»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
00:00 BenGoldberg joined 00:02 cognominal left 00:04 leont left
dalek kudo-star-daily: 2aa91d2 | coke++ | log/ (8 files):
today (automated commit)
 00:04
00:05 aborazmeh joined, aborazmeh left, aborazmeh joined 00:11 ShimmerFairy left 00:12 n0tjack joined 00:22 n0tjack left 00:25 ShimmerFairy joined 00:36 Skarsnik left 00:39 laouji left 00:40 laouji joined 00:41 laouji left, laouji joined 00:46 Zoffix joined 00:50 tokuhiro_ joined
tokuhiro_ how do i pass parameters by reference with nativecall? 00:52
i want to write code like python's byref docs.python.org/2/library/ctypes.h...-reference 00:53
psch tokuhiro_: a thing that used to work is passing an Array[CPointer] 00:54
00:55 rangerprice left
psch tokuhiro_: but i wouldn't bet on me being up to date on current practices 00:55
ugexe CArray[OpaquePointer] 00:57
psch .tell FROGGS your patch and the result makes me thing the jvm build is due to some oversight in the Binder, though i haven't found a place to place a BP to gain more insight, just fyi
yoleaux psch: I'll pass your message to FROGGS.
psch ugexe++: right, that was probably it, thanks
tokuhiro_ thanks. i'll try it 00:58
01:00 lizmat joined 01:04 lizmat left 01:15 n0tjack joined 01:23 n0tjack left 01:24 aborazmeh left
labster m: my $t1 = now; for ^1000 { <1/65536>.perl }; $t1 = now - $t1; my $t2 = now; for ^1000 { <1/65537>.perl }; $t2 = now - $t2; say "<1/65537> is {sprintf("%.2f", $t1/$t2)} times faster than <1/65536>"; 01:38
camelia rakudo-moar 16d366: OUTPUT«<1/65537> is 16.32 times faster than <1/65536>␤»
timotimo m: say <1/65536>.perl; say <1/65537>.perl 01:43
camelia rakudo-moar 16d366: OUTPUT«0.000015␤<1/65537>␤»
timotimo wow, is our algorithm to turn it into a decimal really that bad? 01:44
labster yeah, it's pretty awful. It's looping on $!denominator %% 2
timotimo oh 01:45
so it's basically O(number of 1 bits in denom)?
labster Yes. But then after all of that work, Rational.Str doesn't know how to handle it anyway.
timotimo right, it fails to find something good 01:46
and finding out it'd not work does the maximum amount of work first
perhaps we can count the number of bits set up front and decide based on that that we'd end up bailing out anyway?
labster no, it finds out it will work incorrectly
I'm tempted to try with base 10 instead of 2 and 5, it might actually work. 01:48
timotimo er, right
labster Mainly because Rational.Str rounds at min($denominator.chars + 1, 6) 01:49
er, max(). Not written like that anyway 01:50
timotimo i'm not familiar with the code, actually, so i was just hoping i could inspire you to build something %) 01:51
but you already had that in mind
labster I'm inspired to build, just don't know what to build :P 01:55
timotimo are you "limited" to the rat-to-string thing in this desire/inspiration? 01:56
02:03 noganex_ joined 02:05 dayangkun joined 02:08 FROGGS joined 02:19 FROGGS left 02:22 AlexDaniel left 02:28 lea joined 02:31 ab5tract_ joined
labster timotimo: I think my imagination is less limited after dinner 02:32
timotimo is very late for bedtime :| 02:33
02:35 yqt left 02:36 ab5tract_ left
dalek ecs: 653f803 | skids++ | S32-setting-library/Containers.pod:
Adjust some prototypes to post-GLR implementation state.

Also, write better natural-language descriptions of some methods, since the pseudocode had gone stale anyway.
Remove map(*) as it was removed during GLR.
Grammar and whitespace cleanups
Still much work to do on S32, and sub-style slurpies have yet to be changed.		 02:41
TimToady --> airport & 02:48
02:58 n0tjack joined 03:03 n0tjack left 03:04 kaare_ joined 03:07 dayangkun left 03:12 dayangkun joined 03:42 n0tjack joined 03:47 n0tjack left 03:50 laouji left 04:29 khw left 04:35 FROGGS joined 04:38 laouji joined
moritz
.oO( TimToady returns an airport )
 04:44
good morning * 04:45
04:45 FROGGS left 04:51 BenGoldberg left 04:52 aborazmeh joined, aborazmeh left, aborazmeh joined, FROGGS joined 04:53 xinming left 04:54 xinming joined
tadzik g'morning moritz 04:57
yoleaux 7 Sep 2015 22:18Z <mohij> tadzik: I'm more or less done with putting *env functionality into rakudobrew. Most commands work including local, global and shell switching.
7 Sep 2015 22:19Z <mohij> tadzik: I have only tested on linux so far. Will do some Windows testing soon. What do you think about merging?
7 Sep 2015 22:20Z <mohij> tadzik: Tradeoff: Slight startup time decrease because the shims are perl and not bash anymore.
04:58 diana_olhovik_ joined, laouji left 04:59 laouji joined
tadzik .tell mohij I'd be delighted :) Can you send me a PR? 05:01
yoleaux tadzik: I'll pass your message to mohij.
tadzik thanks yolo
FROGGS[mobile] tadzik: there is also a message by moritz in the clogs 05:03
05:03 diana_olhovik___ joined
tadzik oh 05:04
05:04 diana_olhovik_ left
tadzik FROGGS[mobile]: I don't see it 05:06
FROGGS[mobile] something about rakudobrew
moritz unlikely 05:07
FROGGS tadzik: ups, wasnt moritz: irclog.perlgeek.de/perl6/2015-09-07#i_11181969
yoleaux 00:57Z <psch> FROGGS: your patch and the result makes me thing the jvm build is due to some oversight in the Binder, though i haven't found a place to place a BP to gain more insight, just fyi
FROGGS tadzik: and now I also see that mohij did send it again properly
sorry, still being asleep obviously 05:08
.tell psch thanks for the hint... now I just need to figure out what that means :o)
yoleaux FROGGS: I'll pass your message to psch.
05:11 diana_olhovik___ left
moritz iirc jnthn++ fixed some binder things in JVM 05:13
ShimmerFairy Yeah, I last saw jnthn mention that his JVM hacking should make finding the build issue less scary now for others, or something to that effect :) 05:22
05:29 quester joined, quester left, quester joined, [Sno] joined
quester m: my @a="i i", "j j"; my @b = @a, "k k", "l l", "m m"; say @b.perl 05:31
camelia rakudo-moar 16d366: OUTPUT«[["i i", "j j"], "k k", "l l", "m m"]␤»
quester m: my @a="i i", "j j"; my @b = | @a, "k k", "l l", "m m"; say @b.perl
camelia rakudo-moar 16d366: OUTPUT«["i i", "j j", "k k", "l l", "m m"]␤»
quester I'm fairly impressed... I have about 180 lines of perl6 code written six months ago for $work, and the only thing that the GLR broke was the declaration of a list of keywords. Kudos to everyone involved, jnthn++ in particular. 05:35
Oh, and TimToady++ of course. 05:36
tadzik yay, another production perl 6 :) 05:40
05:42 dayangkun left 05:43 dayangkun joined 05:44 dayangkun left 05:45 dayangkun joined, dayangkun left 05:47 dayangkun joined 05:49 n0tjack joined 05:51 domidumont joined 05:53 laouji left, n0tjack left 05:55 domidumont left 05:56 domidumont joined 05:58 laouji joined, laouji left 05:59 laouji joined 06:14 araujo left 06:15 araujo joined 06:16 araujo left 06:17 araujo joined 06:18 CQ joined 06:24 FROGGS left
tadzik gist.github.com/anonymous/c1b0012f2dd4d3e4ca0e 06:24
...huh?
the :in(True) is the offender
for some reason 06:25
FROGGS[mobile] tadzik: what should it do instead? block for input? 06:35
06:46 laouji left
tadzik FROGGS[mobile]: I guess so 06:46
not bail out would be a good start :)
06:46 laouji joined
tadzik I mean, it's a minimized test case from a code that launches dmenu, writes to its input and reads from its output 06:47
this is the simplest I could come up with that reacts the same way
06:48 FROGGS joined 06:49 rurban joined, Ven joined
FROGGS tadzik: that script does not fail for me 06:50
06:51 FROGGS[mobile] left 06:56 leont joined 06:59 laouji left 07:00 laouji joined 07:10 ab5tract_ joined 07:12 leont left 07:18 rurban left, Loren joined
masak morning, #perl6 07:19
FROGGS morning masak
Loren morning & afternoon. 07:20
07:20 darutoko joined
masak Loren: afternoon, Loren of .cn 07:21
Loren haha..
m: class Line { has $!value; method get-value() { return $!value; } } my Line $line; say $line.perl; say $line.get-value(); 07:24
camelia rakudo-moar 16d366: OUTPUT«5===SORRY!5=== Error while compiling /tmp/yvjh0B5Asv␤Strange text after block (missing semicolon or comma?)␤at /tmp/yvjh0B5Asv:1␤------> 3method get-value() { return $!value; } }7⏏5 my Line $line; say $line.perl; say $lin␤ expecting an…»
Loren why '$line' can access $!value ? In my perl6 it print 'Cannot look up attributes in a type object'. 07:27
FROGGS m: class Line { has $!value; method get-value() { return $!value; } }; my Line $line; say $line.perl; say $line.get-value();
camelia rakudo-moar 16d366: OUTPUT«Line␤Cannot look up attributes in a type object␤ in method get-value at /tmp/IcjD82gF79:1␤ in block <unit> at /tmp/IcjD82gF79:1␤␤»
FROGGS m: class Line { has $!value; method get-value() { return $!value; } }; my Line $line; say $line.perl; say try $line.get-value(); say try $line.new.get-value(); 07:28
camelia rakudo-moar 16d366: OUTPUT«Line␤Nil␤(Any)␤»
FROGGS Loren: attributes are to be used with class instances, so you need to instantiate first
Loren so the 'prototype-based OO programming' is not support now ? 07:29
moritz you can have class attributes too 07:30
FROGGS m: class Line { my $value; method get-value() { return $value; } } my Line $line; say $line.perl; say $line.get-value(); 07:31
camelia rakudo-moar 16d366: OUTPUT«5===SORRY!5=== Error while compiling /tmp/exnwBEQ45H␤Strange text after block (missing semicolon or comma?)␤at /tmp/exnwBEQ45H:1␤------> 3 method get-value() { return $value; } }7⏏5 my Line $line; say $line.perl; say $lin␤ expecting an…»
FROGGS m: class Line { my $value; method get-value() { return $value; } }; my Line $line; say $line.perl; say $line.get-value();
camelia rakudo-moar 16d366: OUTPUT«Line␤(Any)␤»
FROGGS m: class Line { my $value = 42; method get-value() { return $value; } }; my Line $line; say $line.perl; say $line.get-value();
camelia rakudo-moar 16d366: OUTPUT«Line␤42␤»
moritz Loren: there's not much support for prototype-based programming in Perl 6; but the meta object protocol is powerful enough that you could well implement it on your own
Loren camelia, thks
07:32 ab5tract_ left
Loren moritz, thks, now i know. 07:32
07:33 zakharyas joined
moritz hasn't the slightest idea why people would favor prototype-based OO 07:35
07:36 Sqirrel left, Sqirrel_ joined
Loren It seems like better than 'class-based'. 07:36
ShimmerFairy moritz: I'm looking now, and I'm not sure I get what it even is. 07:37
masak ShimmerFairy: may I recommend Steve Yegge's blog post on the matter? 07:38
moritz Loren: better in what way?
ShimmerFairy masak: sure :)
masak moritz: it's not clear to me where protoype-based OO shines. I can use it, but it feels neither better nor worse than class inheritance.
moritz masak: my impression was that most JS OO patterns revolve around emulating class-based OO 07:39
masak only about half of them.
moritz: nowadays when I write stuff in JavaScript, I tend to favor things like Object.extend when creating objects. kind of leaning towards mixins.
ShimmerFairy masak: from what I've seen so far, Perl 6 makes the issues supposedly solved by prototype-based OO a non-issue (but to be fair I'm not sure I get it yet)
moritz ... and about another quarter evolve around the absence of block scoping 07:40
Loren moritz, May be less coding work
moritz Loren: I haven't seen any evidence (or even examples) that support this claim
masak ShimmerFairy: steve-yegge.blogspot.se/2008/10/uni...ttern.html 07:41
moritz: no, I meant that about half of the JS code I see that uses OO, seems to want to first build an insulation layer of class-based OO on top of the prototype system. 07:42
moritz: I take it to be mostly an unwillingness to adapt to prototype-based OO, and a belief that class-based OO works better for what they want to do.
07:43 skids left
moritz masak: I think there's a real lack of educational material on how to design software with prototype-based OO 07:43
masak and while I personally don't have an urge to layer class-based OO on top of what JS already has, I can kind of sympathise with the instinct to put something familiar in place of something unfamiliar.
moritz: yes, that's part of why I like Yegge's post. 07:44
it's definitely a different mode of thinking. in many cases, things get slightly easier.
ShimmerFairy "it's not especially well liked as an academic topic. This is because OO design has no real mathematical foundation to support it" oh no. the horror. save me. aaah.
masak but (as @getify points out), the really treacherous thing is that class inheritance is fundamentally based on *copying* whereas prototype OO is fundamentally based on *linking*. 07:45
Loren it's very convenient in sometimes
masak ShimmerFairy: I could imagine a world where OO had a solid theoretical foundation. I wouldn't mind that.
ShimmerFairy masak: I'm just pointing out that I don't care for the mathematical view on programming. Knuth's DLX paper was the best thing I found on DLX, but that's in spite of the mathematical notation, and if I ever write about NFAs I'll stay clear from the utterly useless mathematical definition and jargon :) 07:46
masak ShimmerFairy: I couldn't disagree more.
it's far too easy to bash theory, not knowing or caring what it has done for computing. 07:47
on the other hand, the fact that people can use all the fruits of theoretical work and be *oblivious* is also a kind of strength, I guess. 07:48
ShimmerFairy masak: I know the mathy bits are important, but for someone like me, I know programming through programming, not through math; mathematical notation and jargon when describing computer concepts is more than unhelpful to me.
moritz masak: I agree with your on the importance of mathematics; I also agree with ShimmerFairy that many theory articles could benefit from less jargon/notation 07:49
masak fair enough.
sometimes it seems that some articles introduce jargon/notation for the *authors*, not for the *readers* 07:50
it's kind of the same issue as with readable source code
ShimmerFairy On the subject of NFAs, it's looking at diagrams and reading through walkthroughs of them that helped me get a grip on them, not this "five-tuple" crap :P
07:51 leont joined, n0tjack joined
moritz ShimmerFairy: a good theory teaches both 07:53
ShimmerFairy: try writing a sound proof while talking about walking through diagrams... 07:54
masak it's also a matter of what level you're getting in on. a really soft and furry tutorial will help a beginner. the five-tuple crap helps me, because I have a bit of math background already.
ShimmerFairy moritz: my main problem with mathematically-bent computer articles is that I'm looking up stuff so I know how to code it, and the mathematical view on programming is just about the most unhelpful thing in the world for that goal (for me at least). I need discussion with snippets of code, not with obscure mathematical expressions.
moritz ShimmerFairy: so you should be looking for cookbooks, not theory articles 07:55
masak I agree with moritz that it's probably about most theory being presented badly.
nine I think, I'll rather read the book the blog post mentioned. Cannot be that much longer anyway. (Hint: if you have trouble with your blog posts becoming too long, don't repeat yourself endlessly and don't write about how your blog posts become too long...)
masak this is especially clear when reading academics who can write, such as SPJ or Eric Meier.
07:55 n0tjack left
ShimmerFairy moritz: when trying to look up NFAs and DFAs for learning regexes, I only found math-based articles. It seemed like no-one can resist the high-level, computer science stuff :/ 07:56
[Tux] test 50000 53.736 53.653 07:57
test-t 50000 55.116 55.032
moritz ShimmerFairy: have you read "Mastering Regular Expressions" by Friedl? 07:58
ShimmerFairy If it's a book, then I surely haven't :) 07:59
07:59 espadrine joined
moritz well, it's rather good for learning regexes :-) 07:59
masak ShimmerFairy: it's a bit like saying "when I tried to learn French, people kept bothering me with details about syntax and grammar :/"
moritz and yes, it's a book
masak "I don't want to learn grammar, I just want to learn French!"
ShimmerFairy masak: well, the issue is that I don't know computer science math at all. I want to learn how regexes work, not the entire foundational base of CS math :) 08:00
Loren I don't want to learn En grammar, i just want to learn english.
masak Loren: :D
Loren masak, :p 08:01
masak well, I wish you both the best of luck. (meant non-ironically.)
nine masak: the irony is, that you don't actually have to know Grammar to learn a language. I know practically no rules for English grammar, I just happen to have read enough English to get it right intuitively most of the time.
masak nine: I know, I know. 08:02
nine Though I somtimes wish, I knew more :)
masak of course you can learn a language without ever talking about its grammar. and of course many people do that, every day.
kind of a "learning by attrition".
grondilu yes but language teacher must justify their job. If people were learning languages as toddlers do, there would be no need for teachers. 08:03
masak "grammar" was never a bad word to me. but at some point I realized that it was a bad word for others, kind of like "math", because they didn't see the point of it and it was just random rules to them.
nine Well knowing the rules can greatly speed up the process.
masak the reason we talk so much about grammar is that it's an unparalleled way to structure ideas about language and structures. 08:04
just like math is.
ShimmerFairy masak: I think I'd like it if CS math was written and talked about in terms of programming languages; I can't help but think traditional mathematical notation/jargon is completely inappropriate for programming :)
TEttinger ShimmerFairy: ... what's hard about five-tuple? are non-native english speakers expected to know every prefix for triplet, quartet, quintet, sextet, septet, etc.?
