for your needs
masak thanks. 12:58
the relevant log is longish, so this will take a few minutes.
I'll post the URL when I've sent the email.
13:01 alc left 13:04 bluescreen10 joined
masak moritz: rt.perl.org/rt3/Ticket/Display.html?id=102256 13:04
jnthn moritz: oh...
moritz: I read the wrong live of p6eval output 13:05
nom: say int.HOW.mro(int).elems
p6eval nom cbc652: OUTPUT«1␤»
jnthn ok, *that* is a problem :)
moritz aye :-)
masak ok, that goes into the ticket.
moritz masak: tested in S02-types/native.t 13:07
dalek ast: 45800f1 | moritz++ | S02-types/native.t:
label and extend tests for RT #102256
masak moritz++
moritz ah, I can add that note myself :-)
moritz adds the 'testcommitted' tag 13:08
13:11 tokuhiro_ left, tokuhiro_ joined
PerlJam Have you guys seen www.infoq.com/presentations/Simple-Made-Easy ? 13:11
masak yep :) 13:12
I mentioned it here the other day.
PerlJam yeah, I'm looking at it again and picking out things I missed the first time :)
masak irclog.perlgeek.de/perl6/2011-10-22#i_4601445
13:13 molaf left
PerlJam Perl 6 fares well as a language that allows for Simplicity as Hickey defines it. 13:14
masak I should hope so! :P
PerlJam: you probably saw this when I wrote it, but I like this example of Perl 6 simplifying my life: gist.github.com/1149245 13:15
PerlJam yes, that is awesome. :)
masak it's like the Perl 6 code only focuses on delivering content, with all the fat trimmed. 13:16
that bit will definitely go in my Dancing Links reprise in London next month. :)
moritz :-) 13:17
moritz likes it
masak now, in terms of complecting things... you might argue that the Perl 5 version is long because it complects the circular list traversal and finding the smallest item. 13:18
moritz reads that blog post as "TLAs are LTA" :-) 13:19
masak because if you took those apart, you could at least use List::Util's &min
moritz does &min from List::Util support the equivalent of &by? 13:20
13:20 sayu joined
masak no :( 13:21
PerlJam no, but there's List::UtilsBy
masak :P
this will make a great set of slides... :)
"...and here's the Perl 6 version..."
PerlJam it literally adds the "by" for all of min, max, uniq, zip, sort, etc. 13:22
moritz zip_by ?
hm, nope :-) 13:23
masak PerlJam++
PerlJam LeoNerd++ (for writing the module so that Perl 5 can sort of compete with Perl 6 ;) 13:24
masak note that Perl 6 de-complects out of the box. 13:25
I remember looking at the loop I had written, getting the realization that it could be made into a sequence operator.
13:27 MayDaniel left
masak lazy lists rock. they allow you to decouple things and make them co-routine-like almost without thinking about it. 13:27
13:27 benabik joined
PerlJam Like past perls, once people get used to using these "higher level" constructs, they'll wonder how they ever got along without them :) 13:29
I just wonder how much of the Awesome Perl 6 will export outside of Perl 13:30
masak well, I think Perl 5 has shown so far that it adapts pretty nicely to the new tricks that Perl 6 provides. sometimes 'smartmatch' turns into 'psychoticmatch' along the way, but by and large, Perl 5 adapts. 13:31
what Perl 6 provides is something of a "full immersive experience".
flussence ... installs ghc 13:33
arnsholt PerlJam: I think grammars as the natural next step to regexes will catch on (eventually) 13:34
13:34 abercrombie joined
arnsholt The full range of awesome in P6 grammars might be hard to put into some languages, but grammars are a more natural way to express these things, on the whole 13:34
PerlJam arnsholt: I tend to agree, but only time will tell. 13:36
13:36 im2ee left
arnsholt Oh yes, definitely 13:36
13:36 im2ee joined
arnsholt Lazy lists are likely as well. The growing interest in Haskell and friends points in that direction as well 13:37
PerlJam I wonder if Perl 6 feature adoption in Perl 5 could be used as an indicator about Perl 6 feature adoption in the wild
masak probably not. Perl 5 people are usually a bit ahead of any such curve.
so maybe it's a good predictor ;) 13:38
PerlJam maybe
arnsholt Yeah, it's not a perfect model, but as a proxy it's probably not entirely useless
(To first order, and all that =)
PerlJam lazy lists will be hard for most other languages to adopt because to be useful, they need to be pervasive. 13:39
masak aye. 13:40
jnthn The hard thing was getting them to be eager enough in the right places so as to avoid surprises.
Which got a bunch easier to do when pmichaud++ came up with the immutable iterator model. 13:41
moritz wonders if it's possible to make a for 1..100 { } in sink context about as fast as iterating with native ints 13:42
masak if you can compile it into a while loop, why not? 13:44
moritz I was more thinking along the lines of haviing hash.map produce some "intelligent" iterator that, when in sink contest, does it at run time 13:45
jnthn The thing is that you have a block at runtime.
The cost is muchly in that invocation too 13:46
Whereas the while loop tends to flatten out.
Well, with --optimize=3 at the moment
moritz well, that's still much better than the call chain we have now
jnthn But maybe we should try brining that one into level 2 for the November release
moritz with eager, eager, gimme, reify, reify, reify or so
13:46 SHODAN left
masak jnthn: no rush :) 13:46
jnthn Yeah, I mentioned to Pm that I suspect we have one too many layers.
masak: well, it's only changing a number to do it ;) 13:47
The next thing the optimizer needs to do is learn that variables exist.
Today it lives in mostly blissful ignorance of them :)
moritz :-)
moritz thinks an NQP optimizer would be more important for now 13:48
masak quite possibly.
jnthn Well, it needs that variable analysis.
In many senses, I need to solve the same kind of analysis problem for both. 13:49
masak similar-ish languages, are they? :P
jnthn Apparently :P
13:52 darkop joined 13:55 mikemol left
arnsholt I'm getting an urge to make something better than make. Someone whack me =) 13:56
13:56 mikemol joined
moritz hugs arnsholt 13:56
arnsholt Or that =D
masak tickles arnsholt with a goose feather
flussence have a look at redo before you go doing that
masak or rake.
benabik cmake? 13:57
flussence github.com/apenwarr/redo - I haven't used it, but it looks nicest of the alternatives I've seen
moritz cmake just creates Makefiles, no?
.oO( Module::Build )
 13:58
benabik moritz: No idea.
arnsholt flussence: Oooh, never heard of redo before
moritz benabik: at least that's a project I used handled it
arnsholt Sounds shiny from the opening paragraph
masak I was just thinking the same :)
13:59 thou joined
moritz remembers reading the original notes, being excited for a minute, and then discarding it 14:00
arnsholt I've had this idea of a better make several times, but each time I'm reminded of a quote from someone I had a chat with at Opera: "make sucks. It's just too bad all the alternatives suck even more." 14:01
PerlJam People have been trying to create a better make for *decades*
mux did you mean, people have been creating better makes for decades, /6􏿽xB0 14:02
make that ? :-) at the end
PerlJam heck ... one of the first disucsisons I had on IRC about programming stuff was how much make sucked and that it needed to be replaced with something better. 14:03
moritz ah well, no Windows support
PerlJam That was in 1992 or so.
benabik moritz: The problem with Windows support is that you either reinvent shell syntax or miss some power of Make because you don't support shell syntax. :-/ 14:04
mux scons is not bad at all
masak ah! 'scons' was the name I was looking for.
