»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.
00:04 ab5tract joined
dalek d: fc92e1e | TimToady++ | / (5 files):
get back into a self-hosting state
 00:05
00:08 jerrycheung joined 00:09 mr-foobar left 00:10 mr-foobar joined
TimToady std: "be happy" 00:10
camelia std fc92e1e: OUTPUT«ok 00:01 138m␤»
TimToady \o/
BenGoldberg std: "don't worry" ~ "be happy"
camelia std fc92e1e: OUTPUT«ok 00:01 140m␤»
TimToady that too altogether Too Long 00:11
BenGoldberg That's one second + 138 milliseconds? 00:12
std:
std: ;
camelia std fc92e1e: OUTPUT«ok 00:01 137m␤»
TimToady no, I mean the patching took too long
BenGoldberg Ahh.
TimToady I've been working on it for most of 24 hours 00:13
diakopter std: my %a{%a} 00:14
camelia std fc92e1e: OUTPUT«ok 00:01 141m␤»
diakopter a rollicking throwback rt.perl.org/Public/Bug/Display.html?id=119369 00:16
TimToady: did you have to work around a perlbug?
TimToady still not entirely sure about that 00:17
there was one place that depending on hash order, but that doesn't seem to have caused the bug 00:18
anyway, I forced infix_postfix_meta_operator to return the correct position, and doctored up viv to allow False to be parsed as a name as well as an identifier, and suppressed experimental warnings
and did a 'make reboot' to get all that into the bootstrap 00:19
00:19 raiph left
TimToady shower & 00:19
diakopter std: s|||a||b||c|| 00:20
camelia std fc92e1e: OUTPUT«===SORRY!===␤Unsupported use of suffix regex modifiers; in Perl 6 please use prefix adverbs at /tmp/_OOybAugfd line 1:␤------> s|||a||b⏏||c||␤Parse failed␤FAILED 00:01 139m␤»
diakopter hm. I'm looking back at this irclog, and I don't understand my quote here, _at_ _all_ irclog.perlgeek.de/perl6/2013-10-25#i_7762134 00:24
oh nm, now I do
the p6weekly thing needs "on this day in each of the last 9 years of #perl6 quotes" 00:26
timotimo diakopter: if you'd like to contribute an amusing/astounding/funny/interesting/... quote per past year for every week, i'll gladly accept your contributions :) 00:27
diakopter well, they'd be impossible to distinguish from villifying snark
timotimo may be 00:28
diakopter well, 1 year ago today TimToady was hired at Craigslist 00:29
tony-o nice
00:34 chenryn left 00:42 ab5tract left 00:43 Ben_Goldberg joined, BenGoldberg left 00:44 Ben_Goldberg is now known as BenGoldberg
timotimo t.h8.lv/p6bench/2014-09-18-pre_rele...akudo.html / t.h8.lv/p6bench/2014-09-18-pre_release_nqp.html 00:49
i may have run the newest benches with different parameters or something
in quite a few benchmarks, performance has tanked dramatically since the previous release :( 00:53
tony-o timotimo: i have this working on jvm but moar isn't 00:54
i have a feeling that i'm hitting some ceiling
timotimo aaw, bummer
tony-o jvm is just as fast once it burns in (and starts up)
01:06 yeahnoob joined 01:10 JimmyZ joined 01:14 slmult0_z joined
tony-o timotimo: added middleware tests, seems to be working on moar at the moment 01:14
01:14 ab5tract joined
tony-o i'm also refactoring this stuff as i go to make it less 'Promise' heavy and have better flow control 01:14
01:14 bjz left, FROGGS_ joined 01:17 ab5tract left 01:18 FROGGS left, Larre_the_cow left 01:29 dayangkun joined 01:34 araujo left 01:35 araujo joined 01:36 leont left 01:38 dayangkun left 01:39 JimmyZ left 01:51 slmult0_z left
retupmoca timotimo, arnsholt: Net::ZMQ fixed: github.com/arnsholt/Net-ZMQ/pull/1 02:03
.tell masak I have a couple of PR's open for you (on Data::Pretty and Text::Markdown) 02:06
yoleaux retupmoca: I'll pass your message to masak.
02:24 xragnar left 02:25 chenryn joined 02:26 xragnar joined 02:34 xragnar left 02:43 dj_goku left 02:44 noganex joined 02:48 noganex_ left 03:06 jerrycheung left 03:09 xragnar joined, xragnar left 03:11 molaf joined 03:28 Akagi201 left 03:29 Akagi201 joined, jerrycheung joined 03:31 molaf left 03:33 Akagi201 left 03:39 tinyblak joined 03:47 BenGoldberg left 03:49 xragnar joined 03:51 Tacit joined, Tacit is now known as TacitSage 04:02 Akagi201 joined 04:03 kaare_ joined 04:05 xragnar left 04:06 xragnar joined 04:09 [Sno] left 04:13 anaeem1_ joined 04:14 TacitSage left
awwaiid I got some perl6 advocacy in today :) 04:16
They always see my DCBPW shirt and ask sarcastically about Perl6. Increasingly, I tell them -- Winter is coming. 04:17
04:23 kaare_ left 04:24 rindolf joined 04:29 anaeem1__ joined, anaeem1_ left 04:31 kaare_ joined
dalek ast: 81291eb | (David Warring [email@hidden.address] | integration/advent20 (7 files):
[integration] 'temp OUT' => 'my OUT'
 04:35
japhb timotimo: Did startup time get worse on nqp? It looks like it from the bar charts at the top of the nqp benchmarks .... It also looks like the only place nqp sped up is JSON parsing and a tiny amount in a couple rc-* tests. 04:36
04:42 chenryn left 04:45 mr-foobar left 04:52 kaare_ left 05:13 chenryn joined 05:19 kaare_ joined 05:20 chenryn left
japhb Oddly it looks like much of the r-m performance regression might be explained by push getting slower. 05:33
05:34 Psyche^ joined 05:36 Psyche^_ left 05:42 [Sno] joined 05:47 tinyblak left 05:51 FROGGS_ is now known as FROGGS
FROGGS .tell masak for ya' release: gist.github.com/FROGGS/78c5203d0b145ca57196 05:52
yoleaux FROGGS: I'll pass your message to masak.
05:53 chenryn joined 05:57 yeahnoob left 06:00 yeahnoob joined 06:03 Ven joined 06:08 Akagi201 left 06:09 mr-foobar joined 06:16 Ven left 06:17 wamiks joined 06:24 FROGGS left 06:25 wamiks left 06:34 xinming left 06:35 SamuraiJack joined 06:36 xinming joined 06:37 SamuraiJack left 06:40 mr-foobar left 06:41 mr-foobar joined 06:50 Akagi201 joined 06:56 Timbus_ left 06:57 Akagi201 left 06:58 cognome left 06:59 cognome joined 07:00 FROGGS joined, leedo left, zakharyas joined 07:01 Akagi201 joined 07:06 leedo joined, cognome_ joined 07:07 cognome left
sergot hi o/ 07:07
nwc10 \o
japhb o/ 07:10
(I'm in your timezones, scaring your residents) 07:11
07:22 darutoko joined
tadzik heh 07:22
07:23 Timbus joined 07:33 firnsy left, firnsy joined 07:34 slmult0_z joined, xragnar left 07:37 xragnar joined 07:39 eiro left, tardisx left, PerlJam left 07:43 virtualsue joined 07:45 cognome joined 07:47 cognome_ left
dalek ast: 3b63d6f | (Elizabeth Mattijsen)++ | S17-procasync/print.t:
Fudge 2 more flapping tests
 07:49
FROGGS # Failed test 'checking X::Proc::Async::AlreadyStarted.WHICH' 07:50
# at t/spec/S02-types/WHICH.t line 313
# expected: 'X::Proc::Async::AlreadyStarted'
# got: 'Failure|1018962658'
:o(
dalek ast: 7321365 | (Elizabeth Mattijsen)++ | S17-procasync/basic.t:
Also test return value
lizmat FROGGS: will check
FROGGS I'll fix t/spec/S10-packages/precompilation.t, which should be easy 07:51
lizmat what's wrong with that?
and where ?
lizmat doesn't see any problems
FROGGS No such method 'at_key' for invocant of type 'BOOTHash' 07:52
in sub postcircumfix:<{ }> at gen/jvm/CORE.setting:2862
in method precomp at gen/jvm/CORE.setting:21972
on jvm only
lizmat ah, ok
I see...
FROGGS here is the list of problems: gist.github.com/FROGGS/78c5203d0b145ca57196
you can scratch the output of moar though 07:53
07:54 jack_rabbit left
dalek ast: bfb5f8a | (Elizabeth Mattijsen)++ | S02-types/WHICH.t:
Fix JVM, Proc::Async related doesn't work there
 07:56
lizmat FROGGS: ^^^ that should fix the pb on JVM, building uptodate JVM now 07:59
(and parrot for that matter)
FROGGS will also rebuild and test
lizmat++
08:00 eiro joined, tardisx joined, PerlJam joined 08:03 slavik left
dalek kudo/nom: a396fa2 | (Tobias Leich)++ | src/core/CompUnit.pm:
fix CompUnit.precomp on jvm
 08:04
08:07 telex left 08:08 telex joined
lizmat FROGGS: I guess we need to encapsulate access to input parameters somehow, to prevent the same issue propping up at different places... 08:09
FROGGS lizmat: yeah, we probably want to hllize it for the user, like we do with %*ENV and others 08:10
dalek ast: dfeebad | (Elizabeth Mattijsen)++ | S02-types/WHICH.t:
Oops, forgot Proc::Async itself
 08:11
08:12 fhelmberger joined, erkan joined, erkan left, erkan joined
lizmat FROGGS: not sure what the issue is with S10-packages/precompilation.t 08:14
ah, I do... same issue
FROGGS lizmat: that's what I just fixed
lizmat indeed... need to pull and rebuild :-) 08:15
lizmat is not quite awake yet
FROGGS lizmat: this also seem to fix the S22-package-format fail
lizmat yeah, they were the same
FROGGS though, t/spec/S32-io/slurp.t still fails on parrot :o( 08:16
m: say +lines()
camelia rakudo-moar d0e09e: OUTPUT«20␤»
FROGGS p: say +lines()
camelia rakudo-parrot d0e09e: OUTPUT«20␤»
FROGGS hmmmm
08:17 slavik joined
nine I wonder how DBIish deals with binary data. 08:17
08:19 dakkar joined
lizmat FROGGS: t/spec/S17-scheduler/every.t seems a flapper that doesn't get enough CPU when run in the spectest 08:22
FROGGS tests about timers are always dangerous... 08:23
dalek ast: bbb789b | (Elizabeth Mattijsen)++ | S17-scheduler/every.t:
Be a little more lenient in interval checking
 08:24
lizmat Hmmm.... the failure of t/spec/S17-supply/uniq.t on JVM I can only explain if sleep 1 sleeps less than a second 08:28
sleep isn't integer based on JVM, is it ?
08:29 kaleem joined
lizmat afk for a few hours& 08:29
08:31 yeahnoob left
FROGGS perl6-j -e 'sleep 3; say time - BEGIN time' 08:31
4
08:38 bjz joined 08:39 pmurias joined 08:44 yeahnoob joined 08:52 sqirrel joined
daxim__ parrot just had 6.8 release a couple of days ago, when are the september releases of moar, nqp, rakudo scheduled? 09:00
nwc10 today, I think.
