»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg camelia perl6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by sorear on 25 June 2013.
[Coke] I'll try to redo the failing gists later tonight. 00:00
colomon woah, that's freaky, lots of files are failing. :( 00:08
nr: say 'abc' ~~ m^b^ 00:10
camelia rakudo b7bbd6, niecza v24-86-g39ab531: OUTPUT«｢b｣␤␤»
colomon nr: say 'abc' ~~ m?b? 00:11
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Regex not terminated at /tmp/N_OQUYyiRg line 1 (EOF):␤------> say 'abc' ~~ m?b?⏏<EOL>␤␤Parse failed␤␤»
..rakudo b7bbd6: OUTPUT«｢b｣␤␤»
TimToady niecza is correct 00:12
nr: say 'abc' ~~ m?b??
colomon TimToady: really? that may be a roast bug, then
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Regex not terminated at /tmp/4sa920Ybsu line 1 (EOF):␤------> say 'abc' ~~ m?b??⏏<EOL>␤␤Parse failed␤␤»
..rakudo b7bbd6: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/UU6Txs0IVc:1␤------> say 'abc' ~~ m?b?⏏?␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ statement modifier␤ st…
colomon std: say 'abc' ~~ m?b? 00:13
camelia std d4cc5ab: OUTPUT«===SORRY!===␤Regex not terminated at /tmp/JEt9M8qISX line 1 (EOF):␤------> say 'abc' ~~ m?b?⏏<EOL>␤Parse failed␤FAILED 00:00 43m␤»
colomon std: say 'abc' ~~ m*b*
camelia std d4cc5ab: OUTPUT«===SORRY!===␤Regex not terminated at /tmp/n1mXeyff99 line 1 (EOF):␤------> say 'abc' ~~ m*b*⏏<EOL>␤Parse failed␤FAILED 00:00 43m␤»
TimToady std: say 'abc' ~~ m* b* * 00:14
camelia std d4cc5ab: OUTPUT«ok 00:00 43m␤»
TimToady std: say 'abc' ~~ m? b? ?
camelia std d4cc5ab: OUTPUT«ok 00:00 43m␤»
TimToady n: say 'abc' ~~ m? b? ?
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Quantifier quantifies nothing at /tmp/8oX9DDI8EZ line 1:␤------> say 'abc' ~~ m? b? ⏏?␤␤Regex not terminated at /tmp/8oX9DDI8EZ line 1 (EOF):␤------> say 'abc' ~~ m? b? ?⏏<EOL>␤␤Parse fa…
colomon so niecza is wrong there, but rakudo is wrong to accept m?b?
TimToady yes
quantifiers are a sublanguage, as it were, so it's like "{""}" 00:15
colomon ... all of which is quite interesting, but completely orthogonal to my exploration of why parrakudo is now failing all the tests in delimiters.t
TimToady n: say "{"stuff"}"
camelia niecza v24-86-g39ab531: OUTPUT«stuff␤»
TimToady n: say 'abc' ~~ m+ b+ + 00:17
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Quantifier quantifies nothing at /tmp/9U6MagvOH7 line 1:␤------> say 'abc' ~~ m+ b+ ⏏+␤␤Regex not terminated at /tmp/9U6MagvOH7 line 1 (EOF):␤------> say 'abc' ~~ m+ b+ +⏏<EOL>␤␤Parse fa…
TimToady std: say 'abc' ~~ m+ b+ +
camelia std d4cc5ab: OUTPUT«ok 00:00 43m␤»
TimToady n: say 'abc' ~~ m[ [b] ] 00:18
camelia niecza v24-86-g39ab531: OUTPUT«｢b｣␤␤»
TimToady n: say 'abc' ~~ m? [b?] ?
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Regex not terminated at /tmp/Kds1faYcvc line 1 (EOF):␤------> say 'abc' ~~ m? [b?] ?⏏<EOL>␤␤Parse failed␤␤» 00:19
TimToady std: say 'abc' ~~ m? [b?] ?
camelia std d4cc5ab: OUTPUT«===SORRY!===␤Regex not terminated at /tmp/A7J_WZFDeA line 1 (EOF):␤------> say 'abc' ~~ m? [b?] ?⏏<EOL>␤Parse failed␤FAILED 00:00 44m␤»
TimToady weird
oh, same deal 00:20
std: say 'abc' ~~ m? [b?]? ?
camelia std d4cc5ab: OUTPUT«ok 00:00 44m␤»
TimToady std: say 'abc' ~~ m? [b?]+ ? 00:21
camelia std d4cc5ab: OUTPUT«ok 00:00 44m␤»
TimToady n: say 'abc' ~~ m? [b?]+ ?
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Quantifier quantifies nothing at /tmp/YpwLlhc3rx line 1:␤------> say 'abc' ~~ m? [b?]+ ⏏?␤␤Regex not terminated at /tmp/YpwLlhc3rx line 1 (EOF):␤------> say 'abc' ~~ m? [b?]+ ?⏏<EOL>␤␤Pa…
TimToady yeah, the terminator should take precedence on that one
r: say 'abc' ~~ m? [b?]+ ? 00:22
camelia rakudo b7bbd6: OUTPUT«␀␀␀[␀␀␀3␀␀␀1␀␀␀m␀␀␀=␀␀␀=␀␀␀=␀␀␀␀␀␀[␀␀␀0␀␀␀m␀␀␀S␀␀␀O␀␀␀R␀␀␀R␀␀␀Y␀␀␀!␀␀␀␀␀␀[␀␀␀3␀␀␀1␀␀␀m␀␀␀=␀␀␀=␀␀␀=␀␀␀␀␀␀[␀␀␀0␀␀␀m␀␀␀␤␀␀␀U␀␀␀n␀␀␀r␀␀␀e␀␀␀c␀␀␀o␀␀␀g␀␀␀n␀␀␀i␀␀␀z􏿽xE2
colomon so.... std is wrong in some cases too? or am I misunderstanding something there?
TimToady no, it was right
colomon okay
TimToady the ? after the [] was quantifying the []
colomon TimToady: it was the m? [b?] ? that caught my attention 00:24
why doesn't that work?
TimToady because the ? quantifies the []
colomon even with the space?
TimToady r: say 'abc' ~~ m? \w + ? 00:25
camelia rakudo b7bbd6: OUTPUT«｢abc｣␤␤»
colomon would it make sense to ban some of these delimiters in the name of sanity?
TimToady no reason to, really, it's just a DIHWIDT
if we nannify everything, the executable will be twice as large as it is 00:26
colomon TimToady: I'm against nannifying everything, but clearly both compilers are getting these wrong right now. I'd be surprised if fixing matters will take less code than banning those delimiters would. 00:27
TimToady nr: say 'abc' ~~ m? \w+ ? 00:28
camelia rakudo b7bbd6: OUTPUT«｢abc｣␤␤»
..niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Quantifier quantifies nothing at /tmp/rH6fq0vjmM line 1:␤------> say 'abc' ~~ m? \w+ ⏏?␤␤Regex not terminated at /tmp/rH6fq0vjmM line 1 (EOF):␤------> say 'abc' ~~ m? \w+ ?⏏<EOL>␤␤Parse …
TimToady well, niecza already effectively bans it :)
but it shouldn't
colomon and rakudo gets the case where you need to use the delimiter as a quantifier wrong. 00:29
TimToady it's checking for a stopper between the atom and it's quantifier, and I think that's wrongish 00:30
colomon and there are a boatload of tests that need to be rewritten, too.
