samcv hmm. i gotta do some more stuff before i get paid. merge the collation stuff. write some documentation. and fix the unicode property name clashes in the database. and Generate Hangul codepoint names algorithmically off the top of my head 00:00
and get the knuth-morris-pratt string search thing merged, tho that wasn't part of the grant, not a bad thing to do 00:01
so fairly close
AlexDaniel samcv++ ? your work 00:06
BenGoldberg Zoffix, is that a pink guitar, or a shiny red one lit with an unfortunately placed camera flash? 01:24
Zoffix Don't know I'm colorblind. The salesman mentioned it being red while picking out the stap tho, so I guess it's red :) 01:26
BenGoldberg Hmm... in that case, is it a dark color, or a light color? Pink just another word for pale red. 01:29
Zoffix dark 01:32
AlexDaniel pink? That's clearly red 01:35
nice red, by the way 01:36
Zoffix Here's a shot with different light: i.imgur.com/S8TV4c9.jpg 01:37
it gets lighter and has a bit of wood pattern visible, closer to the neck 01:38
AlexDaniel Hm, that's more like… maroon?
Zoffix shrugs :) 01:39
ugexe windows failing to build asyncsocket.c 04:25
github.com/MoarVM/MoarVM/commit/5e...d01da8b953 04:26
Geth rakudo/curfs-candidates-refactor: b378a157eb | (Nick Logan)++ (committed using GitHub Web editor) | src/core/CompUnit/Repository/FileSystem.pm
Add candidates method as CURFS recommendation manager

  See #1125 for .candidates details
Creates full Distribution for both -I. and -Ilib, where the later will use the directory recursion it was already doing (to determine id) to automatically populate meta data for provides and files (bin/, resources/).
Makes it possible to do `sub MAIN("version") { $*REPO.need("My::Module").distribution.meta<ver> }` to allow scripts to give the right version regardless of CURI or CURFS being used to load it.		 05:55
rakudo/curfs-candidates-refactor: d0a3fec997 | (Nick Logan)++ (committed using GitHub Web editor) | src/core/CompUnit/Repository/FileSystem.pm
Reimplement files using .candidates
 06:15
nqp: 152094b76b | (Stefan Seifert)++ | 2 files
Add support for %*ENV<NQP_LIB> to MoarVM and JS ModuleLoaders

On the JVM we already support this and this is needed for replacing the hard coded '.' and 'blib' search paths during the build.		 07:26
pmurias nine: you are working on fixing up rakudos build? (I have noticed the nqp commit and I don't want to duplicate work) 08:09
* are you
nine pmurias: I am 08:30
pmurias: already got a fix, but trying to implement a better one 08:31
Geth rakudo/nom: 7e40372456 | (Stefan Seifert)++ | 4 files
Unbreak the build now that nqp no longer searches . and blib by default

nqp commit dd25a924638958a6cdca4c81a0f8798182e52104 removed '.' and 'blib' from nqp's default module search paths. Those were needed only during rakudo's build as after a make install we will find nqp modules in nqp's libpath. This mirrors similar changes to Perl6::ModuleLoader.
Would love to just pass on --nqp-lib to NQP's ModuleLoader, but our command line processing in main.nqp happens only after loading modules alredy.		 08:42
nine Second paragraph explains why my attempts at a better fix just didn't work out :/ 08:43
pmurias nine: the installed perl6-m passes the argument with --libpath directly to the moar executable 09:17
stmuk_ ping 09:21
stmuk something odd on freenode .. seem to be banned on many channels 09:22
yoleaux 16 Aug 2017 21:57Z <AlexDaniel> stmuk: you may be interested in this discussion: irclog.perlgeek.de/perl6-dev/2017-...i_15027543
jnthn stmuk: Yeah, I was klined from the lot overnight 09:24
I'm guessing some accident
stmuk seems like it 09:26
hmm there problem should be a R* security fix release of some sort .. is the nqp execute fix a simple patch or more complex? 09:27
s/problem/probably 09:29
it's just nqp dd25a92 I guess? 09:40
ah no :) 09:43
travis-ci Rakudo build passed. Stefan Seifert 'Unbreak the build now that nqp no longer searches . and blib by default 09:47
travis-ci.org/rakudo/rakudo/builds/265487131 github.com/rakudo/rakudo/compare/3...4037245655
Geth rakudo/atomics: 22e25da2c6 | (Jonathan Worthington)++ | src/core/Atomic.pm
Eliminate Atomic.foo methods for atomic-foo subs.

