»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.
lue timotimo: Notice how S26 cheats this by putting four spaces in front of every line in its explicit code blocks, and its implicit code blocks (the .WHY/.WHEREFORE diagram S26/Declarator Blocks is an example) are not indented 00:03
timotimo ah, whoops :)
lue That's been a problem for a while, not your fault :) 00:04
(I'm looking at perlcabal.org's S26 btw)
I just noticed, thanks to that diagram, that Pod6's HTML converter appears to strip *all* leading whitespace. Huh. 00:05
timotimo hm. 00:06
lue Just from implicit code blocks though. The explicit code blocks have their indentation preserved. 00:08
timotimo that's kind of outside my headspace right now, sorry 00:09
lue I think that WHY/WHEREFORE diagram actually gives a good look at how the indentation gets messed up (I'll post a gist to explain)
timotimo i'll test :allow<...> configuration parsing now and then i'll see if i can get formatting codes inside code blocks working
lue timotimo: that's fine. I should describe the problem for everyone interested anyway :)
timotimo sounds good
timotimo does it seem like a terrible and disgusting hack to set a dynamic variable that's in the grammar from a function inside Perl6/Pod.nqp? 00:18
lue gist.github.com/lue/5872996 How it seems Pod::To::HTML messes up [ I'll present this to #perl6 again tomorrow when there's more of a crowd :) ] 00:22
lue & 00:24
timotimo okay :) 00:25
not sure where B<=begin pod> is matching so that it would cause it to trip up in a block with :allow 00:34
oh, of course! <pod_string> will not match newlines! 00:35
i feel like it still needs to set $*VMARGIN before it'll work though 00:36
BenGoldberg I have a very silly question. Considering that perl6 is very unicode capable, is there any chance that Rat could make use of the unicode fraction symbols, when there is an appropriate one? 01:03
For example:
r: .chr.say for 188..190, 8528..8542
camelia rakudo 5c0dae: OUTPUT«¼␤½␤¾␤⅐␤⅑␤⅒␤⅓␤⅔␤⅕␤⅖␤⅗␤⅘␤⅙␤⅚␤⅛␤⅜␤⅝␤⅞␤»
timotimo you can prototype that in pure perl6 code :) 01:05
r: my class Rat { method new($a, $b) { say "oooohverride!" } }; say <1/2>; say 1/2;
camelia rakudo 5c0dae: OUTPUT«1/2␤0.5␤»
timotimo hm, can't override it like that, eh?
r: class Rat { method new($a, $b) { say "oooohverride!" } }; say <1/2>; say 1/2; 01:06
camelia rakudo 5c0dae: OUTPUT«1/2␤0.5␤»
timotimo which is strange
oh, no it's not
infix:</> will just have the regular Rat in its scope
i'm surprised both pod_textcontent tokens have a <pod_newline> in them, but they are joined with <pod_newline>+ in the pod_content:sym<text> token, too 01:08
doesn't that mean it requires there to be two newlines after each one?
BenGoldberg (Temporarily) ignoring output, how hard would it be so that the 1/2 character will always act as a hardcoded Rat.new(1,2)? 01:20
timotimo macro it? 01:24
gotta run now
dalek rl6-roast-data: a420f1c | coke++ | / (4 files):
today (automated commit)
 01:32
timotimo i was hoping i could get my debugging results faster with only make perl6, but i can't run my tiny test case with the resulting binary :( 01:33
probably need to pass an empty setting ror something 01:34
[Coke] ^^ the rakudo.jvm gain there is illusory - rakudo.parrot is failing tests.
timotimo indeed 01:35
dalek kudo-star-daily: fd88b6c | coke++ | log/ (5 files):
today (automated commit)
 01:36
kudo-star-daily: 61c9e6b | coke++ | log/ (5 files):
today (automated commit)
kudo-star-daily: 0a70dd4 | coke++ | log/ (5 files):
today (automated commit)
kudo-star-daily: e9ff3c7 | coke++ | log/ (5 files):
today (automated commit)
timotimo do you KBOw how you can type with two fingers on GHthe android portrait keyboard? 01:37
it really amazes me ow hthat works
[Coke] whoops. 01:40
timotimo i shall there is no shame in force pushing 01:42
[Coke] that last rakudo.jvm was the wrong one. the right one had a huge # of failures, presumably because I tried TEST_JOBS=2. rerunning 01:45
(and I'll probably force push today's commit once it's right. 01:46
sorear think I'll patch test_summary tonight 01:54
[Coke] sorear++ 02:01
timotimo i don't understand why pod_textcontent:sym<regular> requires strictly more spaces than the VMARGIN 02:22
i think i'd be better off not trying to use <pod_textcontent> in blocks where :allow is activated 02:25
that seems to stretch it a wee bit far.
[Coke] uhoh. seeing lots of aborts on the rakudo.jvm run. wondering if the last run count was right. 02:26
sorear I may be in the process of breaking other tests while fixing rx.t 02:27
[Coke] ... a lot of them, aye. 02:45
up to S32-str, hopefully can post this before sleep 02:46
dalek rl6-roast-data: 1c3e8d8 | coke++ | / (5 files):
today (automated commit)
 02:53
[Coke] ok, that rewrote history. 02:54
rakudo.jvm down to 4.48%. :(
colomon wtf? 02:55
are those numbers broken? I mean, I'm looking at a single file right now with 100 passing tests. Surely that cannot be 9% of the total passes for R-JVM. 02:56
[Coke] There is insufficient memory for the Java Runtime Environment to continue. 02:57
colomon likewise cos.t has 154 passing tests.
ah
[Coke] colomon: you could try the ulimited p6 I'm using to see if that's what it is. 03:01
colomon [Coke]: it's time for me to go to bed, but I'd be happy to give it a try in the morning. 03:02
colomon r: say 1R/2**256 03:08
camelia rakudo 5c0dae: OUTPUT«===SORRY!===␤Confused␤at /tmp/1xhmfMZBUd:1␤------> say 1⏏R/2**256␤ expecting any of:␤ whitespace␤»
colomon r: say 1.FatRat/2**256 03:09
camelia rakudo 5c0dae: OUTPUT«0.0000000000000000000000000000000000000000000000000000000000000000000000000000086␤»
colomon r: say (1.FatRat/2**256).Num
camelia rakudo 5c0dae: OUTPUT«8.63616855509444e-78␤»
colomon r: say 1/ (1.FatRat/2**256).Num
camelia rakudo 5c0dae: OUTPUT«1.15792089237316e+77␤»
dalek : 8449f0d | lue++ | docs/feather/syn_index.html:
[docs/feather] Add S99 to syn_index.html

It's already available if you type the URL directly, this just makes it accessible from the index.		 04:32
dalek ecs: c7e59ef | lue++ | S99-glossary.pod:
[S99] Add definitions, fix links

Adds a number of terms and definitions to the glossary. Also fixes links so they point to parts of this page instead of other (nonexistent) pages.		 05:14
dalek ecs: 0af5ba1 | (Brent Laabs)++ | S32-setting-library/IO.pod:
document IO::Path.succ and .pred, clarify path creation
 05:43
ecs: 78487b0 | (Brent Laabs)++ | S32-setting-library/IO.pod:
specify :parent option in IO::Path.cleanup
ecs: f95923d | (Brent Laabs)++ | S99-glossary.pod:
Merge branch 'master' of github.com:perl6/specs
diakopter .tell masak did you see www.youtube.com/watch?v=oo-gA9Z9SaA 06:37
yoleaux diakopter: I'll pass your message to masak.
diakopter wait, that's not the oo talk
.tell sorear your voice is there
yoleaux diakopter: I'll pass your message to sorear.
sorear diakopter: woohoo the videoproject is not dead 06:49
yoleaux 06:37Z <diakopter> sorear: your voice is there
FROGGS morning 07:18
sorear o/ FROGGS 07:23
FROGGS ohh damnit, GlitchMr splintered /o\ 07:38
lizmat lue++ for fixing stuff in S99 and teaching me the correct way of doing intra document links 07:42
lizmat is wondering whether we should write S99 in pod6 and use the =defn markers 08:00
dalek p: f63cafa | sorear++ | src/vm/jvm/runtime/org/perl6/nqp/jast2bc/AutosplitMethodWriter.java:
Break switches with more than 256 cases into bite-sized pieces to make them more palatable to the list splitter
 08:01
lizmat it would give timotimo something else to test his work on
sorear heehee 08:03
dalek ecs: 9b12b4c | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Some more additions / refinements / explanations
 08:20
dalek p: 4da2c35 | sorear++ | src/vm/jvm/runtime/org/perl6/nqp/jast2bc/AutosplitMethodWriter.java:
transfer try/catch and line number info to the fragments
 08:20
jnthn morning, #perl6 08:42
sorear morning jnthn 08:45
FROGGS morning
sorear jnthn: [Coke] saw the pass rate fall off a cliff yesterday. Might it be something I did?
