»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! | tinyurl.com/p6contest
Set by moritz_ on 28 December 2010.
allbery_b I disagree, actually. RPG makes plenty of sense if you treat it as a batch spreadsheet 00:12
allbery_b instead of as a standard procedural program 00:12
colomon ack, my p4 tests take far too long now.... 00:56
sorear the issue definitely seems to be related to $*LEFTSIGIL and list assignment 02:01
I've golfed it down to 8 lines from 12k
diakopter wonders what could go from 12,000 to 8 lines 02:05
oh, a broken test case
heh
sorear I think I see what's going on here, hehe 02:14
dalek ecza: 4e9a91d | sorear++ | / (5 files):
[v6] Fix my @foo = 1; my $bar = @foo[0] = True misparse
 02:20
sorear What build framework would people most like to see Niecza v2 use? 03:11
sorear ((niecza-compiled-niecza)-compiled-niecza)-compiled-niecza passes all tests 03:49
colomon \o/
sorear++
sorear now all I need is a build system... and so the question 03:54
given all the issues people have been having with odd versions of xbuild, I'm thinking of switching back to make 03:55
colomon And I obviously wasn't able to get the toolchain working for me at all, so I'm no help with the question. I'll probably take another stab at it over the weekend, for what it's worth.... 03:56
sorear colomon: v2 will have binary distributions. 04:00
colomon sweet!
colomon rakudo: my $a = 0 => 21; say $a.perl; say $a > 21; 04:04
p6eval rakudo a350ff: OUTPUT«0 => 21␤maximum recursion depth exceeded␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Num' at line 7361:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in
..'Cool…
colomon interesting bug. most remember to pick it back up in the morning. 04:05
snarkyboojum rakudo: my $a = 0 => 21; +$a 04:13
p6eval rakudo a350ff: OUTPUT«maximum recursion depth exceeded␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Num' at line 7361:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in
..'Cool::Num' a…
TimToady sorear++ 04:14
snarkyboojum rakudo: (1=>2).Int 04:21
p6eval rakudo a350ff: OUTPUT«maximum recursion depth exceeded␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Num' at line 7361:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in
..'Cool::Num' a…
snarkyboojum doing .Real, or .Rat or .Int etc on a Pair like that seems to eventually cause my perl6 executable to core dump 04:22
sorear gives up on doing a portable Makefile for this 04:37
jasonmay sorear: just wondering, have you been to a yapc? 04:45
if not (and even if so), you should definitely come to this year's 04:47
couple of the nethack guys go
moritz_ sorear: ping 05:13
aka "good morning" :-)
sorear jasonmay: no I haven't. 05:17
moritz_: hi, pong.
moritz_ sorear: I'd like to make niecza's spectest architecture similar to rakudo's 05:18
sorear: ie a file which just holds the names of spectest files known to pass 05:19
currently the shell script has the disadvantage of hardcoding paths, and if you forget a \ at the end of the line the rest of the test files aren't run
sorear: what do you think of that? 05:20
sorear moritz_: sounds good 05:24
dalek ecza: 56c80b0 | sorear++ | t/ (2 files):
Merge remote branch 'moritz/spectests'
 06:29
cotto seen jnthn 07:30
aloha jnthn was last seen in #parrot 7 hours ago saying "s/PBCs/PMCs/?".
masak morning, #perl6 08:16
sorear hello masak 08:17
pmurias sorear: re build system, if you want to replace xbuild, Perl (6) seems the best choice 08:20
sorear There is a chicken and an egg there. :) 08:22
pmurias generally there's a lot of those in a bootstraped compiler ;)
the user will have to have a bunch of .NET bytecode to compile niecza for the first time 08:23
sorear What if the user wants to build niecza on a system with only clisp? 08:24
pmurias unless you want to develop the perl 5 or write niecza in a subset supported by v5
sorear: compile the build system to clisp ;)
sorear: he will have to have the version of niecza bootstrapped to common lisp 08:26
we will have to drop supporting compilation by p5 niecza at some point
masak last day today for submitting contest solutions.
sorear wonders how to handle getting the initial compiler 08:29
solution 1: check 6MB of data into git which will change occasionally and bloat the history a lot 08:31
masak that's the nqp-rx route. 08:32
seems to work quite well.
sorear solution 2: require the user to obtain a build from (downloads page, last month)
this is the mono/ghc/gcc route.