mux++
moritz benabik: I'm well aware of the problems. It's solutions I'm looking for :-)
mux I wish it didn't allow full python syntax however, that's quite dumb since people are now able to create SConstruct files that won't work with some python versions 14:05
PerlJam has heard of "cons", but not "scocs"
er, scons
mux they definitely didn't need the full power of the language
PerlJam: well, it's supposed to be cons successor
moritz mux: but I'm sure it made the implementation much easier :-)
[Coke] why are we talking about improving make? don't we have better things to do?
mux moritz: yeah :-)
benabik [Coke]: #perl6 distracted? That never happens. 14:06
[Coke] I have to use ANT regularly, don't get me started. :P
mux oh man, ant
stupid XML configuration files for builds
benabik moritz: I've found Rake to be reasonable since I get the power of Ruby instead of sh. But like everything else, it has its problems.
mux ant is such a nightmare
moritz ah right, I wanted to fix feather for [Coke]++
masak Ant definitely uses XML in a way no-one ever should.
benabik Hm. Redo really is fascinating. 14:07
mux I wish people understood how XML /may/ be approrpiate for exchanging data between possibly very different systems, but that there's no way in hell it's appropriate to present a XML configuration file to a human
moritz [Coke]: trouble is, feather has only about 1G of RAM, and rakudo build on amd64 needs about 1.5G :(
mux: and it was never meant to be a data exchange format, just a markup language 14:08
jnthn wishes he knew where it all went
moritz it's a nice markup language for sure
benabik Although I think a .do file only fails if the last command failed. :-/
moritz can't you do a set -x or so?
[Coke] moritz: to increase the limit? not past the system default. 14:09
moritz [Coke]: sorry, that was to benabik
mux moritz: it's not so bad as an interchange format, because thanks to the huge number of XML related standards and nearly ubiquitiuous language support, you get something that you can use to store data/parse data/validate data for nearly free
[Coke] moritz: ah, yes.
benabik moritz: Yes, although I dislike systems that have "all your files should include ___ at the beginning". Make the default safe. 14:10
benabik will stop pretending this is #redo instead of #perl6 now.
mux oh, and it gives you structure and deals with character encodings itself.
moritz ... which partially relies on autodetection 14:11
mux yeah but that's because OSes didn't catch up and fail to give applications proper MIME types for files :-)
moritz stops pretending this is #xml-rant instead of #perl6 :-) 14:12
mux stops pretending
masak ooh, #xml-rant! 14:13
14:13 cognomore left, cognomore joined
dalek ast: c3a585e | (Solomon Foster)++ | S16-filehandles/filetest.t:
Fudged for niecza.
 14:14
ast: 6a5a339 | (Solomon Foster)++ | S32-num/complex.t:
Fudge for Niecza.
ast: 016a50f | (Solomon Foster)++ | S32-num/exp.t:
Fudge for Niecza.
ast: 7560a7a | (Solomon Foster)++ | S32-num/int.t:
Make Niecza-friendly: don't assume Rats stringify the same as Nums, and fudging.
14:15 tokuhiro_ left
masak perl6: say 5/4 14:16
p6eval niecza v10-61-gbacff10: OUTPUT«5/4␤»
..pugs b927740, rakudo cbc652: OUTPUT«1.25␤»
masak isn't that a Nieczabug?
moritz it is 14:17
masak submits it
any relevant synopsis I can quote? 14:18
PerlJam masak: hey ... supposedly the TPF voted (or whatever) on your grant proposal ... have you heard anything?
moritz masak: you can cite roast :-)
masak PerlJam: haven't heard anything yet, no. 14:19
moritz masak: S02: say 1/5; # 0.2 exactly (not via Num)
masak moritz++ 14:20
PerlJam masak: bummer. Maybe it's time to ping Karen again :) 14:21
masak perl6: say 1/3
moritz ... though by latest spec that's .gist, and not .Str
p6eval rakudo cbc652: OUTPUT«0.333333333333333␤»
..pugs b927740: OUTPUT«0.3333333333333333333333333333333333333333␤»
..niecza v10-61-gbacff10: OUTPUT«1/3␤»
moritz perl6: say (1/2).Str
p6eval niecza v10-61-gbacff10: OUTPUT«1/2␤»
..rakudo cbc652: OUTPUT«0.5␤»
..pugs b927740: OUTPUT«*** No such method in class Rat: "&Str"␤ at /tmp/JaZi9iv517 line 1, column 5 - line 2, column 1␤»
moritz ok, that is wrong at least
masak PerlJam: feel free. in the meantime, I'm finding a few macro tuits here and turning them into commits or design gists. 14:22
moritz: you mean Rat.gist isn't defined in the spec? sounds like something that should be spec'd if you ask me. 14:23
dalek ast: 47573c5 | (Solomon Foster)++ | S32-num/exp.t:
Fix niecza fudging.
 14:24
moritz masak: no. I mean that the 'say 1/2' line invokes .gist on the Rat
dalek ecza: e4121a3 | (Solomon Foster)++ | lib/CORE.setting:
Add log subs and log10 sub and method.
ecza: 9b24a83 | (Solomon Foster)++ | lib/CORE.setting:
Add unpolar and cis.
ecza: 9b9d1c0 | (Solomon Foster)++ | lib/CORE.setting:
Add exp($x, $base).
ecza: 519e0b6 | (Solomon Foster)++ | lib/CORE.setting:
Add Complex.polar.
ecza: 71325b0 | (Solomon Foster)++ | t/spectest.data:
Turn on S32-num files that now work.
moritz masak: and not .Str. So the spec doesn't talk about .Str stringification there
masak moritz: yes, I understand. 14:25
maybe that bit of spec just needs clarifying.
moritz well, it is very silent on most .gist'ifications, and on many .Str'ifications 14:26
14:26 JodaZ left
moritz S02 is a weird mixture of "bits and pieces" and incomplete API documentation 14:26
masak perfect for someone to just guerrilla-spec at some point. 14:35
moritz quick straw poll: should Rat.new(2, 1) stringify to 2.0 or to 2? 14:36
and should 2e0 stringify to 2 or 2e0?
PerlJam 2.0
masak 2.0
moritz: what does 2e1 stringify to?
moritz masak: 20 or 20e0
[Coke] coming from tcl, I'd prefer the .0 on the end.
flussence perl6: say (eval '2.0').perl 14:37
masak moritz: scientific notation feels like Num to me, but if they're Int then maybe they can stringify like Int.
p6eval pugs b927740, rakudo cbc652, niecza v10-66-g71325b0: OUTPUT«2/1␤»
flussence likes round-tripping
rakudo: say (eval '2.0').Str.eval.perl
p6eval rakudo cbc652: OUTPUT«2␤»
flussence hm
perl6: say (eval '2.0').Str.eval.perl
p6eval rakudo cbc652: OUTPUT«2␤»
..pugs b927740: OUTPUT«*** No such method in class Rat: "&Str"␤ at /tmp/mivPVyzw7Y line 1, column 5 - line 2, column 1␤»
..niecza v10-66-g71325b0: OUTPUT«␤Unhandled Exception: Unable to resolve method eval in class Str␤ at /tmp/6bZqfMADY_ line 1 (MAIN mainline @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2193 (CORE C1051_ANON @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2194 (CORE module-CO…
14:38 cognomore left, cognominal_ joined, cognomore joined
flussence perl6: say eval((eval '2.0').Str).perl 14:38
p6eval niecza v10-66-g71325b0: OUTPUT«2/1␤»
..rakudo cbc652: OUTPUT«2␤»
..pugs b927740: OUTPUT«*** No such method in class Rat: "&Str"␤ at /tmp/0q2Ojup5PN line 1, column 10-26␤»
PerlJam moritz: I'm with masak on scientific notation
though I never think of a number in scientific notation as "integer"; they're always some floating point thing in my head. 14:39
(at least historically)
moritz will regret this one 14:40
14:41 cognominal left
masak 'what will 2e0 stringify to?' might be the wrong question. 'what type does it parse to?' sounds like a more immediate question. 14:42
moritz Num 14:43
the spec is pretty clear
14:43 im2ee left
flussence (do we even have any types that stringify to scientific notation?) 14:43
moritz on this account
PerlJam flussence: I hope not 14:44
14:44 im2ee joined
moritz flussence: strings which contain numbers in scientific notation :-) 14:44
14:46 skangas joined
[Coke] hurm. so, I change my mind: Floats should stringify with .0 even if they're intable. Rats should stringify to ints if they can. 14:49
masak agreed. 14:50
dalek ecs: 660b108 | moritz++ | S02-bits.pod:
[S02] define .Stringy and .gist for Num and Rat

leave a bit of wiggle room in the Num stringification fow now.