FROGGS jnthn: you might be interested in my comment at the bottom: rt.perl.org/rt3/Ticket/Display.html?id=116613
jnthn sorear: Well, I've been practically useless for the last couple of days...
sorear: Which means it's possible I've committed next to nothing or something stupid... :) 08:46
sorear 19:54 < [Coke]> rakudo.jvm down to 4.48%. :(
jnthn sorear: Looks like my last commit was on the 23rd.
So I'm guessing falling off a cliff yesterday isn't anything I did. 08:47
FROGGS: augmenting the subtype causes method cache recomputation, which is the thing that needs to happen. 08:48
FROGGS jnthn: that is what I thought :o)
jnthn I didn't forget the issue, it just didn't make it too close to the top of my todo list yet. 08:49
sorear Whose bright idea was it for the type argument to anewarray to be the *component* type, while the type argument to multianewarray is the *finished array* type and the argument to newarray is something COMPLETELY DIFFERENT? 08:50
jnthn Strangely *inconsistent*. 08:51
jnthn tripped over that at least once. 08:51
TimToady A foolish inconsistency is the hobgoblin of large minds. :)
jnthn mmmm...hobgoblin! 08:52
...but not at this time of day :)
sorear ? 08:53
TimToady do they make beer from hobgoblins now? :) 08:53
sorear
.oO( do they make beer from animals, period? )
 08:54
jnthn en.wikipedia.org/wiki/Hobgoblin_(beer)#Hobgoblin :) 08:54
sorear i see. 08:56
FROGGS jnthn: that was not really a reminder for that issue, I just wanted to hear you thoughts... like "yes, that could be the fix, or: no, there is some underlying issue"... 08:57
jnthn FROGGS: Well, the deep issue is it needs some kind of weak reference-y thing.
FROGGS ahh, k
jnthn FROGGS: Otherwise, anonymous subclasses would accidentally live forever.
sorear how hard would it be to add weak references to nqp? 08:58
jnthn sorear: Well, I was pondering a special-cased publish/subscribe mechanism for the MOP to convey things like this, and hiding the details of how it achieves weakness, rather than actually doing real weak references. 08:59
sorear: We could look at doing weak refs generally, though.
sorear I think that weak refs are a pretty important language feature, I'd be +1 to doing them generally if it's not infeasible on Parrot 09:00
jnthn Well, that's kinda what put me off ;)
JVM has 'em though. I didn't put 'em in Moar yet, but only 'cus I haven't needed them so far. 09:01
sorear how would you do pubsub without weakrefs??
jnthn sorear: "great evil" 09:04
sorear: There is a hook in Parrot for when something is being freed. My idea was that when an STable gets freed, it goes and unsubscribes stuff first. 09:05
sorear jnthn: does parrot guarantee non-movement? 09:14
oh wait, the STable isn't garbage collected 09:15
sorear jnthn: the p5way is to mix a role (ok, MAGIC type) into backref targets that 1. maintains a list of unmanaged backrefs 2. clears them on free 09:17
jnthn sorear: yes, it doesn't move 09:22
sorear: STables are GC'd too
There's no global type registry 09:23
sorear jnthn: I mean they're malloced, not PMCs 09:24
s/mean/meant/
jnthn sorear: oh
yes
Well, they're wrapped in a PMC
timotimo good * #perl6
jnthn o/ timotimo
sorear: I'm not a Rakudo JVM spectest run here 09:26
uh 09:27
I'm *doing* a ...
It's failing to look especially bad. 09:29
dalek p: a468f3d | sorear++ | src/vm/jvm/runtime/org/perl6/nqp/jast2bc/AutosplitMethodWriter.java:
Generate wrapper code
 09:37
sorear is now at the point to start trying to run fragmented code
surprise, surprise! VerifyError!
jnthn /o\ 09:38
timotimo does a very simple first attempt at making table cells be parsed with format codes 09:43
also missing for more greatness: allowing =config C<> :allow<...> 09:44
jnthn sorear: The spectest here on my machine doesn't have epic fail like in the nightly run. 09:45
FROGGS unpushed code? 09:46
jnthn nothing here 09:47
FROGGS sad there is no error message in the test summary 09:48
sorear jnthn: Unhandled exception: Can not call method 'load_setting' on a null object 09:49
know anything about that?
oh, huh 09:51
apparently if you do --target=jast it doesn't serialize
so if you jast2bc manually, you wind up with useless classes
jnthn I think --target=jast dumps the JAST text 09:53
yeah, spectest finished here. Actually ran 200 more tests since the last run I did. 09:57
masak good noon, #perl6 09:59
yoleaux 06:37Z <diakopter> masak: did you see www.youtube.com/watch?v=oo-gA9Z9SaA
timotimo did he see it?! 10:00
masak he had not. thanks, diakopter++ :) 10:02
timotimo not from that viewpoint at least 10:03
in my dream last night i held a spontaneous talk about something probably perl6 related and i made many many thinkos and when i saw the recording directly afterwards the video and audio quality was as bad as if it were recorded on VHS in the early 90s and played back today >_> (there is no point to this) 10:04
sorear o/ masak 10:07
masak \o
sorear masak: have you seen en.wikipedia.org/w/index.php?title...on=history ?
masak yes, I have. 10:08
jnthn tries to discern something useful from it 10:09
"yet another"? :) 10:10
masak yes, that doesn't sound filly NPoV :)
fully*
timotimo hum. i have a question.
shouldn't i be able to just use %*LANG inside src/Perl6/Pod.nqp? 10:11
FROGGS jnthn: c'mon, it is just another backend :P
timotimo return %*LANG<main>.parse($t, :rule<pod_string>).ast <- shouldn't this work?
moritz nothing will set the .ast unless you specify actions 10:12
timotimo oooooh
of course
MAIN-actions should do, right? 10:13
and it seems like main is wrong and should be MAIN instead
FROGGS should probalby be MAIN, yes
but if you invoke it like this, it doesnt know about its actions
timotimo yes, i passed :actions(%*LANG<MAIN-actions>) now 10:14
i hope that's the actual instance in use, not the class
FROGGS I'd try $*ACTIONS 10:15
sorear $(NQP_PREIX) not work so well 10:16
timotimo i see %*LANG<MAIN-actions> := $*ACTIONS.HOW.mixin... a few times
FROGGS and then it is bound again to $*ACTIONS in token statement 10:17
sorear ===SORRY!===
java.lang.RuntimeException: java.lang.RuntimeException
FROGGS so maybe there is no difference
jnthn sorear: I think there was a runtime exception? 10:18
;)
FROGGS sorear: well, that explains it
sorear i have heard, probably apocryphally, that there are prolog implementations which output "No." on any syntax error
FROGGS I have heard niecza saying "WTF" :o) 10:19
sorear that one's a special case in succ/pred 10:20
FROGGS I see *g*
sorear Oh hey a new one 10:21
===SORRY!===
java.lang.ClassFormatError: Illegal class name "Ljava/lang/Object;" in class file 18A8CCE068DABA0E352F805EC10407D0764CC67F
slightly more useful
timotimo .o(at least it's not crashing) 10:22
masak is that because of the semicolon at the end? 10:24
sorear masak: it's because the java virtual machine has at least four incompatible syntaxes for class names and I used the wrong one for the context
moritz thought that should be java.long.Object
*lang
jnthn sorear: Argh, that is teh annoying... 10:25
sorear It's actually supposed to be java/lang/Object there
jnthn still can't remember which syntax is wanted in various places :/
masak moritz: notice the 'L' in the beginning. 10:26
moritz: this is some kind of internal VM representation.
moritz oh 10:29
timotimo oh that's why it doesn't finish 10:31
should always increment the index when in while loops.
ah, darn 10:39
i messed up the serialization somehow and am getting Cannot assign a non-Perl 6 value to a Perl 6 container
turns out, serialize_aoaos isn't the right thing to use when you don't have strings, who would have known?
sorear oops. the verifier has decided that Lorg/perl6/nqp/runtime/CallFrame; and Lorg/perl6/nqp/runtime/CallFrame; can only be merged as Ljava/lang/Object; 10:42
oh, do I need to mention that we're now successfully creating a classfile from rx.rakudo? 10:46
timotimo \o/
80% here we come.