ShimmerFairy TEttinger: it's not the word, but the thing itself (which I couldn't be bothered to find back) 08:05
TEttinger tuple?
nine You can learn a language's grammar by reading hundreds of books and watching TV for thousands of hours, or you can learn it by studying a grammar book. Of course latter is less entertaining, but definitely quicker.
ShimmerFairy TEttinger: no, the actual 5-tuple that's discussed when it comes to NFA/DFA 08:06
TEttinger oh
08:06 lizmat joined, ab5tract_ joined
masak TEttinger: it's kind of a thing with five entries that completely describe an NFA. 08:07
ShimmerFairy "NFAs can be described by this five-tuple" ok yes but what _is_ an NFA, actually?
TEttinger ahhhh
masak ShimmerFairy: yes, it sounds to me that you've been starting at the wrong entry level.
reading Wikipedia, perhaps :)
ShimmerFairy masak: and every other webpage I could find on the subject, yes :)
masak at this point, you should stop, take stock, and find a better source text. 08:08
TEttinger but what _is_ a ShimmerFairy? a collection of molecules? a collection of subatomic particles that make up molecules? magic?
nine Definitely magic
masak I would second moritz++' book recommendation.
ShimmerFairy Yeah, too bad it looks like I'd have to buy the book though :/ (to my non-surprise, my county's library system doesn't have it)
masak TEttinger: it's a 2-tuple of a fairy, and shimmer.
ShimmerFairy: another sure-fire way to understand regexes is to re-implement a grammar engine in Perl 6. worked for me. 08:09
TEttinger at this point I like to recommend people to sit in on philosophy courses at an early college level and just watch as people either completely don't get it or get their minds blown
nine ShimmerFairy: the irony is that you throw around words like "allomorphic" which I've never heard in > 20 years of programming and sound very mathy to me ;)
masak heh 08:10
ShimmerFairy nine: to be fair, I didn't come up with the term :P
masak I'm wary of allomorphic things. not because of the term itself, but what the term *means*.
TEttinger I'm wary of allosaurs 08:11
which I assume are allomorphic
masak almost certainly.
ShimmerFairy And I do enjoy math quite a bit (I've looked up some statistics so I can calculate if benchmark timing differences are "statistically significant", for example). Just not the CS math. Never got into it, and I don't see how it could help me with my actual coding
TEttinger ShimmerFairy, yeah, for the most part I'm the same way 08:12
I enjoy "rediscovering" basic math stuff... since I've done it on math tests for some time :)
08:13 dakkar joined
TEttinger "oh crap what's the formula for x-distance of a point on the unit sphere... uh... pythagoras, I choose you!" 08:13
ShimmerFairy masak: funny thing is, I was looking up this stuff to reason about possibly implementing a P6 grammar engine :P (because I'd like to implement an emacs major mode for Perl 6, and I'd like to have a Real Parser™ to figure out P6 files)
masak ShimmerFairy: it may even help you to study the commit history of GGE. 08:14
tadzik FROGGS: huh, very weird
masak ShimmerFairy: it starts very simply (and test-based) and gets more advanced as it goes.
ShimmerFairy The main problem with emacs' default facilities, btw, seems to be that there's no (easy?) way to provide context to them.
masak ShimmerFairy: things like this: github.com/masak/gge/commit/8f5369...f9b691f007 08:16
TEttinger this is about the extent of technical papers that I can follow, and thankfully it's rather computer-not-science-y in terms of the operations it describes: www.pcg-random.org/paper.html
masak (just an example)
TEttinger (absolutely great idea for an RNG, btw)
lizmat commute some further, Basque country, here we come! & 08:17
08:17 lizmat left
ShimmerFairy masak: I'll be sure to keep the book and GGE in mind whenever I revisit the topic :) 08:17
masak :)
ShimmerFairy Also, emacs' regexps are absolutely horrible, so there's also incentive to write a parser for the kind of regexes that aren't mind-numbing :P 08:18
08:24 laouji left
_itz magala airport was flooded yesterday with many cancelled flights but we managed to escape 08:25
tokuhiro_ how do i call c function has varargs like printf with NativeCall? 08:26
08:27 abraxxa joined
FROGGS tokuhiro_: you don't (yet) :o( 08:28
tokuhiro_ :(
FROGGS tokuhiro_: but you are welcome to help implementing that
tokuhiro_: the backends we use (dyncall / libffi) support that 08:29
tokuhiro_: we just don't make use of that currently
tokuhiro_ FROGGS: i see
08:31 laouji joined
ShimmerFairy masak: also, that Yegge post is quite long, so I'll have to read it in-between doing other stuff :) . But from what I've seen from everything so far, I can't help but wonder if prototype-based OO would be a fit for implementing an MDL compiler (there was a mention of Lisp, I saw, so maybe) 08:32
08:37 aborazmeh left 08:39 pmurias joined
BooK just wondering: is there a Perl6 lib for OAUTH yet? 08:42
ShimmerFairy Haven't heard of one, but I don't keep track of modules that well :) try modules.perl6.org and see if something's there 08:43
pmurias do the CREDITS in nqp get updated? the tests seem to use that file 08:44
not sure how much can I assume about that file when writing tests? 08:45
FROGGS pmurias: well, we would recognize test failures after updating it... and we really do not touch that file often
I mean, it is three years old now
nine BooK: or use one of the Perl 5 OAUTH modules
mrf nine++ # Inline all the things 08:46
FROGGS Inline::COBOL /o/
mrf NO!
mrf would be happy if he never has to see COBOL again. 08:47
FROGGS :D 08:48
ShimmerFairy I wonder if S05 should have an operator precedence table like S03 does (right now I want to know if [a | b || c] means [ [a | b] || c ]) 08:49
08:49 leont left 08:50 pmurias left 08:51 Ven left 08:52 Ven joined 08:53 laouji left 08:56 Ven left
FROGGS tokuhiro_: fwiw, I'd be happy to work on varargs (with you or alone) after I got C++ support merged after the JVM build is healthy again... 08:58
08:59 laouji joined
labster Can someone tell me if this commit is a good idea? github.com/labster/rakudo/commit/c...34cd25f794 I'm trying to fix Rat.perl doing too much rounding. 09:00
m: say <1/128> == <1/128>.perl.EVAL
camelia rakudo-moar 16d366: OUTPUT«False␤»
tokuhiro_ FROGGS: i want to implement it but i don't have knowledge around it :( 09:02
FROGGS tokuhiro_: well, that's what collab is for :o) 09:05
09:06 laouji left
tokuhiro_ :) 09:06
FROGGS we can learn it together
09:06 laouji joined
tokuhiro_ yep 09:06
09:15 pmurias joined
dalek p/js: 7844967 | (Pawel Murias)++ | / (4 files):
Implement nqp::setinputlinesep.

Test that it works with a input separator consisting of two characters.		 09:17
p/js: 77d3834 | (Pawel Murias)++ | src/vm/js/bin/run_tests:
Add passing test 19 to run_tests.
09:22 abraxxa left, abraxxa joined 09:25 ab5tract_ left, ab5tract_ joined
ab5tract_ o/ #perl6 09:25
09:26 cognominal joined
ab5tract_ .tell nine I should have clarified further that my "sets are hashes, mmkay" feeling is deeply related to only ever using hashes for sets. This is surely a Perl-ish attitude. Anyway, I totally support efforts to make sets feel more natural to the people who actually have strong opinions about them 09:28
yoleaux ab5tract_: I'll pass your message to nine.
pmurias ab5tract_ \o 09:34
psch hi #perl6 o/
yoleaux 05:08Z <FROGGS> psch: thanks for the hint... now I just need to figure out what that means :o)
psch FROGGS: well, the dispatch for postcircumfix:<{ }> seems to hit the candidate in src/core/hash_slice.pm:49, whereas it probably should hit the "is parcel" one in line 12 of the same file 09:37
09:38 quester left
psch FROGGS: although maybe $tag is to blame after all, i haven't look in that direction 09:40
FROGGS psch: $tag or rather @tags is to blame 09:41
psch: I can potentially explain it in more detail this evening 09:42
09:48 TEttinger left 09:50 rmgk left 09:51 rmgk joined 09:54 abraxxa left
ab5tract_ psch: wouldn't the fact that parcel has been removed invalidate the is parcel candidate? 09:59
10:03 rindolf joined
psch ab5tract_: we still have a parcel trait_mod 10:03
Parcel has been removed
FROGGS: fair enough, i have a tendency to blame nqp-j in general, because i can't shake the feeling that something in there must be somhow wrong - probably mostly because of its slowness... :) 10:04
FROGGS psch: no, it is not just the slowness... 10:05
10:11 laouji left
jnthn "is parcel" is the name of the trait that maps to \foo in a sig 10:14
We probably should rename it at some point :)
But since you always express it as \foo it's not high priority
FROGGS jnthn: btw, what's your feeling of that line: 10:15
my int $SIG_ELEM_SLURPY := ($SIG_ELEM_SLURPY_POS +| SIG_ELEM_SLURPY_NAMED +| $SIG_ELEM_SLURPY_LOL);
FROGGS .oO( We need more dollars, dollars, dollars... )
jnthn Where is it? :)
FROGGS BOOTSTRAP.nqp:99
jnthn Also, I think it's missing a $ :)
FROGGS aye :o) 10:16
jnthn Curious how little it breaks ;)
10:16 laouji joined
jnthn Maybe we don't need the SIG_ELEM_SLURPY_NAMED in there at all? 10:16
If nothing's broken I'd just remote that :)
*remove
10:19 rarara joined
FROGGS lunch & 10:19
JimmyZ we have method pop(Array:D:) is parcel 10:20
Does it mean return \foo?
jnthn Well, means "return exactly what is returned" 10:21
psch named params also need the trait. i also agree that renaming might be a good idea, but don't have a suggestion for the new name
jnthn Aye
Me either
"is raw" is a bit undescriptive
jnthn won't worry about it much for now :)
10:25 xfix joined
ShimmerFairy "is noncont[ainerized]" ? :P 10:25
JimmyZ is original :P
jnthn is not-screwed-with 10:26
nine If I had known that there is such a trait, I'd not have had to fix those Pair creation bugs... 10:27
yoleaux 09:28Z <ab5tract_> nine: I should have clarified further that my "sets are hashes, mmkay" feeling is deeply related to only ever using hashes for sets. This is surely a Perl-ish attitude. Anyway, I totally support efforts to make sets feel more natural to the people who actually have strong opinions about them
nine It's probably a good thing that I didn't know then ;)
10:29 abraxxa joined
abraxxa FROGGS: hi! I'm working on DBDish::Oracle again and wonder if there where NativeCall improvements in the last few releases? 10:30
10:30 virtualsue joined 10:36 andreoss joined
dalek kudo/nom: b6d3180 | jnthn++ | src/core/ (2 files):
A couple of exception types for shaped arrays.
 10:37
kudo/nom: 2f06734 | jnthn++ | src/core/Array.pm:
Add default shape method for unshaped Array.
kudo/nom: 1f3372e | jnthn++ | src/core/Array.pm:
Take (but currently ignore) shape in Array.new.
nine Oh, shaped arrays almost done? 10:38
10:38 pat_js joined 10:39 dayangkun left
jnthn Well, the VM guts are almost done ;) 10:39
Working on the Perl 6 level now
10:40 pmurias left
nine Well I usually consider work that's been started as almost done...so jay shaped arrays! And yes, I'm an optimist ;) 10:40
jnthn During GLR when I re-did the [;] slicing stuff I did it in a way that makes shaped arrays easier 10:41
10:41 pmurias joined
jnthn Or paved the way for them 10:41
So, now to try and write up all the pieces. :)
*wire
nine Yeah, I had a hunch that it's good that you tackled the slicing parts 10:42
ab5tract_ much excitement!
10:44 n0tjack joined
baest jnthn: very nice! 10:44
10:45 virtualsue left 10:50 n0tjack left 10:51 telex left 10:52 telex joined
_itz ok >GLR has is-lazy .. is there a simple way of doing that with pre GLR code? 10:55
jnthn _itz: .infinite may approximate it, though it's different 10:56
_itz ok ty
jnthn perl6-m -e "my @a := Array.new(:shape(5)); @a[5] = 42;"
Index 5 for dimension 1 out of range (must be 0..4) in block <unit> at -e:1
Progress :)
10:59 brrt joined
dalek p: 64a0872 | jnthn++ | src/vm/jvm/runtime/org/perl6/nqp/sixmodel/reprs/MultiDimArrayInstanceBase.java:
Fix off-by-one in error reporting.
 11:00
kudo/nom: cbf85b3 | jnthn++ | src/core/Array.pm:
First steps for shaped arrays.

No multi-dim indexing support yet, but Array.new(:shape(5)) will now enforce the fixed size.		 11:01
jnthn Apparently it's time to eat lunch :) 11:02
bbiab
11:07 laouji left 11:09 Skarsnik joined 11:10 Loren left
psch m: my $s = (:types, :DEFAULT); my @a = flat "ALL", $s>>.key(); say @a.perl 11:11
camelia rakudo-moar cbf85b: OUTPUT«["ALL", "types", "DEFAULT"]␤»
psch $ ./perl6-j -e'my $s = (:types, :DEFAULT); my @a = flat "ALL", $s>>.key(); say @a.perl'
["ALL", ("types", "DEFAULT")]
/o\
TimToady sitting in Oslo, 4 hour delay... :/ 11:14
moritz Oslo is nice. Gardemoen not so much :=)
11:15 Possum joined 11:18 dpk joined 11:23 Sqirrel joined, Sqirrel_ left
dalek p/js: 52d13fe | (Pawel Murias)++ | / (4 files):
Preprocess regex tests (as we can't easily compile them at runtime before the bootstrap).

For now pass rx_basic.		 11:26
11:27 Loren joined
jnthn TimToady: Aww :( 11:28
And Malaga - Oslo can't have been short...
moritz 4h, maybe? 11:30
TimToady a bit more
but we had a headwind
FROGGS abraxxa: I think there were some improvements... the docs/Changelog will tell 11:31
oetiker jnthn did you find the fondue in olten ?
abraxxa FROGGS: in which repo? I always fail to find the correct one on github
FROGGS abraxxa: github.com/rakudo/rakudo/blob/nom/.../ChangeLog 11:32
abraxxa FROGGS: thanks
jnthn oetiker: Yes, we did! :) 11:33
oetiker \o/
abraxxa why does rakudobrew not list 2015.08?
jnthn oetiker: Thanks for the tip...it was a really rather nice one. :-) 11:34
oetiker: We managed to get a nice table outside also.
nine abraxxa: we didn't do a release in August
oetiker beautiful :-)
jnthn abraxxa: There was no 2015.08.
abraxxa jnthn: the rakudo ChangeLog has an entry and I saw the tag 11:35
so only a rakudo release but no moar release?
jnthn oetiker: Indeed. :)
11:36 brrt left
jnthn abraxxa: Don't think there was a Rakudo release actually cut/uploaded anywhere either. 11:36
moritz abraxxa: there's no rakudo 2015.08 tag 11:37
abraxxa: changelog yes, because I started the release process
11:37 dustinm` left
jnthn abraxxa: www.nntp.perl.org/group/perl.perl6....11182.html 11:37
abraxxa the MoarVM repo has a 2015.08 tag 11:38
i saw it when rakudobrew updated the git repos
jnthn I think there maybe was a MoarVM release.
Though not sure the site got updated yet
abraxxa no problem, thanks 11:39
jnthn Seems timotimo++ tagged it
timotimo: Did you by any chance make a Moar 2015.08 tarball?
[Coke] TEttinger++ 11:40
11:41 Urchin joined
[Coke] 1,038 RTs 11:45
(14 GLR tickets still open) 11:46
11:47 n0tjack joined 11:49 dustinm` joined 11:51 n0tjack left 11:54 pmurias left
TimToady m: say 0.91723840987192374893723.perl.EVAL 11:58
camelia rakudo-moar cbf85b: OUTPUT«===SORRY!===␤Cannot find method 'compile_time_value'␤»
[Coke] jnthn: is updating S07 on your list before the september release? 12:00
12:00 domidumont left
[Coke] Be nice if we had a place to point to details about the GLR from the changelog. 12:00
12:00 pmurias joined 12:02 domidumont joined
[Coke] (no pressure, just thought I saw you mention it this weekend) 12:04
andreoss m: say 0.9999999999999999999999.perl.EVAL 12:06
camelia rakudo-moar cbf85b: OUTPUT«0.1␤»
moritz that's bad :( 12:08
m: say 0.9999999999999999999999.perl
camelia rakudo-moar cbf85b: OUTPUT«0.10␤»
moritz that's very bad :(
m: say 0.9999999999999999999999 12:09
camelia rakudo-moar cbf85b: OUTPUT«0.10␤»
moritz m: say 0.9999999999999999999999 - 0.1
camelia rakudo-moar cbf85b: OUTPUT«0.9␤»
moritz m: say 0.9999999999999999999999 - 0.9
camelia rakudo-moar cbf85b: OUTPUT«0.1␤»
moritz m: say 0.9999999999999999999999 - 0.9999
camelia rakudo-moar cbf85b: OUTPUT«0.0001␤»
moritz m: say 0.0999999999999999 12:10
camelia rakudo-moar cbf85b: OUTPUT«0.0999999999999999␤»
moritz m: say 0.09999999999999999999999
camelia rakudo-moar cbf85b: OUTPUT«0.1␤»
moritz so it's only overflow over the decimal point that results in a bad stringification
TimToady exceeds 66-bit denominator, I expect 12:14
[Coke] r: my Int @a=^3; @a[1]:delete; @a.map:{say .WHAT.perl }; @a[1].WHAT.perl.say 12:15
camelia rakudo-jvm cbf85b: OUTPUT«Can't open perl script "/home/camelia/rakudo-inst/bin/eval-client.pl": No such file or directory␤»
..rakudo-moar cbf85b: OUTPUT«Int␤Any␤Int␤Int␤»
moritz m: say 0.09999999999999999999999.^name
camelia rakudo-moar cbf85b: OUTPUT«Rat␤»
ShimmerFairy m: say :16<0.FFFFFFFFFFFFFFF0>; say :16<0.FFFFFFFFFFFFFFF1> 12:16
camelia rakudo-moar cbf85b: OUTPUT«0.99999999999999999913␤0.10␤»
moritz TimToady: still a rat
m: say 0.9999999999999999999999.nude
camelia rakudo-moar cbf85b: OUTPUT«(9999999999999999999999 10000000000000000000000)␤»
moritz m: say 0.9999999999999999999999.Str
camelia rakudo-moar cbf85b: OUTPUT«0.10␤»
moritz it's really a bug in Rat.Str
ShimmerFairy TimToady: unless I'm mistaken, doesn't Rat still have an Int denominator instead of UInt64 in rakudo's implementation?
