»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.
00:01 mberends left 00:03 donaldh left
census Good night rindolf! :) 00:09
00:13 skids joined, cognominal left, cognominal joined 00:17 cognominal left
TimToady lue: why not just use eqv if you really want to check for equivalence? 00:35
(usually one doesn't want that though) 00:36
lue TimToady: I wanted to see if there was some way I could print what the compiler uses to check Name Equivalence
(just working my way through the spec; not actually coding) 00:37
TimToady that's just it, the compiler rarely checks name equivalence, but usually type containment is all that is wanted 00:42
othewise polymorphism doesn't work right
00:59 anuby joined 01:14 woosley joined
[Coke] yawns. 01:18
01:20 hypolin joined
diakopter yawns 01:21
[Coke] hah! 01:27
japhb dangit, you got me too. 01:34
TimToady sips his coffee bemusedly, and wonders if it will keep him awake too late tonight... 01:35
Pleiades` mmmmm coffee 01:39
rjbs misreads Name Equivalence twice as Near Equivalence.
I have written a very simple Perl 6 program to write a very simple z-machine program. I was pleased to get it working. I might keep putzing around. If anybody feels like wasting a little time on me, I'd be curious as to how badly my p5 accent is showing in gist.github.com/rjbs/5001268 01:49
TimToady don't need the parens on assignment to %for 01:51
01:51 apejens joined
rjbs woah 01:52
TimToady don't need the parens on the while either
01:53 adu joined
rjbs I felt like I ought to be using splice as a method there, but I failed to do it the first time 01:53
and used loop. then I did what you see
TimToady might be a way to recast the while as a for, though asking how many elements is left is harder
rjbs I feel really weird ditching those parens on %for=. I'll either have a hanging indent or a line with just ;. Ah, the tribulations of Perl! :) 01:54
TimToady using an accumulator like $result often means you could join '', gather ... instead 01:55
rjbs will consult perl6advent.wordpress.com/2009/12/2...r-of-eden/ for gather.
TimToady dunno why you have the last if, when the splice should go false 01:56
if it doesn't, maybe you found a bug
rjbs Oops, that's cruft from when I'd written it as a loop{...}
TimToady could use for @zchars -> $c1, $c2 = 5, $c3 = 5, then or in the 0x8000 outside the loop 01:59
mebbe 02:00
rjbs Yeah, I had tried that between my first while/splice and loop{...}, but was unsure how to proceed with the 0x8000 so I rejiggered it. Now that I have it working, good time to revisit that.
(@x -> $x, $y) seemed, gut-wise, more p6ish.
what do we call those guys on the rhs of the -> ? 02:01
TimToady parameters
it's just a lambda
rjbs . o O ( I can't believe it's just lambda!™ )
TimToady it's just applied funny so that it chews its parameters off the list and leaves the rest 02:02
well, and the syntax is magical, insofar as its occuring where a term would be expected 02:03
*it's
rjbs my $b = gather for @a -> $x, $y { take $x } 02:05
That's pretty cool.
dalek osystem: 3975e6f | grondilu++ | META.list:
Adding modular arithmetic module

This is a translation of Perl5's Math::ModInt		 02:07
rjbs TimToady: Actually, deleting that last on line 48 leaves me looping forever! 02:08
TimToady r: say so [].splice(0,3) 02:09
p6eval rakudo 6392d6: OUTPUT«True␤»
TimToady oopsie
yes, a bug
r: say +[].splice(0,3)
p6eval rakudo 6392d6: OUTPUT«3␤»
TimToady n: say +[].splice(0,3)
p6eval niecza v24-23-g0520c7c: OUTPUT«0␤»
TimToady rakudobug
census Has anybody worked with Jaro-Winkler distance between strings before? 02:10
TimToady suspects Jaro and Winkler have, for some reason... 02:11
rjbs How do I get the last element in an Array? 02:12
TimToady [*-1]
sorear @foo[*-1]
rjbs Thanks
TimToady nr: @foo[-1] 02:13
p6eval rakudo 6392d6: OUTPUT«===SORRY!===␤Variable '@foo' is not declared␤at /tmp/99m_Ufdgea:1␤------> @foo[-1]⏏<EOL>␤ expecting any of:␤ postfix␤ infix or meta-infix␤ infix stopper␤ statement end␤ statement modifier␤ …
..niecza v24-23-g0520c7c: OUTPUT«===SORRY!===␤␤Unsupported use of [-1] subscript to access from end of array; in Perl 6 please use [*-1] at /tmp/iUPBW6SdMV line 1 (EOF):␤------> @foo[-1]⏏<EOL>␤␤Parse failed␤␤»
TimToady nr: .[-1]
p6eval rakudo 6392d6: OUTPUT«Cannot use negative index -1 on Any␤current instr.: 'throw' pc 333436 (src/gen/CORE.setting.pir:149675) (src/gen/CORE.setting:9303)␤called from Sub 'sink' pc 365333 (src/gen/CORE.setting.pir:162404) (src/gen/CORE.setting:10525)␤called from Sub 'MAIN' pc 585 (src/ge…
..niecza v24-23-g0520c7c: OUTPUT«===SORRY!===␤␤Unsupported use of [-1] subscript to access from end of array; in Perl 6 please use [*-1] at /tmp/1PeW3ivn0D line 1 (EOF):␤------> .[-1]⏏<EOL>␤␤Parse failed␤␤»
rjbs updated gist.github.com/rjbs/5001268 with gather/take; pleased
TimToady that's a lot prettier 02:14
rjbs Could probably just make that a map
TimToady sure
and the ~= could be a join of a map too 02:15
which you could return directly without a $result variable 02:16
cotto Could I get either Rakudo commit bit or someone to apply a patch?
rjbs Hm. How do I map pairwise?
cotto gist.github.com/anonymous/5001423
rjbs I think I'm getting a bit overwhelmed at ==><== and ambidextrous flow.
TimToady rjbs: just supply two parameters to the lambda
well, I dunno if join is smart about Buf 02:18
actually, [~] would likely be better than join '' anyway 02:19
cotto pull request submitted; feel free to continue ignoring the gist 02:21
TimToady :) 02:22
TimToady has no rakudo commit bit, for the record...
[Coke] I do. those semicolons seems reasonable. 02:23
cotto I didn't intend to refer to anyone in particular.
rjbs Works, not useful: my $x <== map -> $_ { word($_) } <== @values;
Doesn't work, would think it useful: my $x = [~] <== map -> $_ { word($_) } <== @values;
02:24 lustlife joined
dalek kudo/nom: c3dc86e | cotto++ | src/ops/perl6.ops:
clean up some op definitions
 02:24
kudo/nom: 134b008 | (Will Coleda)++ | src/ops/perl6.ops:
Merge pull request #105 from cotto/nom

clean up some op definitions
[Coke] Thanks, cotto.
TimToady why not just my $x = [~] map { word($_) }, @values 02:25
rjbs Couldn't get that formulation, groped a lot. Maybe I was always missing the comma before @values.
TimToady r: map {...} 1,2,3 02:26
p6eval rakudo 6392d6: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/9p8rUavgui:1␤------> map {...} ⏏1,2,3␤ expecting any of:␤ argument list␤ postfix␤ infix or meta-infix␤ infix stopper␤ statement end␤ statement …
TimToady n: map {...} 1,2,3
p6eval niecza v24-23-g0520c7c: OUTPUT«===SORRY!===␤␤Strange text after block (missing comma, semicolon, comment marker?) at /tmp/3QVV91MteN line 1:␤------> map {...}⏏ 1,2,3␤␤Parse failed␤␤»
rjbs I saw that error quite a lot.
TimToady it could use an error more like niecza's there
dalek p: f802b49 | cotto++ | src/ops/nqp (3 files):
clean up some ops definitions
 02:27
p: 031801c | cotto++ | src/ops/nqp_dyncall.ops:
use a non-empty default case
p: 2ddd4dc | cotto++ | src/ops/nqp (3 files):
Merge pull request #75 from perl6/ops-fixes

clean up some ops definitions
TimToady map and grep are not magic syntax in p6
rjbs TimToady: must be nice!
TimToady people say they don't use indirect objects in p5 but map and grep are parsed that way, in fact 02:29
rjbs That's true of squintillions of things people say they don't use in p5. :)
TimToady but it does mean we all leave out the comma occasionally
rjbs Updated my gist one more time, feeling better. And now, back to p5. Blockers to track! 02:30
Thanks for the review!
TimToady soitinly
afk & 02:32
02:36 ObseLeTe left
adu hi TimToady 02:38
02:43 jaldhar joined 03:03 gdey left, census left
japhb rjbs, looking at your gist, it occurs to me that write-at could use a little more magic, so that writing the header is less wordy. You already have write-at slurping the buffers, but you don't use that for the START HEADER. 03:07
But it could even be a little smarter than that, by transparently handling Ints as words (since that seems to be the most common thing you need to use). 03:08
03:16 census joined 03:26 orafu left 03:27 orafu joined
rjbs japhb: turning ints into words, I think, would be more likely to lead to me accidentally using word instead of byte 03:35
japhb: I agree, re making most of the header one big call; that's an artifact of the construcion that I could fix
japhb: ...but note that many of the header segments are not contiguous.
It could be done with pairs... 03:36
swarley I got bored, so I'm writing a PASM compiler. I suppose that's one way to spend an afternoon (Making something useless) 03:38
03:39 crazedpsyc is now known as Getly, Getly is now known as crazedpsyc 03:51 adu left
diakopter swarley: O_O 03:51
diakopter nudges you back toward 6model/qast
swarley Don't worry. I'm not moving away from it.
I just wanted to see what it looked like using libjit as a base instead of what was already in place 03:54
diakopter you're a jit 03:57
swarley :( I know
diakopter it was a joke of the absurd.. 03:58
swarley i'd hate to see parrot die, it's such a neat project.
diakopter if my last msg had been 1s later.. 04:00
swarley haha. 04:01
diakopter TimToady: that STD_P5 grammar would be helpful pretty soon 04:03
pugs: { use v6; say 4; { use v5; print 33; { use v6; say 44 } } } 04:05
p6eval pugs: OUTPUT«4␤Error eval perl5: "sub { use ops (':default', 'binmode', 'entereval');; print 33; {use v6; say 44 } ␤}"␤*** 'print' trapped by operation mask at (eval 2) line 1.␤␤Can't call method "" on an undefined value.␤»
diakopter TimToady: that oughtta keep you occupied a week or few :P 04:12
04:13 dayangkun joined
diakopter TimToady: oh, and, it needs to support source filters and extensions that hook the parser ;) 04:19
04:24 xinming_ joined
swarley ew filters.. 04:27
04:28 xinming left 04:38 rking joined 04:41 SamuraiJack joined 04:44 preflex_ joined 04:46 preflex left, preflex_ is now known as preflex 04:52 DrEeevil joined
labster r: class A is rw { has $.foo; }; class B is A is rw { method bar { $.foo = 1 }; }; say B.new.bar; 04:55
p6eval rakudo 134b00: OUTPUT«1␤»
labster r: role A is rw { has $.foo = 2; }; class B is A is rw { method bar { $.foo = 1 }; }; say B.new.bar; 04:56
p6eval rakudo 134b00: OUTPUT«Cannot assign to a readonly variable or a value␤ in method bar at /tmp/B6IdxBKuTI:1␤ in block at /tmp/B6IdxBKuTI:1␤␤»
grondilu tadzik: panda shows way too many error lines when failing to find a module
labster agreed grondilu. 70 some, the last time I checked.
r: role A is rw { has $.foo }; class B is A is rw { method bar { $.foo = 1 }; }; say B.new.bar; 04:57
p6eval rakudo 134b00: OUTPUT«Cannot assign to a readonly variable or a value␤ in method bar at /tmp/5pxKV9ssHV:1␤ in block at /tmp/5pxKV9ssHV:1␤␤»
labster role can't take 'is rw'?