pmurias and have a file in the repo specifing the version of the initial compiler to download 08:34
masak seems to work quite well, too :)
pmurias sorear: i think solution 2 is better 08:35
sorear: if we automate that make could even to the grabbing of the initial compiler automatically
sorear here's another crazy idea: use a second git repo to hold the history of niecza binary releases, and then use a submodule to track it from the source repository 08:44
pmurias sorear: we can try that too and see which option performs better 08:46
pmurias that is if the bytecode is similiar between different versions or is all of it checkout out and git only hurts 08:47
sorear: if we wrap that behind perl scripts we can even switch that transparently 08:50
;)
pmurias sorear: how does the performance of niecza v2 and niecza v1 compare? 08:53
sorear pmurias: v2 is a bit faster 08:54
pmurias that's very good
sorear with -Bnam and amortized over a largish file, it's 2-3x faster 08:55
but that's the best case
pmurias what does -Bnam do? 08:55
sorear "Hello World" takes 8-9 seconds with both versions
writes the intermediate code, and stops
for some reason the NAM->CIL step is much faster when run standalone than when called as a subroutine by the compiler 08:56
like 5s versus 45s for STD.nam
that's one of the things I'd like to fix before v2, if I can figure out what's causing it 08:57
pmurias Hello World takes ~2sec on my machine with the p5 niecza
sorear !!!
pmurias on the second run
sorear ah.
I was counting the entire process
in compile-and-run mode
pmurias me too
sorear erm, in compile-and-run mode with the p5 niecza there shouldn't be much difference from run to run 08:58
pmurias the first run loads all the files from the disk into the caches
sorear oh right.
pmurias .pm's and things
sorear (v2 is different since v2 compiles the setting on the first run, which it can do twice as fast as p5's)
pmurias the ~2sec vs 8-9sec could be caused by differences in hardware 08:59
uni (Logics with counting) & 09:01
sorear I'm used to everyone else having computers 3x faster than me, not 4 :)
sorear mm logic. 09:01
fun stuff.
also niecza-v6 uses TOO MUCH MEMORY 09:10
masak Perl 6 and Perl 5 parsing: perlmonks.org/index.pl?node_id=882311 09:52
masak the only answer I can think of is "yes, things are very dynamic, so you can't get rid of the parser, ever. but don't worry, we've built a nice programming language around it" :P 09:53
masak penned a reply 10:00
sorear ERROR: certificate common name `*.github.com' doesn't match requested host name `github.com'.
fun. 10:01
dalek ecza: 923eae5 | sorear++ | / (97 files):
Redo build system; bootstrapped niecza is live
 10:17
sorear tomorrow I'll integrate tests and versioning with the new Makefile, update the README, and talk to moritz_ about p6eval 10:18
sorear -> sleep
masak nice. 10:19
colomon sorear++
snarkyboojum dang, niecza's build process has been completely re-written :O 10:44
insane productivity :) 10:45
the final two steps of the Makefile which cp to a run/ directory failed, because it the directory didn't exist. mkdir run under the Niecza root and re-running make resulted in a working Niecza.exe :) 10:54
Fox5600 hey does anyone know it filehandles are really supposed to have methods like abs, exp, log, sin, etc ? 11:04
masak Fox5600: I agree that sounds odd.
Fox5600 because they do in rakudo
masak rakudo: say $*IN.abs
p6eval rakudo a350ff: OUTPUT«maximum recursion depth exceeded␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Num' at line 7361:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in
..'Cool::Num' a…
Fox5600 all of then does that
masak rakudo: say $*IN ~~ Cool
p6eval rakudo a350ff: OUTPUT«1␤»
masak that's why.
why does IO need to be Cool? 11:05
rakudo: say IO ~~ Cool
p6eval rakudo a350ff: OUTPUT«1␤»
snarkyboojum saw the same error with (1=>2).Num today 11:06
flussence rakudo: say +$*IN;
p6eval rakudo a350ff: OUTPUT«maximum recursion depth exceeded␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Num' at line 7361:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in
..'Cool::Num' a…
snarkyboojum rakudo: +(1=>2) 11:07
p6eval rakudo a350ff: OUTPUT«maximum recursion depth exceeded␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Num' at line 7361:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in
..'Cool::Num' a…
flussence rakudo: say ~$*IN;
p6eval rakudo a350ff: OUTPUT«IO()<0xa0102f8>␤»
flussence well, the number part seems to be the only problem at least.