masak sorear++ 10:47
sorear ne'er said it *ran*. :D 10:48
timotimo if it exists, it almost runs, dunnit? :)
took only a couple of millenia for single-celled organism to learn to walk on land after all 10:49
sorear if I could stop getting quite so many VerifyErrors, then sure :D
um
java.lang.VerifyError: Expecting a stack map frame in method 912E823799037398A506061657F7E304613B5EB4.qb_1$f4(I[Ljava/lang/Object;)I at offset 38677
but I'm using COMPUTE_FRAMES... 10:50
oh, because it's dead code 10:52
jnthn sorear++ # getting us towards rx.t 11:03
dalek p: b8d3b40 | sorear++ | / (3 files):
Refactor to remove asm-commons dep; install asm-tree
 11:10
sorear Files=1, Tests=739, 1 wallclock secs ( 0.20 usr + 0.00 sys = 0.20 CPU)
Result: PASS
dalek p: 7b14f69 | sorear++ | src/vm/jvm/runtime/org/perl6/nqp/jast2bc/AutosplitMethodWriter.java:
Assorted bugfixes
 11:12
p: cd1ee20 | sorear++ | src/vm/jvm/runtime/org/perl6/nqp/ (2 files):
Automatically use method splitter if needed
kudo/nom: d8a0e12 | sorear++ | tools/build/Makefile-JVM.in:
List asm-tree as a 3rdparty jar
sorear only 739? 11:13
jnthn: curious if it's robust enough to work on someone else's system...:D 11:14
sorear [Coke]: did not get around to test_summary hacking. oopsies. 11:17
sorear -> sleep
sorear looks forward to hearing how many tests he broke
lizmat gnight sorear!
sorear (we're passing rx.t, after all) 11:18
lizmat before you can have any eggs, you need to break some omelette!
jnthn sorear: Will pull and try it :) 11:21
sorear++ # this is awesome!
daxim tell daxim we both, present and future self, are here 11:22
where's the message bot?
jnthn Also...rx.t passed right off without us having to fix the regex engine anywhere?! :)
FROGGS daxim: use .tell 11:23
daxim .tell daxim ho hum.
yoleaux daxim: Talking to yourself is the first sign of madness.
colomon rn: say (1.FatRat/2**256).Num
camelia rakudo 5c0dae: OUTPUT«8.63616855509444e-78␤» 11:24
..niecza v24-86-g39ab531: OUTPUT«8.6361685550944446E-78␤»
daxim .tell raiph learn to write better copy - www.useit.com/articles/inverted-pyr...yberspace/ - in long texts, put the conclusions first
yoleaux daxim: I'll pass your message to raiph.
colomon jvm> (1.FatRat/2**256).Num
0
timotimo is not making progress :|
jnthn .tell jntһn this is madness too 11:25
yoleaux jnthn: What kind of a name is "jntһn"?!
jnthn aww!
lizmat .tell this is madness as well 11:26
yoleaux lizmat: I'll pass your message to this.
tadzik what kind of rapping name is steve
lizmat who's this?
timotimo tadzik: is it because you're a water-dwelling mammal? 11:27
tadzik: i'm glad you're back; do you have a couple of minutes for poor ol' me?
tadzik timotimo: oh, sure 11:28
timotimo i've invented (haha) a function for src/Perl6/Pod.nqp called "parse_formatcode_in_text" and i'm calling it from inside table, so that each cell would theoretically have their formatting codes parsed
but i've run into all sorts of errors while blindly trying to come up with something, anything that would work 11:29
tadzik ah, yes
timotimo the newest, strangest error is this one here: gist.github.com/timo/543f7b8b791cbd617b71
tadzik my understanding was always that either formattingcodes should be entirely post-processed (which would help code blocks as well), or that table parser will need to call back to the grammar at some point 11:30
timotimo i've tried parsing the pod_string rule before, that didn't work (that's where i copied that merge_twines + serialize_array call from)
except formatting codes allow newlines where they may otherwise not be, so a little part of it needs to be in the non-post part
tadzik *nod* 11:31
timotimo if this code starts working, maybe the preprocessing can be slimmed down and formattingcode generation can indeed be made post-processing 11:32
that would certainly make things easier; the code blocks thing you've mentioned had me busy for quite a while with very little results 11:33
timotimo also, it seems like the code i had before that gave me this output for a simple table with cells with and without formattingcodes: Array.new(Pod::Block::Table.new(caption => Any, headers => Array.new(), config => ().hash, content => Array.new(["", "", ""], ["", "", ""], ["", "", ""]))) 11:34
tadzik :|
timotimo it seems like i'm forgetting to reset the @newrow contents in the nested loop below, fwiw 11:35
colomon sorear: with your changes, rx.t works fine on my system. however, there's a turn of debugging output you left live in there. :) 11:36
Ljava/lang/RuntimeException; : 1
calcSize: 0-1 : centry(69) uentry(11) base(1) uexit(14) cexit(39) total(134)
etc
colomon doesn't know if rx.t worked before or not; his Linux box has a *lot* of memory
timotimo it didn't work before because the jvm aborted compilation 11:38
can you make out how many tests in that suite succeed?
dalek ecs: f553b6c | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Some more additions and explanations
 11:39
colomon timotimo: "in that suite"? in roast, or some subset? 11:42
timotimo sorry; rx.t
colomon timotimo: 100%
All tests successful.
Files=1, Tests=739, 22 wallclock secs ( 0.08 usr 0.01 sys + 0.05 cusr 0.04 csys = 0.18 CPU)
Result: PASS
timotimo wow, that's pretty cool :) 11:43
takes us up to only 74% of rakudo, though :( 11:44
can you run it on parrokudo and measure the time there as well? 11:46
lizmat parrokudo ? 11:47
:-)
colomon Files=713, Tests=20529, 589 wallclock secs ( 3.64 usr 0.91 sys + 19.53 cusr 3.89 csys = 27.97 CPU) 11:49
that's not a significant change from the last results I got, I think 11:50
maybe more passing tests? can't really tell from that.
timotimo that's a massive time difference between parrot and jvm
oh
you were running *all* files
timotimo fails communication today
lizmat Files=1, Tests=739, 56 wallclock secs ( 0.20 usr 0.03 sys + 54.62 cusr 1.05 csys = 55.90 CPU) # MacBook Pro i5/Parrot 11:51
54.62 cusr compared to 0.05 cusr? WTF ? 11:52
timotimo tadzik: fwiw, my changes to the pod stuff are in timo/rakuo on the pod6-improvements branch if you want to check it locally
is that "cumulative user time" or something?
colomon lizmat: as I understand it, it has something to do with sorear++'s java harness for the test
timotimo colomon: but it says "parrot" at the end of that line 11:53
lizmat still, 22 wallclock vs 56 wallclock, that's a factor of 2.5
colomon oh.
timotimo you have the same computer specs as colomon?
colomon is going to upgrade his MBP to 10.8.x sometime this weekend, he thinks... 11:54
timotimo: surely not
timotimo in that case the 2.5 number isn't to be trusted
colomon I'm running the tests on my big linux box
I'm rebuilding parakudo right now 11:55
lizmat parakudo ? :-) 11:59
lizmat is watching the evolvement of a new term 12:00
daxim perlmonks.org/?node_id=1040324 # commit stats 12:11
timotimo will hopefully manage a non-spiky contribution pattern 12:15
non-spiky as in: not falling off to nothing for very long periods of time, possibly forever
colomon notice that I have made 400+ p6 commits in the last year, just mostly not to the projects he looked at in that post. 12:20
lizmat it doesn't matter.. it's all lies, damn lies and statistics anyway 12:21
colomon Files=1, Tests=739, 22 wallclock secs ( 0.07 usr 0.00 sys + 21.04 cusr 0.33 csys = 21.44 CPU) # parrakudo 12:22
timotimo although ... a tool could be devised to tally up the commits and stuff over all projects in the ecosystem
there's already a "fetch all the git repos" script for the smoker
so pretty much exactly the same time?
timotimo out of curiosity, can you run the rx.rakudo-jvm.t with just ./perl6 on the jvm? 12:23
it'll probably turn out slower in that case :(
colomon Files=1, Tests=739, 22 wallclock secs ( 0.10 usr 0.01 sys + 0.07 cusr 0.03 csys = 0.21 CPU) # rakudo jvm
timotimo (again: slow startup + warmup time for the jit + many different pieces of code will not let the jit warm up properly)
colomon note that neither test seems to be accurately reporting the wallclock secs 12:24
timotimo that definitely went through the evalserver
colomon > time ./perl6 -Ilib t/spec/S05-mass/rx.rakudo.jvm 12:26
real0m21.091s
user0m45.127s
sys0m0.644s
lizmat: yes, true. I love how he uses committers over time to see how the "live" the repos are, without looking at " 12:27
timotimo that makes more sense. parallel gc and all
colomon commits over time at all, which is the actual interesting thing.
timotimo then again, a "commit" can be small or big, but still be just one commit 12:28
colomon I think git makes it pretty easy to count the lines in a commit
timotimo right, and github also shows that cumulated 12:29
github.com/perl6/nqp/graphs/contri...amp;type=a
JimmyZ Don't care about these details :P
colomon the fact that we've added two entire new backends in the last year suggests that we are not fading into nothingingness!