well, nothing to be done for rewriting the tests. But I don't have time for it now. 00:31
TimToady I assume some of these tests are from P5, where we search for the closing delimiter before reparsing, which is wrong on the face of it 00:32
r: say 'abc' ~~ m? [b?] ** 1..1 ? 00:33
camelia rakudo b7bbd6: OUTPUT«␀␀␀[␀␀␀3␀␀␀1␀␀␀m␀␀␀=␀␀␀=␀␀␀=␀␀␀␀␀␀[␀␀␀0␀␀␀m␀␀␀S␀␀␀O␀␀␀R␀␀␀R␀␀␀Y␀␀␀!␀␀␀␀␀␀[␀␀␀3␀␀␀1␀␀␀m␀␀␀=␀␀␀=␀␀␀=␀␀␀␀␀␀[␀␀␀0␀␀␀m␀␀␀␤␀␀␀U␀␀␀n␀␀␀r␀␀␀e␀␀␀c␀␀␀o␀␀␀g␀␀␀n␀␀␀i␀␀␀z􏿽xE2
TimToady at least rakudo isn't doing that 00:34
dunno why niecza is failing, the <!stopper> in quantified_atom should prevent it from testing the stopper as a metachar at the front 00:39
probably a difference in the whitespace parsing 00:42
since STD does sigspace differently 00:43
nr: say 'abc' ~~ m?b??? 00:48
camelia niecza v24-86-g39ab531: OUTPUT«｢｣␤␤»
..rakudo b7bbd6: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/mRvWd3E67R:1␤------> say 'abc' ~~ m?b???⏏<EOL>␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ prefix or term␤ prefix or meta-prefix␤…
TimToady turns out ?? is a minimal ?
nr: say 'abc' ~~ m? b?? ?
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Quantifier quantifies nothing at /tmp/sDBJfH28Bk line 1:␤------> say 'abc' ~~ m? b?? ⏏?␤␤Regex not terminated at /tmp/sDBJfH28Bk line 1 (EOF):␤------> say 'abc' ~~ m? b?? ?⏏<EOL>␤␤Parse …
..rakudo b7bbd6: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/3SNBqrTBQN:1␤------> say 'abc' ~~ m? b?⏏? ?␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ statement modifier␤ …
TimToady but that's still buggy in both of them
nr: say 'abc' ~~ m! b?! ! 00:49
camelia rakudo b7bbd6, niecza v24-86-g39ab531: OUTPUT«｢｣␤␤»
TimToady er, isn't ! supposed to be greedy? 00:50
nr: say 'abc' ~~ m! b? !
camelia rakudo b7bbd6, niecza v24-86-g39ab531: OUTPUT«｢｣␤␤»
TimToady o_O 00:51
nr: $_ = 'abc'; say m! b? !
camelia rakudo b7bbd6, niecza v24-86-g39ab531: OUTPUT«｢｣␤␤»
TimToady nr: $_ = 'abc'; say m! b+ ! 00:52
camelia rakudo b7bbd6, niecza v24-86-g39ab531: OUTPUT«｢b｣␤␤»
TimToady why isn't ? looking for 1 before 0?
nr: $_ = 'abc'; say m! b* !
camelia rakudo b7bbd6, niecza v24-86-g39ab531: OUTPUT«｢｣␤␤»
TimToady oh, null string at the beginning, duh 00:53
nr: $_ = 'bcd'; say m! b?! !
camelia rakudo b7bbd6, niecza v24-86-g39ab531: OUTPUT«｢b｣␤␤»
TimToady there we go
I mean, just because jetlag makes my brain think it's 2am, that's no excuse... 00:54
01:06 Guest1337 joined 01:17 FROGGS left 01:30 FROGGS joined 01:49 thundergnat left 01:53 SamuraiJack_ joined 02:05 benabik joined 02:48 aindilis joined 02:50 JimmyZ_ joined
JimmyZ_ std: $_ = 'abc'; say m!b*! 02:53
camelia std d4cc5ab: OUTPUT«===SORRY!===␤Regex not terminated at /tmp/Z3joLEkHnl line 1 (EOF):␤------> $_ = 'abc'; say m!b*!⏏<EOL>␤Parse failed␤FAILED 00:00 44m␤»
JimmyZ_ std: $_ = 'abc'; say m|b*|
camelia std d4cc5ab: OUTPUT«ok 00:00 44m␤»
JimmyZ_ the former is bug? 02:54
TimToady no 02:59
std: $_ = 'abc'; say m! b* ! 03:00
camelia std d4cc5ab: OUTPUT«ok 00:00 44m␤»
TimToady *! is a valid quantifier
it is a * that is explicitly greedy, even if the surrounding context is ratchety 03:01
std: $_ = 'abc'; say m!b*!!
camelia std d4cc5ab: OUTPUT«ok 00:00 44m␤»
JimmyZ_ and so *? 03:02
TimToady n: $_ = 'abc'; say m!b*!!
camelia niecza v24-86-g39ab531: OUTPUT«｢｣␤␤»
TimToady *? is explicitly frugal
n: $_ = 'abc'; say m?b*?? 03:03
camelia niecza v24-86-g39ab531: OUTPUT«｢｣␤␤»
TimToady n: $_ = 'abc'; say m?b*? ?
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Quantifier quantifies nothing at /tmp/PbIhFf2Ogi line 1:␤------> $_ = 'abc'; say m?b*? ⏏?␤␤Regex not terminated at /tmp/PbIhFf2Ogi line 1 (EOF):␤------> $_ = 'abc'; say m?b*? ?⏏<EOL>␤␤Pa…
TimToady niecza only has the bug after whitespace
JimmyZ_ good 03:04
03:13 JimmyZ_ left 03:18 Chillance left 03:43 Guest1337 left 03:46 Guest1337 joined 03:52 preflex_ joined, ChanServ sets mode: +v preflex_ 03:53 preflex left, preflex_ is now known as preflex 04:31 Patterner left 04:33 Psyche^ joined, Psyche^ is now known as Patterner 04:44 BenGoldberg left 05:14 SamuraiJack_ left 05:22 kaare__ joined 05:23 dmol left 05:38 raiph left 05:39 FROGGS[mobile] joined
FROGGS[mobile] morning 05:40
FROGGS TimToady: do I understand it right that rakudo is wrong because it cant handle the stopper-char within a regex? 05:52
TimToady: or is there something else wrong too?
05:54 xenoterracide left
FROGGS does a git bisect to track down the commit that break so many tests... 05:56
06:00 mstefanko left 06:05 mstefanko joined
sorear o/ FROGGS 06:06
FROGGS hi sorear 06:07
06:08 xilo left 06:11 crab2313 joined
[Coke] sorear: I'm trying to use eval-server and client, and am getting: 06:18
===SORRY!=== 06:19
This representation can not unbox to a native str
on every invocation. any suggestions as to what I'm doing wrong?
gist.github.com/coke/5894087
sorear [Coke]: nothing seems obvious, but passing --ll-exception will help 06:24
does it work if you use ./perl6 on line 16? 06:25
FROGGS 412307989be6df484710ac7d46bb72d003649594 is the first bad commit
Author: jnthn [email@hidden.address]
Add missing check for 0-count in subst.
that was my guess, it is the only related commit
sorear FROGGS: commit looks very thinkoey 06:27
FROGGS: try with ?? ($replacement.count == 0 ?? $replacement() !! $replacement($m))
FROGGS will do
sorear a replacement Callable with no arguments needs to be called with no arguments, not stringified as Sub()<...>
FROGGS yes 06:29
thanks for the hint, would have taken another cup of coffee for me to figure out...
sorear (fixing this will probably bring us back down from 88...) 06:31
[Coke] sorear: ah, crap. no, ./perl6 ** doesn't work in that situation. thanks for the obvious debug pointer. 06:34
sorear: only if it's a parrot only fix, right?