Fox5600 rakudo: say $*IN.chr 11:08
p6eval rakudo a350ff: OUTPUT«maximum recursion depth exceeded␤ in 'Cool::Num' at line 1862:CORE.setting␤ in 'Cool::Num' at line 1862:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Numeric' at line 1849:CORE.setting␤ in 'Cool::Num' at line 7361:CORE.setting␤ in 'Cool::Numeric'
..at li…
masak mojibake from the dir() function -- twitter.com/VienosNotes/status/25874069228363776 11:58
if anyone's able to reproduce this using a small piece of code, I'm willing to file an RT ticket about it.
flussence "open((0x30a1..0x30e1).chrs.Str, :w).say('aaaa'); dir();" does give me garbage on the REPL 12:07
masak flussence++ 12:10
masak submits rakudobug
masak hm. I'm not getting garbage on the REPL. I'm getting that (very long) file name, just as it should look. 12:13
pmurias what do people use a REPL for? 12:18
flussence I get it with perl6 -e as well... but printing out the .chrs in both and `ls` both show up OK 12:19
arnsholt pmurias: I use it for experimentation
I find it to be more convenient than -e, especially when I'm experimenting with syntax and such
masak never got into the habit using the REPL 12:21
arnsholt Me neither, not completely
My Lisp-head friend is completely addicted to his REPL, though
snarkyboojum I get garbage on the repl - unless I'm misunderstanding what it should print :) 12:22
flussence it *shouldn't* print "ã¡ã¢ã£ã¤ã¥ã¦ã§ã¨ã©ãªã«ã¬ã­ã®ã¯ã°ã±ã²ã³ã´ãµã¶ã·ã¸ã¹ãºã»ã¼ã½ã¾ã¿ããããããããããããããããããããããããããããããããã ã¡"
snarkyboojum then I get garbage too :) 12:23
pmurias arnsholt: the Lisp REPLiness seems very awkward and strange to me
arnsholt How so?
pmurias i can't really explain "write code to a file and then run it" seems much more convienient then modifing your program from a REPL and then build an executable from the image from time to time 12:25
arnsholt Yeah, it's a bit weird 12:26
pmurias it seems to statefull
arnsholt But for experimentation, a good REPL is quite nice
pmurias i use the lisp REPL for that 12:28
erlang forces people to use a REPL but i just piped commands to it with cat ;)
Persouille hello 12:30
pmurias hi 12:31
Persouille i'm trying to configure ilbot but i don't know where to store the /msg nickserv identify command
:)
masak Persouille: this is #perl6. you'll want to talk to moritz_. 12:32
Persouille oh sorry i thought it was the right channel 12:33
moritz_, : can I PM you ?
daxim 김정일bot 12:34
masak moritz_ is the right channel :)
Persouille h􏿽xE9h􏿽xE9
lol
thank you very much
:)
pmurias masak: the ilbot docs mention that you have to ask moritz_ on #perl6 12:35
masak pmurias: yes, I had a vague memory of that. still, if moritz_ isn't here, there isn't much the rest of us can do. 12:53
pmurias masak: based on a look into the source of ilbot what he wanted required changing the source 13:15
pmurias masak: but he left before i managed to tell him what to change 13:15
takadonet morning all 13:16
colomon \o 13:23
dalek ecs: ece3202 | (Kodi Arfer)++ | S32-setting-library/Containers.pod:
[S32/Containers] Require &classify's test argument to be Callable.

An arbitrary Matcher may not have much of a notion of a return value.		 14:16
masak Kodi++ 14:26
masak I also notice that .classify assumes that &test has exactly *one* return value. 14:32
masak S32 should at least mention what happens when a different number of values are returned. 14:33
dalek ast: 2167a01 | (Kodi Arfer)++ | S0 (3 files):
More tests for RT #71462 (smartmatching against a type should yield a Bool).
 14:58
kudo: 7fd3dea | KodiB++ | src/ (4 files):
Made smartmatching with something other than a regex on the RHS return a Bool.
 14:59
Kodi Is there a special reason why some of the @-parameters in the functions at the bottom of Any-list.pm *aren't* slurpy? 15:07
colomon errr... can you pass a regular @parameter in instead of a slurpy? 15:11
rakudo: say roll(3, <a b c d e f g h>) 15:12
p6eval rakudo a350ff: OUTPUT«bae␤»
Kodi rakudo: say first * %% 2, 1, 3, 4, 7, 9, 10; 15:13
p6eval rakudo a350ff: OUTPUT«Too many positional parameters passed; got 7 but expected 2␤ in 'first' at line 1816:CORE.setting␤ in main program body at line 22:/tmp/u_SPyx2Y0Y␤»
Kodi rakudo: say first * %% 2, (1, 3, 4, 7, 9, 10);
p6eval rakudo a350ff: OUTPUT«4␤»
colomon for sure it seems like grep and first should be the same!
Kodi rakudo: .say for grep * %% 2, 1, 3, 4, 7, 9, 10;
p6eval rakudo a350ff: OUTPUT«4␤10␤»
Kodi colomon: That's what I was thinking. 15:14
colomon that's LHF for sure.