09:01 sqirrel left
FROGGS daxim__: today, aye 09:01
xiaomiao oh my, more packaging work for me tomorrow :D 09:02
FROGGS :D
nwc10 xiaomiao: I forget, what OS are you packaging for?
xiaomiao gentoo 09:03
daxim__ I have up-to-date versions of libuv, dyncall, linenoise under my control in rpm-land now, will ask debian.or.at tonight for assistance with dpkg-ising 09:05
I hope until apw2014 full backended rakudos are available on major platforms 09:06
09:09 jerrycheung left
FROGGS daxim++ 09:09
09:12 noganex left, slmult0_z left
timotimo sounds great! 09:13
i can't really tell why push would have become slower :\ 09:15
09:16 rindolf left
dalek line-Perl5: 0235056 | nine++ | / (3 files):
Return binary strings from Perl 5 as Buf to Perl 6

Non-UTF-8 non-7-Bit-ASCII strings are now returned as Buf from Perl 5. If they are strings at all the user can tell Buf.decode the encoding and get a Str back. Otherwise we treat it as uninterpreted binary data.		 09:16
nine .tell leont just because it's impossible doesn't mean I can't do it :) 09:17
yoleaux nine: I'll pass your message to leont.
09:18 rindolf joined 09:20 donaldh joined 09:25 dwarring left, leont joined 09:29 Woodi left, Woodi joined 09:43 yeahnoob left 10:04 chenryn left
masak hi #perl6. 10:28
yoleaux 02:06Z <retupmoca> masak: I have a couple of PR's open for you (on Data::Pretty and Text::Markdown)
05:52Z <FROGGS> masak: for ya' release: gist.github.com/FROGGS/78c5203d0b145ca57196
masak thanks, retupmoca, FROGGS.
I'm going to start the release process once I'm back from dinner, in O(2 hours). 10:29
&
10:34 chenryn joined
FROGGS masak: I'm going to update the gist I posted... 10:35
dalek p: 31bc846 | (Tobias Leich)++ | src/vm/parrot/QAST/Operations.nqp:
read a byte and not a char in EOF check

Reading a char can fail when this char gets decoded to utf8.		 10:37
kudo/nom: 2160632 | (Tobias Leich)++ | tools/build/NQP_REVISION:
bump nqp revision for eoffh@parrot fix
 10:38
FROGGS masak: perl6-p is down to one failing test file: gist.github.com/FROGGS/78c5203d0b145ca57196
10:39 chenryn left 10:45 yoleaux left
carlin nine++ # making Inline::Perl5 handle binary 10:46
... .oO( and if I'd remembered to push my code before leaving home I could have started making use of it )
10:47 woolfy1 joined, woolfy left 10:50 chenryn joined
leont jnthn, lizmat: can a merge (stdout/stderr) option be added to Proc::Async? 10:51
10:52 yoleaux joined, ChanServ sets mode: +v yoleaux
lizmat leont: with our current check for proper order, I think that should be possible 10:57
there just won't be a guaranteed order between stdout/stderr
otoh, after we have Supply.lines, you can just .merge the two supplies 10:58
lizmat starts working on that 11:01
11:03 FROGGS left 11:05 brrt joined
leont I meant merging on an fd level 11:08
yoleaux 09:17Z <nine> leont: just because it's impossible doesn't mean I can't do it :)
leont dup2 and stuff, except on a libuv level 11:09
nine leont: does this look sane to you? github.com/niner/Inline-Perl5/commit/0235056c6e
lizmat leont: probably possible, but that would be up to jnthn 11:12
11:14 SamuraiJack joined 11:17 SamuraiJack left
lizmat leont: wouldn't you then run the risk of mixed incomplete lines? 11:18
11:22 salv0 left 11:26 brrt left 11:34 chenryn left 11:38 salv0 joined 11:39 sqirrel joined 11:40 FROGGS joined 11:50 virtualsue left, chenryn joined 11:51 anaeem___ joined 11:54 anaeem1__ left 11:57 bjz left
lizmat www.savagechickens.com/wp-content/u...mascot.jpg feels strangely appropriate in a Perl 6 context 11:59
leont To some extend, yes, depending on the buffering of the program. Not saying it's always sensible, but some applications like it (which sometimes includes tests) 12:01
12:03 anaeem___ left 12:04 anaeem1 joined
masak greetings, people of #perl6. 12:32
timotimo regreetings, masak of the east
masak your monthly release process is now starting. fasten seatbelts and stow your tray table.
have a nice flight.
FROGGS lizmat: does that also need to be fudged for jvm? 12:34
# at t/spec/S02-types/WHICH.t line 325
# expected: 'Proc::Async'
# got: 'Failure|1438485088'
Failed tests: 551, 571, 573, 575
lizmat I though I fixed that ?
12:34 bjz joined
FROGGS lizmat: I just pulled 12:34
lizmat in my version of WHICH.t, Proc::Async is listed in the moar section 12:35
which should be skipped on jvm ?
FROGGS ohh, I'm not running the fudged version 12:36
lizmat *that* would explain :-)
FROGGS :o)
yeah, works now
phewww :/
err, :o)
pmurias what does 'MVMArray: atpos expected string register 12:41
' mean
timotimo you used atpos and should have used atpos_s
moarvm is stricter about this than others are
FROGGS timotimo: I guess it is the other way around, no? 12:42
he probably used atpos_s when he should not
12:42 virtualsue joined
masak FROGGS: re t/spec/S32-exceptions/misc.rakudo.parrot -- should we fudge it on Parrot, you think? 12:42
FROGGS: what's the failure? how bad is it? 12:43
timotimo FROGGS: might be either way
masak is it a recent regression? has anyone done a bisect on it?
FROGGS masak: I'm running it right now to see...
masak FROGGS++
FROGGS timotimo: hmmm, oaky
masak FROGGS: you are everything a release manager could ask for, and more. <3
[Coke] FROGGS: by hand or using the corpus of data already available at perl6-roast-data ?
just trying to save you time, man. :) 12:44
FROGGS [Coke]: by hand
masak: *g*
12:44 chenryn left
FROGGS # at lib/Test.pm line 309 12:45
# Expected: (X::Obsolete)
# Got: (X::AdHoc)
# Exception message: Cannot get character past end of string
pmurias timotimo: and now I have a segmentation fault, moarvm--
FROGGS # at t/spec/S32-exceptions/misc.rakudo.parrot line 532
m: use Test; my $*foo = 0; throws_like { EVAL '$*foo = 1; say' }, X::Obsolete; 12:46
camelia rakudo-moar 216063: OUTPUT« 1..2␤ ok 1 - code dies␤ ok 2 - right exception type (X::Obsolete)␤ok 1 - did we throws_like X::Obsolete?␤»
FROGGS p: use Test; my $*foo = 0; throws_like { EVAL '$*foo = 1; say' }, X::Obsolete;
camelia rakudo-parrot 216063: OUTPUT«use of uninitialized value of type Nil in numeric context in sub throws_like at lib/Test.pm:294␤␤ 1..2␤ ok 1 - code dies␤ not ok 2 - right exception type (X::Obsolete)␤ ␤# Failed test 'right exception type (X::Obsolete)'␤# at l…»
FROGGS p: say 12:47
camelia rakudo-parrot 216063: OUTPUT«===SORRY!===␤Cannot get character past end of string␤»
FROGGS >.<
golfed to three bytes
ahh, the fix seems easy 12:48
masak FROGGS++! 12:49
so the bare 'say' fix broke Parrot and no one noticed... 12:50
FROGGS no
[Coke] masak: that is not entirely true.
FROGGS my s/substr/eqat/ optimization did
[Coke] many things have been breaking parrot for weeks now.
FROGGS yeah, too many 12:51
[Coke] feels like boy who cried wolf.
masak we need a better volunteering system for these things. one that ends up not burning out or dissapointing volunteers. 12:52
I saw something nice in that vein happen in yesterday's backlog.
[Coke] FROGGS: does the travis system run spectests or just the build? 12:56
FROGGS [Coke]: it just runs 'make test' after building perl6-m 12:57
pmurias nqp-m: nqp::join(',', [nqp::null_s])
FROGGS (and nqp-m and moar every time)
camelia nqp-moarvm: OUTPUT«(signal )»
pmurias is that segfault correct behavior? 12:58
nqp-m: say(nqp::join(',', ["foo","bar",nqp::null,"baz"])) 12:59
camelia nqp-moarvm: OUTPUT«foo,bar,baz␤»
pmurias and the above also seems a bit suspect
12:59 anaeem1 left
[Coke] segfault is never correct 12:59
masak fsvo "never" :) 13:00
13:00 mr-foobar left
masak there are tradeoffs in the MOP where, if you shoot yourself in the foot, your foot ends up well and truly shot. 13:00
dalek kudo/nom: dde0dc3 | (Tobias Leich)++ | src/Perl6/Grammar.nqp:
fix "say" at EOF

nqp::eqat seems to be less tolereant about being oob, compared to nqp::substr.		 13:02
13:03 Alina-malina left
FROGGS masak: this should mean that moar and parrot pass (with todos), and jvm probably flaps t/spec/S17-supply/uniq.rakudo.jvm 13:03
masak ended up doing `git clone --depth=20 [email@hidden.address] because the full clone took *ages* from where he is
FROGGS: awesome.
FROGGS: is there any reason we prefer flapping to fudged?
FROGGS masak: well, I don't 13:04
I prefer passing tests :o)
13:04 Alina-malina joined
carlin nine: would it be possible to pass a Buf back to Perl 5? 13:04
[Coke] ideally, skip flapping tests and open an RT. 13:05
lizmat is looking at uniq.t
[Coke] I could be convinced that a TODO would be OK if the fudge comment specifically says "flapping"
13:05 eternaleye left
masak lizmat++ 13:05
[Coke]: agreed.
nine carlin: planning on doing that this evening 13:06
masak let's make this release squaky-clean for the maximum number of people and backends! :)
lizmat it was always test #7 that was flapping, right ?