PerlJam I'm not sure what it would mean. 05:01
(I'm not sure what "is rw" means on a class even)
The only place "class.*is rw" appears in the spec even is an example that illustrates that traits are applied with "is" 05:03
S12:134 fwiw
TimToady see S12:703 05:06
PerlJam ah
05:10 census left
labster Roles are class-like in a lot of places, but I guess this is not one of them. 05:10
PerlJam labster: They could be. It's only a spec change away. ;) 05:21
sorear i see no reason why "is rw" *shouldn't* be extended to roles
it's not like it means "is open" 05:22
05:32 Chillance left 05:38 hypolin left
grondilu rn: sub foo($i) { state @p := grep &is-prime, 2 .. *; return @p[$i] }; say foo 5; 05:41
p6eval niecza v24-23-g0520c7c: OUTPUT«===SORRY!===␤␤Cannot use bind operator with this LHS at /tmp/J88IlP4BpB line 1:␤------> $i) { state @p := grep &is-prime, 2 .. *⏏; return @p[$i] }; say foo 5;␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza…
..rakudo 134b00: OUTPUT«(timeout)»
grondilu rn: sub foo($i) { my @p := grep &is-prime, 2 .. *; return @p[$i] }; say foo 5;
p6eval rakudo 134b00, niecza v24-23-g0520c7c: OUTPUT«13␤»
moritz \o 05:42
grondilu rn: sub foo($i) { state @p = grep &is-prime, 2 .. *; return @p[$i] }; say foo 5;
p6eval rakudo 134b00, niecza v24-23-g0520c7c: OUTPUT«(timeout)»
grondilu is grep supposed to be eager? 05:43
TimToady no, but assignment is
grondilu why assignement should be eager? I will always hang. 05:45
(well at least I can't see an example where it wouldn't) 05:46
.-1s/I will always/It will always/
moritz because in too many places it would surprise too many people if it were lazy
TimToady I'm trying to figure out why the := doesn't work; maybe sink is not treating is special as it does my? 05:47
grondilu moritz: don't you think it is worse if the program hangs?? 05:48
05:48 SamuraiJack left
moritz grondilu: no. It's easy to find out why, and dealing with infinite lists/ranges is relatively rare 05:48
PerlJam "doctor, it hurts ..."
TimToady r: sub foo($i) { (state @p := grep &is-prime, 2 .. *)[$i] }; say foo 5;
p6eval rakudo 134b00: OUTPUT«13␤»
TimToady r: sub foo($i) { state @p := grep &is-prime, 2 .. *; return @p[$i] }; say foo 5; 05:49
p6eval rakudo 134b00: OUTPUT«(timeout)»
TimToady r: sub foo($i) { so state @p := grep &is-prime, 2 .. *; return @p[$i] }; say foo 5;
p6eval rakudo 134b00: OUTPUT«13␤»
moritz r: sub foo($i) { my @p := grep &is-prime, 2 .. *; return @p[$i] }; say foo 5; 05:50
p6eval rakudo 134b00: OUTPUT«13␤»
TimToady yes, seems to be sinking it
not treating state like my
05:50 benabik joined
TimToady moritz: maybe the desugaring of 'state' is introducing a sink somehow? 05:52
or hiding the state from the 'my' recognizer?
grondilu I guess I should just get used to using ':=' more often.
moritz TimToady: that seems to be the case, yes
05:54 hypolin joined 06:06 kaleem joined 06:20 rindolf joined 06:22 DrEeevil is now known as bonsaikitten 06:45 erkan left 06:46 rindolf left
dalek kudo/nom: cd95e40 | moritz++ | src/Perl6/Actions.pm:
/./ and rx/./ literals now match against $_ in sink context
 06:48
kudo/nom: 5dfb733 | moritz++ | src/Perl6/Actions.pm:
avoid sinking of state initializers
06:49 rindolf joined
TimToady moritz++ for being moritz 06:57
07:01 sqirrel joined 07:04 FROGGS joined
rindolf Hi all. 07:04
TimToady: hi.
TimToady
.oO(I'm not part of all?)
 07:05
tadzik grondilu: could you open a bug please? 07:06
I can look at that in some spare minute 07:07
07:10 rindolf left, rindolf joined 07:16 rindolf left 07:17 rindolf joined
sorear o/ rindolf, sqirrel, FROGGS 07:18
rindolf sorear: hi. 07:22
grondilu tadzik: I don't know how to open a bug about panda
rindolf TimToady: “all for TimToady, and TimToady for all.”
07:22 rindolf left 07:23 rindolf joined, hypolin left
FROGGS morning @all 07:23
07:29 rindolf left, rindolf joined 07:35 hypolin joined 07:37 rindolf left, rindolf joined, salv0 joined 07:43 rindolf left 07:44 rindolf joined 07:56 rindolf left, rindolf joined 07:58 SamuraiJack joined 07:59 mberends joined
sorear mberends! \o/ 08:00
mberends \o sorear!
08:03 am0c_ joined
sorear :D 08:03
08:03 am0c_ is now known as am0c
FROGGS grondilu: github.com/tadzik/panda/issues 08:04
mberends sorear: have you looked at rurban++'s p2 and _why++'s potion underneath it? perl11.org/p2/potion.html
08:07 sitaktif joined, rindolf left 08:08 rindolf joined 08:11 cosimo joined
sorear mberends: not really. 08:11
i've decided to get out of the language design/implementation game. not my cuppa
FROGGS sorear: may I ask what's your "cuppa" nowadays? 08:12
mberends sorear: I understand. I read that you were happily writing in JavaScript nowadays. 08:13
08:15 domidumont joined
sorear FROGGS: i would like to be in a field where programs are the means, not the end. telescope development would be ideal if I can make that happen 08:17
FROGGS: my current $work involves using programs to sell shoes. It's suprisingly fun. 08:18
08:18 rindolf left, rindolf joined
FROGGS sorear: I totally agree 08:19
08:22 anuby_ joined 08:23 anuby left, DarthGandalf joined
FROGGS at work I like that if you made program A, B and C (internal websites), then you have several opportunities to comfort the employees and reduce chores just by connecting A, B and C 08:30
" I love it when a plan comes together."
08:37 domidumont left, domidumont joined 08:39 FROGGS left
timotimo good day 08:41
08:47 fhelmberger joined 08:51 FROGGS joined 08:53 am0c left 08:54 census joined 08:55 am0c joined 09:04 census left 09:36 dakkar joined 09:38 cognominal joined 09:40 ObseLeTe joined 09:43 labster left 09:57 spider-mario joined 10:01 hypolin left 10:05 daxim joined 10:08 marloshouse joined 10:11 anuby_ left 10:19 REPLeffect joined 10:21 donaldh joined 10:30 FROGGS left 10:35 erkan joined, erkan left, erkan joined
arnsholt jnthn: What are you using as the underlying Java type for nqp::list in NQP-JVM; arrays or one of the List<T> implementations? 10:35
tadzik I think it's an array
arnsholt: github.com/jnthn/nqp-jvm-prep/blob...tance.java 10:36
10:36 dayangkun left
arnsholt Cool. A Java array, wrapped with a bit of logic to handle accesses 10:37
tadzik aye
10:37 FROGGS joined
arnsholt I think that'll fit with what I'm pondering 10:38
And Java guarantees the widths of byte, short, int and long. Even better! 10:39
FROGGS well, thats the job of a sane VM 10:40
arnsholt Well, Java isn't a VM, so it could just say that it should be -at least- those sizes 10:41
FROGGS ohh, yeah, thought you're talking about JVM
arnsholt Common Lisp for example only mandates a minimum width for its numerics
masak good forenoon, #perl6 10:42
FROGGS masak! \o/
10:42 REPLeffect left
tadzik hello! 10:43
donaldh arnsholt: what are you pondering? 10:44
arnsholt donaldh: Adding a buffer type to NQP
donaldh oh, it looks like VMArrayInstance reimplements a lot of what is already available in the Java collections.
FROGGS nqp: say("foobar" ~~ MAKE_REGEX("fo+")) 10:45
p6eval nqp: OUTPUT«Could not find sub &MAKE_REGEX␤current instr.: '' pc 46 ((file unknown):38) (/tmp/paionMTqV_:1)␤»
arnsholt Rakudo needs better support to implement Buf properly, and NativeCall needs better Bufs to handle blobs better
jnthn morning o/ 10:52
Or forenooning at least
timotimo got a second to merge github.com/rakudo/rakudo/pull/104 ? :)
arnsholt Just in the nick of time for forenoon =)
FROGGS hi jnthn 10:53
arnsholt A timo-commit to my repo =D 10:54
timotimo untested, alas
though i imagine with in-proc sockets one could build a bit of testing code without even having to make multiple processes
arnsholt I think that gives timotimo++ the distinction of first-ever committer that isn't me to one of my repos =)
Yeah, that might work. I looked a bit at the S32-io socket tests last night for inspiration as well 10:55
timotimo whoa! yays :)
jnthn timotimo: I think it would be more general to put the (...) parsing where STD does - in typename: github.com/perl6/std/blob/master/STD.pm6#L2547 10:56
timotimo hm, i had a look at that, but for some reason i decided not to do that. i should reconsider i guess
jnthn Well, then it'll catch it everywhere <typename> is used 10:57
Also it's in the right place for the luck person who gets to implement the feature ;)
arnsholt timotimo: A couple of comments: The last argument of zmq_setsockopt is a size_t, so the last Perl 6 type should be just int, rather than CArray[int] (get sockopt has size_t* for the last one, thus CArray[int])
timotimo oh, right, because it's not an out parameter 10:58
i'll fix it real quick
arnsholt And you're nulling the values, rather than setting them to $value =)
Do $val[0] = $value instead
timotimo my only excuse is that it's late :) 10:59
er, it was when i copypasted that code
10:59 cognominal left
arnsholt I made some brainos as well, so you're excused I think =) 10:59
10:59 REPLeffect joined
timotimo i ended up copypasting the get function and changed every get to set ... in the get function >_< 10:59
arnsholt =D
Classic copy-pasta
donaldh jnthn: are we committed to BCEL or could we consider alternatives? 11:00
jnthn donaldh: Depends on the alternative and why. 11:01
donaldh jnthn: I'm not sure invokedynamic is fully implemented in BCEL yet. Still digging through it.
jnthn donaldh: Also, I don't much feel like re-writing JASTToJVMBytecode.java...it wasn't much fun to do in the first place ;)
donaldh jnthn: And BCEL seems to be less actively developed than ASM. Also ASM has a lot of supporters 11:02
arnsholt jnthn: I've added my suggestion for NQP buffer ops to gist.github.com/arnsholt/4963237 (no rush to comment though, won't have much time 'till over the week-end I think)
donaldh jnthn: asm.ow2.org/ 11:03
jnthn: I'll keep working on BCEL for now but I might also try an ASM port of JASTToJVMBytecode.java on a branch. 11:04
11:05 cognominal joined
jnthn donaldh: If you're interested in working on that, feel free. Having just taken a look, yes it seems quite active and also the license looks fine. 11:05
donaldh jnthn: yep.