moritz then please change the spec too, not just your mind :-) 14:51
masak actually, I don't mind if 4/1 stringifies to 4/1 either.
moritz pugs: say 1/7
p6eval pugs b927740: OUTPUT«0.1428571428571428571428571428571428571429␤»
moritz it used to answer with 1/7 at one point
I hated it
masak yeah.
moritz I *knew* that I asked it for the value of 1/7.
masak Rats should stringify to their decimal expansion.
we're just talking about the Int case here. 14:52
moritz you mean where .denominator == 1?
masak or rather, the "denominator in the reduced fraction is 1" case.
right.
[Coke] should 4/2 stringify the same as 2/1 ? 14:53
moritz masak: so what changed your mind?
[Coke]: yes
[Coke] ah, catching up, reduced, aye.
14:53 nornagest joined
[Coke] rat's aren't floats. 14:53
masak moritz: correctly understanding the domain we were moving in. 14:54
moritz: I don't like for pi to stringify to something with 'e0' at the end, for example.
moritz masak: and it won't 14:55
S02:3543 constant pi is export = 3.14159_26535_89793_23846_26433_83279_50288;
that's a Rat, not a Num
masak fair enough.
it makes things a bit better...
...but I think I still prefer Rats and Nums to stringify the same, rather than to have Nums append 'eN' to stuff. 14:56
Num is a sort of inaccuracy fallback for Rat, after all.
moritz doesn't care too strongly either way, as long as it's specced and consistent
perl6: say 2e0 ** 35 14:57
p6eval pugs b927740, rakudo cbc652, niecza v10-66-g71325b0: OUTPUT«34359738368␤»
moritz perl6: say 2e0 ** 65
p6eval pugs b927740: OUTPUT«36893488147419103232␤»
..niecza v10-66-g71325b0: OUTPUT«3.68934881474191E+19␤»
..rakudo cbc652: OUTPUT«3.68934881474191e+19␤»
masak I'll have a look at the spec tonight after $dayjob, unless someone beats me to it.
14:57 sivoais left 14:59 slavik1 left, slavik2 joined
masak std: if 42 { say "A" } else if 5 { say "B" } 15:01
p6eval std 580b69a: OUTPUT«===SORRY!===␤Please use 'elsif' at /tmp/kf213KjczR line 1:␤------> if 42 { say "A" } else if ⏏5 { say "B" }␤Confused at /tmp/kf213KjczR line 1:␤------> if 42 { say "A" } else if ⏏5 { say "B" }␤ expecting stat…
masak perl6: if 42 { say "A" } else if 5 { say "B" } 15:02
p6eval rakudo cbc652: OUTPUT«===SORRY!===␤Missing block at line 1, near "if 5 { say"␤»
..niecza v10-66-g71325b0: OUTPUT«===SORRY!===␤␤Please use 'elsif' at /tmp/fgP25RSzdE line 1:␤------> if 42 { say "A" } else if ⏏5 { say "B" }␤␤Two terms in a row at /tmp/fgP25RSzdE line 1:␤------> if 42 { say "A" } else if ⏏5 { say "B" }…
..pugs b927740: OUTPUT«*** ␤ Unexpected "if"␤ expecting "=" or bare or pointy block construct␤ at /tmp/TEvJkbeeLF line 1, column 24␤»
masak rakudo-- pugs-- niecza++
[Coke] occasionally wonders: if we have to recognize "else if" to complain about it, why not just support it instead. 15:04
(instead of complaining, not instead of elsif)
moritz [Coke]: it introduces the dangling-else ambiguty again
15:04 sayu left
moritz if 1 { } else if 2 { } else { } # does the 'else' belong to 1 or to 2? 15:05
that's a pitfall that Perl generally avoids by requiring blocks after if/elsif/else, and by having 'elsif' separate from 'else if'
JimmyZ how does C do it? 15:06
[Coke] I would expect it to work the same as if you had used elsif.
moritz iirc it just disambiguates one way or another
JimmyZ would like to else if = elsif
mux yeah, that's why you often have to wrap macros inside do { ... } while (0); blocks 15:07
to avoid unexpected problems when your macro is used as: if (foo) MACRO;
in moritz's example, I believe C intreprets it as if 1 { } else { if 2 { } else { } }
not sure though 15:08
moritz wonders if the two cases are actually semantically distinct
flussence I don't see any way for that to be ambiguous, unless the language allows one `if` to have multiple `else` blocks attached, which doesn't make sense to me 15:09
JimmyZ hates elsif :)
masak JimmyZ: tell me why.
mux moritz: mmm, I think you're right, it doesn't matter one way or the other :-) 15:10
JimmyZ because I use PHP too. when I use else if in perl, I always got a unexpected wrong
or switch from other languages 15:11
jnthn decommute &
flussence if/else if/else is just a longer way of writing `$if-clause ?? ... !! $else-if-clause ?? ... !! ... # else`, and that's well-defined (well, in most languages...)
masak JimmyZ: if you have that as a reason, you must also respond to moritz's reason for having 'elsif'.
JimmyZ elseif is ok to me too. 15:12
flussence (of course, in PHP, nested ternaries behave as uselessly as possible...) 15:14
(wow, ghc takes a *long* time to compile...) 15:16
15:18 fsergot left 15:20 sivoais joined, sivoais left
sorear good * #perl6 15:21
15:21 sivoais joined, wtw left
masak hi sorear 15:23
masak decommutes
JimmyZ what's different between 'if 1 { } else { if 2 { } else { } } ' and 'if 1 { } elsif 2 { } else { } '
the first is C and the second is Perl6 15:25
abercrombie no difference 15:27
JimmyZ 'if 1 { } else if 2 { } else { } ' in C and 'if 1 { } else { if 2 { } else { } } ' ? 15:28
'if 1 { } else if 2 { } else { } ' in C and 'if 1 { } elsif 2 { } else { } ' in Perl 6? 15:29
abercrombie still no difference since those 2 forms in C are the same. 15:30
moritz I was wrong earlier
the difference really comes from forcing blocks
sorear JimmyZ: the elsif version has fewer braces
moritz not from the 'elsif' vs. 'else if' distinction 15:31
geekosaur blah. mono-sgen cores on my Air w/latest Mono... (mono works but seems like it's slower?)
geekosaur shouldn't be doing this, probably, but wished he'd had a p6 over the weekend 15:32
JimmyZ sorear: but why disallowed 'else if' 15:33
abercrombie In C, if {} else if {} else {} is the same as if {} else { if {} else {} }
moritz because we consistently force blocks after 'if' and 'else' 15:34
JimmyZ why not 'else if' = 'elsif' ? 15:35
moritz because we consistently force blocks after 'if' and 'else'
and we want to avoid exceptions
15:36 am0c joined
JimmyZ does C have this issues? 15:36
these 15:37
15:37 darkop left
abercrombie JimmyZ: you know, in C actually there isn't a seperate "else if" but really just a simple style for else { if } 15:37
moritz C doesn't force blocks after 'if' and 'else', hence it has the dangling else problem.