13:07 eternaleye joined
masak .oO( flapping tests giving us the bird ) 13:07
carlin nine++ \o/ 13:08
13:15 rurban joined
timotimo nine: with what you've learnt so far, would Inline::Python for perl6 be harder than for perl5? and would it be harder than Inline::Perl5? 13:18
13:18 noganex joined
dalek ast: 670d41a | (Elizabeth Mattijsen)++ | S17-supply/uniq.t:
Give expiry test some more time
 13:18
lizmat FROGGS, masak ^^^ hope that fixes it in a less flappy way 13:19
FWIW, I've never been able to complete a 'make spectest' on JVM :-(
FROGGS lizmat: I was able to over night 13:20
with TEST_JOBS=1
pmurias timotimo: I think Inline::Python would be easier, as python is a more sane language then perl5
timotimo haha
nine timotimo: right now it would probably be easier because one can re-use most of Inline::Perl5's Perl6 code and just swap out the p5helper.c with a pyhelper.c. The latter is probably just a lot of copy&paste from Perl5's Inline::Python.
timotimo cool :)
nine timotimo: Perl 6 having a GC instead of refcounting helps a lot here. Getting two subtly different refcounting systems correct was a major headache when writing Inline::Python. I remember lots of hours reading not so useful documentation and source code of both Python and Perl. 13:22
timotimo oh my :S
maybe we can do Inline::Pypy instead, that doesn't do ref counting either
13:23 kaare_ left
nine timotimo: on the other hand Inline::Python is really, really stable. It doesn't leak a single byte in any case I could come up with. 13:23
dalek kudo/nom: 12ca984 | masak++ | docs/announce/2014.09.md:
[docs/announce/2014.09.md] added

Next up: filling in changelog highlights and contributors.
nine I guess that one could port it to Perl 6 over a rainy weekend.
masak people reading through and reviewing ^^ would be really heplful. kthx!
nwc10 masak: startup is now faster than Moose! (At least some platforms) 13:24
gtodd squeak++
oops
squeaky++
lizmat nwc10: it will get 25% faster still once we can get the lazy @*INC population working in precomp 13:25
nwc10 lizmat: that's next month's news? 13:26
lizmat I hope so: totally blocked on figuring out why that doesn't work
[Coke] jvm - been able to complete those test for some time on host07. need to figure out why they are exploding on host08 :P 13:27
lizmat masak: I assume we're on a commit freeze atm, right ?
masak lizmat: I'm leanient... but I expect people not to do anything stupid during release cycle :) 13:28
also, commit with the knowledge that your commits may end up before the release tarball/tag, or after it.
PerlJam greetings #perl6 people! o/
masak no leap seconds this month. nice. :)
PerlJam masak++ release 13:29
lizmat I'll keep Supply.lines for next month's release then
potential users like leont are on HEAD anyway, I would hope
masak *nod*
lenient*
lizmat "leanient" - a tolerance for lean persons 13:30
gtodd why does perl6 -n -e 'Supply.interval(1).tap({say time}); ' not start prtinting out the time right away? 13:31
(p6 stis there until I press return)
PerlJam gtodd: because of the -n
gtodd errm sits
yeah that's what I was trying to work around :-) 13:32
without the -n it prints right away but exits :)
PerlJam gtodd: BEGIN is your friend
gtodd ok just silliness on my part
oh YEAH!
good
thanks
nwc10 masak: current estimate - startup 17% less time than Moose. 13:34
so saying 15% is safe, if you want to say something
gtodd ok one other silly Q .. I have a one liner baby perl 5 versus 6 "speed tests" I do now and then one that picks a random element from a large array read from disk ...
(these are not real tests) ... 5: perl -MIO::All -E 'my @arr = io("/usr/share/dict/words")->slurp; say $arr[rand @arr]; 6: perl6 -e 'my @arr := "/usr/share/dict/words".IO.lines; say @arr.pick' 13:35
I'm pretty sure perl6 -e 'my @arr := "/usr/share/dict/words".IO.lines; say @arr.pick;' used to work fine but now it can't read the file handle .... 13:36
lizmat gtodd: because perl6 is lazy, you will get wildly different values
lizmat looks
gtodd hmm ... 13:37
lizmat: errm wait /usr/share/dict/words is a link to /usr/share/dict/web2
masak nwc10: yes, this might be worth mentioning.
gtodd ok perl6 -e 'my @arr := "/usr/share/dict/web2".IO.lines; say @arr.pick'
works
lizmat gtodd: you may find that IO.slurp.lines is faster 13:38
nwc10 masak: specifically, those figures are for "my" machine, which is someone else's big multicore x86_64 Centos box on which I have an account
but I think that they hold for most x86_64
lizmat apparently slurping the whole file in mem, and then looking for line endings, is faster
than reading line by line
nwc10: I'm seeing similar values on my MBP 13:39
nwc10 masak: I think that it is worth mentioning, as it's a "take us seriously" point, but it needs some sort of subtle caveat about "other stuff isn't faster yet. But we're actively working on it." or something like that 13:40
13:40 bjz left
masak damnit, Oracle! your site doesn't offer a clear, easy way to download a JDK! it's as if you don't *want* people to develop in your antiquated language! 13:40
nwc10 ... it's as if you don't want people to install the Ask toolbar
masak bonus points to anyone who direct-links me a .tar.gz file to download 64-bit Linux JDK, here on the channel. 13:41
tadzik challenge accepted
nwc10 tadzik: were you thinking of coming to Salzburg?
tadzik masak: download.oracle.com/otn-pub/java/jd...x64.tar.gz
nwc10: yes
nwc10 tadzik: please register. t-shirts are being ordered soon 13:42
leont masak: then try finding the right documentation for things…
13:42 Ven joined
tadzik nwc10: oh, okay 13:42
nwc10: thanks for the reminder
nwc10 tadzik: please encourage others who want to attend
*and* want a t-shirt :-)
nine The Austrian Perl 6 Workshop it seems :)
13:43 bjz joined
gtodd lizmat: thanks ... for the IO stuff (Failed to stat in filehandle:) I must have used a different file to construct the array or a different system (with a file rather than link) ... but yes using .slurp.lines gives 3.69s versus 4.42s 13:43
nwc10 nine: 5 is also available
tadzik masak: oh nevermind, that doesn't actually work :/
nwc10 but the hackathon is definately biased
tadzik stupid oracle 13:44
masak tadzik: see? D:
tadzik you can probably deduce some hack from aur.archlinux.org/packages/or/orac...4/PKGBUILD though
13:44 Ven left
nwc10 also, please star the hackathon. we're going to start worrying if it gets near 25 13:44
nine It occurs to my that the relation between Perl5 and Perl6 is a lot like between C and C++
masak my challenge still stands. direct link, .tar.gz, JDK 7, 64-bit Linux.
tadzik there probably isn't any, apart from the mirrors :/ 13:45
13:45 Ven joined
masak I'm not interested in "no" answers, I'm interested in the link... :) 13:45
tadzik heh
Ven o/, #perl6 13:46
jnthn masak: Occasionally, there's benefits to being on Windows ;) 13:47
13:47 virtualsue left
nine jnthn: ironically in situations involving a large UNIX vendor it seems 13:47
FROGGS masak: gist.github.com/P7h/9741922
masak: you have to pass headers... 13:48
gtodd masak: there could be a 1-800 number that you can dial and leave a credit card number ... :-)
13:48 vendethiel joined
[Coke] sees gittip, which never really took off in his circles, is now called gratipay 13:48
masak ok, I think I finally started the download.
ETA: 4 hours... 13:49
gtodd :-D
masak in the meantime, I think I'll build Rakudo without the JVM backend. 13:50
gtodd masak: in the meantime ORA will make 10million$ ;-)
(and spend 9.9million$) 13:51
masak I have no idea what you're talking about. :)
gtodd good! 13:52
masak but I still don't have a `javac`, so it matters little.
moving right along.
gtodd hehe
masak FROGGS++ # I think that would've worked; thanks -- I ended up finding the link in the browser and doing it there 13:53
vendethiel [Coke]: they've lost all their "customers" anyway :p 13:57
13:58 bjz left
nwc10 masak: also, I hope the reaction to "faster than Moose" is "but of course Moose is slow" 13:58
13:59 bjz joined
leont [Coke]: it seems people are getting bored and leaving it. Or at least I'm getting less money than I used to, not that I ever got a lot. 13:59
lizmat nwc10: nonetheles, many people use Moose in production :-) 14:00
*ss
nwc10 Oh, I don't deny any of that :-) 14:01
14:01 fhelmberger left 14:02 fhelmberger joined 14:04 JimmyZ joined
JimmyZ speaking of start up time, perl6 -e0 costs 0.22s, but perl6 -MHTTP::UserAgent -e0 costs 0.56s 14:05
looks like loading a module is slow ... 14:06
lizmat JimmyZ: is it installed with panda (aka precompiled??) 14:08
JimmyZ yeah
masak m: my $perl = 5; say $perl.:<++>
camelia rakudo-moar dde0dc: OUTPUT«6␤»
masak \o/
lizmat also, I noticed that panda adds an @*INC entry for each installed module
this is bad and will need to be fixed (and is on my list as soon as I unblock from S11)
14:10 virtualsue joined
tadzik hmmm 14:11
that sounds really weird 14:12
14:14 sqirrel left
dalek ast: a804241 | (Elizabeth Mattijsen)++ | S17-supply/lines.t:
Add tests for Supply.lines
 14:14
14:15 rindolf left
masak oh, I didn't see ec17e06ba53e643623dc200d4e116fc97323c45a until now. 14:18
...not sure I think it's a good idea.
m: say ().min
camelia rakudo-moar dde0dc: OUTPUT«Inf␤»
masak m: say Int.min
camelia rakudo-moar dde0dc: OUTPUT«-Inf␤»
vendethiel m: say [min] () # masak 14:19
camelia rakudo-moar dde0dc: OUTPUT«Inf␤»
vendethiel seems pretty consistent to me
PerlJam Why not Nil?
vendethiel mumbles something about monoids to PerlJam 14:20
PerlJam: if they can, reduct operators try to give back an indentity value when you don't pass them lists 14:21
(i.e. a value that, when in the lists passed to said reduct functions, would never change its outcome)
so, here, you can have as many "Inf"s in your lists, [min] isn't gonna change its result. (same for [+] and 0)
dalek kudo/nom: 66694e8 | masak++ | docs/ChangeLog:
[docs/ChangeLog] various updates

Hopefully covers all the highlights of this month.		 14:22
masak ^^ extra eyes welcome.
what vendethiel said. 14:23
14:23 sqirrel joined
masak making Int.min be -Inf is the kind of decision that seems pleasant at the surface, but is actually a corner case/exception, just as much as (for example) negative array indices. 14:24
FROGGS masak: $*EXECUTABLE does actually work now on parrot
masak FROGGS: please add.
FROGGS meh
14:25 orafu joined
masak or, put differently, the .min that takes the minimum of a bunch of values is a completely different .min than the one on a type object. they are two different methods with the same spelling. 14:25
14:25 Ven left
masak (and there's no nice unification between the two types of .min that makes it nice or desirable) 14:25
jnthn masak: Seems .lines gets two mentions 14:26
PerlJam m: [min] <foo bar baz> # Not sure Inf is the identity here :)
camelia ( no output )
PerlJam m: [say min] <foo bar baz> # Not sure Inf is the identity here :)
camelia rakudo-moar dde0dc: OUTPUT«===SORRY!=== Error while compiling /tmp/YOzlz1Rfj1␤Two terms in a row␤at /tmp/YOzlz1Rfj1:1␤------> # Not sure Inf is the identity here :)⏏<EOL>␤ expecting any of:␤ argument list␤ postfix…»
PerlJam m: say [min] <foo bar baz> # Not sure Inf is the identity here :)
camelia rakudo-moar dde0dc: OUTPUT«bar␤»
masak jnthn: saw it. fixing. 14:27
dalek kudo/nom: 919b030 | (Tobias Leich)++ | docs/ChangeLog:
mention $*EXECUTABLE fix on parrot
 14:28
masak PerlJam: Inf *is* the identity for Str, too.
PerlJam: by spec.
PerlJam: see recent discussion between me and TimToady in backlog.