These will be the "Texas form" of the atomic operations; will also add Unicode operator forms.		 09:57
roast: e3d700548b | (Jonathan Worthington)++ | S17-lowlevel/atomic.t
Follow atomics API changes in Rakudo.
roast: 1ca17ba664 | (Jonathan Worthington)++ | S17-lowlevel/atomic.t
Test return values of atomic integer subs.
 10:05
roast: a921dbb029 | (Jonathan Worthington)++ | S17-lowlevel/atomic-ops.t
Add tests for operator forms of atomic ops.
 10:06
rakudo/atomics: e7e33173a7 | (Jonathan Worthington)++ | 2 files
Add operator forms of atomic operations.
rakudo/atomics: cdbb085852 | (Jonathan Worthington)++ | t/spectest.data
Run S17-lowlevel/atomic-ops.t.
rakudo/atomics: 027e25fb59 | (Jonathan Worthington)++ | 4 files
Rename to reflect no Atomic class any more.
 10:13
nqp/master: 4 commits pushed by (Jonathan Worthington)++ 10:20
¦ nqp/master: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...4-g88c6851
rakudo/nom: 18 commits pushed by (Jonathan Worthington)++
review: github.com/rakudo/rakudo/compare/7...92707fac6c		 10:33
¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....8-gb084530
nqp: 47c4bc4d14 | pmurias++ | 2 files
[js] Throw and exception if we have too many or too few positional args
 11:00
nqp: 6c4770c736 | pmurias++ | t/qast/01-qast.t
Test that QAST::Block.custom_args is respected
stmuk nine: NativeHelpers::Blob doesn't install any more 11:41
nine stmuk: upgrade to rakudo/nom 11:46
Or downgrade to the last release :)
stmuk this is with 2017.07-161-gf1464e2 11:49
nine That's 29 commits behind rakudo/nom 11:50
stmuk I'll try blead 11:51
odd Exception failure with URI2017.07-190-g92707fa 12:01
test failure I mean 12:02
Geth rakudo/nom: a45ede9caa | (Jonathan Worthington)++ | src/core/atomicops.pm
Don't return out-of-atomicint-bounds results.

Ensures that any wrap-around will be reflected.		 13:11
ilmari wouldn't it be more logical to have the atomic pre-inc/dec operators be ++?/--? 13:15
jnthn How so?
brrt these are 'atom' emoji, not sure if you can see 13:16
jnthn (It didn't even occur to me that we might move the location of the atomic sign, tbh...) 13:17
timotimo öyou mean the atom symbol should be between -- and $foo?
brrt $foo??++ we cold also do
ilmari $foo?++ , ++?$foo
reads better to me than ?++$foo
timotimo: yes 13:18
timotimo i tend to agree, it's a little closer to our hyperops for prefix and postfix ops
ilmari the atom should point at the thing that's being atomically modified
brrt oh, you're already using the atomic emoji :-o 13:19
had not seen
wonder what the texas version of that is
ilmari functions
jnthn atomic-inc :)
timotimo i'm glad it shows up in my irc, though it's a little hard to make out
brrt atomic-inc($foo)
like so?
ilmari github.com/perl6/doc/blob/master/d...icint.pod6
brrt oh man, that's going to make for one cool 'neat tricks in perl6' talk in some future perl conference 13:20
ilmari (not on docs.perl.org yet)
s/perl/perl6/
jnthn Yeah, I can buy the hyper op similarity argument
timotimo i'll sell it to you for very cheap
i'm very proud of this product and i'd like it to get some … exposure 13:21
ilmari atomic exposure sounds dangerous
dogbert17 turns out that github.com/MoarVM/MoarVM/issues/554 still causes trouble, updated the issue 13:30
Zoffix haha, we actually used the ?-- as an op? That's so awesome :D 13:33
m: my int $x = 42; say $x?--
camelia 42?
Zoffix m: my int $x = 42; say $x?--; say $x
camelia 42?41?
Zoffix I guess it's Twitter trolling day today :D 13:34
damn, when will Unicode in logs get fixed?