TimToady could be
moritz submits rakudobug 12:17
unless andreoss++ wants the honor, of course
ShimmerFairy yeah
TimToady: There is a UInt64 in rakudo, but it's defined as a subset, so I'm not sure changing Rat will actually make it any better than a FatRat atm :) 12:18
moritz ShimmerFairy: it does make it better
ShimmerFairy how is an Int with an extra constraint better than an Int? (Or better than a UInt64 linked to a sized native type?)
moritz ShimmerFairy: it avoids quadratic runtime behavior on things like newton's algorithm converting on sqrt(2) 12:19
ShimmerFairy: it does overflow to Num eventually
m: say (1 / 2**65).^name
camelia rakudo-moar cbf85b: OUTPUT«Num␤»
moritz which prevents runaway numerator and denominator sizes 12:20
dalek kudo/nom: b2fea84 | coke++ | docs/ChangeLog:
2015.08 was skipped.
ShimmerFairy moritz: true, having a sized denominator would help cap the Rat at a particular point, what I meant was that I'm not sure the current UInt64 would provide the kind of performance improvement over FatRat that's talked about
moritz ShimmerFairy: it's not UInt64 that provides the performance improvement; it's the fallback to Num that does 12:21
hoelzro o/ #perl6
yoleaux 7 Sep 2015 11:57Z <pmurias> hoelzro: before merging in js into the master branch remember to run the tests for the jvm/moar backends to make sure our improved test suit doesn't uncover any bugs in those
hoelzro pmurias: will do
ShimmerFairy moritz: oh. The way the specs talked about it, it always seemed like it was the sized type itself. Either the specs need to be clarified, or I didn't read the right parts closely enough :/ 12:22
(that is, of what I've seen nowhere in the specs where it mentioned that FatRat was worse in performance did it mention that it was because Rat has a fallback and FatRat doesn't) 12:24
12:24 [Sno] left
moritz RT #126016 filed for the Rat.Str bug 12:25
12:26 brrt joined
moritz ShimmerFairy: perlpunks.de/paste/show/55eed3ed.19dd.15f newton approximation converging on sqrt(2) 12:26
ShimmerFairy: just run it ones as is, and once with a normal Rat as starting value
maybe with only 12 or 15 iterations instead of 20, if you're impatient :-) 12:28
literal m: %foo<bar><baz><quux>
camelia rakudo-moar cbf85b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/hUvZaeU4Zo␤Variable '%foo' is not declared␤at /tmp/hUvZaeU4Zo:1␤------> 3<BOL>7⏏5%foo<bar><baz><quux>␤»
literal is there a way to invoke rakudo-moar in lax mode here?
12:29 [Sno] joined
FROGGS m: no strict; %foo<bar><baz><quux> 12:30
camelia rakudo-moar cbf85b: OUTPUT«X::Multi::NoMatch exception produced no message␤ in block <unit> at /tmp/IXpit3CwEc:1␤␤»
literal I see
psch m: no strict; $x = 5; say $x
camelia rakudo-moar cbf85b: OUTPUT«5␤»
literal and why doesn't this run? stricter autovivification?
psch the failure mode for the hash access looks pretty weird 12:31
m: no strict; %h<foo> = "bar"; say %h.perl
camelia rakudo-moar cbf85b: OUTPUT«{:foo("bar")}␤»
psch m: no strict; %h<foo><baz><quux> = "bar"; say %h.perl
camelia rakudo-moar cbf85b: OUTPUT«X::Multi::NoMatch exception produced no message␤ in block <unit> at /tmp/UUeJyp5Tg0:1␤␤»
FROGGS m: my %h; %h<foo><baz><quux> = "bar"; say %h.perl 12:32
camelia rakudo-moar cbf85b: OUTPUT«{:foo(${:baz(${:quux("bar")})})}␤»
FROGGS m: my %h<foo><baz><quux> = "bar"; say %h.perl
camelia rakudo-moar cbf85b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/eS3Q57_kdt␤Shaped variable declarations not yet implemented. Sorry. ␤at /tmp/eS3Q57_kdt:1␤------> 3my %h<foo>7⏏5<baz><quux> = "bar"; say %h.perl␤»
FROGGS m: our %h<foo><baz><quux> = "bar"; say %h.perl
camelia rakudo-moar cbf85b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/PlQlie5e31␤Shaped variable declarations not yet implemented. Sorry. ␤at /tmp/PlQlie5e31:1␤------> 3our %h<foo>7⏏5<baz><quux> = "bar"; say %h.perl␤»
12:33 bbkr_ left
abraxxa doc.perl6.org/language/5to6 doesn't mention Perl 6s' undef equivalent 12:34
psch abraxxa: "Since Perl 6 does not allow adding/removing symbols in a lexical scope once it has finished compiling, there is no equivalent to Perl 5's undef &foo;" 12:35
moritz m: our %h; say %h.^name 12:36
camelia rakudo-moar cbf85b: OUTPUT«Hash␤»
literal psch: that's a different thing though
abraxxa psch: i guess that would remove the sub foo from the symbol table in Perl 5 although I've never (ab)used it
perl6 -e 'my $foo; say $foo.perl;' returns Any
literal m: @foo[0] 12:37
camelia rakudo-moar cbf85b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/y05Z8I4YWx␤Variable '@foo' is not declared␤at /tmp/y05Z8I4YWx:1␤------> 3<BOL>7⏏5@foo[0]␤»
andreoss m: my &foo; say &foo.Perl
camelia rakudo-moar cbf85b: OUTPUT«Method 'Perl' not found for invocant of class 'Callable'␤ in block <unit> at /tmp/_SG71yxh1x:1␤␤»
literal m: no strict; @foo[0]
camelia ( no output )
andreoss m: my &foo; say &foo.perl
camelia rakudo-moar cbf85b: OUTPUT«Callable␤»
literal m: no strict; @foo[0].perl
camelia ( no output )
literal m: no strict; say @foo[0].perl 12:38
camelia rakudo-moar cbf85b: OUTPUT«Mu␤»
psch abraxxa: i take that to mean you're wondering about assignment RHS undef? i think Nil does that in p6, but you're right that it's not mentioned
TimToady m: &sleep.wrap({ die "I'm undefined" }); sleep 3
camelia rakudo-moar cbf85b: OUTPUT«I'm undefined␤ in block <unit> at /tmp/AobxTMOX5Q:1␤␤»
abraxxa psch: i have to compare a list of returned database values and need the Perl 6 equivalent of SQL NULL 12:39
literal so why is an empty array element 'Mu' and an unassigned scalar is 'Any' ? and where would Nil come in? :P
abraxxa [ Any, Any, 1 ], works
moritz m: my @a; @a[3] = 1; say @a[0] 12:40
camelia rakudo-moar cbf85b: OUTPUT«(Any)␤»
moritz seems to be Any, not Mu
psch m: no strict; @a[3] = 1; say @a[0]
camelia rakudo-moar b2fea8: OUTPUT«(Mu)␤»
literal I guess it's only Mu when @foo itself was not declared
oh, or not
abraxxa but isn't that just wrong when used in tests because it would allow any value instead of just undef? 12:41
moritz I'd say this is a bug in 'no strict'
abraxxa and isn't Any the root type and not a value?
moritz abraxxa: it should test !.defined
abraxxa: types are also values
abraxxa moritz: github.com/abraxxa/DBIish/blob/mas...#L159-L161 12:44
suggestions how to solve this welcome
moritz abraxxa: uhm, what's the problem there? 12:48
12:48 n0tjack joined
abraxxa moritz: I have added rows to that table where not all columns have a value 12:48
e.g. github.com/abraxxa/DBIish/blob/mas...n.pl6#L115 12:49
just fixing the now failing tests
12:50 shinobi-cl joined
jdv79 .tell lizmat why doesn't S22 root a dist archive with the dist+ver like a p5 one? 12:52
yoleaux jdv79: I'll pass your message to lizmat.
12:52 n0tjack left
jdv79 FROGGS: do you know if there's a reason for that? 12:54
most tarballs, p5 and not, usually have a top level dir of the same name as the file itself - at least that i've seen 12:55
_itz is S07-glr-draft.pod still mostly correct? 13:00
FROGGS jdv79: I did not think about that yet tbh...
jdv79 can we just do that then? normally when i tar -zxvf a file i expect it to politely create one dir with all the stuff it spews 13:01
is one nice feature
_itz oh it still talks about Parcels 13:02
_itz looks confused
jdv79 plus if it was the same as the p5 way there is less friction/confusion - type of stuff
ShimmerFairy jdv79: what are you talking about, precisely? Does P6 offer a tar-like thing or something? 13:03
jdv79 no
perl5 distribution tarballs are of the basename $dist-$ver and they contain one subdir of that same name in which aall else is 13:04
the perl6 META6.json detection code at pause is looking for '^META6.json'. this is the problem. 13:05
brrt is wondering if anybody made a list of the great-tools-used-in-japan slides
jdv79 i packaged my dist in the p5 way and the indexer failed
hence why github.com/andk/pause/issues/185 13:06
brrt slideshare has it: www.slideshare.net/junichiishida/gr...-dont-know
ShimmerFairy jdv79: is that why your upload failed with that "duplicate package" warning? 13:07
jdv79 no, that bug was fixed y'day.
so i got past that issue and this is the new one
i would like some consensus on the dist layout before m PR is applied as it seems to take a decent amount of time to get things done over there 13:09
*my
ShimmerFairy jdv79: cool :) and yeah, that seems like a bug. Does that regex allow the META6.json anywhere in the package? Because that seems like a bad idea, I think I'd prefer "top-level, considering one top directory in the tarfile as the 'real' top-level", or something to that effect 13:10
jdv79 right now it pegs it to the top level of the tarball. i would like it to peg it to one subdir down. 13:11
yeah, I just need liz and/or froggs to weigh in. 13:12
ShimmerFairy jdv79: yeah, I personally wouldn't mind enforcing the common-courtesy tarballs are known for (I very rarely see a .tar.* that doesn't unpack as one directory, and such tarballs are very undesirable, imo)
jdv79 yup 13:13
13:14 sufrostico joined
ShimmerFairy btw, I hope the "must specify Perl6/" thing is temporary, since that feels like a bad permanent solution 13:17
jdv79 in what way? 13:20
we could put up a page that does that automatically or hide it behind a tool but cpan using that dir to house p6 content is pretty solid i think 13:21
ShimmerFairy jdv79: not the fact that we're using Perl6, but the fact that I have to manually type it into the "target directory" field 13:22
13:22 n0tjack joined
jdv79 yeah, next iteration maybe;) 13:23
pmurias hoelzro: \o 13:25
ShimmerFairy also, does that "can never upload a same-named file, even if deleted" apply to P6 files too?
jdv79: well, next iteration would possibly be a "Yes, this is Perl 6" checkbox :P
13:25 hernanGOA joined 13:26 tokuhiro_ left, n0tjack left
jdv79 yes and ok 13:27
psch vivifying array elements to Any instead of Mu under "no strict" seems to work with specifying Any as a fallback to $*OFTYPE in the container_type_info in $*W.auto_declare_var 13:31
i'm not sure it's really supposed to be that easy, though 13:32
also, RT #124912 has me a bit stumped
as in, i'm unsure if "no strict" propagating into EVAL is really that sensible 13:33
OTOH, it's probably less surprising if it did, which is probably reason enough
nine _itz: that seems to be the closest we've got to GLR documentation: gist.github.com/niner/d5f825c07ce1f7a6f7bb 13:34
psch but i think that means teaching HLL::Compiler about "no strict"...
13:34 pmurias left
_itz nine: thanks 13:40
13:40 CQ left
_itz nine: I was also experimenting with gist.github.com/stmuk/de108fc081e45876d1e4 as a learning tool for myself 13:41
dalek ast: 0dc659a | TimToady++ | S03- (3 files):
test .is-lazy on xx, .., and ...
 13:43
ast: 14d11da | TimToady++ | S32-num/rat.t:
test (1/128).perl.EVAL trips round
psch hm, except the container_type_info fix mentioned above makes autodeclared hashes weird
$ ./perl6 -e'no strict; %a<a> = 0; say %a.perl'
(my Any % = :a(0))
i'm pretty sure that's not right 13:44
13:44 CQ joined
dalek kudo/nom: 2d6ee41 | TimToady++ | src/core/Rat (2 files):
.base now can take * digits

also .base-repeating no longer fails on negatives		 13:46
ShimmerFairy psch: I'd like to see some spec confirmation before we consider 124912 a bug. (I personally don't think 'no strict' is worth having in the first place, so I'm not inclined to make this work, but that's just me) 13:47
_itz github.com/perl6/doc/issues/116 13:48
13:48 brrt left
psch ShimmerFairy: design is rather thin regarding anything lax mode, unfortunately 13:49
ShimmerFairy psch: because it shouldn't be there in the first place :D . More seriously, I've thought about it a bit today, and I've come to the conclusion that 'no strict' exists in P6 only because P5 had a concept of strictures, so I'm inclined to say "you can't be lazy, there is no lax mode". Perhaps a 6.1 change, if not sooner :) 13:50
jnthn back 13:51
TimToady m: say 1/128 == (1/128).perl.EVAL
camelia rakudo-moar b2fea8: OUTPUT«False␤»
jnthn [Coke]: Yes, I already re-writing S07
TimToady that should fix itself with recompile 13:52
FROGGS m: say $*PERL.build-date 13:53
camelia rakudo-moar b2fea8: OUTPUT«Method 'build-date' not found for invocant of class 'Perl'␤ in block <unit> at /tmp/6xD7kGQQ_9:1␤␤»
FROGGS m: say $*KERNEL.build-date
camelia rakudo-moar b2fea8: OUTPUT«Method 'build-date' not found for invocant of class 'Kernel'␤ in block <unit> at /tmp/j2mnt2zN1G:1␤␤»
FROGGS m: say $*COMPILER.build-date
camelia rakudo-moar b2fea8: OUTPUT«Dynamic variable $*COMPILER not found␤ in block <unit> at /tmp/WlH0s6EESz:1␤␤Actually thrown at:␤ in any at src/gen/m-Metamodel.nqp:2869␤ in block <unit> at /tmp/WlH0s6EESz:1␤␤»
psch ShimmerFairy: but laziness is one of the three virtues! ;)
FROGGS m: say $*PERL.compiler.build-date
camelia rakudo-moar b2fea8: OUTPUT«2015-09-08T13:09:28Z␤»
psch ShimmerFairy: in seriousness though, i doubt "no strict" will go away, but the interaction between strictness and EVAL is definitely underspecified 13:56
_itz s: say (Array).^roles 13:57
jnthn s? :)
_itz star: say (Array).^roles
camelia star-m 2015.03: OUTPUT«(Positional)␤»
TimToady boarding & 13:58
_itz m: say (Array).^roles
camelia rakudo-moar b2fea8: OUTPUT«Method 'iterator' must be implemented by Iterable because it is required by a role␤ in any compose_method_table at src/gen/m-Metamodel.nqp:2666␤ in any apply at src/gen/m-Metamodel.nqp:2676␤ in any compose at src/gen/m-Metamodel.nqp:2812␤ in an…»
jnthn m: say (Array).^roles.map({.^name})
camelia rakudo-moar b2fea8: OUTPUT«(Positional Iterable)␤»
_itz m: say (Array).^roles.WHAT
camelia rakudo-moar b2fea8: OUTPUT«(List)␤»
13:59 perlawhirl joined
ShimmerFairy psch: like I said, perhaps only in 6.1, but I definitely don't know what the value of lax mode is, so atm I kinda would like to make it go away. :) 13:59
jnthn m: say Iterator
camelia rakudo-moar b2fea8: OUTPUT«Method 'pull-one' must be implemented by Iterator because it is required by a role␤ in any compose_method_table at src/gen/m-Metamodel.nqp:2666␤ in any apply at src/gen/m-Metamodel.nqp:2676␤ in any compose at src/gen/m-Metamodel.nqp:2812␤ in an…»
jnthn ah
Maybe .gist needs to go in the no-pun list.
[Coke] "doctor, it hurts when I no strict." 14:00
psch ShimmerFairy: right, the value is purely traditional. and considering we have strict -e there's no efficiency argument either... 14:01
ShimmerFairy psch: I just came to this conclusion earlier today, so I'm not too strong on it (yet) :P . I am willing to be shown Perl 6 examples that are *greatly* improved by lax mode, however. :)
grondilu rewrote rosettacode.org/wiki/SHA-1#Perl_6 for post-GLR. Basically had to add a few 'flat' here and there. 14:02
ShimmerFairy Like I said, the only reason I can think of it being there in the first place is because Perl 5 had a concept of strictures. :)
14:02 n0tjack joined
[Coke] trips over a JSON encoder that isn't encoding chr(18) properly. blah. 14:03
ab5tract_ ShimmerFairy: I appreciated the lax-ness when it was in the repl by default
If there is a strong *technical* reason for removing it, sure. But otherwise, keep it around for dirty one liners
psch ab5tract_: i read that as an argument for "bring back lax -e", mostly 14:04
and i agree with that, i think
ab5tract_ psch: certainly not!