06:35 xenoterracide joined
dalek kudo/nom: b6d197f | (Tobias Leich)++ | src/core/Str.pm:
invoke the callable, sorear++
 06:37
[Coke] sorear: huh. the cd is in the wrong spot.. but it worked before I switched to the server. weird. 06:38
sorear [Coke]: the server won't see cds done after it is started
[Coke]: I think the fix helps parrot more 06:39
FROGGS: I take it it made a difference?
[Coke] sorear: I think the only real effect is that the pwd was the wrong dir
FROGGS sorear: yes, fixes it
sorear oh, right. uh... 06:40
is it working now?
06:44 dmol joined
FROGGS okay, will bisect the macro-test-fails then 06:45
06:45 btyler left
sorear wonders if anyone has played with the new --archive option to test_summary 06:47
[Coke] I have not. 06:48
FROGGS me neither 06:52
sorear multicore test summaries :D 06:53
FROGGS my test_summary script for v5 does that too, though I have stripped it down much 06:55
it is made to see regressions better 06:56
06:57 crab2313 left 07:07 Guest1337 left
[Coke] jnthn, sorear: in no particular order, gist.github.com/coke/5879701 is an updated list of collated jvm failures. 07:18
sorear: java.lang.ClassFormatError may be of interest to you.
sorear ClassFormatError? wha? 07:22
[Coke] java.lang.ClassFormatError: Duplicate method name&signature in class file - S05-capture/caps.rakudo.jvm 07:23
sorear markdown fail: # java.lang.IllegalArgumentException: sprintf only accepts ints, nums, and strs, not class __P6opaque__62 07:24
(minor)
[Coke] yah, hopefully these will get fixed soon enough to not matter. :) 07:25
urk? when I run that formaterror from the command line... I get... a PBC error?
sorear methinks you have some contamination from parrot stuff... 07:26
[Coke] oh, whew, the pbc is user error.
(cd rakudo instead of rakudo.jvm)
definitely seeing a shift in errors as some are fixed. 07:29
Dynamic variable $*CWD not found is becoming more popular, e.g. 07:30
sorear jnthn wants me to add shell(), but that will just leave us with $*EXECUTABLE_NAME problems...
[Coke] zzz 07:31
07:42 xinming left 07:47 xinming joined 07:53 domidumont joined 08:07 domidumont left, domidumont joined
FROGGS t/spec/S06-macros/quasi-blocks.t (Wstat: 256 Tests: 0 Failed: 0) Parse errors: No plan found in TAP output 08:15
a07c211ff178c4a0badeb043b82ab12ccad6d6b4 is the first bad commit
sorear: Generate local_lifetime hints.
sorear FROGGS: so you're saying some of the fails come from that, and others came from the s/// thinko? 08:19
FROGGS sorear: the s/// is fixed already, only the macro-test fails remain 08:20
so it is either your changes to the actions, or the change of the nqp_revision
sorear probably my changes. grr. 08:22
can't investigate right now
FROGGS np, I can give it a try 08:23
08:29 Rotwang joined 08:37 spider-mario joined 08:40 _daniel-s__ is now known as daniel-s, domidumont left
dalek ecs: 0271f65 | (Brent Laabs)++ | S99-glossary.pod:
+blorst, clog, thinko
 08:52
sorear we use blasts now 08:53
blorst is archaic
labster ye olde perl sixe? 09:14
09:21 lue left 09:24 athomason left 09:33 lue joined 09:35 dmol left 09:40 dmol joined 09:41 xenoterracide left 09:46 woosley1 joined
dalek ecs: af9af14 | moritz++ | S99-glossary.pod:
[S99] fill out some blanks
 09:51
sorear might it be useful to somehow flag tests that require a full process state because they do stuff with chdir, etc? 10:24
yes, we COULD virtualize the current directory for testing... but that would defeat the point, since we wouldn't be testing the nonvirtualized chdir
another big question on my mind: what should be $*EXECUTABLE_NAME on the jvm? 10:26
random observation: we have a lot of tests that do is_run when they mean is_no_extraneous_output. we ought to have a way to trap output without launching a whole new process
we have three cases of is_run: calls that really require a whole process because they're doing chdiry envy stuff, calls that need a split GlobalContext but can run within the eval server, and calls that really just need to hook $*ERR 10:28
10:36 woosley1 left 10:39 autumn is now known as Aki-chan, Aki-chan is now known as autumn
FROGGS sorear: when I disallow sinking at the top of actions.nqp, all is fine 10:51
I'm not yet sure what is wrong though 10:52
sorear sinking. always blame sinking. it's stinking. 10:59
11:00 PacoAir joined 11:10 xinming left 11:12 xinming joined 11:16 dmol left 11:19 grondilu joined
grondilu Hello. I tried to compile rakudo-jvm and I got: 11:22
Error occurred during initialization of VM
11:22 SamuraiJack_ joined
grondilu Could not reserve enough space for object heap 11:22
in the README it is written I should try adding -Xms500m -Xmx2g in the nqp/nqp.bat, but I have no such file 11:23
grondilu realizes it's probably in nqp/nqp 11:24
jnthn sorear: One "easy option" for $*EXECUTABLE_NAME is to point it at the runner bat/sh, but that means it can't use the eval client and so will be slow...though slow and working beats not working at all.
11:31 crab2313 joined
grondilu got a java.lang.StackOverflowError at src/Perl6/Optimizer.nqp:685 11:45
timotimo try --optimize=0 to see if it survives the rest of the execution 11:51
11:52 crab2313 left
grondilu Unknown option: optimize 11:53
timotimo oh, huh 11:55
11:55 araujo left 12:14 Rotwang left 12:25 _ilbot left, moritz left 12:27 _ilbot joined 12:29 moritz joined 12:31 snoopy left 12:35 birdwindupbird joined
moritz IRC logs are back with ipv6 support 12:39
12:39 crab2313 joined
moritz (and I happen not to pay for ipv6 traffic :-) 12:40
FROGGS hmm? why? 12:41
12:41 araujo joined, araujo left, araujo joined
moritz because the core routers don't support ipv6 traffic accounting 12:43
FROGGS hehe, nice
timotimo that's awesome :) 12:46
moritz (not that it matters much; I'm far below the 750Gb/month treshold anyway, but it still feels good that way) 12:47
FROGGS and I guess there is not so much ipv6 traffic anyway
timotimo is on ipv6
moritz and there's a debian and a cpan mirror inside the network, and only traffic leaving the network is acounted anyway
FROGGS I'm not 12:48
moritz I'm on ipv6 via sixxs
timotimo my provider offers native ipv6. the router they ship doesn't have the little box ticked by default afaict, but i do get it 12:49
in fact, i believe the ipv4 traffic that goes through my router gets tunneled through ipv6 to my provider
FROGGS: get a sixxs tunnel or a henet tunnel. vote with your packets! :) 12:50
tadzik does sixxs still have their crazy email policy?
FROGGS just tell me for what I need ipv6 first :o)
timotimo perl6 irclogs!
tadzik FROGGS: chicks dig ipv6 users ;)
FROGGS *g* 12:51
tadzik "is that an ipv6 subnet? Yours? Whatcha doing tonight?"
timotimo FROGGS: freedom from NAT is a big thing
FROGGS well, my gurl doesnt know what ipv6 is...
tadzik also, if we all insist that the future is now, it may eventually be... 12:52
and we're all early adopters anyway
moritz tadzik: what's the "crazy email policy"?
tadzik moritz: they had that last time I tried applying for an ipv6 address
"oh, is that gmail? No, that's no good. We don't accept free email accounts" 12:53
so I asked a friend for an account on her server
"no, that's not good enough either"
so I gave up
moritz wtf
tadzik yeah 12:54
basically, they wanted my ISP email at all costs
masak o/ 12:55
FROGGS ho masak
masak today I may actually get something p6-y done.
one can hope.