PerlJam looking now even :)
nine m: 'abc'.encode('latin-1').perl
camelia ( no output )
nine m: say 'abc'.encode('latin-1').perl 14:29
camelia rakudo-moar dde0dc: OUTPUT«Blob[uint8].new(97, 98, 99)␤»
nine shouldn't this be a Buf? ^^
jnthn No
dalek kudo/nom: 0b70d0d | masak++ | docs/ChangeLog:
[docs/ChangeLog] remove a few double mentions
jnthn Note that Buf ~~ Blob 14:30
nine jnthn: according to doc.perl6.org/routine/encode#class_Str it should
jnthn Blob is the immutable one
FROGGS jnthn: are you fine with me merging cglobals after release?
jnthn nine: Then I think it's out of line with S02...
FROGGS: gah, I forgot about that
FROGGS: But I've got a moment now I can use to review it. Where can I find the stuff I should look at? 14:31
vendethiel masak: I'm not sure there's any value I want to be returned for ().min, truthfully 14:32
but I can see it looking more "consistent" that way
m: say (1, 2, 3).min 14:33
camelia rakudo-moar 12ca98: OUTPUT«1␤»
vendethiel if that works; then it should be the same as [min] 1, 2, 3
lizmat masak: you want me to take out the Type.min/max/minmax ?
FROGGS jnthn: github.com/MoarVM/MoarVM/compare/cglobal
14:33 JimmyZ_ joined
FROGGS jnthn: github.com/perl6/nqp/compare/cglobal 14:34
jnthn: github.com/jnthn/zavolaj/compare/cglobal
not much to read
jnthn Danke
FROGGS gern geschehen :o)
dalek ast: ae5b750 | (Elizabeth Mattijsen)++ | S17-supply/lines.t:
Add tests for unchomped lines
lizmat masak: it just seemed like a logical "niche" 14:35
14:35 JimmyZ left, JimmyZ_ is now known as JimmyZ 14:36 FROGGS left, slavik left 14:37 FROGGS[mobile] joined, Ven joined 14:38 raiph joined
jnthn FROGGS[mobile]: Wow, yes, not much code at all 14:39
FROGGS[mobile]: Slight issue: we leak the library handle.
FROGGS[mobile]: Probably good to fix that one... 14:40
FROGGS[mobile]: Also, I suggest die with NYI in the STORE in NativeCall, so folks won't think it's working
arnsholt If it's avoidable, keeping the leaks in Zavolaj to a minimum is probably a good idea, yeah
There are a couple of unavoidable leaks already =/ 14:41
jnthn I guess for now it prints stub code executed or so
But that isn't very clear about what's up
FROGGS[mobile] k
will do what I can soonish :o)
jnthn So, think it's just those two things I have.
We may be able to do better than the Proxy in the future once I do the natives improvements, but I think even then we can keep the API the same. 14:42
Also, nice that it's relatively little code to make this work :) 14:43
FROGGS++
FROGGS[mobile] that's very true 14:44
arnsholt FROGGS++ indeed
It's great that you're hacking on Zavolaj
FROGGS[mobile] :o)
well 14:45
it is awesome
PerlJam speaking of zavolaj, is it a known thing that there are test failures in 05-arrays.t, 06-struct.t, and 08-callbacks.t?
(I was going to play with Inline::Perl5 which requires NativeCall which reports failures when testing) 14:47
nine FROGGS[mobile]: if you're looking for another feature to implement: Buf or Blob or whatever it's called support would be quite nice.
arnsholt PerlJam: If you're on 32-bit it might be 14:48
IIRC there are some annoyances in that area
PerlJam aye, this box is 32-bit
FROGGS[mobile] that's what I like to fix next, 32bit issues 14:49
so OpennSSL becomes portable
and my XML::LibXML
14:50 slavik joined
dalek kudo/nom: ed1d46c | (Elizabeth Mattijsen)++ | src/core/Supply.pm:
Add support for Supply.lines, as per spec
 14:53
kudo/nom: 807dcb5 | (Elizabeth Mattijsen)++ | docs/ChangeLog:
Mention that Str.lines/IO.lines take an :eager
lizmat *ARGHFFHFHGSFHGAF
I didn't want to push ed1d46c5a2 14:54
dalek kudo/nom: 8970769 | (Elizabeth Mattijsen)++ | src/core/Supply.pm:
Revert "Add support for Supply.lines, as per spec"

Not for this release
FROGGS[mobile] dereferencing a shout is not yet supported
14:57 Ven left 14:58 tere joined
dalek line-Perl5: 5789415 | nine++ | / (3 files):
Support passing blobs and Bufs from P6 to P5 as binary strings.
 15:00
nine carlin: ^^^ for you
lizmat gtodd: forgot to mention, IO/Str.lines now take a :eager parameter, which is about 5x as fast 15:03
sergot m: EVAL "die"; say $!; 15:04
camelia rakudo-moar 12ca98: OUTPUT«Died␤ in block <unit> at EVAL_0:1␤ in any <unit-outer> at EVAL_0:1␤ in sub EVAL at src/gen/m-CORE.setting:781␤ in block <unit> at /tmp/gVLo_HlmWD:1␤␤»
sergot m: EVAL "die"; say "$!:", $!;
camelia rakudo-moar 12ca98: OUTPUT«Died␤ in block <unit> at EVAL_0:1␤ in any <unit-outer> at EVAL_0:1␤ in sub EVAL at src/gen/m-CORE.setting:781␤ in block <unit> at /tmp/2pVy7zkPJl:1␤␤»
sergot can I check if EVAL died or not?
somehow
TimToady m: try EVAL "die"; say $!; 15:05
camelia rakudo-moar 12ca98: OUTPUT«Died␤ in block <unit> at EVAL_0:1␤ in any <unit-outer> at EVAL_0:1␤ in sub EVAL at src/gen/m-CORE.setting:781␤ in block <unit> at /tmp/V8KJsUHZuo:1␤␤»
TimToady m: try EVAL "die"; say "alive"
camelia rakudo-moar 12ca98: OUTPUT«alive␤»
masak is back from a slight household emergency 15:07
masak backlogs
sergot m: try EVAL "die"; say "$!:";
camelia rakudo-moar 12ca98: OUTPUT«Died:␤»
sergot TimToady++ thanks!
masak hm, Parrot stresstest yields many many failures.
lizmat masak: I accidentally commited Supply.lines, but reverted immediately
masak of this kind: 'Non-zero exit status: 255'
*nod* 15:08
I think this is unrelated.
but related to certain Linux distros or something.
been a while since I spectested Rakudo-Parrot.
jnthn ICU?
masak oh, possibly.
how, on a Debianesque, do I check for ICU?
or just a simple perl6 command to run? 15:09
jnthn Well, run one of the tests, and it should fail with ICU lib not loaded or so
15:09 sqirrel left
pmurias how can I check if a %*SC_CACHE is set? 15:10
masak 'no ICU lib loaded'
ok, which package is it that I need to install again?
libicu-dev? 15:11
seems so.
15:11 yeahnoob joined 15:13 oetiker left, chenryn joined
masak lizmat: (re do I want us to take out Int.min) -- if it were only up to me, I'd propose it be taken out. what I really want is for #perl6 to arrive at an enlightened consensus about it. (by which I don't necessarily mean "my opinion", but "what's sane/good/useful") :) 15:16
carlin nine++ thanks
lizmat fwiw, I think we need to expand it to 'int8', 'int16', etc, to be really consistent 15:17
masak how come we mark up icu tests in spectest.data, and then they still fail under stresstest when icu is absent?
lizmat icu=1 not being passed ?
dalek kudo/nom: 592899a | masak++ | docs/announce/2014.09.md:
[docs/announce/2014.09.md] fill in changes highlights
 15:18
masak no idea what's being passed and what isn't :) -- just runnning `make stresstest` over here.
I'm asking at a more abstract level than that, I think -- how come it can happen in the first place? 15:19
lizmat perhaps 'make stresstest' doesn't fudge the spectest.data in the same way that 'make spectest' does? 15:22
e.g. is it trying any S17 tests on parrot with 'make stresstest' ? 15:23
15:23 orafu left, mr-foobar joined
masak a good question. also no idea, but I guess staring at the test results would reveal that. 15:23
the release guide is all `make stresstest`.
15:23 orafu joined 15:27 JimmyZ left
FROGGS[mobile] I spectest always 15:30
15:30 sdo joined
sdo hello 15:30
FROGGS[mobile] I know, I must be a bad person :/ 15:31
PerlJam sdo: greetings
FROGGS[mobile] hi sdo
sdo I was just wondering i went to this page www.perlfoundation.org/perl6/index...._the_onion and there are dead links. Isthat normal?
masak on the web? yes :/ 15:32
sdo hello PerlJam and FROGS
masak question is if we can heal the dead links.
maybe they're just "mostly dead".
sdo :)
ok so as far as I understand everything now is on Githbu isn't it? 15:33
s/Guthbu/Github/
masak fsvo "everything", yes ;)
we do try.
I'm not aware that SOTO talks have a place in a github repo somewhere... 15:34
sdo so now I will troll on Github to info
PerlJam notices the "promised land" quote by TimToady from 2000
masak I'd use various incantations of Internet Archive to find them.
gtodd lizmat: perl6 -e 'my @arr := "/usr/share/dict/web2".IO.lines(:eager); say @arr.pick # this is only 1.3 seconds ... ! with perl5 (using IO::All is might slow it down)
15:34 SamuraiJack joined
sdo thx masak 15:34
PerlJam That makes Perl 6 development clearer ... we've been wandering in the desert all these years :)
sdo everything is tip top on my side 15:35
gtodd lizmat: it's 0.45 seconds ... perl -MIO::All -E 'my @arr = io("/usr/share/dict/words")->slurp; say $arr[rand @arr];'
lizmat gtodd: I guess my machine is faster: 15:37
$ time perl6 -e 'my @a := "words".IO.slurp.lines(:eager); say @a.pick'
untestable
real0m0.483s
gtodd :-) or mine is busy 15:38
mines an oldish 4 core but with spinning disks zfs raid etc might slow IO a bit 15:39
lizmat SSD here :-)
15:39 kaleem left
dalek kudo/nom: d3b0bc7 | masak++ | docs/announce/2014.09.md:
[docs/announce/2014.09.md] contributors
 15:40
15:40 raiph left
gtodd lizmat: is one of the things you are working on somehow going to make p6 smart enough to use :eager without being told to ? 15:40
masak please to check for duplicates kthx thx ^__^
lizmat gtodd: that would be the GLR
gtodd right :-)
lizmat S99:GLR
synopsebot Link: perlcabal.org/syn/S99.html#GLR
masak checked the ones he knows, but is fallible
gtodd GLR =~ Xmas 15:41
;)
lizmat This page was generated by Text::SmartLinks v0.01 at 2014-08-27 17:01:08 GMT.