11:07 REPLeffect left
jnthn donaldh: The other BCEL usage is in P6opaque 11:07
donaldh jnthn: that looks fairly hairy ;-) 11:09
11:10 ObseLeTe left
jnthn donaldh: Yeah, it's generating a backing class 11:14
donaldh nice 11:15
timotimo jnthn: good tip, it works well now.
running the spectests now, but i don't see how it even could fail anything :P 11:19
FROGGS the only thing I could image that it fails parsing test files
timotimo yeah, but it would have failed to parse before, so all those parts should have already been fudged for rakudo
11:20 REPLeffect joined
FROGGS no, it might not since your changed the grammar 11:20
timotimo might not what? fail to parse? but if it does parse it would NYI immediately 11:21
FROGGS timotimo: everytime you change the grammar it is possible that there is a piece of code that is unparseable then.... 11:23
11:23 cognominal left
daxim I don't like the location where panda bootstrap put the library files (~/.perl6), how do I override that? 11:24
11:25 REPLeffect left
timotimo FROGGS: with this change, i feel somewhat confident that it would work for any input that was parsable before :D 11:25
tadzik daxim: I don't think panda puts stuff there anymore 11:26
not since we got CUSTOM_LIB and all that
daxim I'm running rakudo 2012.12 and panda from github HEAD
tadzik hm
timotimo (but i'm running the spectest suite because i know from past experience i am dumb) 11:29
tadzik r: say %*CUSTOM_LIB.perl
p6eval rakudo 5dfb73: OUTPUT«("perl" => "/home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6", "vendor" => "/home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6/vendor", "site" => "/home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6/site", "home" => "/home/p6eval/.…
tadzik daxim: do you have that set? Can you confirm that panda is still putting stuff there, or may it be that those are just old files lying around? 11:31
11:32 cognominal joined
daxim yes, I have that set: paste.scsys.co.uk/231735 bootstrap already put stuff into ~/.perl6, looks like panda itself and its deps 11:32
I have no old files, I just installed rakudo 20 minutes ago on this machine, it doesn't come with panda, so I bootstrap from the GH checkout 11:33
tadzik huh, hm
2012.12 sounds fresh enough for %*CUSTOM_LIB and all that 11:34
daxim: above everything, panda should respect DESTDIR if you set that
daxim then how do I set DESTDIR?
tadzik so the install location is always overridable; if not, then it's a bug
DESTDIR=./foolib panda install Stuff 11:35
that ought to work
FROGGS timotimo: I just meant in general that there is only this one chance to break anything... 11:37
tadzik daxim: bootstrap put stuff in ~/.perl6, from fresh git checkout?
grondilu is recompiling rakudo and realizes he will have to rebootstrap panda. This has consistently been failing lately. 11:38
daxim yes. with DESTDIR, the files were installed there, but then I run into a different problem
tadzik that sounds veery wrong. There is no ".perl6" in panda these days, as I see it
are you sure it's github.com/tadzik/panda, master, up-to-date?
moritz r: say Int("42").WHAT 11:39
p6eval rakudo 5dfb73: OUTPUT«Int()␤»
daxim confirmed. now let me paste the follow-up problem
tadzik go ahead 11:40
daxim paste.scsys.co.uk/231736
tadzik r: say @*INC.perl
p6eval rakudo 5dfb73: OUTPUT«Array.new("/home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6/site/lib", "/home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6/vendor/lib", "/home/p6eval/nom-inst/lib/parrot/4.10.0-devel/languages/perl6/lib", "/home/p6eval/.perl6/2013.01-141-g5d…
daxim even though I changed the installation directory, it still assumes ~/.perl6 at run time
tadzik /home/daxim/.perl6/2012.12/lib 11:41
that's odd
timotimo All tests successful. - now my branch can be merged :)
tadzik or may be that custom-lib put that there
moritz?
FROGGS timotimo++
11:41 REPLeffect joined
tadzik daxim: yeah, most puzzling 11:41
daxim: I assume you don't set PERL6LIB yourself in any way?
daxim echo $PERL6LIB is empty 11:42
moritz tadzik: it does that as a fallback if the site dir isn't writable
tadzik ah, there we are 11:43
daxim: so, if you use DESTDIR then everything gets installed to it, but rakudo will not know that it should look for libraries there 11:44
11:44 xenoterracide left
moritz so you have to set PERL6LIB to $DESTDIR/lib ? 11:44
tadzik I think so
moritz or to $DESTDIR?
moritz not sure
daxim aha, let me try
tadzik /lib 11:45
daxim yes, with lib
tadzik but hm, you say moritz that CUSTOM_LIB will fallback to /home/user/.perl6/version, right?
ah, ok
I understand
11:45 REPLeffect left
daxim gosh, there's *still* a ~/.perl6 dir 11:46
it has a ~/.perl6/2012.12/panda/projects.json file in it
moritz tadzik: yes, where $version is really the whole git-describe string
just for releases it simpliefies to the release version
tadzik daxim: ah, yes. It usually goes to the site dir too :/ 11:47
I guess you can't (easily) override that
moritz had an apicoectomy on two teeth today 11:48
tadzik github.com/tadzik/panda/blob/maste.../panda#L96
you can mangle that though
perhaps we should make it configurable too
PANDADIR=somewhere panda make sandwich or such
but no, that should just work
hrm
tadzik hugs moritz 11:49
11:49 xenoterracide joined
daxim you're looking for bootstrap.pl --prefix=somedir, not env variables, that's how everyone else does it 11:49
tadzik yeah, that makes sense
it's not fixing the problem though (yet) 11:50
unless the bootstrap script will modify the binary it's installing 11:51
moritz it could.
tadzik daxim: could you write me a short spec of what would like for panda to be capable of, the interface and the desired behaviour?
I can take a look at that today or tomorrow
daxim "see GNU autotools" 11:52
tadzik think of your perfect... ah
11:52 REPLeffect joined
tadzik that's sort of a different story, I think 11:52
you fire autotools and install some binaries. You don't install autotools themselves for later use
11:53 integral joined, integral left, integral joined
tadzik but I guess bootstrap.pl can be the autotools here, and panda being panda 11:53
11:53 PerlJam joined
daxim ok 11:54
tadzik I agree that the experience should be improved 11:55
if I myself consider panda fragile and confusing, that means stuff is bad :)
moritz but it's a healthy perception 11:56
tadzik there's too much magic there and I stopped completly understanding it at some point 11:57
daxim at least I get decent replies here and gain an understanding what's wrong, you needn't worry because you're still a parsec ahead of python-pip and whatever "sup" forces ruby to do. I'll tell you the horror stories over a beer
tadzik gotcha :)
so I can expect you on the Polish Perl Workshop? :)
daxim I'll know for certain next week
tadzik awesome
seen brianwisti 12:00
dalek kudo/nom: 9c59a13 | (Timo Paulssen)++ | src/ (2 files):
turn LTA error for coercive type declarations into NYI.
moritz timotimo: thank's for your patches. I --squash'ed them all into one and applied
12:00 SmokeMachine joined
timotimo that was acceptable 12:01
no need to have embarassing brainos all over the commit history :)
maybe one day i'll look back at 20 year old commit data and think "wow, i didn't do as many mistakes as i thought!"
moritz and it's only one deletion and two inserts that way :-)
but less karma, so I need to timotimo++ manually :-)
timotimo haha, all right :) 12:02
moritz funny, the pull request on github still shows "applies cleanly" 12:03
12:04 stephen joined, stephen is now known as Guest13313 12:05 Guest13313 left 12:07 cognominal left, census joined
census Hi! Has anybody used EM algorithm before? 12:07
12:09 cognominal joined
tadzik EM? 12:09
census Expectation-Maximization algorithm 12:10
moritz elektromagnetic?
tadzik wonders if he knows the Polish name of it 12:12
moritz wikipedia only has the Russian name :-)
tadzik but it doesn't look familiar
moritz hasn't used it either 12:13
the german wiki page for EM algorithm has visual example data of eruptions of the Old Faithful geyser 12:15
so now I'm reading about that one
census interesting . . . 12:16
moritz I really have to visit yellowstone at some point
and Iceland 12:17
timotimo hm, i think i remember expectation maximization from somewhere. maybe school.
12:18 cognominal left
moritz would have loved to attend a school that does such stuff :-) 12:19
arnsholt EM was mentioned in some of my NLP courses
12:20 REPLeffect left
census i thought EM was common in CS and machine larning 12:20
learning
timotimo oh, no, that was university
moritz census: it probably is, but there just so many things that are common that you can't know them all :-)
12:26 cognominal joined 12:30 shinobicl joined
dalek p: 6f3692d | jnthn++ | src/HLL/Grammar.pm:
Port EXPR to NQP. Much code from kboga++.
 12:32
p: 866304d | jnthn++ | src/HLL/Compiler.pm:
Toss long-broken --dumper option and related code.

Looking at the code, it almost certainly hasn't worked for a long time.
12:35 cognominal__ joined 12:36 cognominal left 12:37 REPLeffect joined
dalek p: 688763c | jnthn++ | src/HLL/Grammar.pm:
Toss outdated comment.
 12:38
jnthn HLL::Grammar is free of Q:PIR now \o/
moritz \o/
Harzilein where is Q:somelang explained btw? 12:39
12:40 cognominal__ left
timotimo in the nqp documentation? :P 12:41
12:41 rindolf left, rindolf joined
jnthn Harzilein: Quoting constructs are probably explained in S02. But Q:PIR handling is just hard-coded into the grammar rather than anything more clever. 12:41
12:43 cognominal__ joined
Harzilein jnthn: well, niecza also has Q:CgOp and i was interested in reading up on that. 12:47
dalek p-jvm-prep: bb49a80 | jnthn++ | lib/QAST/JASTCompiler.nqp:
Fix non-scanning.
 12:55
p-jvm-prep: 6375842 | jnthn++ | t/nqp/45-smartmatch.t:
We now pass 45-smartmatch.t.