JimmyZ i know it
moritz which has a rather arbitrary disambiguation rule, which bites you if happen to have formatted your code the other way round 15:38
15:41 wamba joined 15:50 agentzh left
JimmyZ would like to suggest that 'else if' is a exception, hehe 15:50
moritz it is. It throws one :-) 15:51
SCNR
JimmyZ is a exception that not consistently force blocks after else :)
jnthn hope 15:54
er
jnthn home
geekosaur perl 6 is trying not to be the crazed bundle of exceptions that perl 5 is 15:55
it's perceived as a slippery slope; allow one, then others start to creep in...
15:56 bluescreen10 left 16:03 bluescreen10 joined 16:04 worr joined
geekosaur also, perl 6 is hardly unique here; sh has elif, and IIRC Algol68 also used elif in preference to else if (for a slightly different reason: the syntax unambiguously ends ifs, so you can't have dangling else-s, but you would then have a bunch of fi-s which can be avoided by using elif 16:04
JimmyZ sleeps 16:06
16:10 molaf joined
sorear goodie, the single-domain change didn't increase #failing-spectest-files 16:15
niecza: say <a b c d>[2,3] 16:17
p6eval niecza v10-66-g71325b0: OUTPUT«c d␤» 16:18
sorear apparently I broke this...?
16:19 wolfman2_ left 16:20 im2ee left 16:21 im2ee joined 16:24 JimmyZ left
TimToady how is it broke? 16:40
sorear TimToady: confusingly 16:41
it says "c" locally
TimToady ah
arity of array maybe?
sorear if I copy the definition of Parcel.postcircumfix:<[ ]> into an augment, it works fine
it's being miscompiled but only if it's in CORE.setting
TimToady it's acting like [+(2,3)] maybe? 16:42
sorear maybe 16:43
but I'm hunting the difference in the compilation now
16:46 im2ee left 16:47 im2ee joined 16:53 donri joined 16:59 dakkar left, mberends joined
dalek kudo/nom: 6b78ad0 | jnthn++ | src/Perl6/Actions.pm:
Fix code-gen for assignment to native types.
 17:03
17:05 im2ee left 17:06 im2ee joined 17:09 REPLeffect left
tadzik #phasers in 20? 17:10
jnthn yes
mberends hi tadzik, jnthn, * 17:11
jnthn o/ mberends
tadzik hello mberends
btw, now #phasers happens to be during my classes
17:12 slavik2 left 17:14 Kivutarrr joined 17:15 pernatiy left, daniel-s left 17:17 GlitchMr joined 17:18 mj41_nb left, mj41 left
dalek ast: 11b1be0 | jnthn++ | S02-types/native.t:
Unfudge a couple of tests for natives.
 17:19
kudo/nom: a4602de | jnthn++ | src/Perl6/Metamodel/NativeHOW.pm:
Helps to actually compute the MRO before using it. :-) This fixes various issues with native types.
jnthn That's who of today's tickets dealt with. :) 17:20
*two 17:21
17:21 Kivutarrr left 17:23 jdv79 joined
jdv79 jnthn: thx for the blog post! 17:23
17:23 packetknife joined
jnthn jdv79: Blog...oh, it was a little while back though. :) 17:23
17:23 aber joined
jnthn should create something else exciting so he can write another one :) 17:23
17:24 abercrombie left
dalek ecza/serialize: 87fff3b | sorear++ | lib/CORE.setting:
Rearrange setting to make inlining work better

Also adds missing \s to non-inlined versions of &postcircumfix ops. Fixes <a b c d>[2,3].		 17:24
17:26 slavik1 joined 17:27 abercrombie joined
jnthn tadzik: I just ran setting compilation through the sub profiler and got a curious result. The optimizer seems to account for only 4% of time spent in compiling CORE.setting here. 17:27
tadzik jnthn: right, I measured it today and it's still about 3 minutes
wonder what caused that back then, weird 17:28
I must keep better track of those results
compiling of the Optimizer.pm itself can't be that long, it's just nqp
17:30 packetknife left, aber left
dalek ecza/serialize: 0271633 | sorear++ | lib/CodeGen.cs:
Implement SubInfo.transparent required for &?BLOCK
 17:34
17:34 buubot_backup left
diakopter . 17:36
sjohnson , 17:37
sorear ! 17:39
geekosaur ∭ 17:41
sjohnson heh
dukeleto [~~] 17:44
mberends ⌨ 17:48
sjohnson (='.'=) 17:49
dalek ecza/serialize: 9ee80cd | sorear++ | lib/Kernel.cs:
Fix GLOBAL/PROCESS in indirect name usage (how did this ever work?)
 17:53
17:54 buubot_backup joined
dalek ecza/serialize: 09b8ac7 | sorear++ | lib/CodeGen.cs:
Actually report errors from C3 inconsistency...
 18:00
diakopter std: 4 +| 5 18:01
p6eval std 580b69a: OUTPUT«ok 00:01 120m␤»
diakopter rakudo: say 4 +| 5
p6eval rakudo a4602d: OUTPUT«5␤»
18:09 Reaganomicon joined
[Coke] rakudo: my $¢=32; 18:13
p6eval rakudo a4602d: OUTPUT«===SORRY!===␤Confused at line 1, near "my $\x{a2}=32;"␤»
dalek ecza/serialize: 4c49318 | sorear++ | lib/ObjModel.cs:
Fix stubbing of modules
 18:14
TimToady perl6: [[1,2,3],[4,5,6],[7,8,9]]»[0].say 18:17
p6eval rakudo a4602d, niecza v10-66-g71325b0: OUTPUT«1 2 3 4 5 6 7 8 9␤»
..pugs b927740: OUTPUT«decodeUTF8': bad data: '\187'␤*** ␤ Unexpected "\187["␤ expecting term postfix or operator␤ at /tmp/F3bf2NksVb line 1, column 26␤»
TimToady pugs: [[1,2,3],[4,5,6],[7,8,9]]>>[0].say
p6eval pugs b927740: OUTPUT«*** ␤ Unexpected ">>["␤ expecting term postfix or operator␤ at /tmp/6rI8WNnHT_ line 1, column 26␤»
TimToady pugs: [[1,2,3],[4,5,6],[7,8,9]]>>.[0].say 18:18
p6eval pugs b927740: OUTPUT«␤»
TimToady b: [[1,2,3],[4,5,6],[7,8,9]]»[0].say
p6eval b 1b7dd1: OUTPUT«147␤»
TimToady masak: I'm trying to figure out what you mean by "real code" in irclog.perlgeek.de/perl6/2011-10-26#i_4616132
only b is close to currently specced semantics for ».[0] 18:19
diakopter TimToady: is "It is not necessary to type any of the unicode symbols. They all have an equivalent that is plain ASCII" still valid? (sorear pointed out $¢) 18:21
moritz masak: does markdown support something like meta data/annotations? 18:23
masak not explicitly.
but it supports arbitrary XML... 18:24
18:24 bluescreen10 left
masak TimToady: code that I wrote for a purpose other than just fun. in Niecza. 18:25
moritz wonders what format to write the FAQs in 18:26
PerlJam pod6
sorear moritz: iambic pentameter
masak Klingon.