(syn 6989f5a) [ Index of Synopses ]
15:41 zakharyas left
lizmat whoah, that hasn't been updated in a while! 15:41
gtodd S99 needs to add GLR ? 15:42
lizmat is that feather breakage ?
it's already in there
just not at the page we're pointing to :-(
gtodd yeah I saw it somewhere already
anyway I guess GLR is more like halloween :-) Xmas is Xmas 15:43
15:43 virtualsue left
lizmat :-) 15:43
dalek kudo/nom: 052a511 | masak++ | docs/release_guide.pod:
[docs/release_guide.pod] update with #80 release name
timotimo gtodd: you don't get to =~ in here :) 15:44
gtodd oopsie
I'm approximately smarter than that
15:44 bbkr joined 15:45 kaleem joined
masak std: 2 =~ 2 15:45
camelia std fc92e1e: OUTPUT«===SORRY!===␤Unsupported use of =~ to do pattern matching; in Perl 6 please use ~~ at /tmp/LgyEzyN77u line 1:␤------> 2 =~⏏ 2␤Parse failed␤FAILED 00:01 139m␤»
masak m: 2 =~ 2
camelia rakudo-moar 897076: OUTPUT«===SORRY!=== Error while compiling /tmp/yXio7wZLVd␤Unsupported use of =~ to do pattern matching; in Perl 6 please use ~~␤at /tmp/yXio7wZLVd:1␤------> 2 =~⏏ 2␤»
15:46 rindolf joined
pmurias nqp: say(nqp::ishash(try %*FOO)) 15:47
timotimo are the optimizations to succeed and proceed in this month's release, or were they in last month already?
camelia nqp-jvm: OUTPUT«(signal )»
..nqp-{moarvm,parrot}: OUTPUT«1␤»
gtodd hmm does something like ... time perl6 -e 'my @arr := "/usr/share/dict/web2".IO.lines(:eager); say @arr.pick;' # compared with equivalent in ruby node python2/3 perl5 give useful marketing information ?
bbkr What does "non-invokable object is non-invokable" mean? I get it randomly when running following simple code: my $job1 = start { say "job 1"}; my $job2 = start { say "job 2"}; await Promise.allof($job1, $job2) 15:48
PerlJam gtodd: I dunno. Seems like the bulk of the time would be spent in IO, so that doesn't really say much about the language.
timotimo also, did we mean to point out the existence of Proc::Async? or should that wait until we have proper re-ordering of stdout and stderr?
lizmat timotimo: ordering should be okish now
15:49 fhelmberger left
timotimo oh? well, that's great! :) 15:49
lizmat there's only the matter of whether it is possible to have a "done" appear after a "done"
timotimo okish as in "much better than before, but can still occasionally break"?
lizmat "more" appear after a "done"
so I'm not sure it will always get all lines
dalek p: bba7e83 | masak++ | tools/build/MOAR_REVISION:
bump MOAR_REVISION

To latest release, in preparation of Rakudo release.		 15:50
lizmat I could see a race where a "done" was done, and after that a "more' WOULD APPEAR
oops capslocky
gtodd PerlJam: yeah
15:51 Akagi201_ joined
dalek p: 03d588a | masak++ | VERSION:
bump VERSION to 2014.09
 15:51
lizmat timotimo: changelog has now: + added experimental support for Proc::Async, MoarVM only for now
timotimo nice :)
did dalek forget about that comment, or did you just not push it yet?
lizmat it's in the Changelog, afaics 15:52
brb
15:54 Akagi201 left, sdo left, virtualsue joined
timotimo oh, it was there already 15:54
i just didn't look closely enough. sorry about that!
15:54 kaleem left
masak got a fail in t/spec/S17-lowlevel/lock.rakudo.moar during stresstest. 15:55
PerlJam carlin: you might want to add your name to the CREDITS file.
15:55 telex left
PerlJam leont: you too 15:55
masak Failed test: 7
pmurias jnthn: any idea what can be causing nqp::serialize to return null for the data part?
masak can anyone confirm?
(reports of passes are fine, too)
timotimo masak: that may be one of our flapping tests :(
masak there's still time to fudge it ;)
pmurias jnthn: "data part" - the thing it returns 15:56
masak and to make [Coke] extra happy, feel free to create an RT ticket or so.
timotimo i can't do it right now, i'm going AFK for a bit
TimToady I would propose that numeric types have a .range method, or a Range coercer, from which you could subsequently retrieve the bounds
15:56 telex joined
TimToady int16.range.max is much clearer than int16.max 15:57
masak +1
lizmat .Range makes more sense to me, then .range 15:58
TimToady we can argue about whether it should be -32768 .. 32767 or -32768 ..^ 32768
masak probably the former.
then .max comes out right, no?
lizmat we're coercing a type to its limits
TimToady unless you're thinking pythonically :)
masak no, I'm thinking .max is the maximum :) 15:59
15:59 raiph joined, bbkr left
masak not the first unattainable value 15:59
TimToady I incline toward that interpretation as well
PerlJam and what lizmat said about .Range ? 16:02
pmurias jnthn: ahh, by reading MoarVM it turns out that it's stashed somewhere and a null is returned
masak either .range or .Range makes sense to me. if we're bikeshedding that, we've already won because the corner case for .min is already gone ;) 16:03
lizmat m: 42.Range.perl.say # then maybe this should also work
camelia rakudo-moar 897076: OUTPUT«No such method 'Range' for invocant of type 'Int'␤ in block <unit> at /tmp/Uh1lj0Mry2:1␤␤»
lizmat would be a Range.new(42..42)
masak that... makes less sense to me. 16:04
PerlJam yeah, me too
TimToady can imagine use cases involving mixed integer/range inputs
masak or rather, it feels like a hobgoblin consistency rather than motivated by an actual use case.
TimToady but [min] already has those semantics 16:05
lizmat m: 1.Set.perl.say
camelia rakudo-moar 897076: OUTPUT«set(1)␤»
lizmat similar case ^^^
TimToady m: my $r = 41..43; say [min] 42, $r 16:06
camelia rakudo-moar 897076: OUTPUT«41..43␤»
jnthn pmurias: Yes, see the t/serialize tests for, iirc, how to get things to work out right
masak m: say set(1).Set.perl
camelia rakudo-moar 897076: OUTPUT«set(1)␤»
jnthn pmurias: If you want the data, that is
TimToady er...
m: my $r = 41..43; say [minmax] 42, $r
camelia rakudo-moar 897076: OUTPUT«Range objects are not valid endpoints for Ranges␤ in method new at src/gen/m-CORE.setting:7777␤ in method minmax at src/gen/m-CORE.setting:1915␤ in sub infix:<minmax> at src/gen/m-CORE.setting:2034␤ in sub at src/gen/m-CORE.setting:18424␤ i…»
lizmat masak: coercing a set to a set, returns self
m: "42".Str.perl.say 16:07
camelia rakudo-moar 897076: OUTPUT«"42"␤»
masak lizmat: which things get wrapped in a Set, and which don't? what's a simple rule?
lizmat anything that's not a set 16:08
masak m: say (1, 2, 3).Set.perl
camelia rakudo-moar 897076: OUTPUT«set(1,2,3)␤»
masak nope.
that one coerces, and removes the Parcel.
lizmat m: bag(1,2,3).Set.perl.say
camelia rakudo-moar 897076: OUTPUT«set(1,2,3)␤»
lizmat hmmm....
masak so some things get wrapped, others get coerced.
lizmat anything non-QuantHash gets wrapped 16:09
TimToady m: my $r = 41..43; say 42 min $r
camelia rakudo-moar 897076: OUTPUT«41..43␤»
TimToady m: my $r = 41..43; say 42 minmax $r
camelia rakudo-moar 897076: OUTPUT«Range objects are not valid endpoints for Ranges␤ in method new at src/gen/m-CORE.setting:7777␤ in method minmax at src/gen/m-CORE.setting:1915␤ in sub infix:<minmax> at src/gen/m-CORE.setting:2034␤ in block <unit> at /tmp/WxwHTRxkiJ:1␤␤»
lizmat m: set(set(1,2,3)).perl.say 16:10
camelia rakudo-moar 897076: OUTPUT«set(set(1,2,3))␤»
masak lizmat: based on what I see about .Set, I would stay away from it, since it seems that it behaves differently depending on what it's called on.
16:10 Ven joined
jnthn lizmat: more should not appear after done; it needs to be part of the SequencingSupply, or whatever yo called the object that is handling the ordering enforcement. 16:11
lizmat I'm only following spec, afaik
masak *nod*
lizmat jnthn: so "done" is not cued?
[Coke] github.com/coke/perl6-roast-data/b....out#L3536 - yesterday's known rakudo-parrot failures.
TimToady .Set is a coercion, not a composer
masak I'm suffering quite a bit here from lack of good connectivity, by the way. it's hampering the release process quite a bit.
TimToady set() is a composer, not a coercion
masak had I thought about it more, I might have done it on feather from the start. 16:12
masak tries that now
jnthn lizmat: more* [done | quit] is the grammar for supplies
[Coke] github.com/coke/perl6-roast-data/b..._rates.csv (search this file: parrot); failing at least something since 8/27
masak TimToady: right, but .Set seems to be sometimes compose, sometimes coerce
jnthn lizmat: Anything that violates that is erroneous.
PerlJam
.oO( and then we all disappear as feather crashes ... 0
TimToady Not I, said the pig.
lizmat ok, then sequencing Supply.lines should now be okidoki
jnthn lizmat: And while we can't prevent folks who make a Supply directly violating the contract, we should try and make sure our own sources don't. :) 16:13
carlin PerlJam: github.com/rakudo/rakudo/pull/310
masak please refrain from merging pull requests during releng ;)
jnthn masak: Branch! :P 16:14
PerlJam carlin: if you also show up in commits as "carlin" you'll want to add a U: line for that too.
16:14 yeahnoob left
dalek kudo/nom: b2a10d7 | Carlin++ | CREDITS:
add self to CREDITS
 16:14
kudo/nom: 594b175 | lizmat++ | CREDITS:
Merge pull request #310 from carbin/egotism

add self to CREDITS file
lizmat oops, sorry, saw it too late
masak :S
16:14 chenryn left
lizmat I figured because carlin was specifically *asked* to update it 16:14
I thought it would need to go into this release :-( 16:15
masak oh, it's CREDITS.
lizmat yes
masak never mind.
lizmat++
carlin++
wow, everything internetty is so fast on feather! <3 16:16
masak is behind some kind of... Wall
16:16 chenryn joined
TimToady watch it! 16:16
dalek kudo-star-daily: 00ab731 | coke++ | log/ (14 files):
today (automated commit)
 16:17
TimToady is glad masak++ is behind him, actually...
PerlJam TimToady: just in case you fall? :)
Ven hehe, there are quite a few people in that CREDITS file :)
masak well, it's a Great Wall, so...
TimToady just don't yell Fire! 16:18
jnthn Though Long Wall is probably a better translation...which is appropriate, 'cus all traffic going through it takes a darn long time...
PerlJam for some reason that makes me want to alliterate: Wonderful Western Wall (maybe?)
masak jnthn: maybe all the bits get stuck in the Gansu corridor... 16:19
TimToady why is everyone wailing now?
jnthn
.oO( The great firewall Mings... )
16:19 anocelot joined
PerlJam (the alliterative is for Larry obviously) 16:19
[Coke] rakudo.jvm run today is taking much longer, it seems, though it hasn't died yet.
good news: the parrot fudging that happened happened early enough that rakudo.parrot seems to have had a clear run today! 16:20
masak also, it's rather nice to watch a build happen without the fan spinning up :P
[Coke]: \o/ 16:21
[Coke] I will be very happy to push this out. Thanks everyone!
masak +1
heh, feather1 doesn't have JVM either, it seems. :P 16:22
at least not in PATH.
jnthn masak: Good luck wiht release! 16:24
'night, #pelr6
masak thanks.
lizmat good night, jnthn
masak 'night, jnthn
TimToady I could give you an account on wall.org if that'd help 16:25
it's got jvm 16:26
though it's stuck on an older Ubuntu 16:27
masak I'm making do without for now.