ShimmerFairy ab5tract_: well, aside from the fact that you have to specify 'no strict' in your one-liner, I'll point out that I never had a problem working under strict mode in one-liners
ab5tract_ perl6 -d or so
pink_mist I'd say if you really want it, make it a slang
ShimmerFairy As a P6-only user, I have _never_ thought to myself "gee, if only I didn't have to type two letters and a space to declare a 'my' variable" 14:05
ab5tract_ no need to have it in -e, but it would be nice to have a flag that is lax
pink_mist seems silly to support two different ways of writing perl6
dalek kudo/nom: 7102178 | jnthn++ | src/core/Array.pm:
Ban certain operations on shaped arrays.
ab5tract_ pink_mist: umm... that cat is already out of the bag via TMTOWTDI
psch and many subs being methods, too 14:06
ab5tract_ ShimmerFairy: sure. I see no reason why you're lack of desire should override the desires of others
If it is particularly costly in terms of technical debt -- sure
pink_mist ab5tract_: I worded that badly; I didn't mean not to support however many ways to write perl, but rather having two separate bases from which to build code
ab5tract_ pink_mist: I really don't think the split between strict/no strict is that deep though. But I could be wrong 14:07
ShimmerFairy ab5tract_: I'm just saying that as someone who has not used P5, (and as someone who used P6 back when there was no lax mode), I've never desired what P5 offers, so I can't help but think other people liking it are just familiar with it.
14:07 andreoss` joined
ab5tract_ ShimmerFairy: 'use strict;' is not there because Perl 5 had strictures... it's there because it *didn't* 14:07
grondilu oh, first time I see this one: 'Bus Error' 14:08
pink_mist ab5tract_: I'd very much like to see it as a slang though; that way the interpreter wouldn't need to worry about two different ways to parse code
jnthn pink_mist: The interpreter already doesn't 14:09
14:09 andreoss left
ab5tract_ ShimmerFairy: that could easily be turned around to you not realizing the utility due to lack of familiarity 14:09
jnthn The interpreter has no idea at all about parsing anything execpt its own bytecode format. It couldn't care less if that was made from COBOL or Perl 6. :) 14:10
ShimmerFairy ab5tract_: what I meant is that 'Perl 5 having a concept of strictures' meaning that it had a conception of being strict vs. not. Perl 6 doesn't have that same issue, so I see 'no strict' as just a way to keep P5-ers happy
ab5tract_ You are talking about something that only became default *this year*
jnthn Pushing "no strict" out to a slang would be a good verfication of slangs, once we get far enough along.
ab5tract_ and I really don't understand your readiness to reject things just because they make P5 people happy 14:11
ShimmerFairy jnthn: would you support the idea of kicking out lax mode in 6.1 or 6.x ? I don't think it's quite appropriate to kick it out of 6.0.0
14:11 perlawhirl left
ab5tract_ it's frustrating to say the least 14:11
jnthn ShimmerFairy: Why kick it anywhere?
psch kick it into a core slang i say!
ShimmerFairy ab5tract_: what do you mean by "only became default *this year*" ? Are you talking about P5?
jnthn ShimmerFairy: It can stay implemented as it is for now
14:12 Ven joined
jnthn ShimmerFairy: Maybe in 6.x when we've got slangs further along we'll kick it out of the core compiler codebase into a module, to decrease code size an iota...but it's a drop in the ocean really. 14:12
14:12 Sqirrel left
ShimmerFairy jnthn: I'm just arguing that lax mode is pretty useless, from my view. I've yet to hear a good reason for it aside from "P5 had a 'strict' pragma". 14:12
(which I don't think is a good reason)
jnthn ShimmerFairy: Then don't use it. 14:13
ab5tract_ ShimmerFairy: no, I mean that strict-ness became default for perl6 -e and the repl only this year
jnthn I thought we already changed -e to be strict recently, leaving about zero places where it's the default.
ShimmerFairy ab5tract_: before lax mode existed, things were strict everywhere, so that doesn't mean what you're implying it does. 14:14
ab5tract_ ..... umm, I'm sorry but I was able to do perl6 -e "@a = rand xx 50" and now I can't
ShimmerFairy jnthn: we did, and of course I don't use lax mode, I'm just mentioning that I find no reason it exists in P6, much less as a part of CORE :) 14:15
ab5tract_ that was just changed recently
Ven o/, #perl6!
ab5tract_ and the repl was also changed to lax mode
ShimmerFairy ab5tract_: I recall there being a time where rakudo didn't have "no strict", so of course things wouldn't be lax anywhere :)
ab5tract_ ShimmerFairy: I'm sorry but you are really wrong about that
jnthn It's true, there was a time where we didn't have lax mode. But we have had it for a while.
I suspect multiple years.
psch 5daa5de7e714908342751decf1d535473d10237e 14:16
Date: Sun Sep 28 13:22:50 2014 +0200
is the commit that implemented lax mode
ab5tract_ NYI is not the same as "not in the language"
dalek kudo/nom: 27a3699 | jnthn++ | src/core/Array.pm:
Simpler elems/eager/is-lazy for shaped Array.
psch ab5tract_: design docs still state -e as lax
jnthn psch: Wow, more recent than I thought :)
ShimmerFairy Yeah, I'm just saying that "lax until this year" doesn't mean what you think it does when lax itself didn't exist until last year. Thus implementation-wise the REPL and -e have always been strict, except for that one time it wasn't for a while :P 14:17
14:18 laubster joined
ab5tract_ ShimmerFairy: ok, I stand corrected. I also thought it had been around for a lot longer. 14:19
Regardless, I don't think the test of whether something should be in core or not should come down to whether you personally find an argument for it to be satisfying
14:19 skids joined
ab5tract_ Nor do I think it is fair to any of the P5 people who have been patiently waiting / helping to hear their expectations denigrated 14:20
jnthn Indeed, what's in Rakudo core is largely my call. And it's staying. :)
ShimmerFairy That's why I have to argue my point here instead of just removing it on a whim :)
jnthn As mentioned, when we have slangs, we can make it one of those (and 'use strict' just pulls in that slang) to decrease core size a bit.
Uh, "no strict" 14:21
ShimmerFairy I just think that 'no strict' means nothing useful to P6 on its own. It only exists because Perl 5 had a use/no strict thingy, and I seriously don't think the concept of being lazy with lax mode would exist otherwise.
jnthn: or 'use lax' :)
psch "use nostrict" :)
jnthn Nah, "no" will (by default) be a way to load a pragmatic module too, eventually.
We just don't have our pragmas story straight enough yet, and probably won't this side of 6.christmas. 14:22
_itz "use strict is gay" (old joke)
ShimmerFairy jnthn: oh, interesting. Will pragmas eventually work like they did in P5? (As in, a funny module?) Or will they be lightly different than modules here in P6?
jnthn ShimmerFairy: The pragma/module distinction doesn't especially exist so far as I'm concerned.
PerlJam finds it bizarre that "no strict" has generated so much discussion. 14:23
jnthn ShimmerFairy: Grammar::Tracer has lexical effect in the scope you import it into. Is it a module or a pragma?
ShimmerFairy jnthn: from my years of P6, I think of pragmas as just compiler flags, so I'm unfortunately used to them being very different. If I were more familiar with P5 I probably wouldn't have that distinction so much :P
TimToady the very attempt to make the distinction indicates only the module system is not yet powerful enough to implement pragmas :) 14:24
no, not yet boarded-really
ShimmerFairy I think the big question is what 'no' means in P6 module loading, since I don't think that's been figured out yet.
jnthn ShimmerFairy: It hasn't, and that's why they're handled differently in the compiler at the moment.
Most such hard-coded things in the compiler with use/no are only that way because it's expedient. 14:25
TimToady m: say 1/128 == (1/128).perl.EVAL
camelia rakudo-moar 710217: OUTPUT«True␤»
TimToady there we go
jnthn And because we know the whole area is related to macros, quasis, etc. and so we can't do what we want with pragmas until we deal with those issues.
ShimmerFairy Not to mention that the kinds of changes pragmas are known to make are surely the even more difficult changes to undo :) 14:26
TimToady really boarding maybe :) & 14:27
14:27 tokuhiro_ joined
moritz hopes for TimToady it's a boring boarding (and flight) 14:27
dalek kudo/nom: d881420 | jnthn++ | src/core/Array.pm:
Implement multi-dim access to Array.
 14:28
jnthn TimToady: Good luck!
ShimmerFairy: Thankfully, our entire parser arch is designed for forgetting stuff at the exit of a lexical scope
ShimmerFairy Hm. When pragmas can be implemented as modules, would 'no strict' be some kind of "strict" module that would be able to trickily define itself as adding features upon a "no", or would it be saner for those changes to be brought in by a 'use' statement instead? 14:29
jnthn: that's good at least :) . Perhaps some of that can be used in making 'no' work (whatever it'll mean), assuming "exit a lexical scope" doesn't automatically cover that somehow.
14:30 Ven left 14:31 tokuhiro_ left
jnthn perl6-m -e "my @arr := Array.new(:shape(3, 3)); @arr[1;1] = 42; say @arr[1;1]" 14:32
42
\o/
And yes
perl6-m -e "my @arr := Array.new(:shape(3, 3)); @arr[1;1] = 42; say @arr[1;3]"
14:32 zoosha joined
jnthn Index 3 for dimension 2 out of range (must be 0..2) in block <unit> at -e:1 14:32
ShimmerFairy jnthn++ # \o/ ever closer! 14:34
jnthn I'm getting dangerously close to the point where we have to work out some difficult things. 14:35
pink_mist if lax mode is lexical ... is it really even lax mode at all? I mean in perl5, any variable you used undeclared aside of the globals was package scoped afaik 14:36
jnthn pink_mist: use strict/no strict are lexical in Perl 5 afaik, yes. 14:37
pink_mist hmm, I guess you're right
jnthn TimToady: If you didn't board yet, can you decide what "my @arr := Array.new(:shape(3, 3)); @arr.map(&say)" does during your flight? ;-) 14:38
pink_mist but the variables in a 'no strict' scope need not be
14:41 zakharyas left
dalek kudo/nom: e819dd1 | jnthn++ | src/core/List.pm:
Toss commented out code we likely don't want.
 14:41
14:46 Ven joined
Tux__ m: my @a; my %h = :A(1), :B(2); @a.push: %h; @a.perl.say 14:50
camelia rakudo-moar d88142: OUTPUT«[:B(2), :A(1)]␤»
Tux__ m: my @a; my %h = :A(1), :B(2); @a.push: { %h }; @a.perl.say
camelia rakudo-moar d88142: OUTPUT«[:A(1), :B(2)]␤»
jnthn Has to be $%h or item %h post-GLR 14:51
Tux__ what should I push to end up with [{:A(1), :B(2)}] ?
jnthn++
14:52 khw joined
Tux__ is trying to get CSV back on line 14:56
Tux__ notes that RT#123888 is fixed 14:57
dalek kudo/nom: 83428be | jnthn++ | src/core/array_slice.pm:
Simple cases of :exists/:delete on multi-dim.
jnthn needs to look at CSV to see what the perf regression is down to
If that's what the numbers you've been posting are about, anyways...
|Tux| oh sure 14:58
the test suite currently still fails
jnthn OK
I'll hold off until it works properly again, so I'm profiling correct code. :)
|Tux| .tell lizmat will you look at Text::CSV/t/15_flags regarding the disappearing method? 14:59
yoleaux |Tux|: I'll pass your message to lizmat.
|Tux| I don't know if I can reach "proper" without help
just got rid of 24+ errors
15:01 andreoss` left
Ven jnthn: why is it postcircumfix:<[; ]> instead of <[ ]> ? 15:01
is that the dispatch that should be hit for @a[3;4]?
jnthn Ven: Yes, we decided to distinguish the two forms of slicing syntactically and just call a different candidate. 15:02
Ven: Uh, different set of candidates 15:03
Ven jnthn: so, there'll be a syntactic difference in my code? Then why check if they're not all integers?
shouldn't this candidate only be called if they're not all multidim?
15:03 domidumont left
jnthn Ven: @foo[*;0] is perfectly allowable, for example 15:03
Ven: I plan to re-write various things in the optimizer though
Ven jnthn: but that's multi-dim, right? not all-ints? I guess I don't really understand it :)
jnthn Ven: The "all ints" check is "can we just go straight off to do an indexing" 15:04
Ven: Rather than having to pick apart what we have and do clever stuff
Ven I don't see how that's the case with [*;0], though? The first one is a whatever. 15:05
abraxxa how can a CArray[uint8] hold codepoints < 0?
jnthn Ven: It's not the case, thus why we fall back to the rather more involved MD-ARRAY-SLICE
abraxxa: Um...I'd hope it can't :)
Ven: Rather than just calling AT-POS 15:06
|Tux| jnthn, when you pull T::C, gist.github.com/Tux/7e7d54789ec3d5c9234a
abraxxa paste.scsys.co.uk/498338
|Tux| no need to get a passing test suite to check on performance 15:07
abraxxa that's my code
this works for getting the error text of an Oracle exception but fails for column names
15:08 Ven left
abraxxa jnthn: ^^ any idea what I'm doing wrong? The warn @col_name_ary; outputs 40-10854 15:09
this is the code which works: github.com/abraxxa/DBIish/blob/mas...e.pm6#L282 15:10
jnthn abraxxa: Um, don't immediately see it. Somebody who's worked on/used NativeCall more recently may be able to spot it. 15:11
15:11 Ven joined
abraxxa jnthn: ok, thanks 15:11
FROGGS: you maybe?
nine: or you?
Ven jnthn: aaah, thanks :-).
abraxxa i thought uint8 should be more precise than int8 but it makes no difference regarding the error 15:12
nine abraxxa: can you try warn @col_name_ary.perl; just to be sure we interpret the output correctly?
abraxxa nine: [-104, 25, -119]<> in method fetchrow at /home/abraxxa/perl6/git/DBIish/lib/DBDish/Oracle.pm6:568
nine That's different than before? 15:14
abraxxa yeah, i guess the order of the columns returned by OCI varies 15:15
the length returned is in bytes, the encoding should be utf-8
nine Just 3 bytes? Do you have an idea which column name this could be?
abraxxa the query is 'SELECT name, description, quantity, price, quantity*price AS amount FROM nom ORDER BY name'
the length is 4 so it should be name 15:16
as the array is 0 based i loop from 0 to 4-1
nine So it should rather be for ^$col_name_len;
15:17 pmurias joined
nine ^$col_name_length is already exclusive 15:17
|Tux| is "[ 1, 2, "foo", 3 ].iterator" still the correct syntax for post-glr?
abraxxa exclusive means what?
pmurias hoelzro: any hard problems stopping the merge or just tuits?
hoelzro pmurias: both =/
no tuits for the last 4 or 5 days, but I'm stuck on merging at the point of 18ece1f 15:18
15:18 Sqirrel joined
nine abraxxa: exclusive means it will stop _before_ reaching the limit 15:18
m: use NativeCall; my $c = CArray[uint8].new; $c[0] = 180; say $c[0];
camelia rakudo-moar 83428b: OUTPUT«-76␤»
abraxxa nine: ok
nine ^^^ there's your problem
moritz |Tux|: what do you want to do?
hoelzro jnthn made a change to NQP that uses an optimization in MoarVM, but I can't seem to figure out how to get nqp-js to compile without reverting that optimization
jnthn |Tux|: Well, the syntax is fine, but I'm not convinced you'll want to call .iterator post-GLR
nine uint8 seems to not be unsigned at all
abraxxa nine: because unsigned are still not implemented correctly and wrap? 15:19
nine yep
abraxxa i wonder why using int8 in the errortext works
nine I don't think it should be a problem in your case. Using int8 or uint8 just changes the interpretation of the array's contents, not its values.
But why do you decode('ascii') if it's utf-8? 15:20
abraxxa nine: that was just a try 15:21
because 'name' should be the same in both encodings afaik
nine But you clearly have code points > 127 in there
abraxxa the length returned seem to be ok: 4, 11, 8, 5, 6 15:22
|Tux| jnthn, github.com/Tux/CSV/blob/master/lib...SV.pm#L296 and github.com/Tux/CSV/blob/master/lib...SV.pm#L384 15:23
15:25 coffee` left, coffee` joined
masak no release manager for this month... 15:26
15:27 xfix left
pmurias hoelzro: one possible workaround would be to extract that part into a method and override it in the SerializeOnce role we mix in to inject hacks necessary for cross compilation 15:27
abraxxa nine: the returned bytes are always different
hoelzro pmurias: I tried that, no dice =(
dalek kudo/nom: 90e4dab | coke++ | docs/release_guide.pod:
Claim 2015.09 compiler release
hoelzro [Coke]++ 15:28
15:28 tokuhiro_ joined
|Tux| .tell lizmat ignore: I found it «ok (@x[$_], "text") for 1 .. 5, 7, 19;» → «ok (@x[$_], "text") for flat 1 .. 5, 7, 19;» 15:29
yoleaux |Tux|: I'll pass your message to lizmat.
pmurias hoelzro: I'll try to figure out why things still work after that optimalization and maybe I can figure out a better workaround
hoelzro pmurias: I have an in-progress merge branch on my machine, I can push it up tonight
after that optimization, I think we just need to provide a dummy NQPReentrantLock REPR 15:30
and then fix whatever else remains =)
jnthn |Tux|: Ho, so you're just implementing a method iterator... Where is that being used? Or is it public API?
hoelzro: What was the change? 15:31
|Tux| the iterator in the first url is to enable $object[$index]
Ven today was a great day. The backlog is full of nice stuff to read & learn :-).
|Tux| the second is to enable special behaviour for an error class
15:33 tokuhiro_ left, pmurias left 15:35 muraiki joined
dalek p/js: b015083 | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
Bunch of small regex fixes/features.