+1 on having prefixes as --?$foo ++?$foo 13:35
$x ?+= 42;... so basically the rule is "atom symbol is closest to the variable the op works on" 13:37
m: my int $x = 42; say ++?$x; say $x 13:38
camelia Cannot resolve caller prefix:<++>(Int); the following candidates?match the type but require mutable arguments:? (Mu:D $a is rw)? (Int:D $a is rw)??The following do not match for other reasons:? (Bool $a is rw)? (Mu:U $a is rw)…
Zoffix hehe
jnthn Yeah, I'll tweak that in a moment 13:39
Just finishing the docs for the scalar case
Zoffix jnthn++
lizmat looking at gfldex.wordpress.com/2017/08/17/pa...g-nothing/ , perhaps we should lift 13:42
9501edae4f73a970e3270e3 until after the release and possibly to 6.d ?
commute&
Zoffix m: say grammar { token TOP { \d+ } }.parse: '4' 13:51
camelia ?4??
Zoffix m: say grammar { token TOP { \d+ } }.parse: ''
camelia 5===SORRY!5=== Error while compiling <anon>?Confused?at <anon>:1?------> 3<BOL>7?5<EOL>?Actually thrown at:? in block <unit> at <tmp> line 1??
Zoffix Confused? What is it confused about?
m: my $x = grammar { token TOP { \d+ } }.parse: 'zzz'; say $x.exception.^name
camelia X::Syntax::Confused?
jnthn huh, it's using the X::Syntax::Confused exception? 13:52
Zoffix That's an LTA exception IMO :\
jnthn Yeah, using an exception intended for the compiler isn't going to work out so well...
Zoffix lizmat, yeah postponing for 6.d makes sense, as I recall a 6.c roast test had to be changed to accomodate the behaviour.
timotimo i'd expect many surprised faces when they get a SORRY! at run time 14:00
Zoffix m: EVAL "x" 14:02
camelia 5===SORRY!5=== Error while compiling /home/camelia/EVAL_0?Undeclared routine:? x used at line 1??
Zoffix :)
jnthn "We're only sorry about compiler errors" :)
Zoffix :)
Geth rakudo/nom: c67d7dd5dc | (Jonathan Worthington)++ | 2 files
Place atomic symbol next to target always.

Based on feedback from ilmari++ and timotimo++.
roast: 266d9e6d57 | (Jonathan Worthington)++ | S17-lowlevel/atomic-ops.t
Chase changes to prefix atomic operators.
 14:03
ilmari jnthn++ # atomic power! 14:07
Zoffix I guess now we just need to make a call to The Church of Atom :) 14:08
oh, Church of the Children of Atom :)
timotimo don't worry, a sheet of paper will protect you from alpha radiation 14:09
Zoffix HeHeHeHe :) 14:10
jnthn Tomorrow and/or Monday I'll see about teaching the JIT about these ops some 14:13
Geth roast: c5587b95ee | (Jonathan Worthington)++ | S06-signature/named-parameters.t
Test to cover RT #131857.
 14:22
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131857
jnthn There goes the blocker. :)
Zoffix \o/ 14:26
jnthn I wonder why I see lower memory use numbers locally than others are reporting 14:27
./perl6-m -e 'sub mem { "/proc/$*PID/statm".IO.lines[0].split(/\s/)[5] * 4096 / 1024 }; say mem(); run "echo"; say mem()'
80624
107844
That a waaay smaller increase
(Than the tickets suggest) 14:28
Zoffix On 2017.07-22-g02667bd I get 59168 / 216484
And on 2017.07-22-g02667bd 32-bit box I get 40796 / 68908 14:29
jnthn 2017.07-191-ga45ede9 here
ggoebel backlogging... I'm surprised no one suggested using ? for atomic :-) 14:43
ilmari ggoebel: uh, that's what is used
ggoebel: see github.com/perl6/doc/blob/master/d...icint.pod6
Zoffix m: my int $x = 42; say $x?++; say $x 14:44
camelia 42?43?
ilmari m: my atomicint $total = 0; start { for ^2000 { $total?++ } } xx 4; say $total 14:45
camelia 570?
ilmari uh?
m: my atomicint $total = 0; await start { for ^2000 { $total?++ } } xx 4; say $total
camelia 8000?
ilmari jnthn: ^^ looks like the docs need some 'await' sprinkling 14:46
m: my atomicint $total = 0; await start { for ^2000 { $total++ } } xx 4; say $total
camelia 7200?
ilmari m: my atomicint $total = 0; await start { for ^2000 { $total++ } } xx 4; say $total
camelia 7128?