FROGGS \o/
and the one is moritz?
tadzik hehehe 12:56
moritz :-)
moritz always hopes that masak progresses with td reviews :-)
masak moritz: yeah.
moritz: so let's do a bit of t3 today. 's a good idea. 12:57
tadzik heh, ś is too fancy for sixxs 12:58
talk about the future now
timotimo get hurricane electric instead? 12:59
tadzik I already applied here
timotimo k
12:59 grondilu left 13:00 grondilu joined
FROGGS so, I have to say "Cha!" for now, see you in a bit 13:05
13:11 woosley1 joined 13:12 woosley1 left
jnthn whoa, this weather is sucky 13:15
nwc10 code more! Ignore it! 13:19
or hide under the duvet
13:19 FROGGS[mobile] left
timotimo both can be combined in my experience 13:19
jnthn Grr. My patch to try and fix $*IMPLICIT handling up for JVM seems to regress some tests on Parrot. :( 13:22
timotimo were we able to figure out what caused the big regression in cokes last run? 13:24
jnthn a thinko that ws already fixed, afaict
13:24 FROGGS[mobile] joined
timotimo ah, great 13:24
.o(still it's hard to say goodbye to the 88%. hope they'll come back around soon enough! :) )
and now, i'll depart
13:29 Rotwang joined 13:34 kaare__ left, kaare__ joined
dalek ecs: f0f6642 | moritz++ | S99-glossary.pod:
more glossary entries
 13:38
13:39 kaare__ left 13:40 FROGGS[mobile] left, FROGGS[mobile] joined
dalek ecs: 5b80653 | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Tpyo
 13:41
lizmat moritz++
FROGGS[mobile] jnthn: if you have some tuits, can you look at the macro+sink fails? 13:43
I dont think I am able to fix it 13:44
13:50 Chillance joined
lizmat FWIW, I can confirm that failures are back to the macro/quasi fails in the spectest 13:51
moritz wonders if the glossary entry for "thinko" should be "thinkographic error" 13:53
13:53 ajr joined 13:54 ajr is now known as Guest94866
lizmat not the other way around? 13:54
13:54 Guest94866 is now known as ajr_
moritz s/entry/explanation/ 13:55
lizmat ;-) cycling& 13:58
14:11 konundra joined
jnthn Oh, that evaluate_unquotes thing seems to fail in the same way on Parrot and JVM backends... 14:18
14:19 Danishman joined
jnthn oh...I think I see why. 'twas the sorear++ locallifetime thing 14:19
nqp: say(nqp::isstr('foo')) 14:21
camelia nqp: OUTPUT«1␤»
jnthn nqp: say(nqp::isstr(NQPMu))
camelia nqp: OUTPUT«0␤»
jnthn FROGGS[mobile]: Building/trying a fix now 14:23
14:36 dalek left 14:37 dalek joined, ChanServ sets mode: +v dalek
dalek p: 9fca92f | jnthn++ | src/QAST/Op.nqp:
Don't trip up over strings in QAST nodes.

It's locallifetime that led to the observable explosion, but handle has been around and this code was vulnerable to that also. Fixes the macro test regressions in Rakudo.		 14:41
14:41 rindolf joined
colomon \o/ 14:41
nqp tests to be added because of this?
dalek kudo/nom: 36f4f0b | jnthn++ | src/Perl6/Actions.nqp:
Fix $*IMPLICIT case handling of $_.

This makes sure it is registered early enough, so that JVM code-gen will not find the wrong one. Should fix some things on JVM, and does not seem to cause regressions under Parrot either.
kudo/nom: 2fb22ef | jnthn++ | tools/build/NQP_REVISION:
Bump NQP_REVISION for macros fix.
colomon \o/ \o/ \o/ 14:42
jnthn Hm, locallifetime busted MoarVM's nqp-cc.
Should be an easy enough fix.
14:42 thundergnat joined
jnthn is somewhat amused that Rakudo JVM runs more tests on his box than Rakudo Parrot, 'cus he's missing ICU. :) 14:44
14:45 raiph joined
colomon ;) 14:46
14:46 btyler joined
tadzik hah 14:48
raiph i'm trying to make sense of fperrad's p6 14:55
is it more or less another nqp backend?
www.youtube.com/watch?feature=play...=160#t=22m 14:56
dalek ecs: 1b67957 | thundergnat++ | S99-glossary.pod:
Move roast to alphabetically logical spot. Expanded acronym.
raiph slides are in english, talk is in french
github.com/fperrad/p6jit/blob/master/README.md 14:57
moritz sites.google.com/site/fperrad/YAP6...edirects=0 has the slides
raiph moritz: thx
15:08 daniel-s_ joined
jnthn didn't know about that project 15:12
FROGGS jnthn++ # I never would have come to that conclusion
15:13 kaare_ joined
dalek kudo/nom: 85a4d9a | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Ops.java:
Implement p6getouterctx op.
 15:15
kudo/nom: 7b04775 | jnthn++ | src/Perl6/World.nqp:
Eliminate a Parrotism for an nqp:: op.

Partly gets macro tests working on JVM backend.
colomon jnthn++ # changes faster than I can spectest.... 15:16
jnthn Seems somewhere along the line I may have regressed something, though :( 15:18
15:19 xilo joined
FROGGS :o( 15:19
jnthn urgh 15:20
15:20 rindolf left
jnthn yeah, I see...gonna have to re-test this change on Parrot, though... 15:20
colomon you guys know how you use test summary with jvm? (I just tried and don't have it figured out yoet.) It seems like make spectest borders on useless when there are so many errors. I'd be happy to run it here as often as I can while I'm working on $work.... 15:21
15:23 xenoterracide joined
thundergnat colomon: Hey, I thought you might be tickled by rosettacode.org/wiki/Continued_frac...%29#Perl_6 I added recently. /I/ was pleased with how it turned out. 15:23
colomon thundergnat++ 15:24
flussence colomon: irclog.perlgeek.de/perl6/2013-06-29#i_7267465 15:28
colomon flussence++ # don't think that's the standard way of running it, but it might be a lot *better * than the standard way... 15:30
ooo, lots of fail.
need to install TAP::Harness::Archive... 15:31
and set the executable bit on test_summary 15:32
now it seems to be doing something.
dalek kudo/nom: af39d66 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Implement 'is copy' array params.
 15:41
kudo/nom: e7ad774 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Implement 'is copy' hash params.
15:42 domidumont joined 15:43 FROGGS[mobile] left 15:50 xilo_ left
colomon I get 21861 passing tests. 15:50
oooo, hash is copy! jnthn++ \o/ 15:51
jnthn is working through the [Coke]++ list of ways we fail and picking some things off it :)
15:54 SamuraiJack_ left, ajr_ left, ajr joined 15:55 ajr is now known as Guest4981, Guest4981 is now known as ajr_
moritz
.oO( 24 ways we fail! )
 15:55
15:56 rindolf joined
dalek kudo/nom: 5b5225b | jnthn++ | src/Perl6/Actions.nqp:
Some CATCH/CONTROL JVM fixes, after $_ changes.