TimToady and the pipes aren't quite so fat
masak it's late here, and I just want to generate tarballs and tag stuff.
(but thanks) 16:28
t/serialization/01-basic.t in nqp-parrot failed.
is this known? do I need to care? 16:29
Failed 12/1355 subtests
lizmat looks 16:30
masak lizmat++ 16:31
lizmat is that the name of the file ?
masak copy-pasted, yes.
encountered by running 'make test' on a fresh nqp with parrot configured.
lizmat aahhhh nqp 16:32
masak nqp.
I will time out to "don't worry" in a really rather small time interval.
and continue to push the tag and serve the .tar.gz for upload 16:33
TimToady if rakudo on parrot is okay, I wouldn't sweat it
16:33 camelia left
masak by the way, anyone here able to upload to rakudo.org? I never seem to have my ssh keys handy. 16:33
FROGGS[mobile] I am
masak excellent.
I'll privmsg you a temp link. 16:34
lizmat is still building a fresh nqp-parrot
FROGGS[mobile] wait for me being on my laptop then...
masak FROGGS[mobile]: masak.org/carl/tmp/nqp-2014.09.tar.gz 16:36
16:37 FROGGS joined
masak FROGGS: ^ 16:37
FROGGS yes yes :o)
lizmat masak: All tests successful. 16:38
Files=100, Tests=3551, 32 wallclock secs ( 0.50 usr 0.19 sys + 26.71 cusr 4.55 csys = 31.95 CPU)
masak 'xcellent 16:39
lizmat++
dalek kudo/nom: 7651c85 | masak++ | tools/build/NQP_REVISION:
[release] bump NQP revision
kudo/nom: ea70a10 | masak++ | VERSION:
[release] bump VERSION
FROGGS masak: done
masak FROGGS++ 16:40
FROGGS: will you be around in about 15..20 minutes for the Rakudo tarball?
FROGGS masak: yes
:o)
masak \o/
it's a date.
FROGGS I am around for about five hours from now on 16:41
TimToady well, a time, anyway
FROGGS 'There was a time' - quote from one of my favourite movies
masak which movie, ooc?
[Coke] lucky number slevin? 16:42
FROGGS Lucky Number Slevin
[Coke] woot.
FROGGS *g*
Bruce Willis - the best german actor :P 16:43
[Coke] ISTR really likely that movie, but wonder if would hold up on rewatch.
realyl *liking
timotimo oooh 16:44
i recently watched that movie for the first time
it blew me away
FROGGS I watched it at least five times, and I'd do it again :o) 16:45
timotimo watch it seven times, then stop.
masak marks that movie up as worth watching 16:46
[Coke] installs the latest Xcode, and wonders if he's going to regret this. 16:48
masak FROGGS: masak.org/carl/tmp/rakudo-2014.09.tar.gz 16:49
carlin "This is the last film to feature Willis with hair"
[Coke] *snrk* 16:50
FROGGS masak: done
[Coke] reminds me I need to add looper to my list. 16:51
masak FROGGS++
16:51 Ven left
PerlJam carlin: he had hair in Surrogates. 16:52
16:52 virtualsue left, pmurias left, dakkar left 16:53 Ven joined
masak p6c email sent. 16:54
carlin darn, wikipedia has mislead me
16:54 leont left 16:56 chenryn left
masak en.wikipedia.org/wiki/Rakudo_Perl_6 updated. 16:57
...this concludes your monthly Rakudo release cycle. 16:58
thank you for being a [willing|reluctant|enthralled] participant.
your feedback is important to us. leave it in the backlog, and a feedbackovore will pick it up and process it. 16:59
the captain has switched off the "fasten seatbelt" sign. 17:00
masak celebrates with the appropriate amount of pillow 17:01
'night, #perl6
17:01 slmult0_z joined 17:02 zakharyas joined
timotimo is pleased to see, that the Debugger still works 17:03
nwc10 masak++
alexghacker m: my $target = "gdfedbcad"; say $target.subst( rx/ d /, 'q', :g ) 17:04
[Coke] can't ssh to host08 17:05
error: Socket read error: [54] Connection reset by peer 17:06
alexghacker well, that isn't at all what I was trying to demonstrate
[Coke] host08 is probably getting hammered by the daily jvm tests atm.
alexghacker ah, poor timing on my part then
[Coke] (so camelia is affected as well)
alexghacker I'll check back later. That's only the first of several in a sequence I was planning on doing. 17:07
timotimo alexghacker: at least locally that just outputs "gqfeqbcaq" 17:13
are you working towards the "in the second argument to subst, $/ doesn't work properly" problem?
FROGGS gnight masak++
gtodd PerlJam: goofed around a bit and as you pointed out, since there's IO in the mix, these are meaningless paste.scsys.co.uk/424273 ... and in any case ruby/node/python oneliners aren't really equivalent ( is p6 .pick the same as ruby's .sample? ; some of them set up data structures differently etc. etc.) ... 17:14
(well node.js javascript doesn't really do "oneliners" ..) 17:15
vendethiel wat?
alexghacker timotimo: maybe, I'm not familiar with that named problem (having just started perl6 Monday)
timotimo: actually, in the example I was building up to, a manually constructed Match object was going to 'do the right thing' despite having a different orig than the target 17:16
perl6-m -e 'my $s = "gdfedbcad"; my $s2 = "abcdefg"; $s2 ~~ m/ d+ /; my $m = Match.new( orig => "abcdefg", from => 3, to => 4, ast => Any, list => ().list, hash => EnumMap.new()); say $/.perl; say $m.perl; say $s.subst( $m, "q", :g );'
timotimo .. you can pass a match object as the first argument to subst?! 17:17
gtodd PerlJam: but there must be something interesting going on with ruby's .sample to make it so fast ... 17:18
17:18 grondilu joined
alexghacker timotimo: it seems that way 17:18
grondilu Hello. I've noticed something funny. I was trying time perl6 -e ';' to check out start-up time, and then for some reason I also tried time perl6 -e 'say "hi";' Surprisingly enough, the former was actually slower than the latter. 17:20
alexghacker I'm trying to figure out what devilry^Wwizardry is going on to make that work
grondilu tries again and can't reproduce it. Meh, nevermind. 17:21
timotimo grondilu: warm hard drive cache or something? 17:22
grondilu possible
PerlJam alexghacker: Output your match object as a string. 17:25
alexghacker ah, demote the Match to a Str, then use to construct a Regex? 17:26
PerlJam ~$m is just "d" so, $s.subst( 17:27
d
oops
17:28 [Sno] left
PerlJam ~$m is just "d" so, $s.subst("d", "q", :g); is what you're getting 17:28
As to what's doing that, I'd guess Cool probably
timotimo ah, hehe. 17:29
PerlJam oh, Str's subst() doesn't even type the arg.
17:31 FROGGS[mobile] left
TimToady so how long do we wait for host08 to go non-catatonic? 17:39
and how do we stop waiting? :)
timotimo m: say "how about now?"
TimToady camelia disappeared quite some time ago 17:40
17:40 Ven left
timotimo oh 17:40
weechat probably filtered that for me
bbl.
TimToady and we can't ssh in
lizmat fg
oops
TimToady doesn't give a fg
17:41 brrt joined
PerlJam brrt: greetings! 17:42
brrt \o PerlJam
colomon o/ 17:43
PerlJam brrt: you may want to add yourself to rakudo's CREDITS file (it only really matters for attribution on release)
brrt uh. i could do that. but i've only hacked MoarVM? and added one patch to nqp, iirc
PerlJam brrt: perhaps someone said brrt++ in a rakudo commit then. :) 17:44
brrt oh that could be true
oh, wait, i recall
there was a bug with null pointers in an extop :-)
lizmat fixed it with my patch 17:45
FROGGS see :o)
colomon oh, it's release day! 17:46
PerlJam brrt: in any case, if you want something other than "brrt" to show up in the list of contributors, add yourself to the CREDITS file.
brrt very well
17:46 guru joined, guru is now known as ajr_
brrt should note that Вячеслав Матюхин isn't sorted right :-) 17:47
or wait
is the cyrillic B a B?
17:47 Ven joined
brrt should i add brrt as my username or bdw? 17:49
(bdw is my github username)
PerlJam both if you like.
(with multiple U: lines)
gtodd there's no comparison (yet) between: ruby -e 'puts File.readlines("/usr/share/dict/web2").grep(/\bzygote\b/)' @0.02 seconds and perl6 -e 'say "/usr/share/dict/web2".IO.slurp.lines(:eager).grep({/<<zygote>>/})' @96.52s seconds .... (grep "zygote" /usr/share/dict/words from the shell is the fastest hehe) ... but why would using feeds in perl6 be faster ? GLR?
alexghacker B is [CYRILLIC CAPITAL LETTER VE] while B is [LATIN CAPITAL LETTER B]
brrt ah ok
diakopter m: 1 17:50
gtodd perl6 -e 'my @arr := "/usr/share/dict/web2".IO.slurp.lines(:eager); @arr ==> grep({m/<<zygote>>/}) ==> my $stuff; say $stuff; is ~three times faster (30 seconds)
alexghacker no idea how they should collate, though. I haven't read enough of the Unicode specification yet
diakopter who kilt host08 :) 17:51
brrt yeah, that's what i thought. the cyrillic b looks like (hang on)
PerlJam brrt: perl6 thinks it's sorted wrongly too :)
brrt Б
very nitpick :-)
alexghacker much codepoint
brrt ehm, you want me to do a pull request?
PerlJam that'll work. 17:52
brrt ok
tony-o timotimo: added quite a few more tests, fixed some bugs to http async server 17:53
lizmat which reminds me, I need to look at IO::Async as well 17:54
brrt github.com/rakudo/rakudo/pull/311
lizmat PerlJam: will you do the honors this time, or shall I ? 17:55
diakopter host08 death: i.imgur.com/rV2UtaF.png
lizmat
.oO( now understands why Oracle makes it so difficult to download java )
 17:56
tony-o why?