Zoffix huh, we got a new type too?
jnthn huh, thought I put the await 14:47
ilmari yes, it's a native int that's of the right size to do cpu-atomic ops on
jnthn Oh, precednece maybe?
ilmari jnthn: only the cas example has await
jnthn m: my atomicint $total = 0; await (start { for ^2000 { $total++ } } xx 4); say $total
camelia 7195?
ilmari not the increment loop
jnthn: that last one is without the atom
jnthn oh
ilmari is your IRC client not rendering it? 14:48
install the latest symbola font
jnthn No, my eyes just imagined it :P
ilmari ah
ggoebel pmurias /o 14:52
JimmyZ jnthn: how about calling it 'atomic' instead of 'atomicint', or will it have something atomicnum ? 15:20
AlexDaniel re revert of 9501edae4f73, see this: irclog.perlgeek.de/perl6-dev/2017-...i_15021994 15:21
ugexe consistency probably... we dont have a unum do we?
AlexDaniel 3 days is definitely not enough for this change, so yeah, not in 2017.08 at least…
timotimo atomint for radiantly fresh breath 15:22
jnthn JimmyZ: atomic underspecifies what it is; it matters that it's an int
Also I'd rather not steal the type name `atomic` :)
And huffman arguments don't really make aint worth it :) 15:25
JimmyZ oh, java has many atomic type: docs.oracle.com/javase/7/docs/api/...mmary.html , haha
jnthn Well, of course they would :P
huh, they have AtomicIntegerArray 15:26
my int @a; atomic-inc(@a[0]) works fine enough for us :)
TimToady huh, so how the heck did I get K-lined 9 or so hours ago? 15:27
yoleaux 13:38Z <gfldex> TimToady: canned worms alert up until irclog.perlgeek.de/perl6/2017-08-17#i_15032122
JimmyZ yeah ,and AtomicLongArray :P
jnthn The Java design for it is consistent with itself and with Java though :)
nine AlexDaniel++ # being strict as release manager
AlexDaniel TimToady: almost everyone did
nine TimToady: Some network issue triggered freenode's spam protection which caused a lot of users to be banned. They are working on the issue 15:28
brrt read 'aint worth it' in the american sense 15:29
JimmyZ to me, atomicint makes me thinking it should have atomicnum too, haha
TimToady .tell gfldex $/ is in fact set by .parse 15:30
yoleaux TimToady: I'll pass your message to gfldex.
brrt but i don't think atomicnum is a thing
jnthn brrt: I...uh...guess cas doesn't care, but yeah :) 15:31
atomic inc/dec/add sure ain't gonna work out on it :)
AlexDaniel releasable6: status
releasable6 AlexDaniel, Next release in 2 days and ?3 hours. No blockers. 164 out of 192 commits logged
AlexDaniel, Details: gist.github.com/61c9c2df1a58bd2e0e...0e6dd2515d
jnthn Also, what would you do with it... :-)
AlexDaniel jnthn: thank you for fixing aliased params
jnthn Sorry for busting 'em :P 15:32
Though to be fair, they were already busted, I guess
I just made optimization more aggressive and then it tripped over the issue
We can actually specialize code like that better at some point in the future
But ain't got time for that right now 15:33
jnthn wonders why he gets sporadic DNS failure lookups of the ir clogs 15:34
Anyway, some time back we got CSV's test-t under 4s thanks to spesh work 15:35
Then I made the million lines benchmark faster. And we were so busy celebrating that we didn't notice test-t went back over 4s again
Anyway, just pushed a commit to MoarVM HEAD that means we can have our cake and eat it. :) 15:36
Zoffix buggable: speed 20 15:37
buggable Zoffix, ???????????????????? data for 2017-08-04–2017-08-12; range: 3.587s–4.925s; 18% slower
Zoffix hehe
jnthn What are the numbers, btw? :) 15:39
(like the 20 in speed 20)
Zoffix "use last 20 entries only"
jnthn Oh 15:40
And speed runs?
Measurement of test-t? :)
raschipi m: '???????????????????'.chars.say # Why 19, then?
camelia 19?
jnthn Oh, maybe it just collects it from the channel :)
Zoffix jnthn: yeah, Tux produces them on some site; each day gets two or so
jnthn Alright 15:41
Well, guess next MoarVM HEAD bump brings it :)
Zoffix raschipi: you missed the outlier. The arrow at the end
raschipi Right, I explicitly deleted it, thinking it wasn't part of the data. The arrow means it's an outlier, I remember now.