 15:57
16:00 SamuraiJack_ joined
masak "24 ways we fail" sounds like an interesting advent calendar. 16:00
or maybe it's the title of the YAPC::EU talk I should submit today. 16:01
jnthn ooh, rain's stopped 16:02
jnthn takes a walk
colomon 21925 passing tests before 5b5225b 16:06
16:08 Rix left
colomon just started a run... 16:12
16:13 Rix joined 16:17 SamuraiJack_ left
colomon 22414 passing tests with 5b5225b 16:22
took < 10 minutes for the run
8 minutes (I was looking away)
16:26 pmurias joined
jnthn And yesterday's coke number was 22100 :) Nice. :) 16:34
colomon How did he get the extra 300 tests passing? I must have missed something in the backlog... 16:36
16:37 Rotwang left
jnthn Parrakudo has gained back 100 or so too, though, so only 200 help to increase the % :) 16:38
16:38 kaare_ left
pmurias jnthn: is it known that nqp-jvm raises a NullPointer exception when undeclared lexicals are used? 16:39
jnthn: and stays silent when they are not used
jnthn pmurias: That's...meant to be caught at parse time. 16:40
nqp: say($a)
camelia nqp: OUTPUT«Use of undeclared variable '$a' at line 2, near ")"␤current instr.: 'panic' pc 14721 (src/stage2/gen/NQPHLL.pir:5232) (src/stage2/gen/NQPHLL.nqp:279)␤»
jnthn C:\consulting\rakudo>install-jvm\nqp.bat -e "say($a)"
Use of undeclared variable '$a' at line 2, near ")"
pmurias: Seems to be caught there too 16:41
16:49 BenGoldberg joined 16:51 Rotwang joined 17:09 birdwindupbird left
pmurias nqp: sayx(123) 17:09
camelia nqp: OUTPUT«Could not find sub &sayx␤current instr.: '' pc 46 ((file unknown):170449615) (/tmp/VVOgA4I7Oy:1)␤»
pmurias jnthn: say($foo) is caught, sayx(123) give a NullPointer exception 17:10
on jvm only
jnthn oh...missing subs, not variables.
pmurias just relized subs are handled differently in QAST 17:11
17:12 colomon left
masak .oO( at least they are QAST in a different light ) 17:12
17:15 colomon joined
colomon What happened to IO::Capture::Simple? 17:17
... answer appears to be nothing, yet my smoke tester thinks it's completely gone from the ecosystem. :\ 17:18
t/spec/S32-temporal/DateTime.t has another new error with the hash is copy thing fixed, but I am being commanded to play Star Wars with Legos ("you can be young Anakin!") right now. 17:20
pmurias jnthn: QAST::WVal expects getobjsc != null, where is that guarantied? I'm trying to fix BEGIN
jnthn pmurias: It's up to the compiler to make sure it doesn't emit a QAST::WVal for something that wasn't add_object'd (that is, placed in an SC at some point) 17:22
17:22 grondilu left
moritz t/spec/S02-magicals/config.t fails because 'linux' ne 'Linux' 17:29
but I guess there's no spec for what should actually be returned 17:30
any objections to making the comparison case-insensitive?
dalek ast: 99263cb | moritz++ | S02-magicals/config.t:
case-insensitive comparison of $*OS

the test assumes the p5 insanity of "linux" vs. "MSWin32", but the specs are silent on the exact case of the return value. (And on the exact return value itself, for that matter).		 17:35
nwc10 no, but don't go so far as to think that qr/win/ only matches the OS from the pacific *north* west :-)
dalek kudo/nom: 8ca095c | jnthn++ | src/Perl6/Metamodel/Mixins.nqp:
Missing stringify, needed on JVM.

Fixes a couple of the enum related tests.		 17:36
kudo/nom: 1e07ff8 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Ops.java:
Make p6capturelex not create a busted chain.

It didn't care what it shoved into .outer, meaning that we could put entirely wrong things in there. This tries to do a bit better, though really the longer-term solution is to fix up the code-gen, hopefully by leaving the QAST compiler to work it out. In the meantime, this seems to fix various lexical issues.
jnthn Seems we have no remaining failures in the 99-problems-* integration tests over parrakudo 17:37
moritz pulls again and starts spectesting from the beginning
17:38 ajr_ left
nwc10 jnthn++ 17:38
17:45 aindilis left, btyler left 17:46 ilbot_nextgen joined
dalek kudo/nom: b3ee5fb | jnthn++ | src/core/Num.pm:
Don't explode on NaN.Int.

Somehow "got lucky" here on Parrot; be explicit for JVM's sake.		 17:47
[Coke] safe to kick off the daily jvm run?
jnthn [Coke]: ya
jnthn is gonna go shop/cook now anyways :) 17:48
Will try and fill in a few more gaps later on this evening.
17:49 crab2313 left
[Coke] jnthn++ I see lots of commits in the backlog! 17:49
rakudoparrot is 100% again. 17:51
17:57 dmol joined
jnthn yay 18:08
colomon 22465 passing now. ;) 18:09
18:12 tomyan joined
GlitchMr rn: sub postfix:<km>(Real $a) { "{$a}km" }; say 40.0km 18:16
camelia rakudo b3ee5f, niecza v24-86-g39ab531: OUTPUT«40km␤»
GlitchMr rn: sub postfix:<km>(Real \a) { "{a}km" }; say 40.0km
camelia rakudo b3ee5f, niecza v24-86-g39ab531: OUTPUT«40km␤» 18:17
18:19 araujo left 18:33 araujo joined, araujo left, araujo joined 18:37 abnorman joined
dalek rl6-roast-data: a105058 | coke++ | / (5 files):
today (automated commit)
 18:38
[Coke] rakudo.jvm @ 89.63% - 18:40
lizmat wow! 18:41
[Coke] r: say ceil(25717 * .8 - 23052)
camelia rakudo b3ee5f: OUTPUT«===SORRY!===␤Undeclared routine:␤ ceil used at line 1␤␤»
[Coke] r: say 25717 * .8 - 23052
camelia rakudo b3ee5f: OUTPUT«-2478.4␤»
[Coke] r: say 25717 * .9 - 23052
camelia rakudo b3ee5f: OUTPUT«93.3␤»
colomon where they hey are you getting 600 more passing tests than I am? (on jvm.)
[Coke] 94 more tests to get 90%
jnthn ooh, so close to the 90%!
[Coke] colomon: github.com/coke/perl6-roast-data/b...ummary.out 18:42
pmurias 90% of roast or 90% of tests passing on parrot?
[Coke] the latter. 18:43
rakudo.jvm isn't going to be passing any more tests than rakudo.parrot to start.
nwc10 possibly stupid question - does the gather/take stuff that sorear hacked on work on Rakudo/parrot? Or just Rakudo/JVM? And is it something spec'd in roast? 18:45
pmurias it doesn't work on parrot 18:47
that is, the continuations opcodes don't work
colomon huh. categorize.t is working for [Coke], failing for me. :\ 18:48
lizmat colomon: strange, not real special Perl 6 code there... 18:49
colomon lizmat: weird jvm bug I isolated days ago
labster 89.63% of rakudo spectests passing, work on second 90% slated to begin shortly. 18:53
colomon I thought jnthn++ had fixed it, so [Coke]'s build may be better than mine. :\
18:54 tomyan left 18:56 xenoterracide|2 joined 18:58 Chillance_ joined 18:59 lelf` is now known as lelf, xenoterracide left 19:01 Chillance left
jnthn colomon: I fixed that one only in one of my most recent commits 19:04
colomon jnthn: I am pulling like every ten minutes here.
colomon may be overexcited about rakudo-jvm 19:05
19:06 rindolf left
FROGGS *g* 19:06
colomon I'm trying to do a more complete rebuild atm 19:07
jnthn Confirm that categorize.t passes locally for me.
colomon confirm it's still a huge fail for me. :( 19:08
jnthn ugh
How's it fail
?
colomon confirm I am an idiot
jnthn :D
19:09 tomyan joined
colomon I can confirm that categorize.t works better if you don't have a "die" added after the first test. 19:11
jnthn ORLY? :P 19:13
Seems compared to yesterday we lost some tests... 19:14
S02-names/pseudo.rakudo.jvm appears to have worsened somehow
Same for our.t 19:15
19:19 domidumont left 19:20 abnorman left
colomon that still only gets me to 22511, well short of [Coke]'s 23052 :\ 19:26
masak is a little disappointed the categorize.t thing wasn't a deeper mystery than that... 19:27
19:27 tomyan left 19:30 pmichaud joined
colomon gist.github.com/colomon/5896533 # differences between my run and [Coke]'s, broken down by area 19:31
500 tests in S03, 60 more in "int"
jnthn weird 19:32
colomon yes 19:33
weird
jnthn Think I've tracked down the our/pseudo regression 19:34
FROGGS perl -E 'say( (-50000 >> 8) & 0xFF)' # 60 19:37
perl6 -e 'say( (-50000 +> 8) +& 0xFF)' # 61
where is the thinko?
timotimo two's complement vs one's complement?