Fix emitting of string containing horizontal whitespace. Implement enumcharlist. Implement eol, bol, lwb, rwb anchors. Fix bug in quantifiers.		 15:35
15:35 pecastro joined, pmurias joined
Ven pmurias++ 15:35
nine pmurias++ # still a hero 15:36
|Tux| jnthn, removing the first would result in «Method 'iterator' must be implemented by CSV::Row because it is required by a role»
jnthn |Tux|: Yes, because the class "does Iterable"
nine |Tux|: do you have some is Iterable there?
|Tux| nine, no is Iterable 15:37
jnthn Oh, if it's "is Iterable" it's certainly wanting changing to "does Iterable"
Looks like you already changed that.
|Tux| yes
abraxxa what's the difference between my @col_name := CArray[int8].new; and my CArray[int8] @col_name; 15:39
|Tux| Method 'error' not found for invocant of class 'X::TypeCheck::Assignment' <= does that ring a bell? 15:40
pmurias is it possible to push half merged branches in git?
nine abraxxa: in the first case you get an initialized object
pmurias: what do you mean by "half merged"?
abraxxa nine: and in the later? 15:41
nine You have a variable that can only hold objects of the specified type
abraxxa and what does 'initialized' mean?
pmurias nine: after asking that I figured that would be just merging a part of the commits
s/part/some/
_itz github.com/jnthn/test-mock/pull/7 # bump 15:43
jnthn _itz: Merged
eiro /wg #phear.org
oops sorry 15:44
abraxxa nine: I expected a different output from my @col_name := CArray[int8].new; @col_name[$_] = 0 for ^32; warn @col_name.perl;
_itz ty
eiro hello btw
15:44 Ven left
|Tux| FYI it surprised me that now [[1,2]] != [[1,2],] 15:44
15:45 plicease joined
jnthn |Tux|: Yes, that's the thing we often call the "single arg rule" 15:46
|Tux|: Same rule as for the .push case you had earlier, fwiw
So [$[1,2]] also does that [[1,2],] does 15:47
*what
|Tux| three more flat's fixes 17000 tests :) 15:54
15:55 yqt joined
pink_mist wow, that's some good value per character of code 15:59
16:00 CQ left 16:02 Ven joined
ugexe m: sub foo($a) { once &?ROUTINE($a) }; foo(1) 16:02
camelia rakudo-moar 90e4da: OUTPUT«Too many positionals passed; expected 0 arguments but got 1␤ in sub foo at /tmp/EYb1e48m4u:1␤ in block <unit> at /tmp/EYb1e48m4u:1␤␤»
16:04 dha joined
|Tux| «take $from .. $r.value;» → Type check failed in assignment to '@!crange'; expected 'Int' but got 'Range' 16:06
$r.value can be Inf
timotimo did i o/ yet?
|Tux| so I bet flat is out of the question
hi timotimo
timotimo o/
you may want to take a Slip
ISTR take now always just takes a single thing, or something like that, and we've got to rely on iteration encountering a Slip there if we want things to flatten out 16:07
Ven \o timotimo :-)
16:07 rindolf left, pecastro left
timotimo to me, flat is like flattening "into", while Slip is like flattening "out of" 16:07
|Tux| so take Slip 1 .. Inf; ? 16:08
timotimo Inf?
abraxxa how do I define a private attribute that gets lazy populated?
timotimo could actually work
|Tux| $r.value can be Inf, as I already said 16:09
timotimo abraxxa: AFAIK lazy population relies on accessors existing to do the "magic"
right, it can, that's why "flat" is out of the question, right?
but with Slip, it probably can still work
abraxxa timotimo: so no built-in functionality like Moo's is => 'lazy' sub _build_foo ? 16:10
|Tux| what is the syntax? I keep getting errors
jnthn m: say (|1..Inf).WHAT
camelia rakudo-moar 90e4da: OUTPUT«(Range)␤»
jnthn m: say |(1..Inf).WHAT
camelia rakudo-moar 90e4da: OUTPUT«(Range)␤»
jnthn m: say (|(1..Inf)).WHAT
camelia rakudo-moar 90e4da: OUTPUT«(Slip)␤»
jnthn heh, third time lucky :P
timotimo abraxxa: we had that built-in for a few days, but it was deemed "for the ecosystem" 16:11
abraxxa timotimo: i see
timotimo in fact, you could go back, find the commit that introduced the change, package that up into a module and put it up on the ecosystem right now
nobody has done that, as far as i know
16:12 rindolf joined
abraxxa but that means that you have to access the private attr always through its accessor? 16:12
hoelzro jnthn: something about not using the string heap, and using the compunit's instead
timotimo abraxxa: direct private attribute access is extremely close "to the metal" 16:13
_itz A suggestion for future stars... that 2015-spw-perl6-course.pdf (or similar) replace UsingPerl6-draft.pdf (unmaintained)? 16:14
masak _itz: I like that suggestion. 16:17
Ven ShimmerFairy: FWIW, my use case for no strict usually stems from -n and -p :)
ShimmerFairy: being able to go -e '%h{$_}++; END { say %h.perl }' or something is great.
16:18 rvchangue joined
abraxxa timotimo: i'd add a method that populates the private attribute if undefined (not sure what's the Perl 6 way to check this) and return it's value and always call the method, never the private attribute directly which will be only used for caching. does that sound sane? 16:18
timotimo you can put a state variable into that method
that way putting undefined values into that attribute is also a valid thing 16:19
and checking for defined values is just .defined
dha Which reminds me, I need to look at documenting C<state>....
abraxxa timotimo: that sounds safer than the private attr on the class, yes
timotimo mhm
a thing it does is put data that more or less belongs to the instance "far away" from that instance 16:20
16:20 Ven left
abraxxa not much docs on state ;( doc.perl6.org/language/variables#Th...Declarator 16:20
timotimo yeah, that's really not much 16:21
16:21 FROGGS left
jnthn dinner & 16:24
16:26 spider-mario joined
dha Almost nothing, in fact. 16:26
And, afaict nothing at all on C<$>.
abraxxa i want to loop from 1 to .field_count, what is the correct syntax? 16:27
for 1.. .field_count -> $field_index { ?
dha (By "almost nothing", I appear to mean "it's mentioned a couple of times in spec with no actual explanation")
|Tux| jnthn or timotimo if either of you would clone github.com:Tux/CSV and make t/78_fragment.t pass (CSV.pm line 123) I'd be grateful 16:29
abraxxa timotimo: Natively typed state variables not yet implemented
|Tux| I give up
I already made 18000 extra tests pass today.
16:29 ab5tract_ left
abraxxa |Tux|++ 16:32
16:40 FROGGS joined
ugexe m: sub foo($a?) { once &?ROUTINE() }; foo(1); 16:41
camelia rakudo-moar 90e4da: OUTPUT«Memory allocation failed; could not allocate 6144 bytes␤»
timotimo abraxxa: you mean for 1 .. $_.filed_count? 16:44
field*
16:45 pierrot joined
abraxxa timotimo: for 1 .. self.field_count { 16:47
16:49 Ven joined
timotimo then you need "self." :) 16:49
abraxxa timotimo: i've overcome the non-implemented state for native vars by using an Int and storing the ub4 to it: github.com/abraxxa/DBIish/commit/9...936311804c 16:50
timotimo ah, state variables with natives is difficult 16:51
i think state vars internally rely on setting a c-level null to the variable
which for native ints is a valid value (for num i'm not so sure)
16:54 mohij joined
[Coke] dha: I think $ is just a variable with an empty name. 16:55
m: my $ = 3; say $; 16:56
camelia rakudo-moar 90e4da: OUTPUT«(Any)␤»
dha It's supposedly the anonymous state variable.
m: for 1..5 {say $++} 16:57
camelia rakudo-moar 90e4da: OUTPUT«0␤1␤2␤3␤4␤»
[Coke] it can be used with state, but I don't think it's exclusive to state.
right. that's not state there.
dha Hm. That is not what I've been told, but that may not mean anything.
[Coke] not all info in #perl6 is equal. :) 16:58
dha When I asked "what the heck is C<$>?" TimToady said it was the anonymous state variable.
rindolf Hi all, sup?
[Coke] well, timtoady is moar likely to be right about that sort of thing, of course. 16:59
dha Right. Hence my assumption that it was true. :-)
timotimo [Coke]: $ is an anonymous state variable 17:00
m: sub test() { $++ }; say test for ^5
camelia rakudo-moar 90e4da: OUTPUT«0␤1␤2␤3␤4␤»
pink_mist m: for 1..5 { say $++; say $++; } #it's not quite state ... it's ... weird
camelia rakudo-moar 90e4da: OUTPUT«0␤0␤1␤1␤2␤2␤3␤3␤4␤4␤»
pink_mist an anonymous state variable basically
oh, timotimo already said that 17:01
timotimo++
dha ...and this is why we need documentation. scene. :-)
dha goes off to steal the documentation on C<state> from Perl 5... 17:03
[Coke] timotimo: yes, but I'm saying it works without explicitly declaring "state" 17:06
Are you saying that referencing $ is the same as "state $" ? 17:07
17:07 dbron_ joined
dbron_ .clear 17:07
timotimo yes, that's what i'm saying
17:07 pmurias left
timotimo $ is short for "(state $)" 17:07
17:08 dakkar left
dbron_ At the beginning of the year, Larry intimated we might see a beta 1.0 in Sept .. that still in the cards? 17:13
17:13 rangerprice joined
rangerprice How can i redirect the output of stderr in a variable ? 17:14
FROGGS dbron_: we still try very hard
abraxxa the C function definition for the pointer to the column name is void *attributep,
ugexe rangerprice: IO::Capture::Simple
ambs well, he said something could get out to his birthday in september, yeah
abraxxa is that NativeCall definition ok? CArray[int8] $attributep,
ambs and said that about 5 days ago :D
Ven rangerprice: change the dynamic variable that represents the output 17:15
ugexe: it's not thread-safe :(
ugexe right, but he can look at the source and see how
dbron_ FROGGS: from what I've seen, you guys go beyond "try very hard" to "never sleep"
timotimo abraxxa: why won't a Str do? is this for null-byte-terminated strings?
dbron_ FROGGS: but did you mean to say "we will still try very hard [to get a RC out]"? 17:16
abraxxa timotimo: return by pointer like the whole OCI
using Str results in a segfault because I guess not enough memory is allocated
timotimo oh
dha Re: C<state> - gist.github.com/dha/02f1f41b5f8937c0271e 17:17
abraxxa you pass it a pointer to a CArray[int8] and a Carray[ub4] which contains the length in bytes
timotimo you may want just Pointer instead 17:18
at least for the CArray[ub4]
abraxxa that's the part that works ;)
timotimo right
abraxxa fetching the error message with docs.oracle.com/database/121/LNOCI/...LNOCI17287 works
timotimo the thing about CArray is that it needs to know the size in order to work 17:19
abraxxa fetching the column name with docs.oracle.com/database/121/LNOCI/...LNOCI17130 doesn't
17:19 rangerprice left
abraxxa OraText *bufp, vs void *attributep, 17:19
timotimo i ... don't actually know exactly how we do this
with shaped array declarations, i expect something like casting to a shaped CArray
abraxxa shaped?
17:19 pat_js left
timotimo to be honest, i'd get a Pointer[int8] and increment it length-of-text times and grab each byte individually into another array or directly into a Uni or something string-ish 17:20
shaped arrays have the size(s) as part of their type 17:21
jnthn is working on that right now
abraxxa i see
timotimo 163056 jnthn │ perl6-m -e "my @arr := Array.new(:shape(3, 3)); @arr[1;1] = 42; say @arr[1;3]"
163102 jnthn │ Index 3 for dimension 2 out of range (must be 0..2) in block <unit> at -e:1
abraxxa can i allocate a CArray[int8] with n number of elements right now? 17:22
2015.07.2
timotimo well, you can just .new it and assign to the n-1th element 17:23
abraxxa i do it with my @col_name := CArray[int8].new; @col_name[$_] = 0 for ^512;
timotimo no need to set to 0 for the initial ones; i *think* we allocate CArray from zero'd memory blocks
m: use NativeCall; my $arr = CArray
camelia ( no output )
timotimo loops
abraxxa so that should suffice? @col_name[511] = 0;
timotimo m: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; say $arr
camelia rakudo-moar 90e4da: OUTPUT«IntTypedCArray[int8].new␤»
timotimo huh? 17:24
m: use NativeCall; my $arr = CArray[int8].new; $arr[0] = 0; say $arr
camelia rakudo-moar 90e4da: OUTPUT«IntTypedCArray[int8].new␤»
abraxxa yeah, I also get that output
timotimo m: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; .say for $arr.list
camelia rakudo-moar 90e4da: OUTPUT«IntTypedCArray[int8].new␤»
abraxxa wasn't able to output its contents using .perl
timotimo that's a bug, potentially from GLR
star: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; .say for $arr.list
camelia star-m 2015.03: OUTPUT«IntTypedCArray[int8].new()␤»
timotimo o_O
star: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; .say for $arr
camelia star-m 2015.03: OUTPUT«IntTypedCArray[int8].new()␤»
timotimo star: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; $arr[$_].say for ^10;
camelia star-m 2015.03: OUTPUT«0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤»
timotimo star: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; $arr[$_].say for ^20;
camelia star-m 2015.03: OUTPUT«0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤»
timotimo star: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; $arr[1000 + $_].say for ^100; 17:25
camelia star-m 2015.03: OUTPUT«0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤…»
timotimo hehe
star: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; $arr[1000000 + $_].say for ^100;
camelia star-m 2015.03: OUTPUT«0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤…»
timotimo star: use NativeCall; my $arr = CArray[int8].new; $arr[10] = 0; $arr[10000000000 + $_].say for ^100;
abraxxa note that I use :=
camelia star-m 2015.03: OUTPUT«0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤0␤…»
timotimo yes, but i use $ instead of @
so this doesn't let us access random memory areas
it probably does a bounds check and returns 0 if it's oob
anyway, i gotta go AFK, do some grocery shopping and cooking 17:26
abraxxa ok, thanks
17:26 Ven left
abraxxa although I didn't graps the difference between $ary = and @ary := 17:26
timotimo for one, @foo = ... is list assignment, so you might end up with a list containing the CArray as its only element; if not, it'll likely try to grab the list inside the CArray and put its values into the array 17:27
with := you replace the array completely 17:28
17:28 pierrot left
timotimo m: use NativeCall; my @foo = CArray[int8].new(1, 2, 3); say @foo.WHAT; say @foo[0].WHAT; 17:28
camelia rakudo-moar 90e4da: OUTPUT«Default constructor for 'IntTypedCArray[int8]' only takes named arguments␤ in block <unit> at /tmp/Z0Uzrg509X:1␤␤»
timotimo m: use NativeCall; my @foo = CArray[int8].new(); say @foo.WHAT; say @foo[0].WHAT;
camelia rakudo-moar 90e4da: OUTPUT«(Array)␤(IntTypedCArray[int8])␤»
timotimo there we go
timotimo AFK
abraxxa why not my CArray[int8] @foo;? 17:29
m: use NativeCall; my CArray[int8] @foo; say @foo.WHAT;
camelia rakudo-moar 90e4da: OUTPUT«(Array[IntTypedCArray[int8]])␤»
abraxxa unexpected!
isn't CArray a subtype of Array? 17:30
17:30 tokuhiro_ joined
japhb m: use NativeCall; my CArray[int8] $foo; say $foo.WHAT; 17:32
camelia rakudo-moar 90e4da: OUTPUT«(IntTypedCArray[int8])␤»
japhb abraxxa: ^^
abraxxa ah, typo, thanks! 17:33
japhb abraxxa: You declared an array of CArrays.
abraxxa so @ instead of $ is nothing else as wrapping the type in Array[]?
japhb Well, it vivifies to one. But really the @ twigil is a constraint. 17:34
17:34 tokuhiro_ left
[Coke] m: my ($,$a,$) = 1,2,3; say $a; # timotimo - even here? 17:34
camelia rakudo-moar 90e4da: OUTPUT«2␤»
japhb abraxxa: And 'my Bar @foo' declares that you want an array-like thing containing Bar elements 17:35
m: my @foo = 1, 2, 3; say @foo.^roles;
camelia rakudo-moar 90e4da: OUTPUT«Method 'iterator' must be implemented by Iterable because it is required by a role␤ in any compose_method_table at src/gen/m-Metamodel.nqp:2666␤ in any apply at src/gen/m-Metamodel.nqp:2676␤ in any compose at src/gen/m-Metamodel.nqp:2812␤ in an…»
japhb m: my @foo = 1, 2, 3; say @foo.^roles>>.name;
camelia rakudo-moar 90e4da: OUTPUT«Method 'name' not found for invocant of class 'Positional'␤ in block <unit> at /tmp/sroFYNmsWB:1␤␤»
japhb m: my @foo = 1, 2, 3; say @foo.^roles>>.^name; 17:36
camelia rakudo-moar 90e4da: OUTPUT«List␤»
japhb hmmm
skaji Hi, I have a question about types.
japhb Either there is still some fixup to do in the metamodel post-GLR, or my metamodel skills have become very out of date.
skaji perl6 -e 'sub test(Positional) { say "called" } my $f = gather for ^5 { take }; test($f)' 17:37
prints "called".
But $f is of type Seq, and Seq ~~ Positional is False. Why does test($f) dispatch test(Positional) subroutine?