Kodi "LHF"?
colomon low hanging fruit 15:15
easy to pick
Kodi Ah.
jnthn Good afty, #perl6. 15:18
colomon \o
takadonet jnthn: yo
colomon rakudo: .say for grep * %% 2, (1, 3, 4, 7, 9, 10); 15:20
p6eval rakudo a350ff: OUTPUT«4␤10␤»
flussence rakudo: (100..110).ords.say 15:22
p6eval rakudo a350ff: OUTPUT«49484832494849324948503249485132494852324948533249485432494855324948563249485732494948␤»
flussence for some reason the longer that range is, the slower each separate .say is... I don't get it. 15:23
doing 100..200 on my laptop I can see the numbers printing out one at a time, 100..110 looks a lot quicker.
jnthn Ouch 15:26
That sounds...bad.
colomon On mine, I'd say 100..110 is about twice as fast as that much of 100..200. Just eyeballing it. 15:28
colomon actually, that's not that surprising, is it? 15:30
jnthn colomon: Why should a longer range make things slower? 15:30
colomon it's not a lazy operation, so going into .ords requires converting the entire range to a string 15:31
flussence the .ords method itself doesn't appear to be doing anything visibly wrong: gather self.comb, take .ord
jnthn Oh, I think I'm missing some context. :)
colomon afk
flussence colomon: it has the same slowdown if I ~ the range first
colomon flussence: right, because that's doing the exact same thing. 15:34
Maybe it's just my imagination, but I'd say (100..200).map(*.ords).say is much faster 15:35
s/much/noticeably/
colomon I don't know exactly what is going on, but I'm guessing what you're seeing has something to do with ugly slowness in ~(Range) and comb. 15:38
flussence ok, I've done a bit more experimenting: if I run `for i in {10..20}; do echo $i; time perl6 -e "(' ' xx $i ** 2).map(*.ord).say" > /dev/null; done`, the time is more or less constant, but if I use .comb, it becomes exponential. 15:39
that is to say, "(' ' x 200).comb».ord.say" is much slower than if I'd done "(' ' xx 200)».ord" 15:40
colomon oh!
flussence so yeah, it's probably that bit causing it.
colomon in comb: (my $m = self.match($matcher, :c($c))
colomon that's got to be almost the slowest possible way to generate one character at a time from a string. 15:41
and...
calling self.match doesn't actually convert self to a Str, does it? 15:42
(99% sure the answer must be no)
flussence no idea, but I'll go with your 99%.
colomon in which case, that (my $m = self.match($matcher, :c($c)) in the loop to generate each char in .comb converts the Range to a Str each time through. 15:43
flussence ouch.
jnthn Ouch
It we're calling .match on it, strinfigying self first is maybe a good idea?
colomon seems like.
we need to set up a benchmark so we know if we're talking sense or not. 15:44
and then the optimization is pretty obvious.
masak people, make sure you send in your trailing solutions to the Perl 6 Coding Contest today! 16:10
masak aloha: hi :) 16:23
colomon masak: I'm busy writing the explanatory e-mail for my second p4 submission.
PerlJam good * #perl6 16:27
masak colomon: I mostly ignore email content outside of the .zip files.
PerlJam (obvious optimizations)++ (reading scroll back) 16:28
dalek kudo: 9b41b8f | KodiB++ | src/core/Match.pm:
Whoops, forgot to commit this.
 16:39
[Coke] KodiB, yah, but what is that commit for? 16:41
clearly you remembered to commit it now. ;)
Kodi [Coke]: I couldn't think of a better description, really. It may not even have any net effect. :P 16:44
masak .oO( a null commit? ) 16:49
PerlJam masak: a Mu commit 16:51
jnthn :P 16:51
Kodi PerlJam: That just about describes it. 16:52
PerlJam or maybe a Nil commit, but "Mu commit" fits better with the content of the commit :)
[Coke] muck omit 17:01
TimToady phenny: tell sorear make fails on the last step if run/ doesn't exist yet 17:09
phenny TimToady: I'll pass that on when sorear is around.
TimToady oh, already reported by snarkyboojum++ 17:10
yay for redundancy 17:11
dalek ast: ef6df94 | (Kodi Arfer)++ | S32-list/ (2 files):
Additions and fudging for slurpy &first and &uniq.
 17:34
dalek kudo: 89ff888 | KodiB++ | src/core/Any-list.pm:
[core/Any-list] Added slurpy versions of &first and &uniq.
 17:35
dalek p-rx/nom: 10b0b0b | jonathan++ | / (2 files):
Stub in a DispatcherSub PMC to be used for subs that serve as dispatchers.