FROGGS (EXPR to NQP)++ 13:01
kboga++
13:08 domidumont left 13:09 domidumont joined
moritz seems there are three Q:PIR groups left 13:12
FROGGS well, if I could read pir I would try to translate that, but ... 13:15
pir is like asm to me
moritz in HLL::Compiler, method lineof does a cache lookup, and if it fails, a binary search 13:16
to map an index into a string into a line number 13:17
13:17 dayangkun joined 13:18 tomaw joined
FROGGS moritz: hmm, I may give it a try after my "current project" 13:20
13:21 woosley1 joined 13:28 ObseLeTe joined 13:30 skids left 13:34 stevan_ joined 13:37 mtk left
census rindolf: I'm sorry your friend reset your connection 13:40
13:40 toebu joined
rindolf census: what? 13:41
census: where?
census rindolf [~Shlomi@bzq-109-64-185-17.red.bezeqint.net] has quit [Read error: Connection reset by peer]
rindolf census: are you jumping into conclusions?
13:41 toebu left
rindolf census: yes, it's probably due to lost packets. 13:41
census: I didn't see it naturally.
census: IRC disconnections happen (surprise!).
census oh ok, i'm sorry you experienced that. that is all. 13:42
13:42 domidumont left
mberends peer is not rindolf's friend 13:42
FROGGS (peer is a first name in northern countries)
rindolf use.perl.org/use.perl.org/~Mark+Lei...rnal/35016 - sucks - it's down. :-(
13:42 toebu joined 13:43 domidumont joined
rindolf web.archive.org/liveweb/http://use....rnal/35016 - nothing here either. 13:43
rindolf hunts and kills Pudge.
13:43 daxim_ joined
rindolf web.archive.org/web/*/http://www.shlomifish.org/ - everything from 2004. 13:44
moritz FROGGS: isn't the first name spelled "Per"? 13:45
rindolf web.archive.org/web/20041013231902/...ifish.org/
Wow, what great CSS skills I had back then!
masak -me .oO( friend is not rindolf's peer )
masak .oO( friend is not rindolf's peer )
13:46 mtk joined
masak FROGGS: I can't recall ever seeing the name spelled "Peer", though. only "Per". 13:46
well, there's "Peer Gynt", but he's fictional. 13:47
13:47 daxim left
mberends .oO( any peer of rindolf is my peer) 13:47
timotimo a peer of my peer is my peer.
13:48 toebu_ joined
mberends to peer or not to peer 13:48
mberends peers at #perl6
FROGGS masak: my friend's name is Lars Peer Davidson
moritz and when he talks to another guy named Peer, it's P2P :-) 13:49
FROGGS *gg*
rindolf This reminds me that I have a series of novels in mind where a female adventurer and swordswoman becomes a Peer (= a Lady, or rather a she-Lord). 13:50
nwc10 jnthn: I have no ICU, so t/nqp/46-charspec.t fails. Otherwise, all suitably boring. 13:51
rindolf I was told in .uk, women cannot become Lords, but in my imaginary world they can.
13:52 toebu left
FROGGS female Lords in rindolf's imagination? 13:52
huh
rindolf FROGGS: are you British? 13:53
masak FROGGS: is your friend Norwegian?
13:53 woolfy joined
FROGGS rindolf: no 13:53
masak: no
rindolf In .uk though, females can become dames which is the female equivalent of a knight.
FROGGS: OK.
mberends woolfy !
woolfy Martin...! 13:54
mberends woolfy ! woolfy ! (sorry masak)
woolfy Please ignore me, just "listening" to the wizards of Perl 6...
mberends ignores
woolfy :-)
FROGGS masak: some ancesters might be from there, be he isnt
rindolf: in .de there were Damen too, but I'm not sure they had a sword. I believe they were just the wifes of knights 13:56
jnthn nwc10: yay :)
rindolf FROGGS: OK. 13:57
FROGGS: well, in my imaginary world (an anti-Fantasy setting set in something like late middle-ages / early renaissance), there are some women warriors and there are many lords, both men and women who are scholars. 13:58
FROGGS .oO( first perl 6 rpg made by rindolf? ) 14:00
masak woolfy: oh, a self-confessed lurker! welcome ;) 14:03
pmichaud I'm lurkng also. :)
masak pmichaud: yeah, it's not like big chunks of Rakudo source code are yours, or something :P
woolfy Thanks, masak!
rindolf FROGGS: it's not an RPG.
FROGGS: it's not even an adventure game. 14:04
FROGGS rindolf: not yet :P
masak is not a wizard, more like a warrior/archer: strangelyconsistent.org/blog/perl-6-is-my-mmorpg
FROGGS masak: the link to the learning curve image is borken 14:06
14:06 daniel-s joined 14:07 toebu_ is now known as toebu
woolfy Masak is a wizard in my eyes 14:08
PerlJam woolfy: do you need help having that removed so that you may see better? ;) 14:10
woolfy Michaud, Worthington, Lenz, Berends, Audrey, Larry... all wizards... (please don't feel offended if I did not mention your name).
PerlJam... I should learn how to write Perl 6... I already downloaded the book Using Perl 6. But for some reasons (not having downloaded and installed Perl 6 to name one), I did not get far yet... So people like you are wizards to me. 14:12
FROGGS woolfy: ya, it's simply not enough to have the standard book of spells grade one in the shelf :o) 14:13
PerlJam heh 14:14
14:14 PacoAir joined
mberends woolfy has the world's largest collection of Perl spells on the shelves, srsly 14:16
14:16 PacoAir left 14:17 PacoAir joined
woolfy And hoping to one day... having a lot of Perl 6 books added to those shelves! 14:17
moritz masak: an archer can be a wizard in his own right too :-) 14:18
masak FROGGS: thank you. seems the image is archived here: web.archive.org/web/20090718140949/...gCurve.jpg
moritz: true. I've seen archers that were sufficiently advanced as to be indistinguishable from wizards ;) 14:19
FROGGS hehe 14:20
nice one
masak r: say "gonna show woolfy some spells :)" 14:21
p6eval rakudo 9c59a1: OUTPUT«gonna show woolfy some spells :)␤»
masak r: .say for 10 ... 1; say "LIFTOFF!"
p6eval rakudo 9c59a1: OUTPUT«10␤9␤8␤7␤6␤5␤4␤3␤2␤1␤LIFTOFF!␤»
14:21 Psyche^ joined
masak r: my $third = 1/3; say "A third: $third"; say "One: {$third * 3} -- look, no loss of precision!" 14:22
p6eval rakudo 9c59a1: OUTPUT«A third: 0.333333␤One: 1 -- look, no loss of precision!␤»
pmichaud r: .say for (10 ... 1), "LIFTOFF";
p6eval rakudo 9c59a1: OUTPUT«10␤9␤8␤7␤6␤5␤4␤3␤2␤1␤LIFTOFF␤»
14:23 ObseLeTe left
nwc10 $ perl -E 'my $third = 1/3; say "A third: $third"; say "One: @{[$third * 3]} -- look, no loss of precision!"' 14:24
A third: 0.333333333333333
One: 1 -- look, no loss of precision!
masak r: my @deck = ('ace', 2..10, <knave queen king>) X~ ' of ' X~ <hearts diamonds clubs spades>; say @deck.pick(5).perl 14:25
p6eval rakudo 9c59a1: OUTPUT«("3 of clubs", "8 of clubs", "9 of hearts", "3 of diamonds", "8 of hearts").list␤»
nwc10 masak: you're going to have to try harder than that to demonstrate that it's not just IEEE being really good at it
s/it$/what they do/
masak nwc10: troo.
pmichaud r: my $third = 1/3; say "A third: ", $third; #how about this?
p6eval rakudo 9c59a1: OUTPUT«A third: 0.333333␤»
masak nwc10: do you have a good example, ooc?
pmichaud er... hmm
r: my $third = 1/3; say "A third: ", $third.gist; #how about this?
p6eval rakudo 9c59a1: OUTPUT«A third: 0.333333␤»
pmichaud r: my $third = 1/3; say "A third: ", $third.Str; #how about this?
p6eval rakudo 9c59a1: OUTPUT«A third: 0.333333␤»
masak pmichaud: we've hidden the rattiness pretty deep ;)
nwc10 masak: no, sorry. I'm not sure where the best place to get the IEEE stuff to break down is
pmichaud masak: is that a change from a year ago or so? 14:26
14:26 Patterner left, Psyche^ is now known as Patterner
masak pmichaud: think so. 14:26
pmichaud: it's for people who don't know and don't care about Rats.
pmichaud r: my $third = 1/3; say $third.perl 14:27
p6eval rakudo 9c59a1: OUTPUT«1/3␤»
pmichaud \o/
FROGGS r: my $third = 1/3; say $third * 3
p6eval rakudo 9c59a1: OUTPUT«1␤»
masak learns that Mendeleev was inspired by playing solitaire when inventing the periodic table
Wikipedia++
FROGGS r: my $third = 1/3; say $third.Str * 3 14:28
p6eval rakudo 9c59a1: OUTPUT«0.999999␤»
FROGGS coo
l
masak: well, dont trust wikipedia blindly
masak FROGGS: who said anything about trusting Wikipedia blindly? :) 14:29
FROGGS :o)
14:33 PerlJam left, p6eval left, kaare_ joined, PerlJam joined 14:34 Util joined
masak welcome back, feather. 14:36
14:37 pmurias joined 14:38 [Coke] joined 14:39 p6eval joined, ChanServ sets mode: +v p6eval 14:40 benabik left, domidumont left
pmurias arnsholt: adopt and disown seem something that would be hard/annoying to implement on some platforms (Like V8). Allocating none garbage collected memory seems easier than turning a garbage collected chunk of memory into something that can be free()'ed 14:40
arnsholt If all you have is GCed memory, you're C interop is going to be hard anyways 14:41
s/you're//
14:42 domidumont joined
arnsholt But if those are going to be hard on a particular platform you could make them noops and make sure users know of the limitation 14:42
pmurias I can allocated none GCed memory under V8, but I can't turn GCed memory into none GC one 14:43
s/allocated/allocate/
arnsholt Well, I'm not going to convert non-GC memory into GC memory on Parrot either. What I'll do is free the non-GC memory on object destruction if it's adopted 14:45
pmurias so things that are that can be disowned/adopted come from C land?
s/that are/ 14:46
pmurias can't type today
arnsholt Yeah. Except the buffer stuff, this is all motivated by NativeCall/Zavolaj
The buffer stuff is in there because I need it for other NativeCall things, even though it's a bit different 14:47
Does that answer your concerns? 14:51
14:52 woosley1 left
pmurias arnsholt: yes, if the adopted disown object are explicitly allocated in non GCed memory it seems sane 14:53
* objects
arnsholt Yeah, this is only for NativeCall objects 14:54
pmurias arnsholt: my one concern is that adopt seems to assume that all C land things need to be destroyed by free()
which might not be the case for all of them
arnsholt Well, non-pointer things are marshalled to natively typed stuff, which don't need special handling 14:59
14:59 skids joined
arnsholt This is only for pointer-typed stuff 14:59
timotimo hm, should i give a NYI for this here, too? 15:01
r: my Hash{Any} %a;
p6eval rakudo 9c59a1: OUTPUT«===SORRY!===␤Type 'Hash' is not declared␤at /tmp/_gShenxJjD:1␤------> my Hash⏏{Any} %a;␤Malformed my␤at /tmp/_gShenxJjD:1␤------> my Hash⏏{Any} %a;␤ expecting any of:␤ scoped declarator␤␤»…
timotimo oh, huh 15:02
unfortunately, i haven't yet looked at what exactly the { } do after a typename
moritz r: my %h := Hash{Int}.new; %h{3} = 4; say %h.perl 15:05
p6eval rakudo 9c59a1: OUTPUT«Type check failed in binding; expected 'Associative' but got 'Any'␤ in block at /tmp/ASwXA6I90v:1␤␤»
moritz r: my %h := Hash[Int].new; %h{3} = 4; say %h.perl
p6eval rakudo 9c59a1: OUTPUT«("3" => 4).hash␤»
moritz r: my %h{Int}; %h<a> = 'x'; say %h.perl 15:06
p6eval rakudo 9c59a1: OUTPUT«Nominal type check failed for parameter 'key'; expected Int but got Str instead␤ in method at_key at src/gen/CORE.setting:6756␤ in method postcircumfix:<{ }> at src/gen/CORE.setting:1555␤ in block at /tmp/UxgNY5HBT4:1␤␤»
moritz r: my %h{Int}; %h{42} = 'x'; say %h.perl
p6eval rakudo 9c59a1: OUTPUT«(42 => "x").hash␤»
timotimo i was wondering. there's a rule in STD.pm6 where you can have { ... } in a typename
TimToady it's a WHENCE, I think 15:07
moritz it's for the MyClass.new(:a(49), SuperClass{ a => 23 } ) thing, right?