Ven oh, is everybody adding themselves to credit :p?
lizmat tony-o: look closely at the image diakopter posted 17:57
dalek kudo/nom: e2017c3 | TimToady++ | t/harness:
remove LOLLY hack
ast: d329a22 | TimToady++ | S02-types/multi_dimensional_array.t:
remove LOLLY hack
ast: 57fa591 | TimToady++ | / (19 files):
Merge branch 'master' of git://github.com/perl6/roast
FROGGS jvm killed my laptop today too
brrt somebody has to take the credit :-)
tony-o lol
dalek kudo/nom: a028192 | (Bart Wiegmans)++ | CREDITS:
Add self to CREDITS
kudo/nom: 376950d | duff++ | CREDITS:
Merge pull request #311 from bdw/nom

Add self to CREDITS
diakopter each process wanted 2.4GB memory 17:58
Ven that happens
PerlJam wanders off to a meeting
Ven JAVA ))))
diakopter wait, 5.8GB
ha
also it's suggesting sacrificing children
TimToady note that you have to multiply TEST_JOBS by about 3 to get the Java effect 17:59
lizmat gtodd: $ time perl6 -e 'my @arr := "words".IO.slurp.lines(:eager); @arr ==> grep("zygote") ==> my $stuff; say $stuff'
zygote
real0m0.805s
grep is smarter than you think :-)
TimToady so TEST_JOBS=3 make j-spectest is equivalent to about TEST_JOBS=9 make m-spectest
gtodd lizmat: ok good :-) 18:00
18:01 camelia joined
lizmat diakopter++ 18:01
tony-o is anyone using travis-ci for module build testing? 18:02
18:02 ChanServ sets mode: +v camelia
brrt java gc is very wow 18:03
complexity causes efficiency, you know 18:04
diakopter m: 1 18:06
gtodd lizmat:I was curious why using feeds ("==>") is so much faster than -> perl6 -e 'say "/usr/share/dict/web2".IO.slurp.lines(:eager).grep("zygote");' (62 seconds here so many orders of magnitude) ... are feeds avoiding getting bogged down in the laziness in lists somehow?
camelia ( no output )
diakopter anyone: email me instead of irc if it needs kicked again; it's much faster
lizmat gtodd: I would guess that as a yes
gtodd lizmat: :-) I like it when my guesses are halfway intelligent yay! 18:07
cheers
lizmat doesn't know that much of the innards of feeds, but assumes they're gather / take implemented
and that's exactly the difference between the :eager and non eager version of .lines 18:08
gtodd nice
very ... consistent
dalek kudo/nom: 9850adc | (Elizabeth Mattijsen)++ | src/core/IO/Handle.pm:
Implement IO::Handle.lines(:count)
 18:09
gtodd ... which to me (non engineer) suggests something well designed is going on inside p6
dalek nqp: 5ed99d2 | usev6++ | 3rdparty/dyncall/configure: 18:10
nqp: add CFLAGS=-fPIC to compile dyncall for backend gen-parrot on FreeBSD
nqp: In order to install perl6-p on FreeBSD 10.0 I had to compile dyncall with CFLAGS=-fPIC.
nqp: I think this was the same problem ChoHag++ reported earlier on #perl6:
nqp: * irclog.perlgeek.de/perl6/2014-06-10#i_8847415
18:10 dalek left
lizmat we do try to do as much as possible with the fewest components, yes :-) 18:10
18:10 dalek joined, ChanServ sets mode: +v dalek
gtodd usev6++ errm 18:11
usev6++++
lizmat bartolin++ on IRC actually, I think
18:11 Akagi201_ left
gtodd this will fix nqp-p build on FreeBSD releases 18:11
PerlJam uesv6 "needs" a CREDITS entry too (still in meeting :) 18:12
er, usev6
18:12 Akagi201 joined, bartolin joined
gtodd bartolin++ 18:12
he
bartolin Hi #perl6
FROGGS hi bartolin 18:13
bartolin Thanks for merging FROGGS++
FROGGS bartolin: pleasure :o)
18:13 jack_rabbit joined
gtodd bartolin: FreeBSD nqp-p yay 18:13
vendethiel bartolin++ # `use v6`-ing; ;-). 18:14
bartolin I managed to install perl6-p on FreeBSD. But there is a second tweak with parrot, which I have not coded properly. 18:15
(yet)
dalek kudo/nom: c6c539f | (Elizabeth Mattijsen)++ | src/core/Str.pm:
Implement Str.lines(:count)
18:17 raiph left
gtodd ooh lines(:count) .... 18:17
18:17 Akagi201 left
gtodd bartolin: I will build on some 9.2 and 9.3 boxes and vms this evening 18:20
18:21 raiph joined
gtodd lizmat: is lines(:count) for users or for the GLR ... I'm trying to grok the commit a little ... 18:22
lizmat it's just an unspecced feature I'm throwing in at this moment
gtodd ok so one coudld pass an Int like lines(:5000) or something 18:23
bartolin gtodd: Great. I would like to know about your results (9.2 and 9.3)
lizmat ah, no, that's the "limit" parameter
and that is a positional
gtodd ah ok ... so it helps the lines/list be less lazy? 18:24
lizmat :count just gives you the number of lines
gtodd or could
lizmat no, something else altogether, and quite possibly to be reverted
gtodd :-D
git is great that way
TimToady maybe should be a .linecount method
lizmat yup, and we now have a month to decide whether it will be in the next release 18:25
and also a .wordcount then, TimToady ?
TimToady well, maybe, or perhaps we need a 'count' context to go with eager, lazy, etc. 18:26
count would almost be a sink
lizmat but an numeric sink instead of a Nil sink 18:27
TimToady right
one more thing tossed into the GLR hopper :)
Ven ingy: hahaha just realized your regexp couldnt fail :( 18:28
dalek kudo/nom: c15fc49 | (Elizabeth Mattijsen)++ | src/core/Supply.pm:
Implement Supply.lines
gtodd I was wildly guessing that there would be something we little users could manually with :args to make p6 lists less lazy before the GLR ... but maybe that *is* the GLR ?
TimToady well, an explicit eager helps in some cases already 18:29
lizmat gtodd: no, the GLR should change evert gather / take into an eager push construct autmagically when it can
gtodd :-)
lizmat TimToady: it does? in which cases ? 18:30
TimToady it helps a little with gather/take even
though not as much as the push rewrite
18:31 dwarring joined
gtodd count as a "context" for eager lazy is mentally/linguistically pleasing .. tells p6 just how lazy or how eager to be 18:31
FROGGS jnthn: ummm, there is a thinko in my cglobal code :/ 18:32
dalek kudo/nom: 51986e8 | (Elizabeth Mattijsen)++ | t/spectest.data:
Add Supply.lines tests
gtodd lizmat++ automagic++ ;-) 18:33
FROGGS jnthn: which is a design issue in nativecall me thinks... 18:34
arnsholt: still there?
18:40 PZt left
FROGGS jnthn / arnsholt: nvm 18:41
gtodd bartolin: the parrot build broke somewhere else for me :-) pbc_to_exe.c:1300:22: error: use of undeclared identifier 'CTL_KERN'
bartolin gtodd: I am on that right now.
nine Just to let you know: I'll be afk until Sunday evening. If some issues with Inline::Perl5 come up please post them to github. 18:42
gtodd bartolin: oh ... nice :-) 18:43
lizmat nine: have a nice long weekend!
FROGGS nine: we'll do :o)
nine lizmat: thanks!
bartolin gtodd: You have to add the following to parrot/tools/dev/pbc_to_exe.winxed:
gtodd: ##ifdef __FreeBSD__
# include <sys/sysctl.h>
#endif
gtodd nine: loving Inline::Perl5 :-) no issues :)
FROGGS ohh!
gtodd bartolin: exellent
FROGGS bartolin: that's my fault then
bartolin gtodd: Does that work for you? 18:44
FROGGS: Your fault?
FROGGS bartolin: I added the code to obtain the executable name/path 18:45
just in parrot 6.8.0
bartolin FROGGS: I had the same error yesterday (before parrot 6.8.0).
FROGGS github.com/parrot/parrot/commit/3c...e46a702828 18:47
gtodd bartloin build is continuing
FROGGS that included that patch as well
18:47 woshty left, orafu left
bartolin gtodd: :-) 18:47
dalek kudo/nom: 696a24b | (Elizabeth Mattijsen)++ | src/core/io_operators.pm:
Make sure lines() passes on named params
 18:48
bartolin FROGGS: Ahh. Now I see.
gtodd bartolin: build is continuing with that fix ... did "gmake" inside parrot dir ... and it finished
bartolin FROGGS: Could you add the #include <sys/sysctl.h> as above?
dalek ast: baae261 | (Elizabeth Mattijsen)++ | S32-str/lines.t:
Add tests for Str.lines(:count)
 18:49
gtodd waist for 20 minutes to do a new git pull ... lately when I have lunch perl6 gets faster while I eat :-)
FROGGS bartolin: I also just can pull request it... 18:50
18:50 zakharyas left, wollmers joined
bartolin FROGGS: Ah. Then I will do it. (Getting familiar with p6, parrot, git, freebsd ...) 18:51
FROGGS bartolin++ # :o)
18:55 rindolf left 19:00 darutoko left 19:01 woshty joined 19:02 Ven left 19:06 ilogger2 joined, ChanServ sets mode: +v ilogger2 19:07 slmult0_z_q joined
bartolin FROGGS: I made a pull request: github.com/parrot/parrot/pull/1091 19:11
FROGGS I forwarded it to #parrot :o) 19:12
19:13 slmult0_z_q left
bartolin FROGGS: Ah. Thanks! :-) 19:13
19:18 Ven joined
dalek p: 4b7e77a | (Elizabeth Mattijsen)++ | docs/ops.markdown:
Add documentation stub for nqp::killprocasync
 19:18
bartolin rurban++ FROGSS++ Well, this is amazing (patch committed to parrot)
FROGGS bartolin: welcome to open source *g* 19:19
19:19 spider-mario joined
Ven that was quick :) 19:19
19:19 dalek joined
dalek ecs: f603f96 | (Helmut Wollmersdorfer)++ | S15-unicode.pod:
fix typo
 19:19
19:19 ChanServ sets mode: +v dalek 19:20 slmult0_z_q joined, daxim__ joined
dalek volaj/cglobal: 2cc378d | (Tobias Leich)++ | t/09-nativecast.t:
remove test about old pointer->pointer cast
 19:20
volaj/cglobal: 7e31ed4 | (Tobias Leich)++ | /:
Merge branch 'master' of github.com/jnthn/zavolaj into cglobal
volaj/cglobal: ce12b91 | (Tobias Leich)++ | lib/NativeCall.pm6:
we cannot keep the ptr without leeking the libhandle
19:21 zakharyas joined
rurban bartolin: can you also check openbsd and netbsd? my smokers there are offline for a few weeks 19:21
bartolin rurban: I'd like to do that, but it will certainly take some time (days). 19:22
rurban theoretically it looks good, but we haven't tested it there. thanks
[Coke] (host08) p6eval is currently pretty busy doing jvm stuff. :P 19:23
rurban and note that I changed your commit a bit. just the subject
19:24 [Sno] joined
bartolin rurban: I've seen it. Thanks for making it more clear ;-) 19:24
[Coke] (today's daily run was killed, due to jvm.) 19:25
was camelia running on host07 previously 19:27
? 19:28
Ven ingy: github.com/ingydotnet/pig-latin-ac...tin.rb#L17 shouldn't that downcase m[1]?
ingy: got racket working :) 19:32
19:33 Guest81141 joined, infiniteloop joined, Guest81141 left 19:34 ezra1 joined 19:42 ezra1 left, Akagi201 joined 19:45 Psyche^_ joined
TimToady Ven: so, next project is a racket backend for rakudo? ;) 19:46
Ven TimToady: not sure what that means :P 19:47
19:47 mberends joined
Ven racket doesn't have the best binary/jit out there 19:47
TimToady that's why we call it "research"?
19:47 Akagi201 left
TimToady and it's the darling of a lot of CS professors 19:48
timotimo who funds it?