AlexDaniel jnthn: re “I can't reproduce such an sizable increase at HEAD or MoarVM/NQP/Rakduo”, so what's your previous number? 15:54
jnthn: or do you want me to test again at HEAD?
jnthn AlexDaniel: Grap MoarVM HEAD also if you do; just added a tweak there that may help some cases 15:56
AlexDaniel mmm that's gonna be more difficult 15:57
AlexDaniel reboot 15:58
jnthn *Grab
Either way, I don't think I'll have time to dig much more into it soon 15:59
AlexDaniel There's now RT #131919 to discuss the .parse change
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131919
jnthn And the longer term fix for things like Proc::Async memory use will be to have it spawn less threads
Which needs an overhaul of ThreadPoolScheduler
AlexDaniel .tell Skarsnik by the way, does RT #131914 affect you in any way, or did you notice it by chance? 16:01
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131914
yoleaux AlexDaniel: I'll pass your message to Skarsnik.
AlexDaniel jnthn: alright, thanks! 16:02
TimToady sure would be nice if there were a way to test the whole ecosystem for a given change... 16:03
ugexe we kinda can
well, Zoffix kinda can :)
AlexDaniel TimToady: we had toaster, but it doesn't seem to work right
right now* 16:04
:)
Zoffix Well, it can work if you run the toaster itself on an older (say 2017.07) Rakudo
I'll try doing that 16:05
TimToady: but yeah, there IS a way :) rakudo.party/post/Perl-6-Release-Q...em-Toaster
TimToady 'cuz it really makes a difference whether a change affects 1 module or 3 modules or 30 modules or 300 modules 16:07
if it's 1 module, a revert is overkill 16:08
if it's 300 modules, then it's 6d
most of the middle numbers mean wait a month
Zoffix But we made a promise to users we won't make any changes that break 6.c-errata.
in 6.c
AlexDaniel “wait a month” is not a bad strategy in cases when we don't know 16:09
TimToady Zoffix: sure, but it's a bit of a judgement call whether 6.c-errata is testing something that wasn't intended to be tested 16:11
b2gills my personal take on that code is it was slightly wrong anyway, it was looping once over Any but assumed it was an empty list
TimToady one could argue that the change alread caught a bug, yes 16:12
b2gills sort of like all the bugs that 5.18 dredged up because of the randomized sort order 16:14
sort order of hashes 16:15
TimToady so as usual, we have to pick somewhere on the slippery going from "Damn the torpedoes, full speed ahead" through "No harm, no foul" all the way down to "If it ain't broke, don't fix it."
we've done a lot things in 6.c that fall somewhere on that slope, because otherwise 6.d would have lots more bugs when it comes out :)
b2gills I still think we need to publish 6c-errata as v6.c.1 at some point 16:16
Zoffix It would be nice to fight against the perception that we rip the floor from under our users all the time. Maybe start aiming to confine these changes to some particular releases, where we can say: look we changed untested stuff that might break your code.
Releases like, the one right after R* release.
TimToady didn't we just do an R* release? :P 16:17
Zoffix Yup :)
TimToady I love someone who can argue both sides of a problem :)
Zoffix heh 16:18
lucasb github.com/rakudo/rakudo/blob/nom/...#L202-L252 16:20
^^ Why there isn't a .Complex coercion method in Cool? 16:21
(sorry to bother here, #perl6 is busy in a linenoise discussino :)
Skarsnik jnthn, for #131914 did you check your pagesize? (getpagesize()) maybe the 4096 in the calcul need to be changed if it's not 4096
synopsebot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=131914
Zoffix m: "1+i".Numeric 16:22
camelia Cannot convert string to number: base-10 number must begin with valid digits or '.' in '031+?5i' (indicated by ?)? in block <unit> at <tmp> line 1??
Zoffix m: "1+1i".Numeric
camelia ( no output )
Zoffix lucasb: seems just an omission
lucasb: PR it :) With tests :)
lucasb Zoffix: nice, I'll try 16:23
Zoffix lucasb: "sorry to bother here"... IMO here *is* the right place for stuff like that. 16:24
Zoffix doesn't usually follow #perl6; too much traffic to take in 16:25
Skarsnik it's a good thing :)
Zoffix Yeah :)
Skarsnik hm 16:39
running a run 'echo', '-n' 20 times growns to 350Mbyte. if I do more it kinda stay the same 16:44
[Coke] There's a ticket. 16:46
Zoffix Ohhhh. mystery solved on why Toaster crashes on bleedier rakudos. Messages in screen log: "not enough memory" 16:47
Skarsnik yeah but it was for one run. I was curious if the memory usage grows or it was just one overhead of having Proc being Proc::async now
I ran against a 2017.06 version 16:48
I should look more at what is exactly the value given by statm 16:49
[Coke] rt.perl.org/Ticket/Display.html?id=131915 16:54
AlexDaniel (it's his ticket)
[Coke] OK. 16:56
Skarsnik hm so the data field (the field used in this) is for memory allocated by the process. but it not the memory used. if I understand correctly 17:00
Dunno why it's called data + stack 17:01
AlexDaniel Skarsnik: should we test it again with resident (.[1]) ? 17:12
uh, I wanted to reboot 17:13
Skarsnik not sure 17:16
jnthn Skarsnik: No, but the numbers that I saw correlated with the numbers from /usr/bin/time's maxrss 17:21
AlexDaniel ¦92bd7e4f54a9^: «72632» ¦92bd7e4: «72536» ¦9658dd98c9dd8ec^^: «78084» ¦9658dd9: «78068» 17:22
that's with say "/proc/$*PID/statm".IO.lines[0].split(/\s/)[1] * 4096 / 1024 17:23
so what is “data + stack” exactly?