TimToady yes, an "autoviv closure" too 15:08
timotimo ah, i see.
TimToady it's a dessert wax AND a floor topping 15:09
timotimo i was pretty happy when i realised, that default values in signatures will turn into closures automatically
daxim_ is this in the book?
moritz that's pretty useful
daxim_: no 15:10
arnsholt Yeah. I debugged a friend's Python program by knowing that Python doesn't thunk default values in sigs
timotimo indeed. 15:11
sometimes that is used as a "trick" to have static per-method variables
which will probably make many newbies go: o_O
arnsholt Eww. =)
moritz wonders if default values are documented on doc.perl6.org somewhere
oh, it already does explain it 15:12
"Default values can depend on previous parameters, and are (at least
notionally) computed anew for each call"
in the docs for type Signature
TimToady it's a good thing someone's been working on constant folding... 15:13
jnthn I think if you do sub foo($x = 42) { } then we already avoid the closure 15:14
moritz I'm pretty sure we avoid it, yes 15:15
r: my $sig := :($ = 42); say $sig.params[0].default
p6eval rakudo 9c59a1: OUTPUT«Block.new()␤»
moritz r: my $sig := :($ = 42); say $sig.params[0].default()
p6eval rakudo 9c59a1: OUTPUT«Block.new()␤»
moritz :(
r: my $sig := :($ = 42); say $sig.params[0].default()()
p6eval rakudo 9c59a1: OUTPUT«42␤»
moritz seems we don't anymore :( 15:16
FROGGS we need way more tests
moritz could try to fix that
FROGGS it is too easy to break things you dont test 15:17
TimToady that's because it's too easy to test things you don't break :) 15:18
FROGGS it would be awesome to have ppl write the tests who actually dont know how it is implemented 15:19
15:22 tipdbmp joined
tipdbmp How come JSON::Tiny's from-json cannot parse 'modules.perl6.org/proto.json'? =) 15:23
FROGGS because 'modules.perl6.org/proto.json' is an url? 15:24
(joke)
tadzik :D 15:26
FROGGS tipdbmp: you want it to automatically download and parse that file? you get my +1 for that :o) 15:27
TimToady
.oO(now, if we could only convince sorear++ that Perl 6 is a kind of telescope, or failing that, a kind of shoe...)
 15:28
FROGGS
.oO( a shoe-o-scope? )
 15:29
TimToady we're selling...IR Cloggers!
no wait, we haven't eaten that dogfood yet... 15:30
tadzik FROGGS: hm, should there be a problem with downloading and parsing that?
15:30 CharellKing1 joined
moritz JSON::Tiny will never automatically download anything 15:30
first, it has JSON in the name, not "HTTP". Second, it has "Tiny" in the name 15:31
third, automatically downloading stuff is a terrible idea security wise
fourth: separation of concerns 15:32
TimToady fifth: moritz doesn't like it
FROGGS
.oO( JSON::Tiny.from-json( fetch => url ) ) <--- tiny code to write
masak moritz++ # doesn't like it 15:33
moritz still violates the "tiny" and the "separation of concerns"
masak .oO( the separation of tiny concerns )
tadzik JSON::Tiny.from-json( fetch(url) ) is as tiny
and saner
moritz also note that from-json is a sub, not a method
"Sometimes, the elegant implementation is just a function. Not a method. Not a class. Not a framework. Just a function." 15:34
masak what would it take for a dogfooding of the IRC logger to happen? broad brush strokes, not the 100-point list of "inch stones".
moritz quoteby John Carmack
FROGGS ohh, (John Carmack)++
nwc10 masak: is it in two halves - the logger, and the web publishing side?
FROGGS now I have to agree too :/
moritz and colomon++ for retweeting that quote 15:35
masak: mostly: doing it
masak: and for the web frontend, performance is a concern
nwc10 so, start with the actual logger?
and run a Rakudo logger in parallel with a perl 5 logger
until the diffs are reliably zero :-) ?
masak on my wishlist: someone blogging an analysis of how far we can bring down the time to process a web request, on Rakudo and/or Niecza, what we can do about it, and what the tradeoffs are. 15:37
PerlJam do && and || short-circuit in NQP ?
moritz nqp: 1 && say(42)
p6eval nqp: OUTPUT«42␤»
FROGGS nqp: say(1 && 2)
p6eval nqp: OUTPUT«2␤»
15:37 CharellKing1 left
FROGGS nqp: say(1 && 2 || 3) 15:37
p6eval nqp: OUTPUT«2␤»
moritz nqp: 1 || say(42)
p6eval nqp: ( no output )
moritz nqp: 0 && say(42)
p6eval nqp: ( no output )
moritz yes, they do
PerlJam doh! I coulda done that :)
TimToady ayup
FROGGS minions++ 15:38
PerlJam wizards++ ;)
moritz evalbots++
nwc10 (*reliable* minions)++
TimToady Give a man a fish, and he'll eat for a day; beat a man over the head with a fish, and he'll wander off and start fishing.
FROGGS hehe
TimToady++ 15:39
see ya
15:39 FROGGS left
moritz give a man a fish, and he'll give you a few dollars. Phish a man's bank account data, and he'll give you all his money. 15:39
masak .oO( Let Me Evalbot That For You ) 15:40
tadzik give man a fire, and he'll be warm for a night. Set a man on fire and he'll be warm for the rest of his life.
masak give a man a fish, and you'll feed him for a day. teach a man to fish, and he'll come back the next day asking you about Perl 5, even though that's off-topic on this channel. 15:42
census: no offense :)
TimToady I vote census Most Improved in that regard
PerlJam a single fish would last me about half a day unless it was a really big fish.
tadzik or a very small man 15:43
TimToady a fish that would last me all die might well kill me due to allergies...
*day
gah, fraudian typos 15:44
huf depends on how you use it
geekosaur PerlJam, ever seen a tuna?
huf you all assume you'd eat the fish
PerlJam huf: bait?
huf or just sit on it, or something
TimToady I don't want a bait that lasts all day
moritz or sell it, and buy some real food in return :-) 15:45
masak on behalf of fish, I'm offended by that last remark.
:P
TimToady old cabbage is real food now, is it?
tadzik as a human, I'd be glad to not be considered "real food" :) 15:46
nwc10 they seem to think it is round here
huf at least it's wiser and less prone to panic
TimToady but yes, if I were the bait, I'd prefer to last at least a day
huf maybe you are, and you did. 15:47
TimToady take me to your leader
PerlJam sounds like it could be the beginning of a good AFJ to me. 15:48
masak tadzik: so, I guess no-one told you what Soylent Green is actually made of, huh? :P 15:49
15:54 gdey joined
tadzik what's Soylent Green? 15:58
huf food
a bit like cram
TimToady kurumono desu ka?
tipdbmp rakudo: sub not-working { say "'not-' not allowed as a sub name prefix?"; }; not-working();
p6eval rakudo 9c59a1: OUTPUT«===SORRY!===␤Undeclared routine:␤ working used at line 1␤␤»
TimToady apologies for a multilingual pun 15:59
neither of which is English 16:01
huf you've lost me
TimToady Kuru is from Fore, a New Guinian language 16:02
or it means "comes" in Japanese
so "Is it something that comes?"
census Thank you TimToady!! :) 16:03
16:03 FROGGS joined
TimToady apologizes for his prion-infested brane 16:03
census masak: No offense taken!! You are infinitely nicer than #perl
FROGGS hi all \o/
moritz r: my $param := :($ = 42).params[0]; say nqp::getattr($param, Parameter, '$default_value') 16:05
p6eval rakudo 9c59a1: OUTPUT«Can not get attribute '$default_value' declared in class 'Parameter' with this object␤ in block at /tmp/9qKPUTW7xi:1␤␤» 16:06
16:06 bluescreen10 joined
moritz r: my $param := :($ = 42).params[0]; say nqp::getattr(nqp::p6decont($param), Parameter, '$default_value') 16:06
p6eval rakudo 9c59a1: OUTPUT«Can not get attribute '$default_value' declared in class 'Parameter' with this object␤ in block at /tmp/v5jkQrNS0K:1␤␤»
masak tadzik: the reference is to an old Harry Harrison novel: en.wikipedia.org/wiki/Make_Room!_Make_Room! -- spoiler alert, people are crowded, someone produces "soylent" (soy+lentils) food cheaply, but it turns out to be made of people.
moritz r: my $param := :($ = 42).params[0]; say nqp::getattr(nqp::p6decont($param), Parameter, '$!default_value')
p6eval rakudo 9c59a1: OUTPUT«42␤»
16:07 gdey left
moritz long story short: the default is indeed stored as a literal (and not a closure), it's the accessor in class Parameter that wraps it in a Block 16:07
masak tadzik: "soylent green" is such a well-known meme, that it essentially means "cannibalism". cf Jonathan Coulton's "Chiron Beta Prime", where it gets a quick mention.
moritz: there could be a consistency-based argument for keeping it that way. 16:08
jnthn Oh, hmm. Now you say that, I have vague recollections of an RT ticket complaining about such an inconsistency... 16:09
FROGGS tipdbmp: why is that? 16:10
tipdbmp I don't know. 16:12
census Timtoday: I just wish you had some pull with #perl to get me unbanned there! :)
TimToady I cannot be everyone's mother. 16:13
census (and that you would not mind using that pull :)
I don't know why aperion won't let me back. Apparently I abused him . . .I don't know when or how this occurred.
16:14 ggoebel_ left
PerlJam census: give it a while, the ban will go away. 16:14
geekosaur you are presently banned because you get re-ip-ed periodically and apeiron thinks you are deliberately evading the original ban
census not at all!!