JimmyZ I did commit almost nothing
timotimo very true 12:30
JimmyZ and I'm not nqp/rakudo committer either 12:31
colomon: Is there a rx.rakudo.parrot time? 12:38
lizmat JimmyZ: I posted one a little while ago: 12:39
Files=1, Tests=739, 56 wallclock secs ( 0.20 usr 0.03 sys + 54.62 cusr 1.05 csys = 55.90 CPU) # MacBook Pro i5/Parrot
colomon time ./perl6 -Ilib t/spec/S05-mass/rx.rakudo.parrot 12:40
real0m20.628s
user0m20.165s
sys0m0.400s
FROGGS r: sprintf('%s') # masak, is that known? dunno if it matters thought because your sprintf will handle it right 12:40
JimmyZ thanks :P
camelia rakudo d8a0e1: OUTPUT«Null PMC access in get_string()␤ in sub sprintf at src/gen/CORE.setting:2769␤ in block at /tmp/pgObleiXIk:1␤␤»
FROGGS though* 12:41
Ulti that perlmonks commit stat post is a bit uppety :/ why dont people bothered about perl5 version number just pull a Java and use a number not related to the version, or even just start saying Perl.18, Perl.20 with each release of Perl5 12:44
Java 2 aka 1.2 that's where they got the 2 from afaik? 12:46
lizmat I wonder what the git stats on Perl 5 indicate ;-) if you really want to compare these silly things
Ulti oh I forgot Java did the version number thing twice, with 1.5 just becoming 5.0 instead 12:50
but its still Java2 even with the bigger version number :S how's that not more confusing than Perl 12:51
Ulti though I guess its the api version rather than language 12:51
colomon lizmat: I'm pondering writing a response to that post on my blog 12:53
JimmyZ github.com/mirrors/perl/graphs/contributors 12:54
:D
lizmat JimmyZ++ 12:57
timotimo commute^Wbikeride& 12:58
JimmyZ github.com/mirrors/perl/commits?author=TimToady 13:00
1988 ....
lizmat now you know why Perl 5 has been so stable :-) 13:01
JimmyZ limzat: I see you there too :{ 13:04
:P
lizmat me? in Perl 5 ? I don't (want to) remember :-) 13:05
timotimo tadzik: fwiw, i think github.com/perl6/Pod-To-HTML/pull/2 can be merged. i *think* i have a commitbit, but i'd like someone else to look over it first. would you be so kind? :) 13:36
tadzik timotimo: yes, by all means :)
timotimo: but I'm at work now
I give you my blessing to go ahead and merge it :) 13:37
timotimo thank you :) 13:38
tadzik I've seen the results it produces, it looks cute 13:39
timotimo parts of it come from my rakudo changes as well 13:40
i'll just make one last run over the code to make sure i didn't break things that were okay before
Nom- Evening 13:43
lizmat Afternoon! 13:44
moritz morning!
colomon morning! 13:45
lizmat :q
hugme hugs lizmat, good vi(m) user!
lizmat thank you, but but but :-)
lizmat starts wondering whether she left an i in somewhere 13:46
Nom- Would discussions relating to Moe be in here or somewhere else? Been playing with it, and it looks rather interesting :) 13:48
tadzik Nom-: #moe is probably more appropriate :) 13:49
Nom- I looked earlier and it seemed empty :P
tadzik maybe it's on irc.perl.org then, I dunno
Nom- Yeah, looks like there's one here but it's empty... i'll check in irc.perl :P
lizmat rn: my Int $h; say $h.of # is this a case of NYI, or is there a reason for this omission ? 13:51
camelia niecza v24-86-g39ab531: OUTPUT«Unhandled exception: Unable to resolve method of in type Int␤ at /tmp/DC_9U2SBQO line 1 (mainline @ 4) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4579 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4580 (module-CORE @ 576) ␤ at /home/p6e…
..rakudo d8a0e1: OUTPUT«No such method 'of' for invocant of type 'Int'␤ in block at /tmp/Okl97IVzja:1␤␤»
lizmat rn: my Int %h; say %h.of # works fine for hashes and arrays 13:52
camelia niecza v24-86-g39ab531: OUTPUT«Unhandled exception: Unable to resolve method of in type Hash␤ at /tmp/4Z8XlFIkN0 line 1 (mainline @ 4) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4579 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 4580 (module-CORE @ 576) ␤ at /home/p6…
..rakudo d8a0e1: OUTPUT«(Int)␤»
lizmat well, in the rakudo case, then :-)
dalek : 444b94e | (Tobias Leich)++ | / (3 files):
dispatch postcircumfix:[] to :P5[] and allow negative values
 13:59
: 4b18516 | (Tobias Leich)++ | lib/Perl5/Terms.pm:
have method P5do in Any only
: e9abe60 | (Tobias Leich)++ | / (4 files):
support unpack
lizmat support unpack, ooo wow! 14:01
FROGGS lizmat: stolen from Buf, though I added directive U 14:02
there might be a better (faster) way to do that, but this worx, so I'm fine with it for now
colomon \o/ 14:03
timotimo FROGGS: kind of surprising how few test cases that makes pass :( 14:13
lizmat I thought the pack/unpack tests in p5 ran in the thousands ?
FROGGS lizmat: pack has >14k 14:14
I'm working on that right now
timotimo: yeah, sometimes you get no extra pass when fixing fundamentally things :/
JimmyZ FROGGS: multi method P5unpack(Str:D:), I guess you lost someting
FROGGS for example because the tests fail because 'use version' fails 14:15
colomon FROGGS: isn't that the worst?
FROGGS JimmyZ: nope, that P5unpack gets invoked on the template, and the to-unpack string is optional (and defaults to $_) 14:15
colomon: what is the worst? 14:16
(there are many worse things)
colomon FROGGS: making a significant change which doesn't register in test runs
"worst" is perhaps hyperbole. ;)
FROGGS well, there are things that suck more... 14:17
JimmyZ shouldn't be self.p5unpack(CALLER::DYNAMIC::<$_>) ?
FROGGS yesterday I broke split(//,...) on the way of fixing something else, and my test_summary told me I'm going to pass 80 tests more, which was a false positive :o( 14:18
JimmyZ knows nothing about it, just a guess.
FROGGS JimmyZ: right you are!
FROGGS runs spectest 14:19
colomon FROGGS: I've had situations where I spent a week or two tracking down and fixing a $work issue which seemed like it should have significant impact, only to run it against my full test suite of files and discover that, out of millions of faces, the only one it fixed was the one I was working on. Or worse, it actually broke more things than it fixed. :( 14:26
FROGGS :/ 14:28
colomon It's not actually worse than having your family murdered by pirates, but it still sucks.
FROGGS that happens here often sadly
right
tadzik ....what happens here often? Getting your family murdered by pirates? 14:29
[Coke] sorear: could you changes have made rakudo.jvm consume more memory?
*your
FROGGS tadzik: no :P 14:30
[Coke] When I run the suite, I run it under: ulimit -t 300; ulimit -v 2048576 14:31
colomon like this? ulimit -t 300; ulimit -v 2048576; make spectest 14:32
[Coke] no; I'm using test_summary. 14:37
here:
[Coke] github.com/coke/perl6-roast-data/b...udo.jvm.sh 14:37
dalek : 16d0c3a | (Tobias Leich)++ | lib/Perl5/Terms.pm:
dispatch .P5unpack() to .P5unpack($_), spotted by JimmyZ++
 14:38
timotimo no results for the tests? 14:40
timotimo aaw 14:42
FROGGS timotimo: no relevant changes
Ulti is there a standard POD notation you are meant to follow for your perl6 classes? so that docs can be auto generated/introspected? 14:45
timotimo just put #| and #= based pod comments everywhere, no? 14:46
flussence =NAME, =AUTHOR, =VERSION
Ulti I meant more something to document methods and that sort of thing, so you have a description of what they are and for 14:47
flussence that would be the #= thing 14:47
Ulti k
PerlJam Ulti: See S26:284 for an example 14:48
.oO( Does S26 render properly yet? )
 14:49
tadzik properlier, thanks to timotimo++ 14:50
timotimo PerlJam: t.h8.lv/S26-documentation.pod.html <- more or less
PerlJam timotimo++ nice 14:51
Ulti PerlJam: thanks, I was reading that as you posted ;) 14:52
timotimo tadzik: can you make a bit of time to help me get rid of the wrongness with the table cell serialisation? i really don't think i understand what's involved and what needs to be different
Ulti the fact I was reading the POD about POD made it a bit confusing :S
timotimo yeah, it looks like chaos in the source, but once you slog through the grammar (or the properly rendered version) it all starts making sense 14:53
another thing is that i implemented proper support for different formatting code delimiters, with multiple < or sinlge «
that should allow the pod source to become much nicer to look at 14:54
tadzik timotimo: sure, if I'll be able to :)
timotimo i'll wait for your ping 14:55
tadzik okay
it may be tomorrow :(
timotimo i'll just have to be strong and keep my fingers off for a day ;) 14:56
JimmyZ timotimo++, It's nicer
timotimo maybe i'll find something else in perl6 that needs fixing/implementing?