 17:41
p-rx/nom: bb3de19 | jonathan++ | src/ops/nqp.ops:
Port the ops related to multiple dispatch; the one that actually does the dispatch is incomplete, but the rest are complete (though untested).
dalek nch-scripts: c15faa0 | (Solomon Foster)++ | non-str-comb.pl:
Add benchmark passing a non-string type to comb.
 17:43
dalek ecza: 430e593 | sorear++ | run/.gitignore:
Make sure obj/ and run/ are recorded in git GH-31 snarkyboojum++
 18:19
diakopter * * 18:20
jnthn *, diakopter 18:21
diakopter :) 18:22
moritz_ back
colomon still here
compiling my attempt at optimizing .comb 18:25
really, it's more of a de-pessimizing 18:26
say [+] (100..200).comb>>.Int; # before: 8.7s after: 1.9s 18:29
jnthn colomon++ 18:30
moritz_ \o/
[Coke] nice. 18:31
colomon about to start spectest run. 18:32
dalek ecs: 308c599 | larry++ | S32-setting-library/Containers.pod:
redesign of classify, add categorize

  * classify should take any arity-1 mapper including hash or array
  * classify always assumes 1 return value from the mapper
  * a function that assumes a list value from the mapper is needed
  * that function is called categorize, a bigger name for a slower function
   (it is less efficient than classify since it must loop over mapper values)
  * both classify and categorize now return a hash, since it's stupid to create
   a hash, return its pairs, only to recreate an identical hash outside,
   and it's trivial to extract the pairs from a hash if you need them.		 18:34
p-rx/nom: 194180d | jonathan++ | src/NQP/Grammar.pm:
Bring the various multiple dispatch and signature parsing changes introduced in JnthnNQP into the nqp-rx/nom grammar.
 18:35
p-rx/nom: d3290da | jonathan++ | src/NQP/Actions.pm:
Actions for {*}.
p-rx/nom: 2f51f68 | jonathan++ | src/NQP/Actions.pm:
Pull the old meta-model's handling of method_def out into something we can easily toss later. (Can't quite toss it yet; grammar is not using 6model yet, and there are grammars with methods within nqp-rx itself, so removing it now would break the bootstrap.)
 18:37
diakopter wheh 18:39
jnthn remembers he should probably eat as well as hack. bbs. :) 18:52
TimToady be happily nominative
colomon made homemade sushi (no raw fish, alas) today, for the first time in years. 18:54
[Coke] yum. 18:56
[Coke] makes plans to go eat a big bowl of sashimi tomorrow.
moritz_ (homemad tasty food)++ 18:58
*made
colomon homemad seems just about right 19:04
how are things going today, moritz_?
moritz_ colomon: chaotic 19:05
I nearly collapsed during the surgery, and had to leave 19:06
colomon you okay?
moritz_ and then the epidural anesthesia didn't work
colomon ooo, ouch.
moritz_ so they had to knock her out
and the baby breathed, but not enough 19:07
so it needed artificial respiration (and still does)
colomon hugs moritz_++
moritz_ thanks
TimToady yow, that sucks
moritz_ wife is recovering, baby is stable
[Coke] hugme: hug moritz_ 19:08
hugme hugs moritz_
moritz_ I'm exhausted, but otherwise OK
[Coke] don't let her hear you say that. :)
TimToady well, you know we all expect you to keep your priorities straight wrt your family, so if you have to cheat anyone of your time, cheat us. 19:09
colomon amen
moritz_ TimToady: sure. But #perl6 is not just time drain, but also recovery for me
colomon (though I did find that #perl6 was a happy relief during long hours of baby rocking.) 19:10
sorear good * #perl6 19:11
phenny sorear: 17:10Z <TimToady> tell sorear make fails on the last step if run/ doesn't exist yet
sorear hugs moritz_ too 19:14
moritz_ thanks 19:16
moritz_ oh, has niecza changed from xbuild to make? 19:24
TimToady mostly
still a bit rough 19:25
moritz_ tries
moritz_ sorear: needs to be updated in README.pod 19:26
moritz_ the first time I run run/Niecza.exe I get a warning from the setting 19:28
Potential difficulties: $p is declared but not used at /home/moritz/p6/niecza/lib/SAFE.setting line 903:
diakopter (me2 on windows, a day or two ago)
moritz_ ------> sub take(\⏏$p) { # should be \|$p
jnthn back 19:29
jnthn moritz_: Congrats on becoming a father, and best wishes to you and your (now slightly bigger) family. :-) 19:30
dalek kudo: 9f6a678 | colomon++ | src/core/Cool-str.pm:
Make a string copy of ~self to work with in Cool.comb, so that repeated calls to .match do not do ~self internally each time.