17:38 Ven joined
timotimo coke, have a look at the --target=ast to find out 17:39
skaji m: sub test(Positional) { say "called" }; my $f = gather for ^5 { take }; test($f);
camelia rakudo-moar 90e4da: OUTPUT«called␤»
dha Could someone take a quick look at this and let me know if it's good enough to slot into the docs (documentation for C<for>)? 17:41
gist.github.com/dha/d8ab9b8cf852d358bfaf
skids There's something wrong with MMD in that respect I think, I had to work around similar issues when doing slice = sequence code. 17:42
And others have noted something funny going on but no golfs or bisects yet.
(wrt test(Positional)) 17:43
m: sub test(Positional) { say "called" }; my $f = gather for ^5 { take }; $f.WHAT.say; say $f ~~ Positional; test($f); 17:44
camelia rakudo-moar 90e4da: OUTPUT«(Seq)␤False␤called␤» 17:45
17:45 yqt left
skids m: sub test(Positional $a) { $a.WHAT.say; say "called" }; my $f = gather for ^5 { take }; $f.WHAT.say; say $f ~~ Positional; test($f); 17:46
camelia rakudo-moar 90e4da: OUTPUT«(Seq)␤False␤(List)␤called␤»
skids m: sub test(Positional \a) { a.WHAT.say; say "called" }; my $f = gather for ^5 { take }; $f.WHAT.say; say $f ~~ Positional; test($f);
camelia rakudo-moar 90e4da: OUTPUT«(Seq)␤False␤(List)␤called␤»
FROGGS dbron_: I say that we attempt it, but tbh I doubt we will have a RC ready in time...
_itz japhb: see scrollback from a few hrs back 17:48
re: metamodel
colomon has managed to once again forget how to .grep all strings that DON’T match a regex. 17:50
FROGGS m: say <foo bar baz quux>.grep(* !~ /a/) 17:51
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/cjpruPyvSK␤Unsupported use of !~ to do negated pattern matching; in Perl 6 please use !~~␤at /tmp/cjpruPyvSK:1␤------> 3say <foo bar baz quux>.grep(* !~ 7⏏5/a/)␤»
FROGGS m: say <foo bar baz quux>.grep(* !~~ /a/)
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/CwdZxoaqi6␤Variable '&infix:<!~~>' is not declared␤at /tmp/CwdZxoaqi6:1␤------> 3say <foo bar baz quux>.grep(* 7⏏5!~~ /a/)␤»
FROGGS m: say <foo bar baz quux>.grep({$_ !~~ /a/})
camelia rakudo-moar 90e4da: OUTPUT«(foo quux)␤»
FROGGS colomon: ^^
colomon FROGGS++
colomon was just about to try that. it always seems too long, somehow….
m: say <foo bar baz quux>.grep(* !~~ /a/) 17:52
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/aE6reeski3␤Variable '&infix:<!~~>' is not declared␤at /tmp/aE6reeski3:1␤------> 3say <foo bar baz quux>.grep(* 7⏏5!~~ /a/)␤»
colomon ah, FROGGS++ again
skids (* ~~ /a/).WHAT.say; (* !~~ /a/).WHAT.say 17:53
m: (* ~~ /a/).WHAT.say; (* !~~ /a/).WHAT.say
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/IlsFSTLp3g␤Variable '&infix:<!~~>' is not declared␤at /tmp/IlsFSTLp3g:1␤------> 3(* ~~ /a/).WHAT.say; (* 7⏏5!~~ /a/).WHAT.say␤»
skids m: (* ~~ /a/).WHAT.say;
camelia rakudo-moar 90e4da: OUTPUT«(WhateverCode)␤»
abraxxa is there any difference between Pointer and Pointer[void]? 17:54
FROGGS abraxxa: no
abraxxa FROGGS: i guessed so, thanks
what's the reason to support the later and even document it? 17:55
FROGGS the reason to support Pointer[void]? it just happened to work so we can disallow using the value of a void pointer directly... and it is more explicit which can be nice 17:56
abraxxa ok 17:57
skaji skids: Thanks. Is Seq actually composed by Positional?
skids m: my $a = Int; Array[$a].new().perl.say # that would be a reason though it likely won't work for Pointer/natives yet.
camelia rakudo-moar 90e4da: OUTPUT«Array[Int].new()␤»
abraxxa OCI seems to fill the memory the pointer points to with something, at least it's 0 before and crap after the call
abraxxa wants proper unsigned native types 17:59
dha Are there currently improper ones?
abraxxa dha: they are typed ones
skids skaji: Seq is specifically not Positional IIRC, but it does allow postcircumfix:<[]>, which was not until recently, and before that only Positional did that, so maybe something got a little haywired. 18:00
dha Ah.
abraxxa say; results in Unhandled exception: Cannot unbox a type object instead of printing a linefeed
an exception occuring inside a for loop prints the line number of the for loop beginning 18:02
japhb _itz: thank you for the Metamodel issues pointer 18:03
.ask jnthn What did you mean by "Maybe .gist needs to go in the no-pun list."?
yoleaux japhb: I'll pass your message to jnthn.
18:04 Ven left
skids skaji: I think the problem is that :(Positional) does the same thing that :(@a) does, actually, and manages to turn the Seq into a list, because... MAGIC! 18:06
m: sub test(\a where Positional) { a.WHAT.say; say "called" }; my $f = gather for ^5 { take }; $f.WHAT.say; say $f ~~ Positional; test($f); 18:07
camelia rakudo-moar 90e4da: OUTPUT«(Seq)␤False␤Constraint type check failed for parameter 'a'␤ in sub test at /tmp/QDGsWprplx:1␤ in block <unit> at /tmp/QDGsWprplx:1␤␤»
skids But, I've seen that also fail to work, specifically on invocants. 18:08
18:09 leont joined
skaji Oh, sub test(\a where Positional) failed! 18:09
18:11 telex left
FROGGS the jvm build is fun somehow: 18:11
$ ./perl6-j -e 'enum Foo <foo bar baz>'
===SORRY!===
Method 'role_typecheck_list' not found for invocant of class 'NQPClassHOW'
18:12 sjn_phone_ joined, espadrine left, telex joined
abraxxa i give up, it just doesn't work 18:13
FROGGS :/
abraxxa i can get the data type, the column name length but just not the column name itself 18:16
18:17 tokuhiro_ joined
FROGGS abraxxa: can you show me the code? 18:18
abraxxa FROGGS: github.com/abraxxa/DBIish/blob/mas...e.pm6#L558 18:19
i've pushed what I have so far
18:19 sjn_phone_ left
skaji skids: I understand Positional type checking is a little weird now. Thanks a lot! 18:19
skids skaji: yes that's a workaround, but it may break if you try it on invocants. I haven't got a handle on what is going on there yet. 18:20
np
18:22 tokuhiro_ left
skids skaji: do note that 'where constraint' clauses are definition-order-dependent and have a diffent place in the MMD pecking order. 18:22
18:22 dbron_ left
FROGGS abraxxa: why do you pass it a CArray[int8]? 18:24
skaji m: sub test($ where Positional) {}; sub test(Positional) {}; 18:28
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/HVpEXUmM3K␤Redeclaration of routine test␤at /tmp/HVpEXUmM3K:1␤------> 3 Positional) {}; sub test(Positional) {}7⏏5;␤ expecting any of:␤ horizontal whitespace␤»
timotimo returns
18:29 ab5tract_ joined
ab5tract_ |Tux|: Impressive :D 18:30
timotimo dha: is "use the arrow" with precedent in the docs? if not, i'd rather have it refered to as "a pointy block"
ab5tract_ But that sounds quite frustrating :)
skids m: multi sub test($ where Positional) {}; multi sub test(Positional) {}; 18:31
camelia ( no output )
dha timotimo - I can certainly do that. Although I think pointy blocks remain undocumented.
skaji skids: oh, sorry.
timotimo dha: also, it would probably have been better to use the "edit" function rather than write a comment, because it does strange things with C<for> and @foo in comments
ab5tract_ realizes that Convos++ does not currently demarcate between "cached stuff from hours ago" and "right now" 18:32
timotimo and the thing about "lazily by default" sounds strange
dha Yeah, I didn't notice that at first.
timotimo so in the comment it now reads "The C Loop" 18:33
we also have a c-style loop, so that's very confusing at first glance :D
abraxxa FROGGS: i need to get the bytes for a utf-8 encoded text, the length in bytes is returned in the var after that
dha FWIW, "lazily by default" comes from existing docs. specifically, 5to6.pod
timotimo mhm 18:34
dha timotimo - will look to fix that in the gist. thanks.
timotimo let me have a closer look
hm
it's kinda weird to write ti like that, because perl5 doesn't really have the same kind of "lazy list" that perl6 does 18:35
dha Well, that's the point, isn't it? That it's *now* lazy, so you would use C<for> instead of C<while>
timotimo also, while lines() would just run once 18:36
FROGGS abraxxa: hmmm, I'd try passing it a CArray[Pointer[Str]], and then 'say @col_name[0].deref'
timotimo m: while lines() { say "boing" }
camelia rakudo-moar 90e4da: OUTPUT«boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤boing␤»
timotimo actually, seems like no.
m: while lines() -> $_ { say $_.elems }
camelia rakudo-moar 90e4da: OUTPUT«20␤eof requires an object with REPR MVMOSHandle␤ in block <unit> at /tmp/0Suy4Ariam:1␤␤»
18:36 mprelude joined
timotimo whoops 18:36
FROGGS :D
timotimo m: lines(); lines();
camelia rakudo-moar 90e4da: OUTPUT«eof requires an object with REPR MVMOSHandle␤ in block <unit> at /tmp/RTZmsoJDfC:1␤␤»
FROGGS timotimo++
timotimo that's the golf, it seems
of course lines() without doing anything to it (like say "boing") won't actually eat any of the lines 18:37
abraxxa FROGGS: trying
timotimo this may be another place where ARGFILES isn't sufficiently like a regular file handle
FROGGS abraxxa: note that the param just need to be CArray[Pointer] 18:38
dha Granted, I took the code examples from 5to6.pod, so if they're wrong there, they'll be wrong in my doc. :-/
abraxxa FROGGS: still my favorite exception: Unhandled exception: Cannot unbox a type object 18:39
FROGGS abraxxa: did you @col_name[0].=new? 18:41
abraxxa FROGGS: @col_name[0] = Pointer[Str];
FROGGS abraxxa: no
abraxxa: you need an instance, not a type object
abraxxa my @col_name := CArray[Pointer[Str]].new; 18:42
FROGGS almost
m: use NativeCall; my CArray[Pointer[Str]] @col_name; @col_name[0].=new;
camelia ( no output ) 18:43
FROGGS m: use NativeCall; my CArray[Pointer[Str]] @col_name; @col_name[0].=new; say @col_name
camelia rakudo-moar 90e4da: OUTPUT«[TypedCArray[Pointer].new]␤»
FROGGS m: use NativeCall; my CArray[Pointer[Str]] @col_name; @col_name[0].=new; say @col_name[0]
camelia rakudo-moar 90e4da: OUTPUT«TypedCArray[Pointer].new␤»
FROGGS m: use NativeCall; my CArray[Pointer[Str]] @col_name; @col_name[0].=new; say @col_name[0].deref
camelia rakudo-moar 90e4da: OUTPUT«Method 'deref' not found for invocant of class 'TypedCArray[Pointer]'␤ in block <unit> at /tmp/iSS9MUMGrz:1␤␤»
FROGGS m: use NativeCall; my CArray[Pointer[Str]] @col_name; @col_name[0] = Pointer[Str].new; say @col_name[0].deref
camelia rakudo-moar 90e4da: OUTPUT«Type check failed in assignment to '@col_name'; expected 'TypedCArray[Pointer]' but got 'Pointer'␤ in block <unit> at /tmp/TK8AeLoT1Y:1␤␤»
FROGGS m: use NativeCall; my CArray[Pointer] @col_name; @col_name[0] = Pointer[Str].new; say @col_name[0].deref
camelia rakudo-moar 90e4da: OUTPUT«Type check failed in assignment to '@col_name'; expected 'TypedCArray[Pointer]' but got 'Pointer'␤ in block <unit> at /tmp/usknrEhTH6:1␤␤»
FROGGS hmpf 18:44
18:44 rurban joined
FROGGS m: use NativeCall; my CArray[Pointer] $col_name.=new; $col_name[0] = Pointer[Str].new; say $col_name[0].deref 18:44
camelia rakudo-moar 90e4da: OUTPUT«(Str)␤»
FROGGS abraxxa: I'd do that and get rid of the @-sigils 18:45
abraxxa FROGGS: gladly!
FROGGS psch: araund? 18:46
around*
abraxxa cannot unbox...
FROGGS abraxxa: can you push?
18:46 yqt joined 18:47 domidumont joined
masak wow, this is nice: ttp://okmij.org/ftp/Perl/Scheme-in-Perl.txt 18:47
masak wishes mvuets were on this channel so he would pass it on
18:48 vendethiel joined
pink_mist hands masak a missing 'h' 18:48
FROGGS pink_mist: nono, that's French
'ttp
pink_mist d'oh lala
masak ah, 'm sorry 18:49
okmij.org/ftp/Perl/Scheme-in-Perl.txt
18:50 rurban left
abraxxa FROGGS: a missing ; lead to the error 18:50
FROGGS ohh
psch FROGGS: semi-around, yeah 18:52
FROGGS: i have some room to pay attention, though :)
abraxxa FROGGS: after the call say $col_name[0].deref; triggers Internal error: unhandled target type
in sub nativecast at lib/NativeCall.pm:433
in method deref at lib/NativeCall.pm:102
FROGGS psch: what would you do here? $ ./perl6-j -e 'enum Foo <foo bar baz>'
===SORRY!===
Method 'role_typecheck_list' not found for invocant of class 'NQPClassHOW'
abraxxa: can you push? 18:53
psch FROGGS: uh... probably look who calls that method on a NQPClassHOW and why that should or shouldn't happen... :) 18:54
FROGGS hmpf
:o)
abraxxa FROGGS: pushed 18:55
psch FROGGS: i'm really confused by 376d7273bb344ec63ec1469846e4ef63fce45f57, fwiw
FROGGS: the check for SIG_ELEM_SLURPY_POS seems logically reversed to the equiv spot in BOOTSTRAP.nqp
FROGGS: but turning it around gives a StackOverflowError...
Metamodel/BOOTSTRAP.nqp:724 is what i'm comparing it to 18:56
abraxxa FROGGS: that's how the pointer is passed in C: (dvoid**) &col_name 18:57
18:57 xfix joined, xfix left, xfix joined
FROGGS abraxxa: yes, that would fit... though I cant explain the error you get 18:58
psch: hmmmm, interesting
psch FROGGS: Binder hunches! :P 18:59
abraxxa FROGGS: i guess what is missing is the encoding so it can decode from Buf to Str
FROGGS abraxxa: no, the error message indicates that it does not know what Str is, which is weird in many ways 19:00
abraxxa comes from sub nativecast in NativeCall.pm 19:01
FROGGS: i also tried Str is rw but that results in a segfault, i guess because no memory is reserved 19:04
FROGGS abraxxa: can you just print $col_name[0]?
abraxxa FROGGS: Pointer<0x8a99178> 19:06
why use a CArray[Pointer[Str]] and not Pointer[Pointer[Str]]? 19:07
FROGGS abraxxa: and now print: nativecase(Str, $col_name[0]) 19:08
abraxxa: dunno... might also "work" (for some definition of work, since it doesnt work yet)
abraxxa UNBELIEVEABLE!!!
the column names
FROGGS \o/ 19:09
abraxxa how does it know how to decode?
FROGGS abraxxa: it defaults to utf8
abraxxa: but you can apply a trait called 'is encoded' or so
abraxxa can it be set explicitly?
ämount works too ;) 19:10
FROGGS and btw, 'Str is rw' won't give you a char** AFAIK, so that's why you get a segfault there
:o)
19:11 zakharyas joined
FROGGS psch: I turned it around and build it now... 19:11
psch: it survived the setting...
19:11 rurban joined
FROGGS psch: but that would certainly explain the failure modes I see 19:12
psch FROGGS: nom and with your patch both threw StackOverflowError in Pod::To::Text iirc
FROGGS psch: sadly not... :o( I can build it, but it fails as usual
abraxxa FROGGS: i don't get why we need two pointers here, can you explain that please? 19:13
psch huh?
FROGGS psch: I'm able to build Pod::To::Text fwiw
abraxxa FROGGS: and why doesn't it require the length that's passed back as well?
psch FROGGS: and it's still bailing with the same un-flat @tags? 19:14
FROGGS abraxxa: you allocate a piece of mem to let the function store a value there, right? and it will store a char*
abraxxa: so you end up having a char**
abraxxa char* is an array of chars?
where one char is one byte, not one character? 19:15
19:15 rurban left
FROGGS psch: ohh! that fixes the flat() bug it seems! 19:15
abraxxa: aye, given that C does not know what a character is 19:16
dha So, you've flat()tened that bug?
dha hides
FROGGS :D 19:17
psch FROGGS: curious. as mentioned, for me it ran into StackOverflowErrors... but i guess your gist from yesterday isn't up-to-date with your current working state 19:18
FROGGS psch: I was able to work around the unflat @tags by changing 'flat a, b' to 'a, |b'
aye
19:25 domidumont left 19:31 thou joined
abraxxa FROGGS: and what i had before was just a char*? 19:34
and the values where memory addresses? 19:35
or parts of one memory address
[Tux] which operation returns a (.....).Seq ? 19:36
abraxxa FROGGS: does that declaration make a difference? CArray[Pointer[Str]] $attributep is encoded('utf8'),
pink_mist thought utf8 was already the default 19:38
19:38 cognominal left
abraxxa it is but OCI also allows UTF-16 and if that's what I need to use for whatever reason I'd like to find all places where I have to change the encoding 19:38
FROGGS abraxxa: before you got the char* back split across 8 slots of the CArray[int8] most likely 19:39
abraxxa: yes, I think that signature is correct 19:40
abraxxa and how does it know the length? null terminated?