FROGGS hmm?
colomon timotimo: p6 is supposed to be two's complement 19:38
jnthn Think there's an RT on that 19:39
FROGGS okay, it looks like the internal storage format of Perl 5 is the answer when bitshifting negative values
ohh
will look
colomon rn: say( (-50000 +> +& 0xFF)'
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Non-declarative sigil is missing its name at /tmp/FWc7MRX6wJ line 1:␤------> say( (-50000 +> +⏏& 0xFF)'␤␤Unhandled exception: Unable to resolve method from in type Any␤ at /home/p6eval/niecza/src/STD.pm6…
..rakudo b3ee5f: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name␤at /tmp/qrwoWxhetf:1␤------> say( (-50000 +> +&⏏ 0xFF)'␤ expecting any of:␤ infix noun␤»
colomon rn: say( (-50000 +> +& 0xFF)
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Non-declarative sigil is missing its name at /tmp/tmNOaBhAH3 line 1:␤------> say( (-50000 +> +⏏& 0xFF)␤␤Unhandled exception: Unable to resolve method from in type Any␤ at /home/p6eval/niecza/src/STD.pm6 …
..rakudo b3ee5f: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name␤at /tmp/G86pvUuSyK:1␤------> say( (-50000 +> +&⏏ 0xFF)␤ expecting any of:␤ infix noun␤»
colomon rn: say( (-50000 +> +& 0xFF)) 19:40
camelia rakudo b3ee5f: OUTPUT«===SORRY!===␤Non-declarative sigil is missing its name␤at /tmp/tWYD_TpRHj:1␤------> say( (-50000 +> +&⏏ 0xFF))␤ expecting any of:␤ infix noun␤»
..niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Non-declarative sigil is missing its name at /tmp/oNi2bRARmp line 1:␤------> say( (-50000 +> +⏏& 0xFF))␤␤Unhandled exception: Unable to resolve method from in type Any␤ at /home/p6eval/niecza/src/STD.pm6…
jnthn ttiar ?
FROGGS rn: say( (-50000 +> 8 +& 0xFF))
camelia niecza v24-86-g39ab531: OUTPUT«60␤»
..rakudo b3ee5f: OUTPUT«61␤»
FROGGS ahh
interesting
jnthn [Coke]: Is gist.github.com/coke/5879701 something you generate or update manually? 19:41
FROGGS jnthn: this one? #115958: bigint rshift op is 1s complement, should be 2s complement
jnthn FROGGS: sounds like it, yes 19:42
FROGGS k, htnkas
thanks
-.-
sorry, I run multiple threads here
jnthn blem.It's no pro
FROGGS rn: use Test; is (-4..-1 X+> 1..3), (-2,-1,-1,-2,-1 xx 8), "Right shift is two's complement" 19:43
camelia rakudo b3ee5f: OUTPUT«not ok 1 - Right shift is two's complement␤# got: '-2 -1 0 -1 0 0 -1 0 0 0 0 0'␤# expected: '-2 -1 -1 -2 -1 -1 -1 -1 -1 -1 -1 -1'␤»
..niecza v24-86-g39ab531: OUTPUT«ok 1 - Right shift is two's complement␤»
FROGGS jnthn: is it likely to be a bug in parrot?
colomon rakudo-jvm$ ./perl6
> say( (-50000 +> 8 +& 0xFF))
60
;) 19:44
dalek kudo/nom: 9e85b50 | jnthn++ | src/Perl6/Grammar.nqp:
Prevent a $*IMPLICIT leakage.

Unregresses pseudo.t and our.t, which we slipped backwards on in the daily roast.
FROGGS that answers it
[Coke] jnthn: gist.github.com/coke/5879701 updated. (manually. I have automated some of it, but I'm counting on you to fix things faster than I need to worry about it. :)
jnthn [Coke]: Thanks, I've been finding it quite helpful.
Using it as a kind of todo list... :)
colomon I have a perl 6 script to make the list from the test summary 19:45
;)
oh, never mind, that's a much more sophisticated list than I was making. [Coke]++ 19:46
[Coke] jnthn: excellent. 19:47
You eliminated 2 of the sections since ysterday.
FROGGS nr: say -50000 +> 8
camelia niecza v24-86-g39ab531: OUTPUT«-196␤»
..rakudo b3ee5f: OUTPUT«-195␤»
FROGGS r: say(nqp::bitshiftr_i(-50000, 8))
camelia rakudo b3ee5f: OUTPUT«-196␤»
jnthn [Coke]: Well, I think I eliminated more than 2 but there are some new ones... 19:48
19:50 abnorman joined
FROGGS r: say(nqp::bitshiftr_I(-50000, 8, Int)) # <--- this one is faulty 19:50
camelia rakudo b3ee5f: OUTPUT«-195␤»
dalek kudo/nom: e0bb438 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
Thinko spotted by GlitchMr++.
19:54 abnorman left 20:00 abnorman joined
dalek kudo/nom: 60d2986 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Ops.java:
Add missing ops related to FIRST handling.
 20:02
20:04 abnorman left 20:10 lelf` joined 20:12 lelf left
moritz \o/ new HPMoR chapters 20:19
and /o\ I discovered that, and now I must go to bed :(
pmurias does any one have any good ways to avoid getting distracted while nqp compiles? ;) 20:23
masak \o/ HPMoR 20:24
tadzik interesting. I made rakudo hang 20:25
masak good. I was afraid we'd have to shoot EY.
tadzik maximum recursion degth
moritz pmurias: not reading HPMoR helps a lot :-)
masak not a total guarantee, though. 20:26
20:26 ilbot_nextgen left
pmurias some kind of notification when the compilations end would be usefull 20:26
tadzik make && echo -e "\a" 20:27
masak make && synth-speak "Oi! It's done, mate."
jnthn pmurias: Use the time to think up ways to make compilation faster. ;)
colomon +1 20:31
;)
jnthn hmmm 20:32
why does finddispatcher not find a dispatcher...
20:33 spider-mario left
tadzik liar 20:33
"you had one job"
jnthn ...'cus I was unconditionally writing a null into $*DISPATCHER 20:35
dalek p: 2cb852a | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Only set dispatcher if there is one to set.
 20:36
kudo/nom: 3fa1ed5 | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Ops.java:
First attempt at p6finddispatcher.