I try it out in the hopes that the ban is over
geekosaur which makes me think apeiron is not quite as swift as he thinks he is
census Since nobody told me when it expired
How am I supposed to know when I am unbanned?
16:14 gdey joined
PerlJam census: keep trying to join until it works :) 16:15
census PerlJam: Apparently that only gets me rebanned
geekosaur he also really doesn't like people who autorejoin, and has admitted this
so I think you just have to stop trying for a wek or so
census I do not do autorejoin. I do not know what that means
And I don't try that often. 16:16
geekosaur maybe you're not autorejoining since you said you try it periodically
census I try like once every few days
geekosaur you're trying often enough, apparently
more 0often than that
census Aperion will not tell me what is "enough"
16:16 Chillance joined
geekosaur [21 11:15] <geekosaur> so I think you just have to stop trying for a wek or so 16:16
census Geekosaur: If I stop trying for exactly 1 week, will that get me unbanned?
Because I don't want to magically try in one week and then get rebanned 16:17
Aperion will not tell me ANYTHING ever
TimToady depends on whether you've flipped the bozo bit
census I just want to sit in #perl without talking.
moritz wow, the ban list in #perl is quite long
census I want to learn by osmosis
Yes, aperion loves to ban people
I've been making friends with some of the banned people
geekosaur I'm not saying exactly
apeiron is not a computer 16:18
awwaiid maybe you should start #perlbanned :)
census I know! :)
Aperion just bans but does not say for how long or why
Just change #perl to #aperion and friends
like #foxandfriends
huf why are you still doing this?
census because i want to be unbanned 16:19
TimToady this is starting to get a little OT here
huf talk to an op
and not here.
masak .oO( "#perlbanned. You will never find a more wretched hive of scum and villainy. We must be cautious." )
census I'm sorry TimToady
PerlJam census: quick! Start talking about the Perl 6 program you're writing to help you mitigate you banned status in #perl 16:20
pmichaud
.oO( "These aren't the IRC channels you're looking for." )
PerlJam :) 16:21
census :)
TimToady census: it's not so much the topic, but the perseveration, I think... see en.wikipedia.org/wiki/Perseveration
masak .oO( <pumpkin> Perl 6, I am your father. <perl6> nooooooooooo ) 16:22
TimToady: "...doing the same thing over and over again and expecting different results..."? 16:23
TimToady well, yes <blush> 16:24
masak trying to track down the originator of that quote. 16:25
TimToady the trick is to swap in a different compulsion when you see that your current one isn't working
awwaiid not only that but... I'm not so sure I'd want to hang out with people who were mean to me. Life's too short. Plenty of other places to learn/contribute :)
TimToady well, I have no idea if I'm banned on #perl or not... 16:26
huf :)
masak :P
PerlJam census: another data point: Short sentences or phrases punctuated with newline that appear "rapidly" are likely to pique #perl's ire.
huf "daddy cant come into the couch fort, it's kids only"
16:26 kaleem left
masak according to various sources, Albert Einstein, Mark Silber, Philip Mangano, and Old Chinese Proverb originated the "over and over" quote. 16:26
awwaiid r: say "hello"
p6eval rakudo 9c59a1: OUTPUT«hello␤»
census thank you perljam 16:27
masak r: say "the goggles, they do nothing!"
p6eval rakudo 9c59a1: OUTPUT«the goggles, they do nothing!␤»
awwaiid oh while we're talking about perl6. Can someone remind me (or tell me to do it myself), if we had magical java bindings, would the syntax be "use java:com.foo.Blah;"? I can't remember 16:28
oh wait, it was com::foo::Blah:from<java>, eh? 16:29
moritz awwaiid: right, the package name must be valid Perl 6 syntax for package names 16:30
awwaiid and then :from<java> on the end?
I'm about to implement this for moe is why I ask :)
(about as in a 1-ish-week period) 16:31
moritz I think so, yes
awwaiid k. I'll go with that.
thanks
moritz though of course use ::('com.foo.Blah'):from<java> is also thinkable
though not very handy, if you ask me
FROGGS n: my @foo = ("a+b", "b+c"); say "aaaabbbbbcaaab" ~~ /^<@foo>+/ 16:35
p6eval niecza v24-23-g0520c7c: OUTPUT«｢aaaabbbbbcaaab｣␤␤»
16:37 SamuraiJack left
masak FROGGS: wow. 16:38
I... wow.
FROGGS ?
that's niecza
jnthn Is it LTMing those?
masak did you just *merge* two DFAs!?
16:38 SamuraiJack joined
moritz masak: isn't that what proto regexes do all the time_ 16:39
s/_/?/
jnthn Prety much :)
FROGGS jnthn: yes, by default. but I'm not sure if the algorithm perfect
masak I know, but these are first-class values.
jnthn *pretty
FROGGS n: say "abcde" ~~ /@(<a bc b cde>)+/ 16:40
lee_ !?
p6eval niecza v24-23-g0520c7c: OUTPUT«｢abc｣␤␤»
moritz masak: it's all code and state machines
masak n: my @foo = /a+b/, /b+c/; say "aaaabbbbbcaaab" ~~ /^<@foo>+/
p6eval niecza v24-23-g0520c7c: OUTPUT«｢aaaabbbbbcaaab｣␤␤»
masak lee_: !!
lee_ mistype :P ignore me!
masak ...another lurker... :)
FROGGS jnthn: I my last example would use NFA, it should match the entire string, right?
TimToady n: say "abcde" ~~ /<@(<a bc b cde>)>+/
masak lee_: hi! :D 16:41
p6eval niecza v24-23-g0520c7c: OUTPUT«｢abc｣␤␤»
16:41 cognominal__ left
masak moritz: say what you like, I think that's totally awesome. :D 16:41
jnthn n: say "abcde" ~~ /<@(<a b bc cde>)>+/
p6eval niecza v24-23-g0520c7c: OUTPUT«｢abc｣␤␤»
FROGGS TimToady: first "a", then "bc" because "bc" > "b", then nothing there that start with "d"
TimToady I just said that 16:42
moritz masak: it is
TimToady ah
FROGGS n: say "abcde" ~~ /||<@(<a b bc cde>)>+/
p6eval niecza v24-23-g0520c7c: OUTPUT«｢abc｣␤␤»
masak FROGGS: I love it how you actually explain LTM to TimToady... :P
FROGGS n: say "abcde" ~~ / || @(<a b bc cde>)+/
p6eval niecza v24-23-g0520c7c: OUTPUT«｢abc｣␤␤»
FROGGS masak: *g*
hmm, weird
moritz masak: and the fact that niecza does it made one of the candidates for the p6cc2011 quite pointless :-)
jnthn n: say "abcde" ~~ /<@(<a b bc? cde>)>+/ 16:43
p6eval niecza v24-23-g0520c7c: OUTPUT«｢abc｣␤␤»
masak moritz: which one, ooc?
moritz: also, I don't know if "pointless" is the word you're looking for... :)
jnthn n: say "abcde" ~~ /<@(<a b {}bc? cde>)>+/
p6eval niecza v24-23-g0520c7c: OUTPUT«Unhandled exception: System.InvalidCastException: Cannot cast from source type to destination type.␤ at RxFrame.ArrayHelperC (Niecza.Frame th) [0x00000] in <filename unknown>:0 ␤ at Niecza.Kernel.RunCore (Niecza.Frame& cur) [0x00000] in <filename unknown…
masak just because there are short, elegant solutions...
FROGGS r: say "abcde" ~~ / || @(<a b bc cde>)+/
p6eval rakudo 9c59a1: OUTPUT«｢abcde｣␤␤»
FROGGS there we go
jnthn r: say "abcde" ~~ / | @(<a b bc cde>)+/ 16:44
p6eval rakudo 9c59a1: OUTPUT«｢abc｣␤␤»
16:44 domidumont left
moritz r: say 'abcde' ~~ / | @(<a ab abc>) / 16:44
p6eval rakudo 9c59a1: OUTPUT«｢abc｣␤␤»
FROGGS jnthn: NFA would try more combinations?
16:47 cognominal__ joined, kaleem joined
TimToady nr: say "abcde" ~~ / | [@(<a b bc cde>)]+»/ 16:49
p6eval niecza v24-23-g0520c7c: OUTPUT«｢abcde｣␤␤»
..rakudo 9c59a1: OUTPUT«｢cde｣␤␤»
masak FROGGS: NFAs don't so much "try combinations" as "exist in a superposition of all combinations possible at this point".
TimToady nr: say "abcde" ~~ / | @(<a b bc cde>)+»/ 16:50
p6eval niecza v24-23-g0520c7c: OUTPUT«｢abcde｣␤␤»
..rakudo 9c59a1: OUTPUT«｢cde｣␤␤»
TimToady nr: say "abcde" ~~ / | <@(<a b bc cde>)>+»/
p6eval niecza v24-23-g0520c7c: OUTPUT«Unhandled exception: System.InvalidCastException: Cannot cast from source type to destination type.␤ at RxFrame.ArrayHelperC (Niecza.Frame th) [0x00000] in <filename unknown>:0 ␤ at Niecza.Kernel.RunCore (Niecza.Frame& cur) [0x00000] in <filename unknown…
..rakudo 9c59a1: OUTPUT«#<failed match>␤»
TimToady wow 16:51
masak FROGGS: if you haven't come upon swtch.com/~rsc/regexp/regexp1.html, I can heartily recommend it. it's enlightening.
TimToady I think I see about three bugs there
masak TimToady: I don't, so feel free to enumerate them. 16:52
I see that Niecza and Rakudo differ on the first one.
TimToady well, n and r disagree on the first
masak :)
TimToady and neither works on the second
masak what's that » doing there?
word boundary, right? 16:53
TimToady nod
forcing backtracking
masak oh!
TimToady which n does but r doesn't for the literals
masak so... Niecza is right on the first one?
TimToady and neither seems to work for the regex forms
yes, because nothing committed to the current LTM inside the @ 16:54
masak submits rakudobug
that's one.
"neither works on the second" may just be NYI.
FROGGS masak: will read, and yes, that is what I meant by "trying" for NFA 16:55
moritz an NFA doesn't try. It just does :-)
TimToady nr: say "abcde" ~~ / (a | b | bc | cde)+»/ 16:57
p6eval rakudo 9c59a1: OUTPUT«｢abcde｣␤ 0 => ｢a｣␤ 0 => ｢bc｣␤ 0 => ｢b｣␤ 0 => ｢cde｣␤␤»
..niecza v24-23-g0520c7c: OUTPUT«｢abcde｣␤ 0 => ｢a｣␤ 0 => ｢b｣␤ 0 => ｢cde｣␤␤»
masak moritz: exactly. like Yoda. :)
TimToady ooh, another bug 16:58
backtrack isn't taking out the [bc] in rakudo
16:58 kaleem left
FROGGS is voting for more regex tests! 16:59
masak submits rakudobug
FROGGS: well volunteered!
FROGGS masak: can you assign it to me?