JimmyZ parse pod6 completely? 14:57
tadzik that way lies madness
tadzik timotimo: typed arrays? :} 14:58
timotimo that way lies madness, too
PerlJam shaped arrays! :)
JimmyZ S09 :D
timotimo madness has many franchises all over the semantic territory nowadays
tadzik hm
timotimo they even offer drive-in service
"madness in passing"
tadzik I know, I know, oh-oh-oh 14:59
look into select()/poll() on JVM :)
timotimo i think i may need to test my [..;..;..] changes a bit further. 14:59
timotimo r: my $aref1 = (1, 2, 3); my $aref2 = [4, 5, 6]; my $multi2 = [ $aref1, $aref2 ]; say $multi2[0] ~~ Parcel; say $multi2[0] ~~ List; # why should this be wrong? 15:45
camelia rakudo d8a0e1: OUTPUT«True␤False␤»
timotimo the tests seem to require the result to be a List rather than a Parcel and it's tagged as a nom regression
that doesn't make sense to me.
dalek kudo/nom: 9880c7b | (Elizabeth Mattijsen)++ | src/core/Any.pm:
Small optimisation in creating pairs

No reason to copy things if we can alias them.		 15:49
timotimo my @a = <foo bar baz quux>; say @a[*]; 15:57
r: my @a = <foo bar baz quux>; say @a[*];
camelia rakudo d8a0e1: OUTPUT«foo bar baz quux␤»
timotimo r: my @a = <foo bar baz> <qx qux quux>; say @a[*][0]; 15:58
camelia rakudo d8a0e1: OUTPUT«===SORRY!===␤Alphanumeric character is not allowed as a delimiter␤at /tmp/MP0ynTIixZ:1␤------> my @a = <foo bar baz> <qx ⏏qux quux>; say @a[*][0];␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta…
timotimo r: my @a = <foo bar baz>, <qx qux quux>; say @a[*][0];
camelia rakudo d8a0e1: OUTPUT«foo␤»
timotimo r: my @a = <foo bar baz>, <qx qux quux>; say @a[][0];
camelia rakudo d8a0e1: OUTPUT«foo␤»
timotimo r: my @a = <foo bar baz>, <qx qux quux>; say @a[0,1][0];
camelia rakudo d8a0e1: OUTPUT«foo␤»
timotimo hm.
locally, my @a = <foo bar baz>, <qx qux quux>; say @a[0,1;0] # -> foo bar 15:59
should @a[*;0] behave just like 0,1 in this case?
timotimo i find the specs on this matter insufficiently slushed out, but i won't dare just tread onto pm's territory 16:01
dalek ecs: 9e0e656 | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Added lemma for "gimme"
 16:05
lizmat which somebody could enlighten me on that, as the word "gimme" is not in the spec 16:06
specifically, what does gimme(*) mean?
[Coke] lemma?
timotimo "reify as many elements as you find sensible right now"?
lizmat [Coke]: en.wikipedia.org/wiki/Headword 16:07
timotimo: but, would it make sense then to later do gimme(pos+1) ? 16:08
timotimo that question i cannot anscer 16:09
i thought gimme(n) was "please reify at least n more elements" rather than "please reify up to the elements up to the index n"
but i am absolutely not sure
masak o/, #perl6 16:16
Ulti if I have several "multi method new" defined do they all use the same BUILD submethod? 16:17
I'm having a hard time just having a different constructor for no params and a single positional parameter
do I need multi method new (Nil) ? 16:18
colomon multi method new () 16:19
Ulti notices ALL of his troubles go away when he doesn't specify int32
colomon dunno if you need it or not
but you don't need or want the Nil, as far as I know. 16:20
Ulti colomon the root problem is just the int32, are the native types immutable? 16:20
masak no, they're mutable.
Ulti hmmm
lizmat masak!
masak lizmat! \o/
r: my int $i = 0; say $i; $i = 32; say $i 16:21
camelia rakudo 9880c7: OUTPUT«0␤32␤»
masak mutable.
Ulti Cannot modify an immutable value Mt.pm:23 in block MT.pm:36 <-- only thing that changed was I gave has int32 $.seed instead of has $.seed gist.github.com/MattOates/5877912 16:25
I think I'm just being super dense
if I use $!seed things start to work out 16:26
Ulti also using 2013.05 R* 16:28
jnthn Well, $.foo means a readonly accessor 16:30
lizmat r: say $Inf 16:31
camelia rakudo 9880c7: OUTPUT«Inf␤»
Ulti jnthn and because of that you would expect that code to work by changing int32 to Int? 16:32
because it does :S
lizmat thinks $Inf is some kind of implementation leakage 16:33
jnthn Ulti: It'd need an "is rw". That doesn't owrk on natively typed attributes yet.
daxim enwp.org/Role_(computer_science) enwp.org/Trait_(computer_programming) what's the exact difference? 16:34
masak lizmat: yes. I guess it falls out of lexical scoping. 16:35
lizmat: ISTR '$Inf' was some temporary hack.
lizmat intentionally so, it seems
lizmat :# XXX: Temporary definition of $Inf and $NaN until we have constants ava 16:36
masak we do have constants ava 16:37
rn: say Inf.^name
camelia rakudo 9880c7, niecza v24-86-g39ab531: OUTPUT«Num␤»
lizmat r: constant Inf = nqp::p6box_n(nqp::inf()); say Inf 16:38
camelia rakudo 9880c7: OUTPUT«Inf␤»
lizmat r: constant Inf = nqp::p6box_n(nqp::inf()); say Inf.WHAT
camelia rakudo 9880c7: OUTPUT«(Num)␤»
lizmat so I guess I could fix that 16:39
lizmat r: constant Inf := nqp::p6box_n(nqp::inf()); say Inf # does binding make a difference? 16:40
camelia rakudo 9880c7: OUTPUT«Inf␤»
[Coke] will do the daily rakudo.jvm run today with no memory ulimit and see what happens. 16:50
masak lizmat: the 'constant' declaration is essentially always a binding, IIUC. even if you use assignment in the syntax. 16:53
lizmat ok
nwc10 do we have enough points yet to make pretty graphs? 16:55
masak probably.
dalek ecs: 2f382ff | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Add IIUC lemma
 17:03
ecs: a17b887 | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Add ISTR lemma
masak hehe. 17:05
AFAIK, DIHWIDT. :)
lizmat DIHWIDT ?? 17:07
masak "doctor, it hurts when I do this" 17:10
daxim DDG.GG ??
masak LMGTFY.
daxim bbq 17:11
lizmat sigh
daxim I got warnocked with my question about traits and roles
masak daxim: roles in Perl 6 correspond more or less exactly to traits in Smalltalk. 17:12
the former is based on the latter.
daxim that doesn't help explain the distinction enwp makes
is that in error?
masak the reason they were not called "traits" in Perl 6 is that that term was already taken for something else.
daxim aha 17:13
masak daxim: hold on, I'll skim the enwp links.
daxim: to be honest, I'm not 100% sure Perl 6 embodies a "role-oriented programming" paradigm. 17:14
at the very least, that's something someone would have to prove by listing what ROOP requires, and point-by-point what Perl 6 provides.
does that answer your question? :) 17:15
daxim hmmm
the table in lemma Role_(computer_science) implies ROP is a subset of OOP 17:16
I'm almost satisfied. what are perl6 traits, then? 17:17
masak ...which was either never clearly defined, or defined too many times by too many people. :)
(OOP, that is)
daxim: Perl 6 traits are the things where you go 'is awesome' on some declaration, fsvo 'awesome'.
r: class A { has $.x is rw }; say "that 'is rw' is a trait"
camelia rakudo 9880c7: OUTPUT«that 'is rw' is a trait␤»
daxim is that the thing where you attach classnames to object instances? 17:18
ah, to attributes
masak for example. 17:19
there's also 'is export' on subroutine declarations.
and 'is copy' on parameters.
and so on et cetera
daxim I'm intrigued and wish to subscribe to your newsletter 17:20
masak :)
dalek ecs: 5300d63 | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Add DIHWIDT, LMGTFY and warnocked
 17:25
lizmat is sad: 17:28
Stage optimize : 31.366
Code ref 'Numeric' does not exist in serialization context
this is the diff and the error: gist.github.com/lizmat/5878505 17:32
suggestions anyone?
timotimo it would seem your comparisons with == would cause that, no? 17:36
is there === in nqp?
r: say(1 === 1)
camelia rakudo 9880c7: OUTPUT«True␤»
timotimo maybe that'll work?
lizmat I actually changed a lot from === to ==, as I don't see why that in this case would make a difference 17:38
timotimo probably because it tries to coerce? :) 17:39
timotimo makes postcircumfix:<[ ]> aware of empty cells inside semilists 17:46
lizmat working on a patch 17:50
timotimo does it seem sensible for postcircumfix:<[ ]>(\SELF, LoL $foo) to get a Nil for empty cells? 17:54
it's not distinguishable from writing an actual literal Nil into the LoL 17:55
lizmat alas, replacing all == with === just gives another error, giving up for now 18:00
timotimo aaw :( 18:01
do you have intuition for the [ ] issue? 18:02
timotimo one: should @a[*;1] give a list of [@a[0, 1], @a[1, 1] ... @a[+@a, 1]]? 18:03
two: should @a[*][1] give that same result?