Makes the various tests that depend on it run, though they have some failures still.		 20:39
20:41 snoopy joined
masak r: sub days($n) { say "OH HAI" }; days 10 20:48
20:48 xenoterracide|2 left
camelia rakudo 60d298: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/FtfusZX2hJ:1␤------> sub days($n) { say "OH HAI" }; days ⏏10␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ statement… 20:48
20:49 xenoterracide joined
masak is this correct by spec? 20:49
colomon n: sub days($n) { say "OH HAI" }; days 10
camelia niecza v24-86-g39ab531: OUTPUT«Potential difficulties:␤ $n is declared but not used at /tmp/o2m4HbdmcH line 1:␤------> sub days(⏏$n) { say "OH HAI" }; days 10␤␤OH HAI␤»
masak should a prefix win over a listop in this case?
colomon std: sub days($n) { say "OH HAI" }; days 10
20:49 ggoebel left
camelia std d4cc5ab: OUTPUT«Potential difficulties:␤ $n is declared but not used at /tmp/MaOYVNsyIy line 1:␤------> sub days(⏏$n) { say "OH HAI" }; days 10␤ok 00:00 45m␤» 20:49
masak submits rakudobug
colomon days is an enum in datetime?
masak oh! 20:50
yes.
dunno what I was thinking.
colomon so std and niecza might not know about it?
masak so it's the moral equivalent of a constant declaration.
ok, so we make a better discriminator.
hold on.
colomon rn: sub pi($n) { say "Blue!"; }; pi 5
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Two terms in a row (listop with args requires whitespace or parens) at /tmp/eFwP8kn2Ln line 1:␤------> sub pi($n) { say "Blue!"; }; pi ⏏5␤Other potential difficulties:␤ $n is declared but not used at /tmp…
..rakudo 60d298: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/AyZpAI2WDc:1␤------> sub pi($n) { say "Blue!"; }; pi ⏏5␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ statement modi…
colomon std: sub pi($n) { say "Blue!"; }; pi 5 20:51
camelia std d4cc5ab: OUTPUT«===SORRY!===␤Two terms in a row at /tmp/v_uuzZOcYG line 1:␤------> sub pi($n) { say "Blue!"; }; pi ⏏5␤ expecting any of:␤ feed_separator␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Other potential difficulti…
masak rn: enum E <gna gne gnu>; say gnu; { sub gnu($x) { say $x }; gnu "OH HAI" }
camelia niecza v24-86-g39ab531: OUTPUT«===SORRY!===␤␤Two terms in a row (listop with args requires whitespace or parens) at /tmp/h3sh0Qsp6c line 1:␤------> say gnu; { sub gnu($x) { say $x }; gnu ⏏"OH HAI" }␤␤Parse failed␤␤»
..rakudo 60d298: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/OjxIDNG5Qk:1␤------> say gnu; { sub gnu($x) { say $x }; gnu ⏏"OH HAI" }␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤ …
20:51 ecocode joined
masak std: enum E <gna gne gnu>; say gnu; { sub gnu($x) { say $x }; gnu "OH HAI" } 20:51
colomon looks to me like Rakudo gets your example correct.
camelia std d4cc5ab: OUTPUT«===SORRY!===␤Two terms in a row at /tmp/0XKq2S00EF line 1:␤------> say gnu; { sub gnu($x) { say $x }; gnu ⏏"OH HAI" }␤ expecting any of:␤ feed_separator␤ infix or meta-infix␤ infixed function␤ statement modifier loop␤Parse fai…
masak yes.
rn: enum E <gna gne gnu>; say gnu; { sub gnu($x) { say $x }; gnu("OH HAI") }
camelia niecza v24-86-g39ab531: OUTPUT«gnu␤Unhandled exception: No match␤ at /home/p6eval/niecza/lib/CORE.setting line 1531 (die @ 5) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2198 (EnumMap._lookup @ 4) ␤ at /home/p6eval/niecza/lib/CORE.setting line 1336 (CommonEnum.postcircumfix:<( )> …
..rakudo 60d298: OUTPUT«gnu␤OH HAI␤»
masak hehehe.
rn: enum E <gna gne gnu>; say gnu; { sub gnu($x) { say $x }; &gnu("OH HAI") }
camelia rakudo 60d298, niecza v24-86-g39ab531: OUTPUT«gnu␤OH HAI␤»
masak ok. 20:52
only remaining question, then:
who's right in the gnu("OH HAI") case, Rakudo or Niecza?
Niecza tries to make the call on the enum. Rakudo calls the sub.
FROGGS I'd say it should prefer the invokable... 20:53
masak why? 20:54
Ayiko n: enum E <gna gne gnu>; say gnu; { sub gnu($x) { say $x }; &gnu("OH HAI") }
camelia niecza v24-86-g39ab531: OUTPUT«gnu␤OH HAI␤»
masak Ayiko: R and N agree on that one.
FROGGS masak: because you tried to call it :P
masak FROGGS: hmm. but the *real* name of the sub is &gnu 20:55
FROGGS: whereas the real name of the enum is 'gnu'.
FROGGS yes, and doing gnu( should look for a sub &gnu
masak I hear what you're saying, I just don't see you providing a strong rationale for it. 20:57
syntactically, it would make more sense for it to look for the enum. since that's the name of the enum.
20:57 PacoAir left
FROGGS IMO a name followed by opening parenthesis should always look for a symbol starting with an & 20:57
masak I'm half playing devil's advocate here, and half actually thinking that it "makes sense" for it to find the enum.
FROGGS if there are no parens it is ambitous of course
masak FROGGS: not at all. we have typecasting syntax that also uses name + ( 20:58
20:58 PacoAir joined
FROGGS hmmm, I was thinking it is name + [ 20:58
21:02 raiph left
Ayiko r: say "I'll be back in " ~ 10*minutes # might take longer ;) 21:02
camelia rakudo 60d298: OUTPUT«I'll be back in 40␤»
masak x.x
FROGGS hehe 21:03
21:06 Timbus left, Timbus joined 21:09 raiph joined 21:10 Danishman left
sorear jnthn: so the sorear/FROGGS subst fix is sane? 21:13
o/
dalek kudo/nom: 718f45a | jnthn++ | src/vm/jvm/runtime/org/perl6/rakudo/Ops.java:
Missing null check.

Fixes at least S32-array/delete.t.		 21:14
jnthn sorear: Yes, total thinko on my part. :(
masak .oO( "I cultivated this, to the point of occasionally checking in bad (working, but crappily written) code on purpose, partially to give the code gnomes something to gnaw on and partly to enhance my rep as an adequate but not great coder." ) 21:19
no. on second thought, that's totally not jnthn. ;) 21:20
jnthn++
jnthn can be decent one day and awful the next, though in the last month I've mostly been awful... 21:21
masak fwiw, I was always distraught by that part of www.sidhe.org/~dan/blog/archives/000435.html -- not wanting to believe that anyone would ever want to make that tradeoff.
thundergnat masak: any thoughts, observations or criticisms on my proposed Text::CSV stuff?
masak thundergnat: looking now. 21:22
jnthn The thinko wasn't to give people something to gnaw on, anyway. Just getting wrong something "too easy to get wrong". :)
21:24 colomon left
sorear it really works much better if you clone and capturelex at the beginning of the block instead of the point of use 21:26
(although in niecza clone and capturelex are the same operation, the point stands)
jnthn sorear: Agree, that's the model I want to move to. 21:28
sorear: I suspect it may need to be the same operation for thread safety reasons.
sorear is now trying to make sense of the $*IMPLICIT changes 21:30
jnthn sorear: The key change is what (iiuc) you suggested: $_ is always set up in finishpad.