TimToady (which can't happen in a pure Cursor model because the info should be attached only to the cursor that is thrown away in order to backtrack)
masak FROGGS: I'll try. 17:00
FROGGS: done. 17:02
FROGGS thanks 17:05
masak well, Perl 6 regexes are in some ways like a perfectly crafted building of LTM and FDA-ishness, that someone crashed a Wolkswagen bus of procedural actions into :P
FROGGS masak: I generally care about regexes these days btw
Volkswagen
masak oops, thanks.
indeed.
FROGGS: I was actively interested in Rakudo's parsing back in 2009. a lot of that interest remains, even if it's more passive today. 17:06
rjbs r: say so [].splice(0,3) 17:09
p6eval rakudo 9c59a1: OUTPUT«True␤»
rjbs What's "so" there?
moritz rjbs: boolification
FROGGS masak: I'm more interested in getting tickets closed and TODOs done :o) 17:10
moritz doc.perl6.org/language/operators#prefix+so
masak r: say [].splice(0, 3)
p6eval rakudo 9c59a1: OUTPUT«use of uninitialized value of type Any in string context in block at /tmp/H85DOXJPHT:1␤␤use of uninitialized value of type Any in string context in block at /tmp/H85DOXJPHT:1␤␤use of uninitialized value of type Any in string context in block at /tmp/H85DOXJPH…
rjbs Is there any mnemonic?
moritz say [].splice(0, 3).perl
r: say [].splice(0, 3).perl
p6eval rakudo 9c59a1: OUTPUT«Array.new(Any, Any, Any)␤»
masak rjbs: yes, it's the opposite of "not".
moritz rjbs: opposite of 'not'
masak faster again
masak rjbs: "I'm *not* kidding" -- "I'm *so* kidding"
moritz: quit post-predicting me! :P 17:11
rjbs Thank you.
FROGGS r: sub so-thing { }; so-thing()
p6eval rakudo 9c59a1: OUTPUT«===SORRY!===␤Undeclared routine:␤ thing used at line 1␤␤»
FROGGS r: sub not-thing { }; not-thing()
p6eval rakudo 9c59a1: OUTPUT«===SORRY!===␤Undeclared routine:␤ thing used at line 1␤␤»
rjbs filed a rakudobug about the result of splice in Rakudo. 17:12
17:12 SunilJoshi joined 17:13 zby_home_ joined
masak r: my @a = << >>; my @c = splice @a, 0, 3; say ?@c 17:13
p6eval rakudo 9c59a1: OUTPUT«True␤»
masak r: my @a = << >>; my @c = splice @a, 0, 3; say @c.elems
p6eval rakudo 9c59a1: OUTPUT«3␤»
masak rjbs: you're getting three undefined elements back. 17:14
moritz masak: postditicting? :-)
masak rjbs: ?3 is True.
rjbs n: says o [].splice(0,3)
p6eval niecza v24-23-g0520c7c: OUTPUT«===SORRY!===␤␤Undeclared routines:␤ 'o' used at line 1␤ 'says' used at line 1␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/boot/lib/CORE.setting line 1443 (die @ 5) ␤ at /home/p6eval/niecza/src/STD.pm6 line 1147 (P6.comp_…
rjbs n: say so [].splice(0,3)
p6eval niecza v24-23-g0520c7c: OUTPUT«False␤»
rjbs masak: TimToady called Rakudo's behavior a bug, last night.
masak reads the spec for .splice
moritz yes, the return value should be the empty list, IMHO
masak rjbs: ah, ok.
rjbs it prevents: while my $x = splice @a, 0, 3 {...} from terminating 17:15
moritz nr: say so (Any, Any, Any)
p6eval rakudo 9c59a1, niecza v24-23-g0520c7c: OUTPUT«True␤»
moritz that part is correct, no doubt
masak agreed.
rjbs I'd expect that to be true, and for the splice in question to return an empty list.
masak While it is illegal for the offset to be larger than the size of 17:16
the array, it is allowed for the final position to be off the end of the array;
in this case the entire rest of the array is spliced, whatever is available.
moritz rjbs: then your expectations are aligned with the specs, and with niecza
masak this makes it sound like Niecza is right and Rakudo is wrong.
17:17 census_ joined, census_ left, gdey left 17:18 census_ joined, census left, census_ left
moritz and I'm pretty sure I know why rakudo does it wrong 17:19
it simply slices the array with the arguments
17:19 census joined
moritz and an array-slice past the end of the array is non-empty (otherwise you couldn't assign to its elements) 17:19
masak then it shouldn't simply slice :) 17:20
moritz presumably, yes
or convert to List first
r: say (my @).list.WHAT
p6eval rakudo 9c59a1: OUTPUT«Array()␤»
masak heh. 17:21
bug?
jnthn No.
moritz no
jnthn r: say Array ~~ List
p6eval rakudo 9c59a1: OUTPUT«True␤»
masak ah.
moritz Array is a subcl... what jnthn said
masak hugs p6eval
moritz pity that it violates Liskov
masak what does?
FROGGS r: sub and-thing { }; and-thing() 17:22
p6eval rakudo 9c59a1: ( no output )
moritz slicing a List beyond its end simply returns Nil or the empty list or so
but slicing an Array doesn't
FROGGS r: sub do-thing { }; do-thing()
p6eval rakudo 9c59a1: ( no output )
masak grumbles about subclasses being an attractive but ultimately ugly idea in programming languages...
moritz nr: say <a b c>[1..5].elems
p6eval rakudo 9c59a1: OUTPUT«2␤»
..niecza v24-23-g0520c7c: OUTPUT«5␤»
17:23 SunilJoshi left
moritz nr: say <a b c>.list[1..5].elems 17:23
p6eval rakudo 9c59a1: OUTPUT«2␤»
..niecza v24-23-g0520c7c: OUTPUT«5␤»
moritz ... at least in Rakudo 17:28
17:30 census_ joined
pmurias is niecza or rakudo correct? 17:30
17:30 census left 17:31 census_ left, alester joined, census joined
moritz I'm pretty sure rakudo's behavior is intentional. Not sure about correct though 17:32
nr: say <a b c>[1..*].perl
p6eval niecza v24-23-g0520c7c: OUTPUT«(timeout)» 17:33
..rakudo 9c59a1: OUTPUT«("b", "c")␤»
moritz since * is bound to the number of elements, with niecza's behavior id should produce ('b', 'c', Any), which is quite inconvenient 17:34
nr: say (1..*).WHAT
p6eval rakudo 9c59a1, niecza v24-23-g0520c7c: OUTPUT«Range()␤»
moritz oh wait, it's not even a WhateverCode
but iirc index ranges ending with * were one of the reasons for rakudo's current behavior
huh, seems that dalek hasn't joined us :( 17:35
17:36 daxim_ left 17:38 dalek joined, ChanServ sets mode: +v dalek 17:44 SamuraiJack left 17:53 cognominal__ left 17:59 cognominal__ joined 18:03 dakkar left 18:05 cognominal__ left 18:21 prammer_ joined 18:23 odoacre_ joined 18:24 dayangkun left 18:25 dayangkun joined 18:29 baest_ joined 18:31 DrEeevil joined 18:32 lustlife left, bonsaikitten left, baest left, s0rear joined, lustlife1 joined, ObseLeTe joined
dalek p-jvm-prep: b25f471 | jnthn++ | / (12 files):
Refactors in preparation for exceptions support.

This adds in some pieces that will be needed for exceptions in the more general case, then switches the existing next/last/redo exceptions over to using the new approach. (Note, they need integrating so CONTROL, which may run on the stack top and resume, will work out.)		 18:35
diakopter the baa clog is .. amusing 18:37
moritz
.oO( blah clog )
 18:38
jnthn *sigh* Implementing Perl 6 exceptions considered tricky...
nwc10 s/ exceptions// ? 18:39
jnthn yeay, but I find many parts of the tricky fun
nwc10 do you have to do all of Perl 6 exceptions for NQP?
jnthn Exceptions aren't. They're just annoying.
nwc10 people should just stop making errors
jnthn No, but I don't want to have to do the whole thing over again.
diakopter Chuck Norris decides NP-Complete questions in O(0) 18:40
(time and space) 18:41
moritz and Turing-Complete questions in O(whatever he wants)
japhb diakopter, he's already decided them ...? 18:42
moritz, that's Bruce Schneier's job. :-)
diakopter japhb: :D as soon as they are posed, it is realized they are answered
nwc10 jnthn: still only the ICU test failing 18:47
pmurias jnthn: did you find a better way for resumable exceptions than keeping your own exceptions stack? like I have to do in nqp-js?
jnthn pmurias: Well, the main thing that makes them do-able at all is that exception handlers run on the stack top, so really it's just exiting the appropriate handler and returning to where the throw was, without doing any stack unwinding. 18:53
But I guess you may be in multiple throws at once so a handler stack is likely needed. 18:54
18:56 census_ joined
TimToady that's what SIMPLECATCH in S04:1035 is supposed to handle so that the "stack" is just the unhandled exceptions returned from the handler 18:57
18:57 census left
TimToady and I'm not sure that control exceptions stack at all... 18:58
18:58 census_ left, census joined
diakopter do not pass Go; go directly to Forth 18:58
TimToady though I guess you could have a new CATCH nested inside the SIMPLECATCH, but then it would have a new piece of activation stack to run on anyway 18:59
19:02 domidumont joined 19:03 cognominal joined
FROGGS the parsers in nqp and rakudo that parse the setting are NFA based? 19:04
jnthn Yes 19:05
See src/QRegex/NFA
FROGGS k
ya, seen that
jnthn Also !protoregex and !alt in Cursor 19:06
jnthn noms 19:15
19:21 raiph left, raiph__ left 19:22 raiph_ left 19:31 raiph joined, ObseLeTe left, WAB0GL7 joined, raiph_ joined
moritz ~- 19:32
masak new operator? :) 19:33
looks like string subtraction...
moritz I was just unable to type <return>~.