FROGGS psch: do you want to push your gotcha?
psch: it really solved the flat() bug
abraxxa: aye
nine Seqs dispatching on Positional is intentional 19:41
[Tux] «return [ @in.map (-> @r { $%( @h Z=> @r ) }) ];» used to return [[< a b c>], [1,2,3], ...] but after GLR returns [[<a b c>], ([1,2,3],...).Seq]
nine Same as assigning Seq to Positional doing the right thing
[Tux] is that intentional?
nine m: my @a = <a b c>; my @b = 1, 2, 3; say (@a Z=> @b).perl; 19:43
camelia rakudo-moar 90e4da: OUTPUT«(:a(1), :b(2), :c(3)).Seq␤»
psch FROGGS: it didn't seem to fix anything when applied to nom HEAD, so feel free to group it with your other fixes
nine m: my @a = <a b c>; my @b = 1, 2, 3; say ($%(@a Z=> @b)).perl;
camelia rakudo-moar 90e4da: OUTPUT«{:a(1), :b(2), :c(3)}␤»
jnthn skaji: We simultaneously wanted to be clear about what are consumable produces of values and also allow a sub foo(@x) { } to be called on something like foo(@a.grep(...)). Seq itself can't be sanely Positional because you only get one shot at the values. 19:44
yoleaux 18:03Z <japhb> jnthn: What did you mean by "Maybe .gist needs to go in the no-pun list."?
FROGGS psch: k, will do
jnthn skaji: So we notice the types didn't match, and then fail over to binding the result of Seq.list; this actually memoizes the list it returns too
skaji: We actually have a role (PositionalBindFailover) that marks out Seq and HyperSeq as the types that have this behavior. 19:45
.tell japhb Normally if you call a method on a role then we pun it into a class and call the method on the pun; there's a small list of exceptions. 19:46
yoleaux jnthn: I'll pass your message to japhb.
jnthn |Tux|: .map returns a Seq after GLR 19:47
[Tux] just add .list at the end or something like that?
skids jnthn: down the road are we likely to be able to (/will we want to) discriminate between (Positional $foo) and (@foo) for PositionalBindFailover? 19:48
mohij Is there a dynamic variable telling the exact version of perl6 (the thing perl6 -v reports)? 19:49
yoleaux 05:01Z <tadzik> mohij: I'd be delighted :) Can you send me a PR?
timotimo m: say $*VM.version 19:50
camelia rakudo-moar 90e4da: OUTPUT«v2015.7.108.g.7.e.9.f.29.e␤»
FROGGS m: say $*PERL.compiler.version
camelia rakudo-moar 90e4da: OUTPUT«v2015.7.1.768.g.90.e.4.dab␤»
jnthn [Tux]: If you actually want a List, yes
FROGGS mohij: ^^
timotimo oh, right, VM.version is the moarvm version
jnthn skids: We don't at the moment; I don't see a strong reason to
[Tux] .list still leaves in the ()
nine skids: if you add a candidate that takes a PositionalBindFailover, it will probably bind more strongly than Positional
FROGGS [Tux]: then you perhaps want .Slip 19:51
skids nine: that was my only reason to want to, so if that works... but isn't it just a role too?
mohij Thanks!
FROGGS psch: damn, now I get the StackOverflowError too :o(
skids nine: OIC, I'll try that sometime. 19:52
FROGGS psch: but rerunning that single line made it pass -.-
jnthn Generally you just want to take Iterable :)
nine m: multi sub foo(PositionalBindFailover $a) { say "Seq"; }; multi sub foo(Positional $a) { say "Array"; }; foo((1,2,3).map({$_}));
camelia rakudo-moar 90e4da: OUTPUT«Ambiguous call to 'foo'; these signatures all match:␤:(PositionalBindFailover $a)␤:(Positional $a)␤ in block <unit> at /tmp/j9f8xBJ2lc:1␤␤»
nine Funny. Locally this works 19:53
[Tux] This Seq has already been iterated, and its values consumed 19:54
nine Oh, I don't have the dispatcher fix yet
skids jnthn: not when choosing an implementation that differs based on whether "user expects it to be memoized anyway."
dalek kudo/UnbreakJVMBuildFactory: 7315a8a | FROGGS++ | src/core/Variable.pm:
flatten export trait list, like we do for Routines
mohij .tell tadzik PR is out. Windows testing now...
yoleaux mohij: I'll pass your message to tadzik.
dalek kudo/UnbreakJVMBuildFactory: 1c313d2 | FROGGS++ | src/core/Inc.pm:
box -I string, we'll .split it later
 19:55
mohij Poll for Windows users: How do you develop and deploy Perl6 on Windows? What would your ideal develop/deploy workflow look like? 19:56
dalek kudo/UnbreakJVMBuildFactory: 91b7725 | FROGGS++ | src/vm/jvm/runtime/org/perl6/rakudo/Binder.java:
psch++, align JVM's slurp/slurp-flat code to moar's
FROGGS mohij: I dunno... deploy an msi? 19:57
mohij how would you update? Do you need more than one version installed?
jnthn skids: Do you have a concrete use case? 19:58
mohij *version of perl6
FROGGS mohij: more than one version of rakudo? usually not
mohij: upgrade would mean to run a newer msi, hopefully it would not replace my panda state, so I could rebootstrap
jnthn skids: My gut feeling is trying to deal with something other than Iterable or Positional is kinda odd
e.g. you typically want to iterate over the thing *or* index into the thing 19:59
psch FROGGS: hm, the SOE was consistent for me, no matter if called from make or manually via the CLI...
FROGGS psch: it was the first time I'd seen it
psch i had played with raising Xss to ridiculous values, but that just kept eating memory...
FROGGS: maybe your other changes push it somewhere less deterministic... :/ 20:00
FROGGS psch: yeah, I had quite some more say statements... and the SO appeared after removing them
skids m: perl6 -e 'multi sub a (Iterable $a) { 42.say; a($a.list) }; multi sub a (@a) { 43.say }; multi sub a ($scalar) { 44.say }; a((1,2,2,3).grep(2));' 20:03
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Yy9GNwTZ0x␤Two terms in a row␤at /tmp/Yy9GNwTZ0x:1␤------> 3perl6 -e7⏏5 'multi sub a (Iterable $a) { 42.say; a(␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ …»
skids m: perl6 -e 'multi sub a (PositionalBindFailover $a) { 42.say; a($a.list) }; multi sub a (@a) { 43.say }; multi sub a ($scalar) { 44.say }; a((1,2,2,3).grep(2));'
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/CgiHw7FzIN␤Two terms in a row␤at /tmp/CgiHw7FzIN:1␤------> 3perl6 -e7⏏5 'multi sub a (PositionalBindFailover $a␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ …»
skids oops.
m: multi sub a (PositionalBindFailover $a) { 42.say; a($a.list) }; multi sub a (@a) { 43.say }; multi sub a ($scalar) { 44.say }; a((1,2,2,3).grep(2)); # this one works
camelia rakudo-moar 90e4da: OUTPUT«Ambiguous call to 'a'; these signatures all match:␤:(PositionalBindFailover $a)␤:(@a)␤ in block <unit> at /tmp/4Pq2IrL1r2:1␤␤»
skids m: multi sub a (Iterable $a) { 42.say; a($a.list) }; multi sub a (@a) { 43.say }; multi sub a ($scalar) { 44.say }; a((1,2,2,3).grep(2)); 20:04
camelia rakudo-moar 90e4da: OUTPUT«X::Multi::Ambiguous exception produced no message␤ in block <unit> at /tmp/R0LkU1wES6:1␤␤»
skids So, nine++
FROGGS psch: I recompiled the setting and there is still no SOE
psch FROGGS: during Pod::To::Text is where i saw it usually 20:05
FROGGS psch: and it also seems like the next-in-split-hack isnt needed anymore
psch FROGGS: as in, only there, actually
FROGGS psch: aye, it compiled P2T and exploded in NativeCall
20:05 dha left
skids Oh weird, that frst one worked locally. I must be behind. 20:06
abraxxa is there an int64 or is long or longlong what i need?
20:07 darutoko left
FROGGS abraxxa: depends... what *do* you need? there are all of them 20:07
abraxxa signed bytes 8
FROGGS m: use NativeCall; say int64 20:08
camelia rakudo-moar 90e4da: OUTPUT«(int64)␤»
FROGGS m: say int64
camelia rakudo-moar 90e4da: OUTPUT«(int64)␤»
FROGGS abraxxa: long and longlong are variable in size
abraxxa the oratypes.h file says: ifdef __LP64__ define orasb8 signed long else define orasb8 signed long long
20:09 dha joined
nine So you want int64 20:09
abraxxa nine: thanks. can you document it please?
labster m: say (98/99).base(10, 1)
camelia rakudo-moar 90e4da: OUTPUT«010.0␤»
20:10 rindolf left 20:11 zakharyas left
dalek c: da7dc52 | FROGGS++ | lib/Language/nativecall.pod:
mention int64, abraxxa++
 20:12
FROGGS abraxxa / nine: done
abraxxa dalek++ 20:13
oh, dalek seems to be a bot
FROGGS++
mohij .botsnack
yoleaux :D
FROGGS aye, dalek reports commits to github 20:14
moritz .humansnack
those pesky, unreliable and unresponsive humans... :-) 20:15
mohij
.oO(send-me-a-beer-license)
nine :D
FROGGS :D 20:16
[Tux] t/90_csv.t .. ok
All tests successful
almost there
skids nine,jnthn: So that PositionalBindFailover trick worked on 16d366a2 and stopped working sometime thereafter. FWIW.
FROGGS nice
jnthn skids: You sure? 9fdaf03b0a4 is the one I'd have expected to change it, when I fixed the multi-dispatcher. 20:17
skids checks.
jnthn Positional and PositionalBindFailover are going to conflict
They have no relationship and so are equally narrow
20:18 tokuhiro_ joined
skids Oh, I never build that merge, so it must have been earler. 20:19
jnthn goes to rest &
skids 13f829c then
20:20 zoosha left 20:21 zoosha joined 20:22 tokuhiro_ left
FROGGS jnthn: rest well 20:22
[Tux] tux.nl/Files/20150908222424.png 20:24
20:24 cognominal joined
[Tux] rest later 20:24
skids should maybe try re-multifying the postcircumfix:<[]> stuff post-MMD changes. 20:27
ab5tract_ github.com/ab5tract/Terminal-Print...1729f4bd88 20:28
[Tux] test 50000 54.734 54.647 20:29
test-t 50000 54.701 54.614
ab5tract_ ^^ This commit reminded me of the days where I would figure out some kind of unexpected, list related solution and then post it here. The response was always "this will hopefully be better after the GLR"
20:30 Loren left
ab5tract_ It's nice that it's a memory, and even nicer that our hopes were well-founded! 20:31
nine Now we just have to get performance to where it should be :) 20:32
ab5tract_ nine: Text::CSV is a good canary for that, but give t/basics.t a spin in Terminal::Print and we have another good candidate for "something that could use speeding up" 20:33
tadzik computer, messages
yoleaux 19:54Z <mohij> tadzik: PR is out. Windows testing now...
tadzik mohij: how's it going? :)
20:34 popl joined
popl congrats on the release everyone 20:34
ab5tract_ there is a @colors.pick in there that is quite costly, but even without it .. well, it's a visual representation of some gains we can make :)
mohij Fine. :-) Hacking rakudobrew lifts the mood quite some. It's a nice thing hacking on.
tadzik I'm glad to hear that :) 20:35
smash .tell TimToady more interesting combinations tweag.github.io/HaskellR/ 20:36
yoleaux smash: I'll pass your message to TimToady.
timotimo popl: what, a release? 20:37
mohij Do you know whether rakudo has something like local::lib built in? Can panda install to a (different) local folder?
masak popl: not quite there yet ;) 20:38
popl Well, in October apparently.
tadzik mohij: yeah, panda respects DESTDIR iirc 20:39
popl timotimo: perl6releasetalk.ticketleap.com/per...lk/details is what I am referring to.
*to what I am referring
tadzik mohij: I'm actually curious if anyone ever used that feature though :)
except myself for testing
mohij halfway to a carton on perl6, yay \o/
masak popl: that talk is in October. the release is still scheduled (?) for Christmas. 20:41
...but feel free to have a peek at Perl 6 before that! :D
popl okay then, I retract my congratulations
timotimo ah
popl better? :P
timotimo no worse feeling in the world than being wrongly congratulated!! 20:42
popl apparently
leont Install destinations is something that will need more attention AFAIK (though I haven't looked at it in detail recently) 20:43
ugexe they exist already
timotimo is there a good reason against giving CArray a "new" candidate that takes values and constructs itself from them?
ugexe CompUnitRepo::Local::Installation.new($dist-dir) 20:44
20:44 espadrine joined 20:45 popl left
timotimo m: Buf[int64].new(123456).gist.say 20:46
camelia rakudo-moar 90e4da: OUTPUT«Buf[int64]:0x<1e240>␤»
timotimo ah, nice.
ab5tract_ timotimo: Maybe we've been waiting on shaped arrays for it?
ugexe i use it to test install/uninstall without touching rakudos default module locations
timotimo very unlikely 20:47
masak 'night, #perl6
timotimo m: Buf[num].new(123456).gist.say
camelia rakudo-moar 90e4da: OUTPUT«MVMArray: bindpos expected num register␤ in any at src/gen/m-Metamodel.nqp:1679␤ in block <unit> at /tmp/tR_jBDGgeb:1␤␤»
timotimo m: Buf[num].new(123456e0).gist.say
camelia rakudo-moar 90e4da: OUTPUT«This type cannot unbox to a native integer␤ in any at src/gen/m-Metamodel.nqp:1679␤ in block <unit> at /tmp/NbsSm2n0Ag:1␤␤»
timotimo interesting
buf expects to contain only inty things, but doesn't complain early when being created with nums
though we have num arrays 20:48
m: say 15.5e0.fmt("%02x").say
camelia rakudo-moar 90e4da: OUTPUT«0f␤True␤»
timotimo m: 15.5e0.fmt("%02x").say
camelia rakudo-moar 90e4da: OUTPUT«0f␤»
timotimo m: 15.5e4.fmt("%02x").say
camelia rakudo-moar 90e4da: OUTPUT«25d78␤»
timotimo that's not so happy, eh
m: 15.5e4.fmt("%f").say 20:49
camelia rakudo-moar 90e4da: OUTPUT«155000.000000␤»
skids wonders if a convenience for "make a GC-safe buf from the CArray and length parameters coming back from this nativecall" would be awesome, or just end up ugly.
timotimo m: 15.5e4.fmt("%s").say
camelia rakudo-moar 90e4da: OUTPUT«155000␤»
japhb timotimo: Buf is notionally serialized data ready for writing to the wire. Perhaps it doesn't support non-integers because floating point doesn't have a well-standardized wire format? Or maybe it's just a NYI.
yoleaux 19:46Z <jnthn> japhb: Normally if you call a method on a role then we pun it into a class and call the method on the pun; there's a small list of exceptions.
japhb Ah 20:50
timotimo japhb: i agree it's a language design question
japhb still wonders why given all the imperfect standards people *have* been willing to make, that we haven't just standardized on IEEE floating point bit patterns as the default standard for floating point on the wire 20:52
timotimo i didn't know we hadn't, tbh
japhb Aside from endian concerns (which are there for two's-complement integers anyway), it's basically about as "de facto standard" as you get.
timotimo oh, yeah, endianness 20:53
japhb timotimo: All sorts of standards assume silly things like taking the IEEE data, unioning it with an appropriate int, and then serializing that. 20:54
20:54 Ven joined
japhb Or going "screw it, we're just going to output scientific notation in ASCII" 20:54
timotimo m) 20:55
japhb (Although oddly, it turns out that scientific notation in ASCII + gzip can compare well with IEEE bit patterns for file size on real-world inputs) 20:56
(The Collada 3D interchange format does this, FWIW. Arrays of floating point are just space-separated ASCII, and then it's assumed you gzip the file.) 20:57
timotimo oh, interesting 20:59
21:00 kaare_ left, kaare_ joined
timotimo perl6 -e 'use NativeCall; my $buf = CArray[int8].new; $buf[5] = 10; $buf.gist.say' 21:03
IntTypedCArray[int8]:0x<00 00 00 00 00 0a> 21:04
21:04 skids left
timotimo (this also makes .elems and .list work) 21:04
i hope people are OK with these change 21:05
changes
21:05 [Sno] left 21:07 kaare_ left, [Sno] joined 21:08 rurban joined
timotimo and a .new that you can pass values to 21:09
21:09 rurban left 21:10 xfix left 21:12 FROGGS left
timotimo perl6 -e 'use NativeCall; my $buf = CArray[int8].new(1, 2, 3, 4); $buf.perl.say' 21:12
IntTypedCArray[int8].new(1, 2, 3, 4)
^- just like this
21:14 FROGGS joined
dalek kudo/nom: 10dcbad | timotimo++ | / (2 files):
give CArray more methods you'd expect from arrayish things

  .gist and .perl for int and num, .elems and .list and also
a method new that noms values into the CArray.
comes with tests.		 21:17
timotimo abraxxa: ^ you may like this
abraxxa timotimo: looking 21:18
(comes with tests)++
timotimo :)
abraxxa timotimo: i wonder why you tested 4 out of the 5 array elements ;) 21:19
timotimo that's ... a secret!
abraxxa my number 1 wish is to improve the 'Unhandled exception: Cannot unbox a type object' error message to point at a line number
21:19 TEttinger joined
timotimo oh 21:20
can you give me a golf for that?
psch .tell FROGGS fwiw, the SOE golfs to 'unit class Foo; sub foo is export { }'
yoleaux psch: I'll pass your message to FROGGS.
abraxxa timotimo: the most boring car ever? 21:21
psch .tell FROGGS and the "unit" isn't even needed
yoleaux psch: I'll pass your message to FROGGS.
timotimo is golf that boring?