PerlJam all of the above?
sorear masak: does that count as a format?
moritz doesn't know any Klingon, and doesn't see much advantage in using it 18:28
PerlJam masak: that code you wrote for purposes other than fun ... you also wrote it for fun too, right?
masak Pod6 could use the exercise, I guess. 18:29
Markdown is less hassle for you.
jdv79 jnthn: yeah, i only remember to check once in a while. 18:30
TimToady masak: my point is that it doesn't work in niecza, unless maybe in /serialize 18:31
it still has the old semantics
which preclude using » and .[0] together, other than as a nop on the leaves
sorear Is there a connection here between ».[0] and masak's real code? 18:32
TimToady irclog.perlgeek.de/perl6/2011-10-26#i_4616132
perl6: [[1,2,3],[4,5,6],[7,8,9]]»[1].say 18:33
p6eval niecza v10-66-g71325b0: OUTPUT«Use of uninitialized value in string context␤ at /home/p6eval/niecza/lib/CORE.setting line 641 (CORE warn @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 21 (CORE Mu.Str @ 12) ␤ at line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/lib/CORE.setting…
..rakudo a4602d: OUTPUT«.[1] out of range for type 1␤ in method at_pos at src/gen/CORE.setting:1013␤ in method postcircumfix:<[ ]> at src/gen/CORE.setting:991␤ in method reify at src/gen/CORE.setting:4052␤ in method reify at src/gen/CORE.setting:3959␤ in method reify at src/gen/CORE.…
..pugs b927740: OUTPUT«decodeUTF8': bad data: '\187'␤*** ␤ Unexpected "\187["␤ expecting term postfix or operator␤ at /tmp/b9Rijo08qs line 1, column 26␤»
TimToady b: [[1,2,3],[4,5,6],[7,8,9]]»[1].say
p6eval b 1b7dd1: OUTPUT«258␤»
[Coke] b,nom: say "be nice if this worked."
18:34 mberends_ joined
TimToady diakopter: $¢ is just experimental error :P 18:35
18:35 mberends left
moritz [Coke]: not too hard to do, I guess 18:35
18:35 mberends_ left
[Coke] then we can remove "pugs" from perl6 and easily add it back in if we want it. 18:36
18:36 slavik1 left
moritz well, that's just a one-line patch anyway :-) 18:36
18:37 slavik1 joined
TimToady well, obviously someone just needs fix pugs 18:37
*to fix
masak :)
18:38 molaf left
TimToady it's easy, just add 6model, and STD parser, and fix all the hacky parts to work around the absence of regexes, and maybe another minor thing or two... 18:39
18:39 fsergot joined
moritz and the compiler middleware, I guess 18:39
masak sounds like fun! 18:40
TimToady or just backend one of the other implementations to emit .hs, and call it pugs
PerlJam TimToady: would it be $ɛ or something if it were experimental error? :)
18:41 bluescreen10 joined
TimToady no, it would be $ε instead :P 18:42
moritz .u ε
phenny U+03B5 GREEK SMALL LETTER EPSILON (ε)
sorear hrrm. seems I've broken return-pass 18:43
TimToady .u ɛ
phenny U+025B LATIN SMALL LETTER OPEN E (ɛ)
PerlJam I did say "or something"
18:45 Chillance joined 18:48 mberends joined
sorear needs to write some kind of debug dumper for .ser files 18:49
18:49 ksi joined
cognominal_ This may interest jnthn. A mere use of github.com/ihrd/uri/blob/master/li...Grammar.pm triggers Malformed $*PKGDECL at line 46, near "=> $init_g" 18:51
this package contains a require, but that should be runtime... 18:52
I would not be surprised if that was related to the serialization stuff 18:53
PerlJam it looks to me like it's treating "grammar => $init_grammar" as a grammar declaration. 18:55
(or trying to anyway)
moritz :grammar($init_grammar) would be a workaround
masak we have an RT ticket about such behavior. 18:56
I don't remember if we closed it. I think we did.
dalek q: 6a76d0b | moritz++ | answers.md:
link to source
[Coke] now if I could just get moritz to work on this java for me. 18:58
moritz masak: that ticket is still open in nom 18:59
cognominal_ Perljam, moritz. indeed this keyword is not protected by an <end_keyword>
PerlJam Is that a LHF? Sounds like it. 19:00
masak moritz: ah. thank you.
PerlJam: wouldn't surprise me if t'were tied to LTM somehow.
moritz nom: say (trusts => 1)
p6eval nom a4602d: OUTPUT«"trusts" => 1␤»
PerlJam masak: ah, could be. 19:01
moritz I'll try it with end_keyword
nom: say (grammar => 1)
p6eval nom a4602d: OUTPUT«===SORRY!===␤Malformed $*PKGDECL at line 1, near "=> 1)"␤»
moritz cognominal_: the <.end_keyword> is at the start of package_def 19:02
cognominal_ I dream of a proto token package_declarator is keyword{ <...> }
19:03 slavik1 left
cognominal_ that would append a <end_keyword> after the <sym> 19:03
moritz how does STD do it? 19:04
GlitchMr print STDERR "lol"
How can I do it in Perl 6?
PerlJam GlitchMr: just like that?
well, not quite like that I guess 19:05
GlitchMr Could not find sub &STDERR
...
PerlJam use $*ERR
cognominal_ moritz, meaning it appears to late because the 'grammar' token is already reduced at that stage?
sorear print $*ERR: "lol"
19:05 slavik1 joined
GlitchMr perl6: print $*ERR: "There is something wrong on line .\n"; 19:05
p6eval niecza v10-66-g71325b0: OUTPUT«===SORRY!===␤␤Invocant handling is NYI at /tmp/xXoPnwelMx line 1:␤------> : "There is something wrong on line .\n"⏏;␤␤␤Unhandled Exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 766 (CORE die…
..rakudo a4602d: OUTPUT«===SORRY!===␤Confused at line 1, near "print $*ER"␤»
..pugs b927740: OUTPUT«There is something wrong on line .␤»
GlitchMr ... 19:06
sorear why do you ... so much
use words, they're clearer
moritz sorear++
GlitchMr :)
Yeah. Right.
sorear also pugs++
GlitchMr I don't know English.
[Coke] also, please don't use "perl6:" unless you mean it. it's quite verbose.
moritz GlitchMr: then don't use ... either. It doesn't tell us anything
PerlJam GlitchMr: $*ERR.print: "foo" 19:07
cognominal_ anyway you gave me the workaround
GlitchMr rakudo: $*ERR.print "There is something wrong on line .\n";
p6eval rakudo a4602d: OUTPUT«===SORRY!===␤Confused at line 1, near "$*ERR.prin"␤»
jnthn Need the colon
GlitchMr rakudo: $*ERR.print: "There is something wrong on line .\n";
p6eval rakudo a4602d: OUTPUT«There is something wrong on line .␤»
GlitchMr Thanks :)
PerlJam I wonder why the dative form didn't work. 19:08
maybe my brain is just comingling perl 5 and perl 6 too much
TimToady perl6: note "lol" 19:09
p6eval pugs b927740: OUTPUT«*** No such subroutine: "&note"␤ at /tmp/o3IAVDLOm5 line 1, column 1 - line 2, column 1␤»
..rakudo a4602d, niecza v10-66-g71325b0: OUTPUT«lol␤»
sorear PerlJam: print $*ERR: "foo" is *supposed* to work but only implmeneted by pugs atm. 19:10
TimToady perl6: note lol "lol"
p6eval niecza v10-66-g71325b0: OUTPUT«===SORRY!===␤␤Undeclared routine:␤ 'lol' used at line 1␤␤␤Unhandled Exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 766 (CORE die @ 2) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1136 (STD P6.comp_unit @ 36) ␤ …
..pugs b927740: OUTPUT«*** No such subroutine: "&note"␤ at /tmp/MYhSSs7gma line 1, column 1 - line 2, column 1␤»
..rakudo a4602d: OUTPUT«===SORRY!===␤CHECK FAILED:␤Undefined routine '&lol' called (line 1)␤»
sorear TimToady++ # I didn't think of that at all
TimToady: say eval '&?BLOCK.WHAT' # this has just come up 19:11
GlitchMr (\N*) ::: { fail "Invalid diff hunk: $1" } 19:12
What is \N?