moritz_ jnthn: thanks 19:31
dalek ecza: e23237d | sorear++ | Makefile:
On second thought, don't pass -I to bootstrap

This means the compiler has to cope with an old setting, but the setting need not cope with an old compiler.
ecza: 4539bcd | sorear++ | README.pod:
Update README with new build info
colomon flussence++: optimization suggested by your (100..200).comb case now in Rakudo. :) 19:35
flussence \o/ 19:36
sorear wonders if a "get compiler base directory" function makes any sense in standard Perl 6 19:39
moritz_ sorear: what would you use it for? 19:40
moritz_ I mean, what would a Perl 6 programmer use it for? 19:41
colomon flussence: Hmmm... I"ll be the exact same thing happens in split, too. And probably elsewhere. 19:43
sorear is annoyed at how coupled niecza is to the CLR 19:58
diakopter :P 19:59
sorear like, you run Niecza.exe, it has to find lib/ somewhere
so it does Q:CgOp { (box Str (rawcall get_BaseDirectory (rawscall System.AppDomain.get_CurrentDomain))) }.IO.append("..").append("lib").realpath 20:00
moritz_ so embed it into a NIECZA::base_dir class 20:01
s/class/sub/
and provide a facility for other backends to override it 20:02
dalek kudo: d5b9d90 | moritz++ | src/core/Pair.pm:
Pair.Numeric
 20:09
colomon moritz_++ # elegant solution to the "How to Numify a Pair?" question. 20:12
moritz_ well, better than infinite recursion :-)
flussence or "method Numeric { 2 }" 20:20
flussence ducks 20:21
flussence spectest passed on one machine, running it on my slow one now. wonder if it'll break sub-2h this time... 20:27
dalek p-rx/nom: e0ac587 | jonathan++ | src/pmc/dispatchersub.pmc:
s/inherits/extends/ for PMC inheritance. D'oh.
 20:31
p-rx/nom: 08ccc35 | jonathan++ | src/ops/nqp.ops:
Fix a name-o that made most of the multi-dispatch related ops busted.
p-rx/nom: 75cf4dc | jonathan++ | src/NQP/Actions.pm:
Align nqp-rx/nom's method_def more closesly with the nqpclr one. This gets setup of proto methods - at least at the code generation level - in shape; the * or {*} also calls the stub that will eventually enter the multi-dispatcher.
rindolf Hi all. 20:34
Building Rakudo git HEAD ( 9f6a67881173bcedaf63603e1cfd485a08b112db ) fails here. 20:35
x86-64 Linux.
jnthn o/ rindolf
rindolf: Please could you paste the error it fails with? 20:36
mberends moritz_++ # best wishes for wife and baby too
rindolf jnthn: sure, wait a sec.
jnthn o/ mberends :)
mberends yo jnthn :)
rindolf jnthn: pastie.org/1461482 20:37
plobsing rindolf: looks like a linking problem 20:39
plobsing run 'nm /home/shlomif/Download/unpack/perl/Perl6/rakudo/git/rakudo/parrot_install/lib/libparrot.so | grep Parrot_api_string_export_ascii' 20:40
it should give a result like '<digits> T Parrot_api_string_export_ascii' 20:41
rindolf plobsing: it doesn't return anything. 20:42
plobsing then your parrot is messed up somehow. is this build failure from a clean checkout? if not, try removing parrot_install and rebuilding. 20:45
dalek p-rx/nom: d7ccb41 | jonathan++ | src/pmc/dispatchersub.pmc:
Make sure we mark the dispatchees.
 20:47
p-rx/nom: 25f7ff9 | jonathan++ | src/metamodel/how/NQPClassHOW.pm:
First piece of the multi-method handling in NQPClassHOW. Direct copy-paste from nqpclr's NQPClassHOW.
 20:49
rindolf plobsing: OK. 20:51
pmurias hi 21:10
rindolf Hi pmurias
dalek p-rx/nom: c6b29ec | jonathan++ | src/metamodel/how/NQPClassHOW.pm:
Port over incorporate_multi_candidates. No algorithmic changes, just VM munging (yes, I hope to have the nqpclr and nqp-rx versions the same some day soon - for now this is not too far off).