i think golf is a good car
i did my driver's license training in a golf, then a passat. both good cars 21:22
abraxxa good !~ boring
timotimo ah 21:23
of course
psch m: sub f { say &?ROUTINE.name }; f
camelia rakudo-moar 90e4da: OUTPUT«f␤»
psch m: sub f { say &?ROUTINE }; f
camelia rakudo-moar 90e4da: OUTPUT«sub f () { #`(Sub|61128792) ... }␤»
abraxxa not that i think that Golf is a good car...
timotimo: an undeclared variable lead to the 'Unhandled exception: Cannot unbox a type object' error 21:24
labster Golf is the minimum viable car? 21:25
21:25 nowan_ is now known as nowan
abraxxa m: die "undeclared var $var"; 21:25
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Gs9lSmyBJT␤Variable '$var' is not declared␤at /tmp/Gs9lSmyBJT:1␤------> 3die "undeclared var 7⏏5$var";␤»
abraxxa m: use NativeCall; die "undeclared var $var";
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/UELAXa3noF␤Variable '$var' is not declared␤at /tmp/UELAXa3noF:1␤------> 3use NativeCall; die "undeclared var 7⏏5$var";␤»
21:26 audy joined, TEttinger left
labster m: no strict; die "undeclared var $var"; 21:26
camelia rakudo-moar 90e4da: OUTPUT«Use of uninitialized value $var of type Any in string context in block <unit> at /tmp/jaGnXolcx2:1␤undeclared var ␤ in block <unit> at /tmp/jaGnXolcx2:1␤␤»
shinobi-cl Hi all 21:27
labster hi shinobi-cl
abraxxa is there a syntax to say if $var == 'A' or 'B' { ? 21:28
21:28 TEttinger joined
shinobi-cl i have a small module i want to upload, that allows to have an array and reference it by rows and columns. Also allows to use headers for the columns instead of numbers. And it is static. You can search on the columns, and generate an indes (basically, a hash of arrays) . What do you think on naming it "Data::StaticTable" ? Too generic maybe? 21:28
colomon abraxxa: you mean like if $var == 1 | 2 { ? 21:29
or possibly if $var eq ‘A’ | ‘B’ { ?
abraxxa colomon: if that works, yes
i tried: elsif $dty == (SQLT_INT, SQLT_NUM) {
those are constants
seems to work 21:30
colomon m: my $a = 10; if $a == 9 | 10 | 11 { say “yup”; };
camelia rakudo-moar 90e4da: OUTPUT«5===SORRY!5=== Error while compiling /tmp/nZAjT0ljdZ␤Bogus statement␤at /tmp/nZAjT0ljdZ:1␤------> 3my $a = 10;7⏏5 if $a == 9 | 10 | 11 { say “yup”; };␤ expecting any of:␤ prefix␤ term␤»
21:30 muraiki left
colomon Not sure why camelia doesn’t like that, but except for the quotes around yup it is correct. :) 21:31
pink_mist the ctrl+h character might be the reason
abraxxa can i define a variable without a type and change the type later? 21:32
psch m: say “yup”; 21:33
camelia rakudo-moar 90e4da: OUTPUT«yup␤»
pink_mist colomon: that is, you embedded a literal backspace character in your line
21:33 Ven left
leont Fun fact: until very recently dollar control-character was a valid variable in perl 5. Not a feature to steal ;-) 21:36
vendethiel remembers leont talking to people about this at APW14, and our eyes getting bigger and bigger *g* 21:37
TEttinger the unicode byte order marker, aka zero-width space, is a valid identifier part in java and a valid identifier in clojure and scala
I strongly suspect it would need to be valid in JVM perl at least 21:38
psch TEttinger: why? 21:39
TEttinger because there would be JVM names you couldn't call otherwise in perl6
is there a way to eval on the jvm perl6 instead of moarvm with camelia? 21:40
psch TEttinger: i'd say you'd just have to call them with the quoted method name syntax
TEttinger: jvm is broken currently, didn't quite survive the glr merge i think
TEttinger ah 21:41
psch TEttinger: it also depends on whether java introspection (and indy) properly reproduce the zero-width space
TEttinger m: say "y\ufeffay"
camelia rakudo-moar 10dcba: OUTPUT«5===SORRY!5=== Error while compiling /tmp/fICSnk6Vdy␤Unrecognized backslash sequence: '\u'␤at /tmp/fICSnk6Vdy:1␤------> 3say "y\7⏏5ufeffay"␤ expecting any of:␤ argument list␤ double quotes␤ term␤»
psch TEttinger: if they don't we don't get it while building the name, if they do... i'm not actually sure what happens then :)
TEttinger what... is the unicode 4-hex-digit escape in perl6? 21:42
psch m: say "y\x[feff]ay"
camelia rakudo-moar 10dcba: OUTPUT«yay␤»
TEttinger m: my $yay = 10; say $yay; 21:43
camelia rakudo-moar 10dcba: OUTPUT«5===SORRY!5=== Error while compiling /tmp/suoI05cxzn␤Bogus postfix␤at /tmp/suoI05cxzn:1␤------> 3my $y7⏏5ay = 10; say $yay;␤ expecting any of:␤ constraint␤ infix␤ infix stopper␤ postfix␤ …»
psch m: say "y\c[ZERO WIDTH SPACE]ay"
camelia rakudo-moar 10dcba: OUTPUT«y​ay␤»
TEttinger it appears to be invalid in moarvm
psch TEttinger: i don't think the BOM is the zero width space, though
TEttinger unfortunately, it is a whole mess of unicode history 21:44
psch m: say chr "\c[ZERO WIDTH SPACE]"
camelia rakudo-moar 10dcba: OUTPUT«Cannot convert string to number: base-10 number must begin with valid digits or '.' in '⏏​' (indicated by ⏏)␤ in block <unit> at /tmp/Ycyy_f6mxZ:1␤␤Actually thrown at:␤ in block <unit> at /tmp/Ycyy_f6mxZ:1␤␤»
psch ehh, how was that
m: say unival "\c[ZERO WIDTH SPACE]"
camelia rakudo-moar 10dcba: OUTPUT«NaN␤»
psch m: say ord "\c[ZERO WIDTH SPACE]"
camelia rakudo-moar 10dcba: OUTPUT«8203␤»
psch m: say (ord "\c[ZERO WIDTH SPACE]").base(16)
camelia rakudo-moar 10dcba: OUTPUT«200B␤»
TEttinger www.fileformat.info/info/unicode/ch.../index.htm 21:46
m: say ord "\c[ZERO WIDTH NO-BREAK SPACE]" 21:47
camelia rakudo-moar 10dcba: OUTPUT«65279␤»
psch ah, no-break
21:48 n0tjack left
TEttinger and it's in an arabic presentation forms block, of all things 21:48
timotimo abraxxa: can you give me a test case that leads to that error?
m: my int $foo = Int 21:49
camelia rakudo-moar 10dcba: OUTPUT«Cannot unbox a type object␤ in block <unit> at /tmp/srIZT25vKA:1␤␤»
timotimo like that? that has a line number to go with it
21:50 rangerprice joined
rangerprice How can i redirect the stderr stream to stdout ? 21:51
timotimo you can my $*ERR = $*OUT
moritz $*ERR = $*OUT
colomon pink_mist: I assure you it was not intentional; indeed, I wouldn’t know how to do it if I wanted to. 21:54
21:54 rurban joined
pink_mist colomon: I'm not surprised; the only reason I went into so much detail is because a lot of irc clients won't even show you it happened; take a look at the irclog linked in the topic for example; you can't see that it's there at all in that ... but it's right before the 'if' 21:55
colomon: which is also where camelia reported the error
abraxxa timotimo: didn't find out what causes it, but if you edit DBIish you can force it easiliy 21:57
i've pushed all i've done today to my repo 22:00
timotimo: it happens for example if you remove the ; in line 688 in lib/DBDish/Oracle.pm6
timotimo abraxxa: something i just remembered, you could also have a native int "is rw" parameter for the size that the call fills in for you 22:02
abraxxa i'm done for today, bye!
timotimo gnite abraxxa! 22:03
22:03 Possum left
rangerprice And if i want to redirect the stderr to a variable, how i can do ? 22:03
abraxxa timotimo: yeah, that should work too!
22:04 rurban1 joined
abraxxa timotimo: also good night and thanks! 22:04
22:04 abraxxa left 22:05 rurban left, rurban1 left 22:06 rurban joined
japhb rangerprice: ISTR there is an IO::String module available via panda (or zef, I suppose) 22:06
22:09 cognominal left 22:11 dha left 22:12 laubster left
psch .tell FROGGS the mentioned golf calls sink from EXPORT_SYMBOL, goes through a bunch of pull-, push-, and reify- methods and then ends in a new, new, bless, BUILDALL loop... 22:13
yoleaux psch: I'll pass your message to FROGGS.
22:14 dha joined 22:16 FROGGS left, hernanGOA left 22:20 tokuhiro_ joined, FROGGS joined 22:22 nightfrog joined 22:24 tokuhiro_ left 22:25 lolisa joined 22:33 espadrine left 22:40 rurban left, mohij left
thou is 'use v6' deprecated now? 22:50
Trying to understand why i got a pull request removing that from files
gfldex r: use v6;
camelia ( no output )
..rakudo-jvm 10dcba: OUTPUT«Can't open perl script "/home/camelia/rakudo-inst/bin/eval-client.pl": No such file or directory␤»
ShimmerFairy thou: it's not deprecated or discouraged; I'd like to see this PR, sounds interesting 22:54
thou github.com/softmoth/p6-Template-Mu...he/pull/11
ugexe rakudobrew doesn't seem to work on OSX anymore 22:55
thou mj41 is Michal Jurosz
ShimmerFairy thou: I'd change those C<v6;> lines to C<use v6;> (just my preference, and it's also a use statement vs. Version object in sink context), but aside from that I'd reject the PR, esp. without a rationale given. 22:56
labster didn't 'v6;' previously mean "no strict;" and does it still? 22:58
23:00 leont left
thou labster, I think you're right. I can't seem to find that in the synopses though. 23:00
23:01 vendethiel left
ugexe v6 used to turn strict on (when it was not default for -e for example) 23:01
timotimo i think it still means that
m: v6; $x = 10
camelia rakudo-moar 10dcba: OUTPUT«5===SORRY!5=== Error while compiling /tmp/EhZz3isSlG␤Variable '$x' is not declared␤at /tmp/EhZz3isSlG:1␤------> 3v6; 7⏏5$x = 10␤»
timotimo apparently it doesn't any more
ugexe maybe it was `use v6` only then 23:02
probably remembering -Mv6
ShimmerFairy thou: I definitely recall something about v6 being in sink context but not getting warned about that, or something :) 23:03
23:05 bbkr_ joined
thou Yeah. I can't find any reference to bald 'v6;' on design.perl6.org now; I'll reject that pull request but update all to 'use v6;' 23:06
ShimmerFairy I'm surprised to not see the bare 'v6' mentioned anymore, but it does make sense that we wouldn't want that to be magical :)
rangerprice Do you know a generic approach to see what shell execute the command inside the backstick (``) ?
23:07 skids joined, bbkr left
thou rangerprice: Does 'set -x' do what you want? 23:07
maybe I misunderstood your question... 23:08
rangerprice I would like to know what shell execute the `command` 23:09
pink_mist presumably the shell in your /etc/passwd file if on *nix 23:10
maybe?
geekosaur you can test that easily enough with something like perl6 -e '`ps -fp$$`.say', Id' think 23:11
I'd also think /bin/sh but...
hm, might need to escape the $s
geekosaur should update rakudo on this machine at some point >.> 23:12
ShimmerFairy I don't think we even have backticks
nope, you'd have to do qx`...` if you really wanted to use backticks in P6 :)
rangerprice Is backstick more efficient than exec() or system() ? 23:13
ugexe m: say %*ENV<SHELL>
camelia rakudo-moar 10dcba: OUTPUT«/bin/bash␤»
japhb built all just before the glr merge, and is afraid to rebuild his main perl6 tree until the ecosystem fallout dies down, lest he break it and be sad
geekosaur less, I'd expect, since it connects to a pipe and captures output
japhb rangerprice: backtick is for capturing stdout.
rangerprice i didn't yet learn perl 6, actually i'm on perl 5 ! 23:14
geekosaur using `` as a shorthand for "system" is an old perl5ish code smell
japhb rangerprice: I answered for Perl 5, because I guessed as much. :-)
ShimmerFairy japhb: interesting, I'm in pre-glr because the allomorphs branch I've been working on started before the merge. It's not by choice that I don't have GLR yet :P
pink_mist rangerprice: you might want #perl then, not #perl6 ...
japhb ShimmerFairy: Yeah, understood. Looking forward to getting the allomorphs in the main tree, that'll be one of my really old wishes come to fruition. 23:15
rangerprice pink_mist: oh thank ! but i will also stay here, after all, i'm a great lambda camel 23:16
japhb rangerprice: Pugs devotee?
ShimmerFairy japhb: I "just" need to change how I do val() at compile-time (last time I tried I got too annoyed with one problem that cropped up), and then I'd definitely merge it in. I could merge it in now, in fact, it's just not handling compile-time val() in a DRY way
japhb nod, saw something about using World lookup to get the setting version of val() in the backlog 23:17
ShimmerFairy japhb: if you didn't catch the recent commits (I pushed the branch a while back), I did put back in your original code, which did improve things by a couple seconds :)
rangerprice japhb: yes
23:18 rangerprice left
ShimmerFairy (And doing that showed me that val()'s implementation wasn't the issue with the slowdown anymore, just the sheer number of calls to val() that happen) 23:18
23:19 rangerprice joined 23:23 ab5tract_ left
ShimmerFairy Also something that's come up is whether certain <>/«» lists, like %hash<subscripts>, should have :v done on them like everything else, or if :v should only be done in places where you as the user could coerce resulting values 23:23
(if the latter, then you'd need %hash{<subscripts>} to get val processing on a <>/«» list, because there you could do stuff like %hash{<subscripts 42>».Numeric}) 23:25
Certainly something I'd like TimToady's opinion on, I think it might be a nicer way than the mostly-implied "<>/«» does :v everywhere all the time" :) 23:28
japhb ShimmerFairy: Thank you for putting my code back, I appreciate that. I wonder how much my original tuning has rotted over the years.
ShimmerFairy japhb: keep in mind that my original code ran my one sorting benchmark to 90s on average, so the "couple seconds less" is after I did a bunch of optimizing changes to my version :) 23:29
japhb ShimmerFairy: I do think it's sane to say that hash subscripting (and perhaps all places that <> is pretty clearly just syntactic sugar) does not deserve :v without a couple extra characters of effort.
ShimmerFairy japhb: and when I put it in, I only changed your code enough to work in val() (of course), and made \i required on NaN\i and Inf\i :) 23:30
japhb: there are some tests that fail because your code is more permissive on "slash rationals", but I haven't done anything about that (fixing neither tests nor code) pending some sort of decision on that 23:31
japhb A decent change. I hadn't bothered, mostly based on "until someone cares one way or the other, don't force the decision"
ShimmerFairy The \i required was just to fix some tests I had un-TODOd back when my version was still in place, and I figured making \i required was better than re-TODOing (and it was a simple enough change, so no big deal) :P
japhb The slash rational code was intentionally permissive (I happen to like that, because I like .Numeric and constant folding to produce the same value if possible), but it's more preference than requirement 23:32
It just seems easier to explain that way.
japhb attempted to think ahead to not causing jnthn, masak, et al. pain in their classes. :-) 23:33
ShimmerFairy japhb: I personally wouldn't mind if the only limit was "only one slash in a slash rational", like we only allow one point in a point rational :)
And if for some reason we wanted val() to handle "<1/2>", then we'd want that to be as restrictive as actual <1/2> in code, so that's where the more restrictive view could go :P 23:36
japhb nodnod 23:37
23:39 Urchin left 23:41 lolisa left 23:43 spider-mario left
ShimmerFairy japhb: in terms of tuning, I wonder how much unboxing might contribute to that; you have a lot of literal numbers in your version, which are mostly likely read as Ints by the parser at first, which then have to be unboxed for the nqp::ops. If that is contributing greatly to it, I'm not sure how that could be fixed, though :( 23:46
japhb Hmmm. Back in The Old Days(tm) ISTR it used to be that literal numbers directly used as nqp::op arguments where just handled directly, no box/unbox. But it may be that that was only an optimization which we may not do anymore. :-/ 23:52
Might be worth looking at the --target=optimize on that.
23:53 mprelude left
japhb And as for how to work around that, we could create some scoped native constants and use them that way, I suppose. Seems like we ought to fix the optimizer though, if indeed it broke. 23:54
ShimmerFairy japhb: wouldn't know, that's just my naive guess based on what I know of parsing, which isn't everything :) . It may very well still be optimized for nqp:: calls (and perhaps any parameter marked as a native type, who knows?)
japhb rangerprice: Ah, missed your response amongst the val() discussion. Glad to see there are still Pugs herders about, I still have a soft spot for it. :-) 23:55
geekosaur still wants to poke at it for real someday 23:56
geekosaur still keeps getting butt kicked by it >.>
ShimmerFairy japhb: just checked perl6 -Mnqp --target=optimize -e 'say nqp::add_i(42, 0)' , and I see Int WVals just like in --target=ast :(
japhb imagines that a brain dump from au++ might result in melting the surrounding brains
ShimmerFairy: Aww, dang
That would certainly explain less than awesome performance. 23:57
(well, at least contribute to it)
ShimmerFairy japhb: more precisely, Wants with a WVal(Int) and a IVal(42)
japhb: my naive "implict unboxing will be happening" guess is why I didn't bother switching to less-readable codepoint numbers in my version. I figured implied unbox_s couldn't be much worse than implied unbox_i :P 23:58
23:59 rmgk left

Powered by Cro and the Raku® Programming Language.

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

Thank you!