timotimo thr: should @a[;1] behave like the proposed solution for @a[*;1] above? 18:04
dalek kudo/nom: 6eab1e9 | (Elizabeth Mattijsen)++ | src/core/control.pm:
Some error message corrections
 18:07
kudo/nom: 55025c5 | (Elizabeth Mattijsen)++ | src/core/control.pm:
Leave note for other people trying to use Inf/NaN as constants
lizmat timotimo: I have no idea
ssutch hallo #perl6 18:17
timotimo hey ssutch 18:19
ssutch hallo timotimo 18:20
timotimo irclog.perlgeek.de/perl6/2013-06-27#i_7259945 - may i ask you to provide an opinion?
ssutch yikes 18:22
ssutch [@a[0, 1], @a[1, 1] ... @a[+@a, 1]] 18:22
dalek ecs: a8ed7cf | (Elizabeth Mattijsen)++ | S29-functions.pod:
Specify "interval" function
 18:22
timotimo For all multidimensional array types, it is expected that cascaded subscripts: [...] will either fail or produce the same results as the equivalent semicolon subscripts: [...]
ssutch ^ does that mean a list composed of elements 0, 1 from @a? 18:23
colomon ssutch: not in perl 6...
[Coke] rakudo.pvm still failing 29 tests. 18:24
S06-macros/opaque-ast.rakudo ; S06-macros/quasi-blocks.t ; S06-macros/unquoting.t 18:25
ssutch timotimo, unfortunately i dont know
timotimo er, i mistyped pretty horribly
apologies
ssutch not sure what a cascaded subscript or a semicolon subscript is 18:26
timotimo it was meant to read [@a[0][1], @a[1][1] ... @a[+@a][1]]
ssutch ahh
timotimo cascaded subscript is like what i wrote one line up ^
semicolon subscript is the same with ][ replaced by ;
ssutch ok, simple enough
in that case, what you propose seems reasonable 18:27
i especially like the 2nd syntax @a[*][1] - seems pretty obvious whta you're trying to do
timotimo good 18:28
ssutch and so does @a[*;1] - now that i know what a semicolon subscript is
timotimo because that's what i just implemented
how do you feel about @a[;1]? the specs say @a[1] is the same as @a[1;]
so it would only make sense that @a[;1] is also the same as @a[*;1]
ssutch what is @a[1;] ?
ssutch @a[1;*]? 18:29
ssutch if so, that makes sense enough 18:29
ssutch what ever happened to your pod parsing for rakudo? 18:29
colomon is engaging in therapeutic p6 coding 18:31
ssutch FROGGS, pmichaud, moritz: anyone have a quick tuit to merge this PR or request changes: github.com/rakudo/rakudo/pull/170
FROGGS ssutch: I'm not the right guy for that, sorry 18:32
ssutch FROGGS: no problem :)
lizmat ssutch: I'm afraid I can't oversee the consequences either 18:37
ssutch that's cool 18:38
timotimo ssutch: i'm waiting for tadzik to help me with a tremendously hairy (at least to me) issue and then we may end up getting formatting codes in cells parsed 18:39
timotimo and if that works, we may even end up with working :allow for code blocks 18:49
FROGGS timotimo: I mentioned that the quote language does the same? :P 18:51
timotimo yes, and i didn't really figure out how to translate that knowledg over 18:52
FROGGS :/
well, have you spotted the 'tweak' methods? 18:53
timotimo no, don't think so
tadzik suggested to stub out the formatting code parsing in the grammar and parse them "in post" 18:54
FROGGS I think I would make tokens for the things to allow/disallow, like for C<...> 18:55
then I would put these tokens in a role
timotimo i don't think that'll fly
FROGGS and if it sees a :allow rule, it would mix the right roles in
timotimo there's also =config C<> :allow 18:56
FROGGS hmmm, I dont know much about pod6...
timotimo ;)
it's easy to allow/disallow single formatting codes with a bitfield
so i don't think i need to create complicated roles, especially if they end up making combinations of classes+roles at run-time and thet could be slow 18:57
FROGGS from seeing your pastes it looks like you are doing too much in a single token...
that is not runtime
timotimo not slow as in "i want to prematurely optimize this!" but as in "prohibitively slow" 18:58
FROGGS maybe, yeah
timotimo github.com/timo/rakudo/compare/pod...provements 19:00
timotimo it looks like much, but it's only a few things :| 19:04
FROGGS yeah, not too bad
colomon rn: my $a = (a => 4, b => 12).Bag; my $b = (a => 1, c => 2).Bag; $a (+)= $b; say $a
camelia niecza v24-86-g39ab531: OUTPUT«bag("a" => 5, "b" => 12, "c" => 2)␤»
..rakudo 55025c: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/gvCwivAvgU:1␤------> ).Bag; my $b = (a => 1, c => 2).Bag; $a ⏏(+)= $b; say $a␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤ statement end␤…
colomon rn: my $a = (a => 4, b => 12).Bag; my $b = (a => 1, c => 2).Bag; $a = $a (+) $b; say $a
camelia rakudo 55025c, niecza v24-86-g39ab531: OUTPUT«bag("a" => 5, "b" => 12, "c" => 2)␤»
colomon rakudobug
timotimo what really blows up is when i try to come up with a way to allow code text blocks to contain formatting codes
colomon wonders if the masak bot is around. ;) 19:05
timotimo also, strange weirdness hapens when i try to call back from src/Perl6/Pod.nqp to the grammar
FROGGS timotimo: what do you mean by "call back to the grammar" 19:06
?
lizmat alas, the masak bot seems to be absent :-)
timotimo gist.github.com/timo/543f7b8b791cbd617b71 like that, FROGGS 19:07
FROGGS sounds like you have to box something, but I dont know what 19:11
timotimo box or perhaps serialize? 19:11
weird thing is: nothing i put there is different from when the rule is entered normally
i just give it a string to be parsed 19:12
dalek rl6-roast-data: ea81904 | coke++ | / (4 files):
today (automated commit)
FROGGS box I guess
[Coke] the memory limit seems to be the issue. I left in a time limit so we don't completely go out into the weeds. 19:13
(rakudo.jvm)
# 06/27/2013 d770a94 rakudo++ (25969); niecza (82.09%); pugs (36.6%); rakudo.jvm (78.13%)
timotimo oooooooh yeah! 19:13
(did rakudo regress even further? :P)
[Coke] no, rakudo is holding steady at 29 failures. 19:14
timotimo that's a mazing
[Coke] 508 more passes for rakudo.jvm to hit 80%, assuming rakudo.pvm recovers.
dalek osystem: 988e978 | segomos++ | META.list:
Update META.list

changed CSV::Parser to github repo instead of shitbucket		 19:15
timotimo the diff is too big to be displayed on github. sounds good to me :)
shitbucket? wow, that's kind of harsh 19:16
huh, it seems like you added a header for the latest change, but no details?
oh, the diff is only so huge because of the gigantig regression in between 19:17
[Coke] you can manually compare to the revision before that
PerlJam wonders what people have against bitbucket 19:19
timotimo i've done that
interesting, lots of regex tests files cancel too early because java chokes on unknown unicode stuff
+java.util.regex.PatternSyntaxException: Unknown character property name {Arabic} near index 9 19:20
colomon Does DateTime parse things like this? Sun Jun 9 23:10:56 2013 -0400 19:29
lizmat in p5 or p6 ? 19:30
colomon p6
masak no.
nor should it, IMO.
colomon why not?
masak that's ecosystem stuff.
it's culturally dependent enough not to be a core concern. 19:31
[Coke] sorear++ #moar jvm passes
I could see core dates having to be iso dates, even.
masak *nod*
that's basically what we have.
DateTime proper parses ISO 8601 datetimes. 19:32
segomos PerlJam: i like bitbucket but my module isn't getting added to the list
PerlJam segomos: It might be because the parser is biased towards github. 19:33
colomon grumble, grumble, I can parse it myself, I suppose.
[Coke] colomon: ... and put it in a module! ;)
colomon [Coke]: it's one line that doesn't do anything useful other than let me extract the year. 19:34
segomos PerlJam: yea that's what i was thinking too so i added github repo to my list of where to push to and i'm testing that theory 19:35
sorear o/ 19:36
colomon \o 19:37
lizmat sorear! 78%+
[Coke] r: "axe bat cat".wordcase.say
camelia rakudo 55025c: OUTPUT«Axe Bat Cat␤»
[Coke] on the jvm, that outputs the original string.
sorear something cheap we could do to get a lot more JVM passes: 19:42
PerlJam segomos: start reading at github.com/perl6/modules.perl6.org...nfo.pm#L47
:-)
dalek kudo/nom: 437d5e2 | (Elizabeth Mattijsen)++ | src/core/List.pm:
Simplify .elems, we have $Inf!
