»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.
TimToady someone needs to bump NQP if jnthn's fix is to get in 00:06
00:07 telex left 00:08 telex joined
dalek kudo/nom: 01bba75 | TimToady++ | tools/build/NQP_REVISION:
bump NQP revision
 00:10
TimToady I'll bump it, please revert if that screws up rakudo release
00:17 telex left 00:18 telex joined 00:25 araujo joined 00:26 laouji joined, telex left 00:28 telex joined 00:34 telex left, telex joined, telex left 00:36 telex joined, BenGoldberg joined 00:38 laouji left 00:39 laouji joined 00:41 laouji left, laouji joined 00:42 telex left 00:44 telex joined 00:47 aborazmeh joined, aborazmeh left, aborazmeh joined 00:51 telex left 00:52 telex joined 00:53 vendethiel left 00:55 tokuhiro_ joined
[Coke] I have to rerelease nqp and retag things anyway. :| 00:57
TimToady okay
[Coke]++ 00:58
00:59 xinming_ joined 01:01 xinming left 01:18 adu joined 01:23 mprelude left 01:26 telex left, telex joined
dalek p: 64f4dae | coke++ | VERSION:
bump VERSION to 2015.09.1
 01:40
01:42 telex left 01:43 travis-ci joined
travis-ci NQP build passed. Will "Coke" Coleda 'bump VERSION to 2015.09.1' 01:43
travis-ci.org/perl6/nqp/builds/80944296 github.com/perl6/nqp/compare/8fd2e...f4dae96d88
01:43 travis-ci left 01:44 telex joined
[Coke] nqp release 2015.09.1 uploaded. 01:45
01:45 larion left
dalek kudo/nom: 06d132a | coke++ | docs/announce/2015.09.md:
add contributors for 2 months
 02:05
kudo/nom: 258f54c | coke++ | docs/release_guide.pod:
mark release as released in docs
kudo/nom: aee232e | coke++ | docs/announce/2015.09.md:
use moar unicode. add changelog info
kudo/nom: d22b3b9 | coke++ | tools/build/NQP_REVISION:
[release] bump NQP revision
 02:07
kudo/nom: cdc2981 | coke++ | VERSION:
[release] bump VERSION
02:11 telex left 02:12 telex joined
[Coke] ponders trying to run the Inline::Perl5 tests as part of the release process. :( 02:24
02:25 kid51 left
[Coke] is thinking that the I:P5 tests shouldn't be in roast if I:P5 is not in core. 02:25
02:26 uvtc joined
[Coke] does panda know how to deal with perl6-m vs. perl6-j ? 02:26
uvtc How can I create `['a', 'b', 'c', 'd', '1', '2', '3', '4']`? This doesn't work: 02:33
m: my @a = 'a' .. 'd', '1' .. '4'; say @a;
camelia rakudo-moar cdc298: OUTPUT«["a".."d" "1".."4"]␤»
TimToady m: my @a = flat 'a' .. 'd', '1' .. '4'; say @a;
camelia rakudo-moar cdc298: OUTPUT«[a b c d 1 2 3 4]␤»
uvtc Ooh. Thanks. 02:34
I like that the repl shows me what expressions evaluate to. But why do I have to type `say @a.WHAT` instead of just `@a.WHAT`? 02:36
ShimmerFairy [Coke]: I personally think rakudo is too happy to have required P6 modules for functionality: I::P5, Linenoise so the REPL is functional, and the default interface for the debugger come to mind. 02:37
uvtc ShimmerFairy, not sure if you're replying to me, but I do indeed have Linenoise installed. 02:38
ShimmerFairy uvtc: not you, just mentioning it in general :)
ShimmerFairy afk
dalek ast: b0f743e | coke++ | / (2 files):
capture test reality during 2015.09 release
 02:41
[Coke] ShimmerFairy: If it's optional, the tests probably shouldn't be in roast. 02:43
02:43 telex left 02:44 telex joined
[Coke] so many jvm failures. :| 02:47
02:49 noganex_ joined 02:52 noganex left 02:53 telex left 02:54 telex joined 03:01 kaare_ joined
adu serialization is a beast 03:07
[Coke] I am tempted to put in a note in the release that says "Sorry, there are some failures in the JVM backend this month, please bear with us." 03:10
adu I like JVM, it's how I plan to deploy Perl6 in a few years, but for development I always use Moar 03:14
03:18 skids joined 03:19 dj_goku joined, thou left
dalek kudo/nom: c0e0c97 | coke++ | docs/announce/2015.09.md:
Too many JVM failures to fix or fudge before rel
 03:24
flussence [Coke]: IMO that's perfectly fine for this month, it's supposed to be a beta after all :)
[Coke] Yup, we can only do so much. if it were a smaller number of failures, I would try harder, but running stresstest on JVM is already crazy. 03:25
We can fudge or fix a few a day, hopefully.