Skarsnik resident is supposed to be data + code 17:24
but it's weird
jnthn Stack is presumably the C stack where function's invocation records live
It'll typically be around a meg per thread 17:25
Skarsnik well if it's stack that get that big it's still not fun ^^
jnthn Well, the thread pool will allow up to 16 threads, then there's the initial thread, and one created for running the specializer, so that's 18MB at most 17:26
Each thread then has a thread-local nursery 17:27
Which is 4MB semispaces, so 8MB total
m: say 9 * 18
camelia 162?
jnthn The thread pool scheduler is fairly un-smart, though 17:28
It just spawns threads up to the limit each time there's work
It thus can end up with more than it really needs
Skarsnik I wonder if there is a good api/lib somewhere to get clear information 17:33
Zoffix hmmm, a 2017.07 toaster running a toast of 2017.07 release died due to not enough memory :( looks like it needs some sort of memory protection against rogue modules 17:38
Zoffix changes from 24GB to 100GB RAM and tries again 17:39
Skarsnik lol 17:40
jnthn I'd imagine the toaster itself was using Proc::Async before, though, so I don't know it woulda ended up with extra use? But yeah, a rogue module could get hungry :) 17:41
Guess can run it with ulimit or some such
ugexe zef uses proc 17:42
Zoffix Yeah, it always used Proc::Async and last release I toasted on near-2017.07 and everything worked. I'd guess it's some module 17:43
Skarsnik Writing a binding to libxml2 with gptrixie to use it in gptrixie is interesting x) 17:44
Geth rakudo: lucasbuchala++ created pull request #1131:
Add .Complex coercion method to Cool		 17:47
rakudo/nom: c9de2184dc | (Lucas Buchala)++ | src/core/Cool.pm
Add .Complex coercion method to Cool
 17:50
rakudo/nom: a30ce6bca8 | (Zoffix Znet)++ (committed using GitHub Web editor) | src/core/Cool.pm
Merge pull request #1131 from lucasbuchala/coomplex

Add .Complex coercion method to Cool
lucasb Zoffix: thank you 17:51
Zoffix lucasb++ thank you! 17:52
TimToady apparently :dba('foo') has been broken in user-level grammars since forever...it parses it but never does calls reduce on quote_EXPR for some reason 17:58
Zoffix hehe, URI.pm fails its tests do to .parse changes 18:09
(it's used by Toaster and failed install :)
s/do/due/; 18:10
bah. Toaster itself doesn't seem to want to start. Says "Confused"; guessing the X::Syntax::Confused from .parse somewhere :/ 18:16
Skarsnik hm I am curious what should be expected with that 18:18
m: my $hash = {a => "b"}; class FHash { method AT-KEY($key) {"b"}}; say $hash<a>:exists; my FHash $fh .= new; say $fh<a>:exists;
camelia True?False?
Zoffix AlexDaniel: I give up again. 18:20
But I'd argue the fact that I can't even start toasting when running it on HEAD commit due is an indication the .parse() change has large impact.
timotimo Skarsnik: if you don't implement EXISTS-KEY, you'll get that :)
Zoffix Toaster doesn't even do much networking.
Skarsnik Haaa
is there the same for defined-keys? 18:21
timotimo defined-keys?
Skarsnik m: my $hash = {a => "b"}; class FHash { method AT-KEY($key) {"b"}}; say $hash<a>:defined; my FHash $fh .= new; say $fh<a>:defined;
camelia Unexpected named argument 'defined' passed? in block <unit> at <tmp> line 1??
timotimo i don't think that's a thing
Skarsnik m: my $hash = {a => "b"}; class FHash { method AT-KEY($key) {"b"}}; say $hash<a>:defined;;
camelia Unexpected named argument 'defined' passed? in block <unit> at <tmp> line 1??