sorear not a newline
TimToady not \n
GlitchMr OK :)
TimToady and \R is not \r, unlike in p5
p5's \R is just \n in p6 19:13
moritz GlitchMr: read S05 for more info on regexes
jnthn sorear: Guess it's Block since eval implies a lexical scope 19:14
moritz cognominal_: I have a patch for the <.end_keyword> things, spectest now...
TimToady is there any reason it needs to be differentiated from Block?
GlitchMr: perhaps if you translated the synopses into Polish you wouldn't have to read them in English :) 19:15
sorear well, it's got to provide $! $/ lexicals
GlitchMr :P
dalek q: 2695668 | moritz++ | / (2 files):
answer another question
 19:16
TimToady seriously, you keep asking questions that could be answered with a grep of the synopses
[Coke] ponders a nicely formatted hardcopy of the SYN. 19:17
cognominal_ TimToady, maybe there should be a keyword "keyword" used in place of "token" that would imply an implit <.end_keyword>. 19:18
*implicit
TimToady and where would .end_keyword go?
moritz [Coke]: it would be out of date before the printing press spit out the last page :-)
[Coke] moritz: sometimes it's nice to read things on paper. 19:19
jnthn actually met somebody with a printed copy of the Perl 6 synopses once :)
jnthn tries to recall who/where
GlitchMr m:P5/(?mi)^(?:[a-z]|\d){1,2}(?=\s)/
... ok...
masak GlitchMr: if you don't say more, we don't know what you mean. 19:20
GlitchMr :)
cognominal_ jnthn, you need to buy a RV or a truck to carry the synopses?
or an iPad? :)
TimToady, the <.end_keyword> would be implied after the declared token within the rule that define it. 19:22
GlitchMr | (.**1..10)
{
$*ERR.print: "There is something wrong near " ~ (~$0).perl ~ ".\n";
return 0.42 # just to end parsing of grammar
}
(I wanted to type ...)
moritz GlitchMr: please don't paste more than one line of code here. Use a pastebin 19:23
GlitchMr That's probably bad coding... yeah... I guess I cannot make good code...
masak GlitchMr: gist.github.com is one.
cognominal_ token terminator:sym<when> { 'when' <.end_keyword> } would be keyword terminator:sym<when>;
because the block that defines it would be redundant.
moritz is not sure that use case is general enough, but it reads nicely 19:24
sorear moritz: macro! :D
TimToady that is currently written: token terminator:sym<when> { <sym> } with an implied .endsym 19:25
cognominal_ going that way on want to be wild and do write keyword terminator:sym<if when unless while…>
when a factorization lead to another, that a good sign.
TimToady: in nom, the .endsym is currently explicit 19:26
TimToady in which case you'd *have* to use use <sym> since it's generic, not 'when'
so the current way is more powerful, if we allow slices
19:27 chromis joined
TimToady
.oO(sliced declarations, sheesh...)
 19:27
dalek ecza/serialize: 3c4b14d | sorear++ | src/NieczaGrammar.pm6:
Prevent mainline Routines from eating return exceptions
TimToady well, more like hyper declarations
dalek ast: 7f92c99 | moritz++ | S02-literals/pairs.t:
unfudge passing tests
kudo/nom: 158bd07 | moritz++ | src/Perl6/Grammar.pm:
protect "grammar" etc with <.end_keyword>

  cognominal++ noted that it is applied too late (after the package_declarator:sym<...>
action method has fired)
TimToady but a slice is a sort of hyper-y thing, I guess 19:28
sorear niecza should be down to 10 failing files
TimToady don't want to distract you, but it'd be nice if duck mapping S03:4276 could get in before your relase 19:30
*lease 19:31
nom too, for that matter
jnthn Do we have an RT for it? If not, plesae somebody make one...RT is my extra memory at the moment :)
TimToady then masak++'s Real Code™ would work
masak it *did* work... 19:32
TimToady on nom?
I don't see how
I showed ^^ that it doesn't work
masak on Niecza.
19:32 bluescreen10 left
TimToady doesn't work there either 19:33
masak with Match objects.
moritz submits rakudotodo
TimToady niecza: [[1,2,3],[4,5,6],[7,8,9]]»[1].say
p6eval niecza v10-66-g71325b0: OUTPUT«Use of uninitialized value in string context␤ at /home/p6eval/niecza/lib/CORE.setting line 641 (CORE warn @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 21 (CORE Mu.Str @ 12) ␤ at line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/lib/CORE.setting…
TimToady I'd say Match objects are only working accidentally 19:34
masak niecza: "oh %foo% and %bar% but %baz% too".comb(/ '%' (\w+) '%' /)>>[0].perl
p6eval niecza v10-66-g71325b0: ( no output )
masak niecza: say "oh %foo% and %bar% but %baz% too".comb(/ '%' (\w+) '%' /)>>[0].perl
p6eval niecza v10-66-g71325b0: OUTPUT«["%foo%", "%bar%", "%baz%"].list␤»
masak \o/
nothing accidental about it ;)
sorear looks like niecza is checking ~~ List | Parcel | Range
TimToady niecza: say "oh %foo% and %bar% but %baz% too".comb(/ '%' (\w+) '%' /).list»[0].perl 19:35
p6eval niecza v10-66-g71325b0: OUTPUT«["%foo%", "%bar%", "%baz%"].list␤»
TimToady niecza: say "oh %foo% and %bar% but %baz% too".comb(/ '%' (\w+) '%' /).list».perl 19:36
p6eval niecza v10-66-g71325b0: OUTPUT«"%foo%" "%bar%" "%baz%"␤»
TimToady anyway, it's supposed to try the [0] before checking for recursion (and ignore the [0] from Any somehow) 19:38
masak I can't decide if I like or dislike the discontinuity of recursing for some things and not for others... :) 19:39
19:40 zamolxes left, zamolxes joined 19:41 manildomin joined
manildomin hi 19:41
moritz hello manildomin 19:42
dalek ast: 918275a | moritz++ | S02-types/hash.t:
unfudge another test for rakudo
masak manildomin! \o/
manildomin hi
masak manildomin: how's it going with Perl 6? 19:43
moritz -> sleep 19:44
manildomin it's giving me some difficulties but I think Perl6 is interesting so I keep learning it
masak 'night, moritz 19:45
manildomin: if you have any favourite difficulties, we're here to help.
manildomin thanks
I assume that Rakudo compiles perl6 code to bytecode for parrot, am I right?
cognominal_ TimToady: the problem with the hyperop-y syntax to fold a proto regex is that is conflict with the multitokenish convention used for circumfix. 19:46
masak manildomin: right. 19:47
well, at least to PIR.
cognominal_ I seee this hyperop-y syntax would apply in many places in the grmmar.