 21:11
moritz_ mberends: thanks 21:20
rindolf Hmm.... github is slow here.
pmurias sorear: why do you print out info in grey? 21:24
sorear because I don't know how not to 21:25
hi pmurias
moritz_ TimToady: re www.perlmonks.org/?node_id=882415 don't type declarations change the grammar too?
pmurias so what does that?
moritz_ well, more like the data the drives the parsing
pmurias s/does/causes
TimToady moritz_: read the parenthetical remark
pmurias sorear: if it's something beyond our control we could reset the terminal after the offending code 21:26
moritz_ TimToady: right 21:27
moritz_ should really go to bed
TimToady alternately, constants are degenerate type declarations :)
pmurias TimToady: is there a way to disable the unused parameter warning? 21:28
rindolf I gave up on git cloning the parrot repository.
It's too slow here.
I guess my ISP sucks again.
Hi TimToady
TimToady in STD, put #OK not used
dunno if sorear's clone does that though 21:29
would be low hanging fruit if it doesn't
sorear it does.
fwiw, anything STD parses but niecza/src/tryfile doesn't, I consider a bug
TimToady or, if you really don't want to use the parameter, just put the sigil
sorear though tryfile is probably a bit bitrotted after the last two weeks
TimToady std: sub foo ($a) {...} 21:30
p6eval std 625303c: OUTPUT«Potential difficulties:␤ $a is declared but not used at /tmp/2Nj3DJbUS6 line 1:␤------> sub foo (⏏$a) {...}␤ok 00:01 120m␤»
TimToady std: sub foo ($) {...}
p6eval std 625303c: OUTPUT«ok 00:01 120m␤»
TimToady std: sub foo (:a($)) {...} 21:31
p6eval std 625303c: OUTPUT«ok 00:01 120m␤»
TimToady std: sub foo ($a) {...} #OK 21:32
p6eval std 625303c: OUTPUT«ok 00:01 120m␤»
pmurias sorear: i get unable to resolve method keys in class Array when doing mono run/Niecza.exe -e 'say 1' 21:33
sorear *%#@ 21:35
I think I just heard the death of the project
TimToady the flag still seems to be waving... 21:37
pmurias sorear: ? 21:38
sorear: corrupted bootstrap? 21:40
TimToady maybe it means that we have to turn Perl 6 into C# for it to work...
CLR seems to have this effect on other languages too :) 21:41
if it's only a currupted bootstrap, sorear++ has recovered from those before 21:42
or maybe he thinks mono is Just Too Flakey now... 21:43
or maybe he just needs to take a step back and breathe for a bit 21:44
TimToady is far too good at "maybes" 21:45
rokoteko Haha. yea. start building perl6 designed to run on MS windows, which 80% of the world uses. :) 21:50
erm. well.. 80% percent of the cyberspace. 21:51
jnthn rokoteko: AFAIK, Niecza *does* run on Windows too. As do various other Perl 6 implementation efforts. :) 21:52
plobsing rokoteko: rakudo runs on windows SFAIK
rokoteko jnthn: hey, I have absolutely no idea. I barely know perl6.
jnthn plobsing: It sure does - because I do just about all my development on Windows. :)
rokoteko I got debian on my work laptop. :) 21:53
sorear pmurias: right, something is seriously wrong.
plobsing then you are one of the lucky 20% or so
sorear fwiw mono run/Niecza.exe -e 'say 1' works for me 21:53
TimToady incompatible mono? 21:54
colomon To be fair, I've never managed to successfully build Rakudo on my Windows box.
jnthn colomon: Really? Oddness. 21:54
TimToady or something different in %*ENV?
jnthn colomon: Unless you did it with Cygwin, where I can well imagine the Rakudo build is very flakey. 21:55
TimToady sorear: as another datapoint, I get pmurias's error here too
I'm running 64-bit, if it makes any difference
diakopter \/ 21:56
sorear rokoteko: Until yesterday, niecza was a mono-only C# program
colomon jnthn: I tried building it with whatever it tried to use. I've got Cygwin and two different versions of Visual Studio on the machine. 21:56
jnthn colomon: Ah, OK. 21:56
sorear TimToady: does mono boot/run/Niecza.exe -e 'say 1' work?
jnthn colomon: I've found ActivePerl + Visual Studio + build from the Visual Studio command prompt = success.
TimToady sorear: no, I get pmurias's error
rokoteko sorear: wasn't niecza the perl6 implementation of perl6?
diakopter colomon: wfm same setup as jnthn... 21:57
pmurias sorear: try make in git clean'ed mono
TimToady pmurias: are you 32 or 64-bits?
pmurias 32
sorear pmurias: "make in git clean'ed mono" ?
rokoteko I should probably follow more closely to the discussion to get a hang of it. :)
TimToady alternately, we could have something from prior builds out here in the hinterlands
pmurias mono boot/run/Niecza.exe -e 'say 1' works
mono run/Niecza.exe -e 'say 1' works
s/works/doesn't work/
sorear: git clean removed all the untracked files 21:58
TimToady sorear: would the whole backtrace help you?