Zoffix It's .defined
jnthn TimToady: I had :dba in Rakudo as an NYI, fwiw 18:22
timotimo you had me at "NYI" 18:24
Zoffix But if anyone wants to run Toaster, just grab a brand new debian ISO, run it in some VM, and just copy-paste these instructions: github.com/zoffixznet/perl6-Toaste...e-vm-setup 18:27
You don't need a 64-core box :) You'd just have to wait longer. And I don't have the patience to do these builds that seem to crash in the middle of it :/
And on HEAD Toaster doesn't start. Says confused pointing to "Connection: closed", which I'm guessing some Grammar.parse failure in URI or some networking module 18:28
TimToady jnthn: but it just calls down into nqp's regex engine to parse, so it should at least call reduce on quote_EXPR, unless it's misparsing somehow...hmm.... 18:35
jnthn Hm, true 18:36
AlexDaniel Zoffix: it's ok, thank you for your effort. I simply didn't have enough time to get my hands on it, but will do after release
TimToady it IS misparsing! 18:37
m: say "abc" ~~ /:dba("foo") abc/
camelia Nil?
TimToady m: say "abc" ~~ /:dba("") abc/
camelia ?abc?? 0 => ???
TimToady m: say "abc" ~~ /:dba("abc")/ 18:38
camelia ?abc?? 0 => ?abc??
TimToady that would definitely explain why it's not reducing the argument :) 18:39
jnthn lol
Geth nqp: f48e972ad3 | (Robert Lemmen)++ | 3 files
support hardware_concurrency op from moar and implement for jvm
 18:43
nqp: 5ab70b5834 | (Robert Lemmen)++ | 3 files
match renamed opcode "cpucores" in moar
nqp: fd1fefaff5 | niner++ (committed using GitHub Web editor) | 3 files
Merge pull request #362 from robertlemmen/master

support hardware_concurrency op from moar and implement for jvm
rakudo/curfs-candidates-refactor: d533d450da | (Nick Logan)++ (committed using GitHub Web editor) | src/core/CompUnit/Repository/FileSystem.pm
Fix CURFS.id when files are added after first load

Fixes spectest failures from caching too aggressively.
This also returns to using a cached `$!id`		 18:48
rakudo: ugexe++ created pull request #1132:
Add candidates method as CURFS recommendation manager		 19:14
Zoffix .in 1hr check on Toast 19:44
yoleaux Zoffix: I'll remind you at 20:44Z
Zoffix relocates
Skarsnik interesting feature ^^ 19:45
nine 5.898s for 100000 rows! 20:07
Less than a year ago it was ~9 seconds for just 10000 rows, i.e. 1/10 of the iterations :) 20:13
brrt badass
Skarsnik and I think xml parsing take half the time than last year or something like that 20:14
TimToady jnthn: I think I figgered it out: rakudo's regex grammar mixes in STD, which defines starter/stopper as <!> rather than using dynvars, so quote_EXPR will never match 20:43
yoleaux Zoffix: check on Toast 20:44
Skarsnik yeah burned toast is not good 20:55
samcv releasable6, status 21:17
releasable6 samcv, Next release in 1 day and ?21 hours. 1 blocker. 164 out of 194 commits logged
samcv, Details: gist.github.com/7b517b4462b475f813...e801fa20ef
jnthn TimToady: Yeah, quote_EXPR is kinda the NQP way of parsing quotes; Rakudo started out that way but took on STD's Q language approach 21:23
TimToady changing nqp's starter/stopper rules to dynstarter/dynstopper makes :dba work in rakudo 21:28
Zoffix greppable6: 030_default-log-level.t
greppable6 Zoffix, Found nothing!
Zoffix AlexDaniel: feature request: ability to find to which distro a filename belongs too; aka tree -f | grep XXX :) 21:29
also 020_ff.t 21:30
jnthn TimToady: Ah, cool :)
Zoffix phew killed them in a nick of time. The 030_default-log-level was using 50GB of RAM and 020_ff.t had two instances using 20GB each. 21:31
jnthn o.O
Zoffix (nick being the VM having only 100GB of RAM and it was about to kill the Toast run :))
Note: this is on 2017.07, so I assume the problem is in module(s) 21:32
TimToady jnthn: though nqp's idea of :dba is a bit too dynamic, insofar as it only lists the alternatives you actually tried after filtering through LTM, not all the possible alternatives at that spt
Zoffix ./moznion/p6-Log-Minimal/t/030_default-log-level.t
./moznion/p6-Log-Minimal/t/020_ff.t
buggable: eco Minimal 21:33
buggable Zoffix, Found 11 results: JSON::Tiny, App::Mi6, Frinfon, Log::Minimal, PDF::Lite. See modules.perl6.org/s/Minimal
Zoffix Hm, nope. No changes in the module's code since ages ago...