manildomin ok, then is it possible asking rakudo not to run the program but to give a pir output file instead? 19:48
19:48 chromis left
masak manildomin: been a while since I tried it -- but yes. 19:48
manildomin: why -- are you experiencing long compile times?
manildomin I was just wondering if it was possible 19:50
cognominal_ token circumfix:sym<[ ]> { '[' <semilist> ']' } # on the other hand, there would be no hyperoperated sym here
19:50 bluescreen10 joined
masak wah, ok. 19:50
PerlJam manildomin: perl6 --target=pir -e 'say "hi"'
manildomin ok, thanks
I have an issue with a code found in the book "Using Perl6" (again :)) 19:58
masak which one?
nom: given sqrt -1 -> $_ is copy { .=subst(/N$/, "m!"); say .subst("N", "Wh"); say .subst("N", "B"); say .subst("N", "Thank you M") }
p6eval nom a4602d: OUTPUT«Wham!␤Bam!␤Thank you Mam!␤»
masak :) 19:59
nom: given sqrt -1 -> $_ is copy { .=subst(/N$/, "m!"); say .subst("N", "Wh"); say .subst("N", "B"); say .subst("N", "Thank you M").subst("am", "a'm" }
p6eval nom 158bd0: OUTPUT«===SORRY!===␤Unable to parse postcircumfix:sym<( )>, couldn't find final ')' at line 1␤»
masak nom: given sqrt -1 -> $_ is copy { .=subst(/N$/, "m!"); say .subst("N", "Wh"); say .subst("N", "B"); say .subst("N", "Thank you M").subst("am", "a'm") }
p6eval nom 158bd0: OUTPUT«Wham!␤Bam!␤Thank you Ma'm!␤»
manildomin this one : pastebin.com/nbnhhrrZ 20:00
masak looks
PerlJam nom doesn't grok that syntax yet, does it? 20:01
masak nom: my @a; my :($first, *@rest) := \(|@a);
p6eval nom 158bd0: OUTPUT«===SORRY!===␤Malformed my at line 1, near ":($first, "␤»
masak seems not.
b: my @a; my :($first, *@rest) := \(|@a); say "alive"
PerlJam (I'm not even sure b did :)
p6eval b 1b7dd1: OUTPUT«===SORRY!===␤Malformed my at line 22, near ":($first, "␤»
masak seems not :)
jnthn std: my @a; my :($first, *@rest) := \(|@a); say "alive" 20:02
p6eval std 580b69a: OUTPUT«===SORRY!===␤Malformed my at /tmp/FTwI75S8RI line 1:␤------> my @a; my ⏏:($first, *@rest) := \(|@a); say "alive"␤ expecting scoped declarator␤Parse failed␤FAILED 00:01 120m␤»
manildomin my version of Rakudo is Rakudo Perl 6, version 2011.04 built on parrot 3.8.0 0
maybe it's too old?
jnthn not even STD :P
masak uhm.
maybe the colon simply shouldn't be there?
PerlJam then why do I remember that syntax? 20:03
masak yeah, good question.
jnthn doesn't think it should be there 20:04
masak checks S06
S06:1480: my (@a,@b,@c) := |$foo;
S06:1732: my Dog ($fido, $spot) := twodogs(); 20:05
PerlJam S03:280L Signature literal :(Dog $self:)
masak S06:3220: :(:who($name), :why($reason)) := (why => $because, who => "me");
PerlJam s/L/:/
masak PerlJam: I don't think :() works with, my, is all. 20:06
s/,//
20:06 cooper joined
masak 'night, #perl6 20:06
PerlJam masak: S03:4882: my :(Dog $foo is woof = 123); # same thing (full Signature form)
manildomin 'night
20:07 jlaire left
PerlJam The text starting from S03:4952 too 20:07
especially where it says "Signatures are expected after declarators such as C<my>, C<sub>, C<method>, C<rule>, etc. In such declarators the colon may be omitted. But it's also legal to use it" :-) 20:08
manildomin if I remove it the code works flawlessly 20:09
20:15 GlitchMr left
im2ee Good night! :) 20:16
20:17 im2ee left 20:20 jlaire joined 20:25 MayDaniel joined 20:32 Je joined 20:35 Je left 20:38 dorlamm joined 20:39 MayDaniel left, envi_ left 20:44 jlaire left, jlaire joined 20:52 pat____ joined
pat____ hi 20:52
have been playing around with "but" and some questions arose.
20:52 manildomin left 20:54 y3llow_ joined, pothos_ joined 20:56 pothos left, pothos_ is now known as pothos, y3llow left, y3llow_ is now known as y3llow
pat____ if I do “my $foo = *+* but role {method Str {"+"}}” 20:57
20:57 kaare_ joined
pat____ it seems to me that it get Stringified to early when I try it on interactive perl 20:58
benabik pat____: It may be that the REPL is using either .gist or .perl to perform the output instead of .Str 20:59
pat____ yeah let me specify
I want a variable ($foo) that returns e.g. 7 if called with $foo(3,4) and "+" if called ~$foo 21:00
benabik Oh that's interesting... 21:01
pat____ yeah
benabik nom: my $foo = *+* but role {method Str {"+"}}; say $foo(3,4); say $foo
p6eval nom 158bd0: OUTPUT«+␤WhateverCode.new()␤»
21:01 DataLinkDKT left
pat____ interesting, my nom here at home says "+" 21:02
benabik b: my $foo = *+* but role {method Str {"+"}}; say $foo(3,4); say $foo
p6eval b 1b7dd1: OUTPUT«7␤_block1093␤»
benabik I would say that both b and nom are wrong here... 21:03
21:03 DataLinkDKT joined 21:04 DataLinkDKT left, wamba left
pat____ nom: my $foo = *+* but role {method Str {"+"}}; say "{~$foo} || {$foo(4,3)}" 21:04
p6eval nom 158bd0: OUTPUT«_block1411 || +␤»
jnthn You need a method gist in the role, not a method Str 21:05
pat____ what's gist, don't know that
jnthn nom: my $foo = *+* but role { method gist {"+"}}; say $foo(3,4); say $foo
p6eval nom 158bd0: OUTPUT«+␤WhateverCode.new()␤»
pat____ it's the same
jnthn hmm 21:06
oh
nom: my $foo = (*+*) but role { method gist {"+"}}; say $foo(3,4); say $foo
p6eval nom 158bd0: OUTPUT«+␤WhateverCode.new()␤»
jnthn no, not that.
pat____ yeah though about that, but if the brackets were wrong they would be *+(* but role{…}) 21:07
jnthn Yes, that's what I was wondering about.
pat____ so, can we say that wasn't considered yet in the implementation, but my intetion was right 21:09
?
or am I missing something with that "but"
jnthn pat____: Looks like it should work. Not quite sure what the underlying bug is - I can't think of any reason off hand why it would fail to work.
benabik I'm also curious about the fact that $foo(3+4) actually returns "+" 21:10
jnthn Right. :)
pat____ me too
jnthn I really want to know what on earth is going on there.
pat____ that's why I came here
jnthn Given I did the mixins implementation, guess that's one for me to fathom... :)
benabik pat____: If jnthn doesn't know what's happening, then us mere mortals are doomed. ;-)
jnthn OH! 21:11
jlaire nom: my $foo = { ($^a + $^b) but role { method gist {"+"} }; say $foo(3,4); say $foo
jnthn bwaha
p6eval nom 158bd0: OUTPUT«===SORRY!===␤Unable to parse blockoid, couldn't find final '}' at line 1␤»
jlaire nom: my $foo = { ($^a + $^b) but role { method gist {"+"} } }; say $foo(3,4); say $foo
p6eval nom 158bd0: OUTPUT«+␤Block.new()␤»
jnthn nom: my $foo = *+*; $foo = $foo but role { method gist {"+"}}; say $foo(3,4); say $foo 21:12
p6eval nom 158bd0: OUTPUT«7␤+␤»
jnthn Yeah, currying gone wild
jlaire .oo( $foo but= ... )
jnthn The whatever currying is transitive.