pmurias git clean -f;make;mono run/Niecza.exe -e 'say 1'
plobsing colomon: there are instructions on how to set up parrot on windows @ trac.parrot.org/parrot/wiki/Platforms/Windows . building rakudo is probably much more platform-independant.
pmurias mono boot/run/Niecza.exe -e 'say 1' works 21:59
sorear: did you git push?
jnthn plobsing: Until we accidentally something platform dependent in any of the C bits of Rakudo (binder/dynops/dynpmcs) anyways. :)
colomon plobsing: thanks. It would be nice to have Rakudo working on that machine, it's a powerful machine that spends most of its time these days running Lacuna Expanse while I do development work on my OS X laptop. 22:00
TimToady boot/run/Niecza.exe works here too
plobsing jnthn: you have far less C than parrot. assume an identical C-code defect rate.
diakopter 'maybe it means that we have to turn Perl 6 into C# for it to work...
' 22:01
diakopter wants to add that to a quotebot somewhere 22:01
jnthn plobsing: :)
sorear pmurias: Does git clean make things better or worse? 22:02
dalek ecza: 4b9b5fa | sorear++ | lib/SAFE.setting:
Fix obsolescent use of (l) in &take
 22:03
ecza: ff486bd | sorear++ | / (5 files):
Add abstract primitives for path and bit ops
TimToady git clean does not help here 22:04
dalek ecza: 2c25f13 | sorear++ | / (4 files):
Add bit and path ops to setting
 22:05
sorear ok, let me try a build from a clean checkout :/
TimToady traceback at gist.github.com/780361 22:06
pmurias sorear: git clean simultates a clean checkout 22:10
TimToady after pulling latest push, the traceback is longer 22:12
[Coke] realizes he's currently at home on a windows laptop and about to "checkout" from work, and should probably try to build rakudo. 22:14
TimToady gist of longer traceback: gist.github.com/780377 22:14
TimToady running Mono JIT compiler version 2.6.7 (Debian 2.6.7-3ubuntu1), Architecture: amd64 22:17
pmurias: what's your version of mono? 22:18
sorear I cannot reproduce from a clean checkout 22:19
TimToady I'm guessing version skew
sorear 2.9/869fd03 ...
I need to set up some kind of multiversion mono env 22:20
TimToady maybe we just need to run with a new mono, like we try to do with parrot 22:25
diakopter www.chacha.com/question/do-monkeys-eat-parrots 22:26
colomon Does niecza work with .NET? It just occurred to me I spent all that time trying to get a version of mono that worked, and never considered using one of my .NET installs... 22:27
diakopter a snapshot of a built one worked for me yesterday 22:29
I haven't tested whether it builds lately
sorear the build process requires a unix toolchain (wget, gmake, cp, rm, unzip) 22:30
the v1.5 snapshot from the github downloads page should work though
diakopter surely the Makefile could be made dmake/nmake compatible 22:31
colomon sorear++. I'll try it when I'm not up in the train room watching the little guy play with his Little Blue Engine, Santa Fe Super Chief, and Hiro.
diakopter (and perl/extensions instead of the rest)
sorear I'm not entirely sure I can live without $(patsubst) :)
sorear notices that his mono is over a month old, fixes 22:32
pmurias sorear: 2.8.1 22:39
sorear: did the mono from run/ work for you? 22:40
diakopter O.O
pmurias sorear: the one in boot/ work for me too 22:46
sorear: what we could use is do have niecza automatically build and tested on different mono versions 22:47
sorear pmurias: yes, run/ works for me. 23:13
pmurias: that would be nice though I still need to debug this somehow 23:14
[Coke] goes "woof" at trying to build parrot on windows. 23:24
[Coke] will try to clean that up a bit for strawberry perl users this weekend. 23:25
jnthn [Coke]++
dalek p-rx/nom: 0499684 | jonathan++ | / (2 files):
Turns out we need some kind of lightweight signature object just for multis, to store the types but also definedness constraints (thus can't easily use an RPA). This should do it for now.
 23:36
p-rx/nom: 3a7f05b | jonathan++ | src/ops/nqp.ops:
Add an op for putting a multi sig in place.
p-rx/nom: d34f3a0 | jonathan++ | src/NQP/Actions.pm:
Emit code that builds and installs multisig objects that contain the type and definedness constraint for each multi candidate.
pmurias sorear: why are you using an development mono 2.9.*? 2.8.2 is the stable one? 23:55
sorear pmurias: 2.9.* has very valuable bugfixes 23:58

Powered by Cro and the Raku® Programming Language.

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

Thank you!