jnthn TimToady: Oh, interesting... :) 21:34
TimToady: So maybe fixing that gets us better errors, or more spurious suggestions? :)
Zoffix Opened an Issue on author's module: github.com/moznion/p6-Log-Minimal/issues/8 21:39
TimToady also, doing it dynamically probably recalculates the same list over and over, so maybe we could find a good peg to hang it on and calculate expectations once per set 21:41
Zoffix stmuk: re banned on Freenode: just got a user notice saying they're battling some massive spam so they kline was temporary
TimToady as for spurious, I'd rather list them all and then allow people to specify :dba('') to suppress the noisy ones
jnthn True 21:43
geekosaur the new blog says they had a bot go slightly haywire and try to kline all active users 21:44
*news blog
Skarsnik wtf are these VM with 100Gb? 21:45
Zoffix So the robot uprising begins?
Skarsnik: Google Compute Engine. 100GB is nothing. Here's my current maximum: i.imgur.com/ceRzOZm.png 21:46
bah
.tell Skarsnik RE wtf are these VM with 100GB.... It's Google Compute Engine. 100GB is nothing. Here's my current maximum: i.imgur.com/ceRzOZm.png And I've seen news they plan to expand it to 128-core VMs (presumably with even more RAM) some time soonish 21:47
yoleaux Zoffix: I'll pass your message to Skarsnik.
Geth rakudo/curfs-candidates-refactor: 877e7dc0ca | (Nick Logan)++ (committed using GitHub Web editor) | src/core/CompUnit/Repository/FileSystem.pm
Use version * for -Ilib; populate resources field
 21:50
AlexDaniel Zoffix: github.com/perl6/whateverable/issues/210 21:53
Zoffix cool 21:57
stmuk wibble
Zoffix oh crap. This was actually a HEAD run, not 2017.07 (and 2017.07 finished without issues apparently) 22:05
holy shit 22:08
buggable: toast
buggable Zoffix, Between 2017.07-194-ga30ce6b and 2017.07: 48 (5.60%) modules got burnt; 55 (6.42%) got unsucced; 367 (42.82%) out of 857 modules appear unusable. See toast.perl6.party/ for details.
Zoffix AlexDaniel: ^ lots of burnage
I'm hoping those results aren't bogus due to some issues in Toaster itself 22:09
Half the ecosystem is unusable :/
AlexDaniel “Aborting due to fetch failure” ?
Zoffix Which one is it? 22:10
Ah 22:11
It's the version thing, innit?
nope
jnthn I've looked at 5 so far and they are all that
Zoffix No idea why. The fetch failure is likely an error from zef
Zoffix upgrades rakudo locally and tries to install Test::META 22:12
jnthn Also Can't extract non-existent path: /home/cpan/.zef/tmp/File-Find.git
ugexe sounds like rate limiting to me 22:23
it could also be that the toaster is using multiple zef processes, all of which are using the same tmp/ 22:32
Zoffix ugexe: then that is it. The toaster is using like 30 zef processes and if it's reporting ~/.zef/tmp then they're all saving there; I guess only --install-to is different 22:34
Yeah, no issues installing Test::META locally. :( 22:35
ugexe github.com/ugexe/zef/blob/33c18c90...t.pm6#L277 (if you comment this out, I think it'll work again) 22:38
it looks like tacking on a random directory in tmp/ works, which should fix it. testing that right now 22:52
Zoffix huggable: hymn of the butterfly :is: www.youtube.com/watch?v=aPy0hAmHu90 22:58
huggable Zoffix, Added hymn of the butterfly as www.youtube.com/watch?v=aPy0hAmHu90
Zoffix \m/
Zoffix calls it a night
Gonna try to re-run Toast on HEAD some time tomorrow. 22:59
ugexe i'll have a workaround in zef by then
Zoffix ugexe++ thanks
ugexe I think I fixed the localcache to be multi zef friendly too... so might be worth trying without disabling it 23:00
suspected fix in 0.1.27 23:25

Powered by Cro and the Raku® Programming Language.

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

Thank you!