The rest is just coping with the fallout of that change. 21:31
sorear what's p6takedispatcher et al for? 21:33
jnthn nextsame etc.
masak rn: say val($_).^name for <a 5 2/3 -0 7+i> 21:34
camelia niecza v24-86-g39ab531: OUTPUT«Str␤IntStr␤RatStr␤IntStr␤Str␤»
..rakudo 3fa1ed: OUTPUT«===SORRY!===␤Undeclared routine:␤ val used at line 1. Did you mean '&eval'?␤␤»
jnthn If the current method is entered through a dispatcher, it makes sure to set it into $*DISPATCHER
masak no I did not mean &eval, rakudo... 21:35
jnthn To be fair, it did suggest the closest implemented thing :P
wow, nqp::attrinited is completely busted... 21:36
jnthn wonders how we got away with that.. 21:37
lol 21:38
It always returend 1. :)
21:38 bruges joined
FROGGS hehe 21:38
easy to test though 21:39
jnthn That explains why defaults.t was failing
And perhaps a good bit more of S12. 21:40
21:40 bruges_ left
jnthn Anything that uses has $.x = blah; didn't get the blah. 21:40
FROGGS cool
sorear wat. 21:41
jnthn Yeah. I'm surprised we didn't catch this earlier...
sorear you mean all the isinit code I tiptoed around in P6Opaque was not actually doing anything?
dalek p: bb508c4 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/P6Opaque.java:
Fix nqp::attrinited.

Accidentally always returned true.
jnthn sorear: Well, it apparently was all correct...apart from the "not initialized" return value was inverted... 21:42
Wow. autothreading.t was failing 'cus of that, apparently...
sorear Think we're over 90 now? 21:43
jnthn Fairly likely. I regressed our.t and pseudo.t in that run, and I think fixing those alone shoulda got us close, if not there. 21:44
[Coke] does a rebuild.
21:44 abnorman joined 21:47 lelf`` joined 21:49 lelf` left
pmurias sorear: you are working on creating a persistent rakudo? 21:50
sorear pmurias: ...I hope rakudo lasts a while? 21:51
nwc10: Did you get a satisfactory answe to your gather/take q?
pmurias sorear: having one instance run the whole test suit? 21:52
jnthn pmurias: That bit already works :)
sorear pmurias: yeah, we already have the whole test suite running in one JVM
jnthn sorear++
sorear it's a bit of a hack but it's necessary to not go insane waiting
jnthn We're down to 184 test files that need attention. 21:53
(That is, they explode or something in them fails.) 21:54
Anyway, enough Rakudo hacking for me today. 21:56
Go sorear! ;)
lizmat jnthn++
sorear++
FROGGS++
21:56 logie left
lizmat masak++ for filing a rakudobug, it's been a while 21:57
sorear jnthn++ but I'm unlikely to do much today 21:58
lizmat gnight #perl6! 21:59
jnthn sorear: np :)
masak lizmat: in the end, it didn't turn into a rakudobug... :) 22:11
still something unresolved in there.
I bet TimToady will know more.
22:12 logie joined
masak logie: oh hai. :) 22:12
sorear welcome! 22:13
masak: you know em?
masak nope.
logie masak: hi!
masak oh hello hi welcome :)
22:13 colomon joined
masak rn: say "it is us, the sawdust!" 22:13
camelia rakudo 718f45, niecza v24-86-g39ab531: OUTPUT«it is us, the sawdust!␤»
sorear o/ colomon
colomon \o 22:14
masak rn: enum Result <horrible excellent>; say "Well that's just {Result.pick}" # :)
camelia niecza v24-86-g39ab531: OUTPUT«Well that's just horrible␤»
..rakudo 718f45: OUTPUT«Well that's just excellent␤»
masak :D 22:15
22:15 FROGGS left
masak rn: enum Imism <full empty>; say "The glass is definitely half {Imism.pick}" 22:15
sorear sawdust?
camelia niecza v24-86-g39ab531: OUTPUT«The glass is definitely half empty␤»
..rakudo 718f45: OUTPUT«The glass is definitely half full␤»
masak niecza is not in a good mood today.
tadzik yay, I'm given an ipv6 address by sixxs
masak sorear: www.amazon.com/The-Cyberiad-Stanisl...156027593/
jnthn Dammit, if they go giving out ipv6 addresses like this we're gonna run out of them, just like we did ipv4 ones! :P 22:16
masak only when the sawdust starts applying individually.
and all the sand. and various atoms and subatomic particles. 22:17
masak .oO( oh well. there's always ipv8, I guess )
sorear tadzik: how many did you get?
tadzik sorear: I'm now requesting a tunnel. Not sure 22:18
sorear tadzik: you need a tunnel for ipv6? I thought the US was the only country backward enough to not offer native ipv6 to everyone
tadzik sorear: hehehe. No, I don't think so :) 22:19
frankly, I've never seen native v6. Anywhere 22:20
Cyberiada \o/
masak yeah. it's a really nice book. 22:21
wonderfully translated to English, too.
sorear tadzik: I;ve heard that the norm for ipv6 is to give random people a huge number of addresses 22:22
like a /64 or so
tadzik yeah, I've heard so 22:25
and I have feeling similar to jnthn's about this thing
they say "of course we'll never be out of ipv6 addresses" 22:26
guess what ipv4 were thinking a couple dozen years ago
* ipv4 folks
pmurias jnthn: is there a smart way of doing QAST::Var.new(:decl('static'),:value($value)) with a $value that may not be a sixmodel object? 22:27
jnthn: like an nqp number or string?
jnthn pmurias: No
pmurias: Though technically those are all 6model objects away from Parrot...
It has to be something that can go in a WVal
pmurias so they can go into a WVal on jvm? 22:28
jnthn Didn't try ('cus it'd be unportable), but I don't see any reason why not.
22:28 PacoAir left
pmurias we could always make that portable ;) 22:31
[Coke] wonders how many whole class A's still exist. 22:32
22:39 raiph left 22:41 abnorman left
masak 'night, #perl6 22:43
22:48 Rotwang left
[Coke] jnthn, sorear: I have a fix for cwd() on the jvm. requires java.io.File; do I need to declare in import somewhere 22:53
? (trying now)
sorear every file needs to list the classes it uses in the import list at the top 22:54
[Coke] I just realized this fix needs to go into the nqp ops, not the rakudo invocation of them. if it's straight java, that's fine, I can do that. 22:55
sorear we seem to be following a style of 1. ASCIIbetical import sorting 2. not using wildcard imports (java.io.*) 3. one blank line between standard library imports, nonstandard library imports (if any), and org.perl6.nqp imports
[Coke] ... oh. the nqp fix is already there.
... we're just die'ing in rakudo instead of calling it. 22:56
even easier.
jnthn [Coke]++ 22:58
[Coke] jnthn: now I have to wait 30minutes for a build cycle. ;) 23:00
23:02 FROGGS joined
jnthn Yeah, but if you fix all the $*CWD hungry tests... ;) 23:08
23:08 abnorman joined 23:09 FROGGS left 23:14 logie left
jnthn 'night, #perl6 23:14
[Coke] r: say $*CWD 23:18
camelia rakudo 718f45: OUTPUT«IO::Path</home/p6eval>␤»
[Coke] ok, cwd() works (though I need to change the java function we're calling), but $*CWD doesn't seem to be automagically set on startup. 23:19
I only see it set in src/core/IO.pm
(and then, only in chdir) 23:20
ah, found a $CWD reference. checking... 23:21
is src/core/terms.pm something that requires using low level constructs? (there's a duplication of the cwd() method in here, wondering if I can reasonably expect to use IO's cwd() 23:25
23:26 yves left, yves joined
sorear [Coke]: check the order of files in the Makefile 23:28
23:29 benabik left 23:30 ecocode left 23:31 benabik joined
[Coke] yay, first pass works. 23:33
dalek kudo/nom: 5679414 | coke++ | src/core/ (2 files):
enable $*CWD & cwd() on rakudo.jvm
 23:40
23:44 jaldhar joined 23:52 abnorman left

Powered by Cro and the Raku® Programming Language.

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

Thank you!