Thus
* + * but blah
becomes
-> $x, $y { $x + $y but blah } 21:13
benabik !!
PerlJam rakudo needs a --deparse option :)
jnthn e.g. it actually sucks the but into the closure too :)
jlaire yay, I guessed right
jnthn Well, when you do * + * / 2 or something it's what you want :)
benabik There's no way to limit the extent of *, is there?
jnthn But given it just took me 5 minutes to realize what on earth was going on, maybe we should shove does and but on the exclude list for * currying...
Opinions welcome.
TimToady: ^^ 21:14
benabik: Don't think so
OK, so at least it's not actually a bug in mixins. *phew* :)
jlaire maybe we'll be able to define a special type of parens that does that as a macro some day
pat____ sorry for even considering that 21:15
benabik I wonder if * should stop at {}s. 21:16
jnthn pat____: Don't worry, it surprised me too.
benabik my $foo = {*+*} but "+"
jnthn benabik: Well, but now you have a closure that returns a closure...
jlaire you need to call it twice
tadzik 'evening 21:17
PerlJam jlaire: nah, just call the result of calling it
benabik jnthn: That's what it does _now_ sure.
jnthn o/ tadzik
benabik: I don't think I want to break that, or so goes my initial reaction.
21:17 cooper left
jnthn benabik: I agree some mechanism for saying "that's quite enough curry for today, thanks" could be handy. 21:17
But then you could always just write it another way.
PerlJam jnthn: I like your idea about does/but 21:18
jnthn: and will would have to be on that list too
jnthn nom: my $foo = {$^a + $^b} but role { method gist {"+"}}; say $foo(3,4); say $foo
benabik jnthn: My thought was that {} already generated a closure, so why do a closure inside a closure? If you still needed it you could do {{*+*}}
p6eval nom 158bd0: OUTPUT«7␤+␤»
jnthn PerlJam: will?
PerlJam: There's a will infix? 21:19
benabik nom: my $foo = {$^a + $^b} but "+"; say ~$foo; say $foo(3,4)
p6eval nom 158bd0: OUTPUT«+␤7␤»
jnthn I thought that was just a trait...
PerlJam oh, maybe.
jnthn PerlJam: I can happily add does/but to the list. Want to give @other time to ponder the issue first though.
benabik pat____: {$^a + $^b} but "+" seems to DTRT, although you'll still need the role { method gist {}} if you want to change how it responds to say instead of just ~. 21:20
pat____ okay, thanks a lot, so there's no way to do that with the whatever “*”? 21:21
benabik wonders if $foo but "foo" should override .gist in addition to .Str
pat____: Apparently only the two statement way (although that may change): my $foo = *+*; $foo = $foo but "+" 21:22
jnthn $foo but= role { ... } should work too
benabik nom: my $foo = *+*; $foo but= '+'; say ~$foo 21:23
p6eval nom 158bd0: OUTPUT«+␤»
benabik golfing is fun 21:24
pat____ nom: my $foo; ($foo = *+*) but= "+"
p6eval nom 158bd0: ( no output )
pat____ nom: my $foo; ($foo = *+*) but= "+"; say ~$foo, $foo(2,3) 21:25
p6eval nom 158bd0: OUTPUT«+5␤»
pat____ got it
benabik nom: (my $foo = *+*) but= '+'; say ~$foo, $foo(2,3)
p6eval nom 158bd0: OUTPUT«+5␤»
jnthn oh, cute. :)
yeah, = is on the no-curry list.
benabik Cute is one word for it. :-D 21:26
21:26 bluescreen10 left
jnthn Cute is good :) 21:27
benabik I can't think of a situation where $foo = {*+*} needing $foo()(1,1) is useful. 21:28
I can see why it's the case, but I don't see why that would be intended.
21:29 soh_cah_toa joined
pat____ maybe some "stop currying brackets= 21:29
like 《》 21:30
I can't think of a way macros could do that in an easy way 21:31
oh
21:31 pat____ left
benabik Rewrite 《foo》to {my $temp = foo; $temp}() ? 21:32
21:35 cooper joined
benabik nom: $_ = {*+*}; .()(1,1).say; $_ = {{*+*}}; .()(1,1).say; $_ = { {*+*} }; .()(1,1).say 21:36
p6eval nom 158bd0: OUTPUT«2␤2␤2␤»
benabik Strange to me that any number of {}s only produce one additional closure. 21:37
jnthn Well, it's an inline block inside the next one.
Same reason that
nom: { say 1 }; { say 2 }
p6eval nom 158bd0: OUTPUT«1␤2␤»
jnthn does that.
benabik But the first ones in an assignment create a closure. Okay. :-/ 21:38
$_ = {my $a={*+*}}; .()()(1,1).say 21:39
nom: $_ = {my $a={*+*}}; .()()(1,1).say
p6eval nom 158bd0: OUTPUT«2␤»
benabik Okay, enough (ab)using that.
21:47 pernatiy joined 21:53 benabik left 21:57 ab5tract left 22:01 kaare_ left 22:17 MayDaniel joined 22:20 dorlamm left 22:21 whiteknight joined 22:22 cooper left 22:23 ksi left
cognominal_ in nqp/src/NQPQ/Grammar.pm, I can't parse the <?[<[]> in | <sigil> <?[<[]> <postcircumfix> 22:23
from the context, I understand it is a zero width assertion that is true is the current char is either '<' or '[' but nevertheless I can't visually parse it 22:24
ho, noe I can! 22:25
*now 22:26
with the '?', I was unable to recognize the character class
jnthn heh, it looks...weird :) But yes :) 22:27
std: / <?[<[]> <postcircumfix> /
p6eval std 580b69a: OUTPUT«===SORRY!===␤Couldn't find terminator ] at /tmp/zUvDuqCkCw line 1 (EOF):␤------> / <?[<[]> <postcircumfix> /⏏<EOL>␤Parse failed␤FAILED 00:01 121m␤» 22:28
jnthn hm
too weird for STD...
Which probably means nqp's parser is being a bit too liberal.
Time for some rest & 22:36
22:39 replore_ left 22:41 envi_ joined 22:43 wolfman2000 joined
TimToady perl6: ({ say "hi"}) 22:51
p6eval rakudo 158bd0, niecza v10-66-g71325b0: ( no output )
..pugs b927740: OUTPUT«hi␤»
TimToady ({ * + * }) ought to work, since {} should be bare brackets
22:53 jlaire left
sjohnson TimToady: that sorta looks like a emoticon of someone with starry eyes kissing 23:09
donri or an owl 23:13
23:14 jlaire joined 23:15 jlaire left 23:16 jlaire joined
sorear good * #perl6 23:20
nornagest Good night. 23:25
23:25 nornagest left 23:27 aindilis left 23:28 MayDaniel left 23:29 aindilis joined
dalek ecza/serialize: 05b5bb6 | sorear++ | / (2 files):
Fix crashes when trying to compile CORE:: name references
 23:30
23:31 Chillance left 23:35 thou left 23:37 envi_ left
dalek ecza/serialize: 878afab | sorear++ | src/niecza:
Fix compile-time references to OUTER etc
 23:38
23:48 aindilis left, aindilis joined 23:52 thou joined 23:59 japhb_ joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!