PerlJam where does the output of build-project-list.pl go when it runs? A page with the latest output might help people (like segomos) diagnose problems. 19:43
[Coke] gist.github.com/coke/5879701 - top "this test file won't run" for rakudo.jvm 19:44
dalek kudo/nom: 2b77d6d | sorear++ | t/harness:
JVM does not need ICU for any tests
[Coke] ah, that'll at least try to run more of them. nice. 19:45
segomos PerlJam: nice find !
what's going on with github right now..i keep getting an angry unicorn 19:46
[Coke] r: print 28690-27844 19:47
camelia rakudo 55025c: OUTPUT«846»
[Coke] so, that'll try that many more tests max.
sorear: I don't think that impacts test_summary. 19:48
lizmat r: my @a= 0,[1,2],3; my @b=<a b c d>; say @b[@a] # not sure what to expect
camelia rakudo 55025c: OUTPUT«a c d␤»
[Coke] hurm, it might.
lizmat r: my @a= 0,[1,2,1],3; my @b=<a b c d>; say @b[@a] # not sure what to expect 19:49
camelia rakudo 55025c: OUTPUT«a d d␤»
timotimo sorear: java hardfails if it doesn't find a unicode class thingie it seems
lizmat r: my @a= 0,[1,2,1],3; say @a.flat
camelia rakudo 55025c: OUTPUT«0 1 2 1 3␤»
masak lizmat: that rhymes with what I expect, fwiw.
lizmat: Array != LoL 19:50
colomon masak! \o/
timotimo +java.util.regex.PatternSyntaxException: Unknown character block name {ArabicPresentationFormsA} near index 29
masak colomon! \o/
timotimo and then it says + 660 tests aborted (missing ok/not ok)
xilo_ compiling rakudo froze my VM :(
lizmat the thing is that postcircumfix[] is doing a .flat on the positions passed
masak xilo_: how much memory does your VM have available?
xilo_ 1.5gb
lizmat r: my @a= 0,[1,2,1],3; my $a = @a.flat; say $a.map( { $_ } ).eager.Parcel 19:51
camelia rakudo 55025c: OUTPUT«0 1 2 1 3␤»
lizmat hmmm 19:52
[Coke] timotimo: looks like java 7 should support unicode 6.0 - is ArabicPresentationFormsA newer? 19:53
jnthn Which tests is tht, ooc?
jnthn guesses one of the S05-mass prop ones 19:54
[Coke] S05-mass/properties-block.t
does perl6 have a way to say which version of unicode it's currently supporting? 19:55
because that would be awesome for properly fudging tests. 19:56
masak [Coke]: I don't know about any such mechanism. 19:57
[Coke] seems like it would be nice, no? 19:58
jnthn tries to take a look at it
timotimo r: my @a = <foo bar baz>, <qx qux quux>; say @a[*][0]; # should this be what it is?
camelia rakudo 55025c: OUTPUT«foo␤»
[Coke] jnthn++
timotimo and if so, should @a[*;0] be the same thing? 19:59
[Coke] my suspicion is that we're testing something that doesn't exist in the jvm's version of unicode.
masak timotimo: it should return <foo qx>, for sure. 20:00
PerlJam timotimo: and @a[*;0] should be the same as @a[*][0] according to my mental model of things 20:01
timotimo the second one, for sure. the first one, though?
i mean i know why [*][0] does what it does 20:02
and i don't know how to make it work properly without internally parsing multiple postcircumfix:<[ ]> as a single [] with a semilist inside 20:03
(in that case i suppose you can still use [*].[0] if you want the "old" behaviour)
sorear [Coke]: I wonder if it might work better to edit perl6 and add -Xmx400m or something 20:05
[Coke] sorear: sure, probably. the ulimit stuff was a straight ripoff of how pugs did it
anyone wants to tackle that, I'd love a pull request. 20:06
sorear pull request to what? I don't see a ulimit in test_summary
timotimo maybe it's in the perl6 "script"?
it could be that that only works if you declare a shaped array? ("that" being making @a[...][...][...] behave exactly like [...;...;...]) 20:08
dalek p: b20fe7e | sorear++ | src/vm/jvm/runtime/org/perl6/nqp/jast2bc/AutosplitMethodWriter.java:
Turn off autosplit debug output, colomon++
 20:08
jnthn Seems it's a canonicalization issue 20:11
[Coke] sorear: github.com/coke/perl6-roast-data/b...udo.jvm.sh 20:12
sorear is hopeful we'll pass 80% today 20:18
nwc10 and start on the second 80%? :-)
timotimo at this pace, it could totally be doable in another week or two 20:19
nwc10 anyway, progess is awesome. Thanks for all your hard work (and jnthn's too, and everyone else)
[Coke] sorear: today already ran. tomorrow. :) 20:21
oh, right, not the official one, nevermind.
s/official/github/ 20:22
colomon thinks there's a 100 ripe for the plucking in fatrat.t... 20:23
[Coke] colomon: doooo it. 20:26
colomon [Coke]: gotta $work and take care of my family. 20:27
dalek p: 3cfa6c3 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Canonicalize property names for Java regex \p{...}

This is enough to get S05-mass/properties-block.t running.		 20:31
jnthn That takes that file from 10 tests before exploding to getting through all 670.
nwc10 awesone
er, awesome
timotimo 660 more attempted tests!
jnthn Passing, even :) 20:32
timotimo and they all work?
amazing!
jnthn There's one skip in there
Dunno if JVM needs it
[Coke] jnthn: that's probably actually 4 test files. 20:34
timotimo that puts us just above 80%
nwc10 quick, somebody write more (spec)tests 20:35
timotimo (even if rakudo unregresses)
[Coke] timotimo: try s05-mass/properties-{block,derived,general,script}
jnthn [Coke]: No, derived/genera/script still explode
timotimo [Coke]: will do 20:36
lue hello world o/
lizmat lue!
timotimo oh, sad to hear that
but it's all still pretty awesome!
xilo_ has anyone got rakudo installed on arch? i've tried the aur package but it fails with some error while building. i tried editing the package to gen-parrot and gen-nqp so i can use git version, but get weird ass permission errors
timotimo "you don't have ass permissions" 20:37
xilo_ <_< 20:38
seriously 20:40
sorear timotimo: "at this pace"? looks like we just passed 80 a few minutes ago :D 20:47
colomon sorear++
colomon jnthn++ 20:47
flussence
.oO( at this rate we'll be passing 150% of the spectests by august )
 20:51
timotimo sorear: yes, we've passed 80%, even if the rakudo regression is fixed 20:52
sorear flussence: I made that joke last week 20:54
jnthn
.oO( at this rate we'll have made 150% of the jokes by September )
 20:55
sorear and probably sometime in july we'll need to think about JStart
JStar
flussence side question: will JVM/Moar do anything to change the thing now where I have to remember to update $PATH every time there's a parrot version change to keep panda working? 20:58
timotimo oh, my. java.lang.IllegalArgumentException: sprintf only accepts ints, nums, and strs, not class __P6opaque__62 - missing some unboxing somewhere?
jnthn timotimo: No, missing a proper sprintf implementation, but masak++ has been working on that
timotimo ah, we've been using the one provided by parrot or something? 20:59
jnthn yeah
timotimo that makes a whole lot of sense
flussence that annoyed me early on, when I tried to use a format string with '%1$s' in it and ended up digging in the code to see why it wouldn't work... the parrot one never did get fixed 21:01
timotimo Method 'submethod_table' not found - does that sound like LHF? 21:02
dalek p: 48d9276 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Handle Unicode script properties.

Gets S05-mass/properties-script.t to pass to the same degree as on Rakudo Parrot.		 21:04
jnthn timotimo: It sounds...weird
> 300 or so more passes there
timotimo t/spec/S04-declarations/my.rakudo.jvm - this is where it happens, but it' sonly worth 11 more attempted tests in this file. 21:05
dalek ecs: 2eb828a | (Elizabeth Mattijsen)++ | S99-glossary.pod:
Add lemma LHF
 21:05
masak oh, there's no definition of "lexotic"? that's a shame, it's fairly well-defined in S04. 21:12
in short: "a dynamic operation with a lexically scoped target".
for example, "return" is lexotic. why? because it has a dynamic effect (it peels back the stack), but the thing it's returning *from* (the surrounding routine) is a lexically delimited thing. 21:13
timotimo i think this is the first time i understand lexotic. 21:14
masak the loop controllers like 'next' and 'last' are also lexotic.
timotimo: \o/
timotimo i didn't understand it when someone told me about lexotic take
lizmat masak: patches welcome!
dalek ecs: b24e94b | masak++ | S99-glossary.pod:
[S99] explain lexotic
 21:17
: 9b30933 | (Tobias Leich)++ | lib/Perl5/warnings.pm:
added FATAL
: c2a8429 | (Tobias Leich)++ | lib/Perl5/Terms.pm:
added .P5[ ](int), I hope that it gains speed
: 0e3b756 | (Tobias Leich)++ | / (3 files):
support ${x}-variables and fix for-loop var declaration
sorear FROGGS: have you tried using multi subs instead of methods for the P5stuff? Seems advantageous to not pollute the global namespace 21:22
FROGGS sorear: but I would directly pollute nested perl6 blocks when I'd use subs 21:23
sorear also, compile time multi dispatch with the optimizer 21:24
masak philosophical question: isn't the idea of "methods" -- attaching behavior to (classes of) data -- fairly weird? 21:26
FROGGS I think it more likely that someone would call a sub by accident, then a method starting with P5 on classes like Any
masak: hmm, no? 21:27
[Coke] masak: not after this many years of OO programming, no.