Ven yeah, racket is used a lot in unis
I mean, even naughty dog uses it :P
(yeah, the game-makers) 19:49
TimToady: a racket slang would be much better! 19:51
actually, all racket slangs. Racket isn't just ONE language.
19:52 raiph joined 19:53 mberends left
FROGGS huh, the cglobal stuff even works on windows... right away O.o 19:54
19:59 jaffa4 joined
jaffa4 hi all 19:59
how to fix this: sub AppendClipboard_3_c(Str $format,Value $data,(&cb(Value $data2)--> Str) render) is symbol('AppendClipboard_3_c') is native($dllname) { * }
dalek volaj/cglobal: ec7dcf2 | (Tobias Leich)++ | lib/NativeCall.pm6:
properly die in cglobal.STORE
 20:00
FROGGS jaffa4: have you tried '&render' ? 20:02
Ven .tell ingy github.com/ingydotnet/pig-latin-acme/pull/3
yoleaux Ven: I'll pass your message to ingy.
jaffa4 FROGGS: that did not work 20:04
20:05 xragnar_ joined, xragnar_ is now known as xragnar
FROGGS jaffa4: why do you give it names anyway? why not just strip 'render'? 20:08
20:08 brrt joined
jaffa4 it has informational value 20:09
FROGGS jaffa4: the replace &cb with &render
then*
dalek ast: ea2cd25 | (Elizabeth Mattijsen)++ | S17-procasync/kill.t:
Simplify the Proc::Async.kill tests

Unfortunately, the actual kill can not be done here, as it causes the current process to lose the remote process, so the P::A promise is never kept or broken. This even affects it being a part of an anyof with a timeout, so we cannot make the test continue. :-( Hope this will help jnthn find what the heck is going on.		 20:11
jaffa4 ok
FROGGS jaffa4: and I guess you also have to strip the return type of the callback
but I'm not so sure about that
[Coke] who has access to camelia's runtime?
j: say 3 20:12
camelia rakudo-jvm 696a24: OUTPUT«Can't call method "syswrite" on an undefined value at /home/p6eval/jvm-rakudo/eval-client.pl line 32.␤»
dalek kudo/nom: c2ca930 | (Elizabeth Mattijsen)++ | src/core/Kernel.pm:
Make sure we return Int's with .signal()
kudo/nom: 3176369 | (Elizabeth Mattijsen)++ | src/core/Proc/Async.pm:
Remove premature optimization

  $*KERNEL.signal has all the mapping we need
kudo/nom: 7c9a978 | (Elizabeth Mattijsen)++ | t/spectest.data:
Add Proc::Async.kill tests back again

They shouldn't hang anymore
[Coke] ^^ for something that is chewing 4 cpus, that's a pretty unsatisfactory answer.
whoops, 3. 20:13
m: say .WHAT 20:15
camelia rakudo-moar 696a24: OUTPUT«Nil␤»
diakopter [Coke]: yeah
[Coke] diakopter: I'm looking on host08 ~p6eval; looks like the evalbot.log isn't registering anything new. 20:16
diakopter oh hm
20:17 camelia joined
[Coke] ah, we've got log/*.log 20:17
diakopter m: ;
well I put it to ~/evalbot/eval.log
20:18 ChanServ sets mode: +v camelia
camelia ( no output ) 20:18
diakopter m: say &say;
camelia rakudo-moar 696a24: OUTPUT«sub say (Any |) { #`(Sub|62277840) ... }␤»
diakopter cool
jaffa4 FROGGS: experience shows it is not necessary 20:19
[Coke] evalbot/eval.log ? empty 20:20
20:21 camelia left, slmult0_z_q left, camelia joined
diakopter m: say &; 20:22
20:22 ChanServ sets mode: +v camelia
camelia rakudo-moar 696a24: OUTPUT«(Callable)␤» 20:22
[Coke] rakudo jvm seems to be building, anyway. 20:23
but the build, even though it's using 'make install', not 'make -j', it's still using most of the box. if I launch a rakudo-jvm spectest at the same time, I suspect that'd kill us both.
dalek p: 1f0d522 | (Tobias Leich)++ | / (2 files):
map op nativecallglobal and bump moar rev
 20:25
kudo/nom: 314e122 | (Tobias Leich)++ | tools/build/NQP_REVISION:
bump nqp rev for moar with C global support
 20:26
brrt .tell timotimo i wanted to ask him about what he'd think would be the next step for the JIT. i want to refactor the params functions but this needs some work as i'll need to have a detailed map of coercions done by these
yoleaux brrt: I'll pass your message to timotimo.
brrt .hugbot 20:27
.botsnack
yoleaux :D
20:27 brrt left 20:28 infiniteloop left 20:29 Ven left
dalek volaj: 4cc56c3 | (Tobias Leich)++ | lib/NativeCall.pm6:
expose sub to fetch C globals

This will partial resolv issue #11.		 20:30
gtodd rurban: bartolin: with the FreeBSd changes from github.com/parrot/parrot/commit/4d473824e2 and the nqp change to use -fPIC in more places github.com/perl6/nqp/commit/5ed99d2 I stil lcan't build parrot on FreeBDS 9.2 ... I get the same error about libdyncall_s and -fPIC 20:31
output here paste.scsys.co.uk/422888 ...
FROGGS gtodd: and you are sure that patch is still applied and that no old files are flying around? 20:32
gtodd possibly a PEBKAC ... it seems to work on FreeBSD-10.0 will get something set up ti try open and net 20:33
FROGGS: gmake clean everywhere ... I will erasing everything a git cloning anew then pulling those revs 20:34
FROGGS: hmm I did perl Configure.pl --backends=parrot --gen-parrot=master --gen-nqp=master 20:35
bartolin gtodd: Regarding old files: While debugging the problem I did rm nqp/3rdparty/dyncall/dyncall/*.o and rm nqp/3rdparty/dyncall/dyncall/libdyncall_s.a between my trys with Configure.pl
gtodd: likewise with dyncall/dyncallback and dyncall/dynload 20:36
gtodd: Otherwise it looks exactly like the problem I had before my patch to nqp (5ed99d2) 20:37
gtodd hmm I am rm ing a lot of stuff but :-) 20:38
20:38 jaffa4 left
FROGGS gtodd: git clean dxf # that should be enough 20:38
gtodd: that deletes everything not checked in, but has troubles with submodules... (but there are none in nqp) 20:39
gtodd hmm ok 20:40
bartolin gtodd: Hmm. And does nqp/3rdparty/dyncall/configure acutally contain the line with -fPIC? 20:42
gtodd if [ -z "${CFLAGS}" ]; then 20:43
printf "CFLAGS=-fPIC\n" >>$C
fi
20:44 Akagi201 joined 20:46 zakharyas left
gtodd so yes 20:46
bartolin gtodd: I'm trying with 10.0 and a fresh copy of rakudo.
gtodd maybe it is my system's libffi libuv etc. 20:48
20:48 Akagi201 left
gtodd everything is recently built 20:48
20:51 anaeem1_ joined
gtodd the layout of modules always gets me reremembering in a suboptimal way :-) (for me) but I assumed it was due to parrot and moar nqp needing to build differently depending on what else is built at the same time --> basically something complicated that was easier if automated ... 20:52
tony-o is there something similar to %INC in p6? 20:53
lizmat tony-o: not really 20:54
20:54 spider-mario left
lizmat there *is* currently internally a has that keeps the files loaded 20:54
but that is temporary until the implementation of S11 and S22
tony-o ahh okay, thank you
lizmat but even then, it probably won't be system wide 20:55
each CompUnit (type) will be responsible for remembering if it has been loaded before already 20:56
dalek kudo/nom: 71de9d1 | (Elizabeth Mattijsen)++ | src/core/IO/Socket/Async.pm:
Normalize (chars|bytes)_supply handling

Step #1 in ensuring the order in which we "more" values.		 20:58
tony-o i was more curious from a plugin perspective if i could 'discover' stuff that has been loaded by another module
lizmat that info may or may not be available yet 20:59
but that type of introspection *should* be possible, I would think
bartolin gtodd: This worked for me on FreeBSD 10.0 21:01
git clone github.com/rakudo/rakudo.git
cd rakudo
perl Configure.pl --gen-parrot --gen-nqp --backends=parrot
## quickly applying the parrot patch manually after parrot repository is cloned
make
rurban —gen-parrot=master should be easier 21:02
bartolin rurban: Ah, that way I would get the parrot patch as well? 21:03
lizmat
.oO( read that as "parrot sketch" :-)
 21:06
bartolin ;-)
dalek kudo/nom: 36badaa | (Elizabeth Mattijsen)++ | src/core/IO/Socket/Async.pm:
Apply ordering logic, same as with Proc::Async
 21:09
lizmat ^^^ should hopefully fix many of the issue with async sockets 21:10
*issues
21:16 travis-ci joined
travis-ci [travis-ci] Rakudo build failed. Elizabeth Mattijsen 'Apply ordering logic, same as with Proc::Async' 21:16
[travis-ci] travis-ci.org/rakudo/rakudo/builds/35675489 github.com/rakudo/rakudo/compare/7...badaa42aeb
21:16 travis-ci left
lizmat ^^^^ failed because of no space left on device 21:17
"Error while writing to 'Makefile': No space left on device at Configure.pl line 272."
where does travis run ? 21:18
host08 ?
lizmat is not sure how to fix this :-(
21:22 erkan joined, erkan left, erkan joined
lizmat calls it a night 21:25
good night, #perl6!
bartolin good night, lizmat++
21:27 jlaire joined
ingy . 21:28
yoleaux 20:02Z <Ven> ingy: github.com/ingydotnet/pig-latin-acme/pull/3
21:29 Larre_the_cow joined
ingy .tell Ven applied,thanks! 21:29
yoleaux ingy: I'll pass your message to Ven.
21:29 haroldwu joined, avuserow joined, apejens joined
timotimo no spec left on device %) 21:29
yoleaux 20:26Z <brrt> timotimo: i wanted to ask him about what he'd think would be the next step for the JIT. i want to refactor the params functions but this needs some work as i'll need to have a detailed map of coercions done by these
ingy .tell Ven join #lingy 21:34
yoleaux ingy: I'll pass your message to Ven.
21:42 raiph left 21:44 Akagi201 joined
ggoebel11111111 nine: any chance Inline::Perl5 can be made to handle the default case for Perl6 which assumes Perl5 if use v6; is absent at the top of a script? 21:46
see perlcabal.org/syn/S01.html 21:47
21:49 Akagi201 left
ggoebel11111111 just caught up with backlog... looks like nine is afk until Sunday evening :-( 21:56
timotimo interesting idea 21:58
22:03 leont joined 22:20 telex joined 22:23 anaeem1_ left 22:34 ezra1 joined 22:37 ezra1 left 22:40 guru joined, guru is now known as Guest25747 22:44 leont left 22:45 Akagi201 joined 22:50 Akagi201 left 22:57 cognominal joined 23:06 AndChat-188244 joined 23:08 cognominal left 23:09 Guest25747 left 23:15 erkan left 23:16 erkan joined 23:22 ivanshmakov joined 23:46 Akagi201 joined, AndChat-188244 left 23:50 Akagi201 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!