nwc10 mmm, yes, ssh escapes 19:34
moritz though string subtraction is interesting
if $a starts with $b, give me the aprt of $b that's after $a
and $a otherwise
diakopter part of $a that's after $b you mean? 19:35
masak .oO( two minutes after the operator was created, there was already uncertainty about what it meant... ) 19:36
moritz diakopter: yes
nr: sub infix:<~->($whole, $prefix) { my $l = $prefix.chars; $whole.substr(0, $l) eq $prefix ?? $whole.substr($l) !! $while }; say 'foobar' ~- 'fo' 19:37
p6eval niecza v24-23-g0520c7c: OUTPUT«===SORRY!===␤␤Variable $while is not predeclared at /tmp/rbCcI_VvpV line 1:␤------> $l) eq $prefix ?? $whole.substr($l) !! ⏏$while }; say 'foobar' ~- 'fo'␤␤Unhandled exception: Check failed␤␤ at /home/p6eval/niecza/bo…
..rakudo 9c59a1: OUTPUT«===SORRY!===␤Variable '$while' is not declared␤at /tmp/oXKL67hxRK:1␤------> q $prefix ?? $whole.substr($l) !! $while⏏ }; say 'foobar' ~- 'fo'␤ expecting any of:␤ postfix␤»
moritz nr: sub infix:<~->($whole, $prefix) { my $l = $prefix.chars; $whole.substr(0, $l) eq $prefix ?? $whole.substr($l) !! $whole }; say 'foobar' ~- 'fo'
p6eval rakudo 9c59a1, niecza v24-23-g0520c7c: OUTPUT«obar␤»
moritz nr: sub infix:<~->($whole, $prefix) { my $l = $prefix.chars; $whole.substr(0, $l) eq $prefix ?? $whole.substr($l) !! $whole }; say 'foobar' ~- 'bar' 19:38
p6eval rakudo 9c59a1, niecza v24-23-g0520c7c: OUTPUT«foobar␤»
dalek rl6-roast-data: e522d98 | coke++ | / (3 files):
today (automated commit)
 19:39
[Coke] wow, roast is going days at a time without a commit! 19:42
masak rn: sub infix:<~->($whole, $prefix) { $whole ~~ /^$prefix(.*:)/ ?? ~$0 !! $whole }; say "foobar" ~- "fo" 19:44
p6eval rakudo 9c59a1, niecza v24-23-g0520c7c: OUTPUT«obar␤»
masak hm, probably didn't need to ratchet that quantifier... 19:45
tadzik in the middle of Higher Order Perl I feel a strange urge to rewrite Panda in functional style instead of oop 19:46
I know, I know, closures are just objects
nwc10 maybe Perl 6 is stable? ship it! 19:48
moritz nah, if it's stable, it's dead 19:49
masak tadzik: certainly there's a balance there. closures are like wonderful mini-objects.
tadzik: I believe most code will have a healthy mixture of OO and closures. 19:50
moritz ad-hoc mini objects
masak yes! :D
I'm currently in love with JavaScript's common pattern: function () { /* privates */; return { /* publics */ } } 19:51
19:51 prammer_ left
masak if you take that pattern seriously, JavaScript's lack of private attributes doesn't feel so serious anymore. 19:51
(and all languages should have an easy way to create an object that just closes over its environment like that) 19:52
tadzik urgh. Somewhere, somehow, Panda is throwing an exception that's not a SixModelObject
wtf
ohh. The "$! instead of $_ in CATCH" again 19:53
masak I do that one too often, too.
tadzik and I need a when {}, alwyas 19:55
jnthn back 19:56
dalek nda: b7b966b | tadzik++ | bin/panda:
Catch Panda.resolve exceptions correctly. Fixes GH-32
 19:57
nwc10 sucessfully nommed?
jnthn yes :)
20:02 ObseLeTe joined 20:05 fhelmberger left 20:18 cognominal left 20:19 labster joined
dalek osystem: 18a5136 | (Brent Laabs)++ | META.list:
Adding Lingua::Number
 20:23
20:25 ObseLeTe left 20:27 gdey joined 20:28 cognominal joined
moritz I just stumbled over the same bug as daxim++ earlier 20:28
panda installed to ~/.perl6/lib/, which isn't in path 20:29
and the site dir is writable
ah, and I know why 20:30
$PATH is wrong, it picks up an old panda first
20:33 alec__ left 20:38 zby_home_ left
GlitchMr Just wondering, does Rakudo with JVM mean that it would be possible to make Java applet that would be online REPL? 20:43
20:46 domidumont left
[Coke] ... yes. but ew, applets. 20:49
masak there's more than one way to do it.
your way is *wrong*. :P
GlitchMr I know. 20:50
jnthn I suspect it'll be possible, with some amount of effort.
GlitchMr Still, Java applet is better than nothing. Or perhaps I'm wrong. 20:51
nwc10 but it's not clear who would be the attacker, and who would be pwned?
20:51 Liz_ joined, Liz_ is now known as lizmat
masak GlitchMr: make a REPL in JavaScript instead ;) 20:51
GlitchMr Hmmm... couldn't Silverlight (I know, even worse than Java applets) work with Niecza.
Anyway, I'm currently compiling Rakudo with emscripten. 20:52
I wonder if I will get anything.
"make" script appears to work
20:52 applicative_ joined
GlitchMr But I'm not sure what would be size of it. 20:52
And I'm almost sure that 100MB wouldn't be too practical. 20:53
I don't even mean server space.
cc -c -o perl6_group.o -Isrc/pmc -I/usr/include/parrot/5.0.0 -I/usr/include/parrot/5.0.0/pmc ... 21:01
wait...
Why it's using cc?
21:05 dolmen1 joined
[Coke] what is it? 21:05
masak it's a popular C compiler, written by the GNU people. 21:15
masak hides
masak watched "Spy Hard" recently :)
FROGGS nqp: my $nfa := QRegex::NFA.new; $nfa.addnode( QAST::Regex.new( "foobar" ) ); say( $nfa.run("foobar", 0) ); 21:16
p6eval nqp: OUTPUT«[ 0 ]␤»
FROGGS I'm not supposed to use it that way, no?
jnthn: ?
you have a hint for me?
masak it returned you a 0. 21:17
looks quite promising.
FROGGS nqp: my $nfa := QRegex::NFA.new; $nfa.addnode( QAST::Regex.new( "foobar" ) ); say( $nfa.run("abcdef", 0) );
p6eval nqp: OUTPUT«[ 0 ]␤»
masak nqp: my $nfa := QRegex::NFA.new; $nfa.addnode( QAST::Regex.new( "foobar" ) ); say( $nfa.run("le foobar", 0) );
p6eval nqp: OUTPUT«[ 0 ]␤»
masak hm.
FROGGS ya
ó.ò
jnthn Generally, you don't fun an NFA directly but instead merge a bunch of them together, giving each possiblity a fate. 21:20
See what happens in, iirc, !protoregex_nfa
(So you run a merged NFA.) 21:24
In fact, looking at !alt_nfa is probably more useful.
And the way it calls merge_substates
FROGGS so I have to call mergesubstates?
k
jnthn yeah
A regex will already have an NFA generally, which you can get by calling .NFA on it. 21:25
21:25 SmokeMachine left 21:26 benabik joined
FROGGS ohh, nice 21:26
nqp: say (/a/).NFA
p6eval nqp: OUTPUT«Confused at line 2, near "say (/a/)."␤current instr.: 'panic' pc 14734 (src/stage2/gen/NQPHLL.pir:5249) (src/stage2/gen/NQPHLL.pm:282)␤»
FROGGS nqp: say((/a/).NFA)
p6eval nqp: OUTPUT«Method 'NFA' not found for invocant of class 'NQPRegex'␤current instr.: '' pc 119 ((file unknown):155890185) (/tmp/8CAjkQUo1m:1)␤»
FROGGS okay, within rakudo then
in Cursor
dalek p-jvm-prep: 2e88e42 | jnthn++ | / (12 files):
First pass at exception handling.

Makes basic try/CATCH work. Enough to pass 3 out of 8 tests in 44-try-catch.t.		 21:41
[Coke] jnthn++
FROGGS jnthn: do I just run the nfa and return that from INTERPOLATE? 21:43
21:43 benabik left
jnthn FROGGS: No 21:44
FROGGS: You (presumably) have an array of things to run.
You use the NFA to tell you which indexes of the array to try and in which order.
So your fates in this case are array indexes. 21:45
21:46 shinobicl left
masak oh! nice. 21:49
21:51 sqirrel left 21:54 ObseLeTe joined 21:55 ObseLeTe left 21:56 bluescreen10 left 21:58 bluescreen10 joined 21:59 rindolf left 22:03 skids left 22:04 lustlife1 left 22:06 applicative_ left 22:25 PacoAir left 22:28 kaare_ left 22:33 dolmen1 left 22:35 census left
masak 'night, #perl6 22:35
22:35 census joined 22:39 woolfy left 22:46 woolfy1 joined
dalek p-jvm-prep: f4ee919 | jnthn++ | lib/QAST/JASTCompiler.nqp:
Fix a thinko, winning one more test.
 22:47
p-jvm-prep: da77c72 | jnthn++ | lib/QAST/JASTCompiler.nqp:
Toss an unused parameter.
p-jvm-prep: aa2d439 | jnthn++ | / (3 files):
Arg re-ordering to simplify stack handling.
22:49 wk joined 22:55 s0rear is now known as sorear 23:01 bluescreen10 left 23:02 FROGGS left
dalek p-jvm-prep: 909516f | jnthn++ | docs/ROADMAP:
Update ROADMAP.
 23:05
jnthn sleep & 23:07
23:16 woolfy1 left 23:33 hash_table joined 23:36 skids joined, mberends left 23:42 hash_table left
grondilu sub count(@a) { my %count; %count{$_}++ for @a; return %count } # Any more elegant way to write this?? 23:49
sorear that's almost, but not quite, what classify is for. or is that categorize now 23:53
23:54 cognominal left
[Coke] r: my @a = <a b c d e a b c>; say @.classify(); 23:56
p6eval rakudo 9c59a1: OUTPUT«===SORRY!===␤Variable @.classify used where no 'self' is available␤at /tmp/yfn1Ob4War:1␤------> @a = <a b c d e a b c>; say @.classify()⏏;␤»
[Coke] r: my @a = <a b c d e a b c>; say @a.classify();
p6eval rakudo 9c59a1: OUTPUT«Not enough positional parameters passed; got 1 but expected 2␤ in method classify at src/gen/CORE.setting:6066␤ in block at /tmp/Hr7ZlaWp08:1␤␤»
[Coke] r: my @a = <a b c d e a b c>; say @a.classify($_); 23:57
p6eval rakudo 9c59a1: OUTPUT«Nominal type check failed for parameter '&test'; expected Callable but got Any instead␤ in method classify at src/gen/CORE.setting:6066␤ in block at /tmp/KXcCBqmYnK:1␤␤»
[Coke] r: my @a = <a b c d e a b c>; say @a.classify(.Str);
p6eval rakudo 9c59a1: OUTPUT«use of uninitialized variable $_ of type Any in string context in block at /tmp/iJIJhfkd4d:1␤␤Nominal type check failed for parameter '&test'; expected Callable but got Str instead␤ in method classify at src/gen/CORE.setting:6066␤ in block at /tmp/iJIJhfkd4d:1…
[Coke] ... classify. You're on your own after that. :P 23:58
spider-mario r: my @a = <a b c d e a b c>; say @a.classify(*.Str);
p6eval rakudo 9c59a1: OUTPUT«("a" => ["a", "a"], "b" => ["b", "b"], "c" => ["c", "c"], "d" => ["d"], "e" => ["e"]).hash␤»
spider-mario r: my @a = <a b c d e a b c>; say @a.classify(*); 23:59
p6eval rakudo 9c59a1: OUTPUT«Nominal type check failed for parameter '&test'; expected Callable but got Whatever instead␤ in method classify at src/gen/CORE.setting:6066␤ in block at /tmp/rEvMpeASRe:1␤␤»
spider-mario r: my @a = <a b c d e a b c>; say @a.classify: {$_};
p6eval rakudo 9c59a1: OUTPUT«("a" => ["a", "a"], "b" => ["b", "b"], "c" => ["c", "c"], "d" => ["d"], "e" => ["e"]).hash␤»

Powered by Cro and the Raku® Programming Language.

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

Thank you!