FROGGS attaching behaviour to things it pretty nice, not weird
masak Google Go presents an interesting alternative, if you ask me. 21:28
it doesn't really have methods. it just makes it look like it does.
flussence s/presents/stole from p5/ :P
FROGGS can you explain?
masak FROGGS: best I can do is probably link to some tutorial somewhere that explains it. 21:29
FROGGS hmmm, okay, I can google that by myself :o)
masak FROGGS: here: golang.org/ref/spec#Method_expressions 21:30
note how the methods are outside of the struct.
tadzik because they are not "struct" things
you can define methods on any type
flussence hm, I think D works in a similar way... 21:31
masak right. they are not data.
tadzik like, function types
masak which means that methods are added extensively, like an open set. and so they are in fact more like functions, even if golang calls them methods.
I... I think I like it. 21:32
tadzik I really like Go. And I have the wicked satisfaction, that their module system is light years behind Perl 6
maybe I expect too much of it
masak in what sense is it lightyears behind? 21:33
tadzik it has a list of github repos, nothing else 21:34
timotimo we have a list of github repos and nothing else, too :P
tadzik you're looking for a library? Which one to pick? Well, good luck looking
timotimo they probably even have 100x as many as us? :)
tadzik it's like googling for java libraries, pretty much
timotimo: about 10x :)
timotimo oh, you mean panda
tadzik but we have smoketests 21:35
badges
timotimo indeed we do
tadzik you know what to expect
dalek kudo/nom: 3e58687 | (Elizabeth Mattijsen)++ | src/core/Any.pm:
Make []:(kv|p|k|v) work according to spec
 21:41
lue likes Perl 6 for not having "use Math;" 21:43
lizmat has done enough commits for today, sleep& 21:44
FROGGS gnight lizmat 21:45
timotimo lue: agreed. math is way too useful to push it into some weird namespace
lue good ♞ lizmat o/
timotimo although it would be neat if the compiler could throw away unused things or something
if that even makes sense in the architecture of rakudo
lue
.oO(Strangely, Go also has "use fmt;")
 21:46
timotimo: but how can else can I rely on eval(prompt("Enter a Perl 6 command: ")) ‽
timotimo lue: eval would of course throw that out 21:47
lue
.oO( fine then. qqx"perl6 -e '{prompt("Enter a Perl 6 comand: ")}'" )
 21:48
timotimo qqx is eval, too.
FROGGS nope 21:49
that is shell execute
dalek p: 4bcbdc6 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
More canonicalizations.

Gets us most of S05-mass/properties-general.t (runs all that Rakudo Parrot does, but 4 fail).		 21:50
jnthn Another > 500.
timotimo whoa.
FROGGS cool!
dalek : 3698c7b | (Tobias Leich)++ | / (2 files):
let +@array return the number of elements
 21:57
FROGGS r: say (97700 / 40515) ~ '%'
camelia rakudo 2b77d6: OUTPUT«2.411453%␤»
timotimo pythex.org/ - such a tool for perl6 should be very doable, no? 22:10
(once we get the js backend, even easier!)
[Coke] jnthn: RAWR
jnthn Unfortunately, the remaining one - for derived properties - is kinda harder. 22:15
lue timotimo: emacs comes with one for elisp too (which there is *very* helpful) 22:22
timotimo the coolest way to do such a thing surely is Rakudo::Debugger anyway :) 22:23
lue
.oO(There'd be a lot more modifier buttons under the P6 regex input)
 22:24
timotimo no, there would be not. there would be an input field before the /. 22:26
dalek rl6-roast-data: 91a30ed | coke++ | bin/rakudo.jvm.sh:
ignore memory limits for now
sorear jnthn: howso? 22:33
masak 'night, #perl6
jnthn sorear: They don't appear to be represented directly in any of the built-ins. 22:34
timotimo i'm supposed to be programming something entirely different, but my head is generating lots of ideas for a web frontend to perl6 grammars :| 22:35
hey, how good is perlito6's grammar support?
apparently ... zero 22:39
that's pretty weak if it's supposed to be "finished" :( 22:40
sorear timotimo: pmurias is gsoccing a rakudo-js backend 22:42
if he fails, I'll do it
[Coke] I don't think perlito6 is "finished". where did you see that? 22:43
jnthn
.oO( what *won't* sorear++ do?! :D )
 22:45
timotimo i think i saw it in lizmats talk about the perl6 diaspora?
[Coke] timotimo: I think if perlito6 was finished, I'd be spectesting it.
jnthn heard the same as timotimo somewhere
It's possible we heard it from the same source ;)
sorear [Coke]: do you still spectest pugs? :p 22:47
timotimo i think it was supposed to only cover a limited subset of perl6 and perl5 respectively 22:48
and it reached that
timotimo I'll be bumping to latest pygments very soon, so we'll see this in shortly. 22:52
YYYYYYYEEEEEEAAAAAHHHH
\m/ pÿgṁëṅtṡ.ṙḃ \m/ 22:53
sorear ṗẏġṁėṅṫṡ.ṙḃ? 22:54
flussence
.oO( shouldn't that be :ṙḃ? )
 22:55
timotimo (heavy metal umlauts everywhere!) 22:59
[Coke] sorear: yes. 23:02
I'm trying to get all 4 tests running on the same box.
(pugs, niecza, rakudo, rakudo.jvm) 23:03
colomon Files=739, Tests=24123, 697 wallclock secs ( 4.22 usr 0.80 sys + 20.09 cusr 4.14 csys = 29.25 CPU) 23:10
!!!!!!!
timotimo wait. what. how? 23:13
that's 93%!?!
[Coke] that's tests run, not passed. 23:14
no? 23:15
ah, that's after all jnthn++'s updates? 23:16
r: print 24123-20291 - 23:17
camelia rakudo 3e5868: OUTPUT«===SORRY!===␤Two terms in a row␤at /tmp/8wu2z_AqRf:1␤------> print 24123-20291 -⏏<EOL>␤ expecting any of:␤ postfix␤ infix stopper␤ infix or meta-infix␤»
[Coke] r: print 24123-20291
camelia rakudo 3e5868: OUTPUT«3832»
[Coke] has to be more than that, I think - that's a 3.8K difference in passing tests vs. the run earlier today.
timotimo 3.8K more tests passed? 23:26
[Coke] Based soley on colomon's send above. no clue. 23:29
flussence heh... I just did `make spectest` and it freaked out because I didn't have File::Slurp:from<perl5> installed 23:31
sorear could change that to use only core modules I guess 23:40
sorear uses File::Slurp in everything he writes pretty much
timotimo what's so special about that compared to regular slurp from perl6
flussence ...it runs in perl5?
sorear: don't worry about it - I found it more odd that I *didn't* already have it installed 23:41
timotimo er. 23:42
i meant regular slurp from perl5
are they not the same? 23:43
flussence the main difference is File::Slurp exists and slurp doesn't
sorear timotimo: p5 doesn't *have* a slurp builtin 23:44
timotimo it ... it does not? 23:45
sorear timotimo: you can either use File::Slurp, or you can do the open, $/=undef, readline, close || die $! dance
sorear no, it doesn't. 23:45
timotimo :o 23:49
perl6 is a nice place.
flussence and yeah, those are definitely "tests run" and not "tests passed" - I'm watching a few hundred fails scroll by 23:51
Files=739, Tests=21324, 1317 wallclock secs ( 7.35 usr 0.46 sys + 29.14 cusr 3.25 csys = 40.20 CPU) 23:53
timotimo that's a couple less 23:54
ah, that's the mysterious 3k
flussence my nqp was using whatever commit the parrot build was on, that's probably why 23:55
flussence updates
timotimo maybe 23:57
sorear yeah, there's an nqp fix that allows it to compile very 'long' subs 23:59
it still has issues with subs that have more than about 3,000 local variables

Powered by Cro and the Raku® Programming Language.

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

Thank you!