TimToady JVM is allowed to play catchup between now and Christmas
[Coke] is the phrasing in that last commit acceptable? 03:26
I feel like we should hang a lampshade on it, but I don't want to apologize too much or promise too much in terms of fixing it, either.
03:29 telex left, telex joined, telex left 03:30 telex joined
ShimmerFairy [Coke]: I have to say, it's always weird to see "You don't have Inline::Perl5" in the stresstest run. Since I don't plan on working with P5 (much less working on 5/6 interop), I'm not sure when I'd ever need to run I::P5 tests. 03:33
[Coke]: If I::P5 represents something that's supposed to be part of core in the future (that is, the ol' "use v5" emulation thingy mentioned in S01), then perhaps it should be core-ified soon :) 03:34
[Coke] ShimmerFairy: as someone who commits to the core and wants us to deliver a working product each month, that's when. :)
ShimmerFairy [Coke]: true, but that should be an R* concern more than a just-the-compiler concern, no? 03:35
03:42 _28_ria joined 03:45 samb1 joined
[Coke] if you only run your tests once every N months, that's not awesome. 03:46
anyway, I think the crux is: "are they core or not". 03:47
03:50 vendethiel joined 03:54 perlawhirl joined
perlawhirl hi all 03:55
can i get a little help/clarification on regexes
[Coke] Perl 5 or Perl 6? 03:56
We're mostly about Perl 6 here.
perlawhirl 6
i've made a cli tool that accepts user input... some of which is converted to regex... so building a regex from a string 03:57
ShimmerFairy [Coke]: The way I see it, if Inline::Perl5 is meant to become 'use v5' eventually, then keep it. Otherwise, its tests should get its own repo :)
perlawhirl i'll explain more... but here's my prob laid out: pastebin.com/7M8Z1Nqj
[Coke] perlawhirl: ok.
perlawhirl it seems that even compiling a regex is rx:s//, i still can't use it in a regex match
unlike perl5's qr, which could have a non /x compiled regex inside a /x match and it would still work 03:58
03:58 aborazmeh left
perlawhirl so as a workaround i have to subst all spaces with \s before compiling. am i doing something wrong or is that a bug? 03:58
by "compiling" i mean... "creating regex object" 03:59
03:59 nucp6 joined
[Coke] OOC, why are you starting with strings and not... regexes? 04:00
nucp6 Good morning perl6 !
[Coke] ah. because it's a user input.
perlawhirl yes D
:D
ShimmerFairy perlawhirl: have you tried just an :s (that is, remove the :i and see what happens). Also try switching around the :s and :i 04:01
perlawhirl fyi... it emulates a Cisco IOS cli, if you are familiar... "show run | i pattern"
flussence m: say anon regex { ^ match this\s+please$ }
camelia rakudo-moar c0e0c9: OUTPUT«Potential difficulties:␤ Space is not significant here; please use quotes or :s (:sigspace) modifier (or, to suppress this warning, omit the space, or otherwise change the spacing)␤ at /tmp/zCp1Vtbe7s:1␤ ------> 3say anon regex { ^ match7…»
flussence m: say anon rule { ^ match this\s+please$ } # this might work... 04:02
camelia rakudo-moar c0e0c9: OUTPUT«rule { ^ match this\s+please$ } # this might work...␤»
ShimmerFairy perlawhirl: also, try it without the m// around $re, since $re already has a regex object in it
flussence m: say " match this please" ~~ rule { ^ match this\s+please$ }
camelia rakudo-moar c0e0c9: OUTPUT«｢ match this please｣␤»
flussence m: say " matchthis please" ~~ rule { ^ match this\s+please$ }
camelia rakudo-moar c0e0c9: OUTPUT«Nil␤»
flussence does that help?
perlawhirl hmmm... didn't know about the 'rule' wrapper... will try that 04:03
flussence rule is basically regex with more "floaty" whitespace rules; anything literal that matches \s+ in input will match \s+ in output too 04:04
perlawhirl ShimmerFairy: can you please repeat your first suggestion, my client converted some of your text to emoticons
which are now disabled
[Coke] jvm got worse in the tarball test. j-test failing with: 04:05
===( 119;31 0/? 0/? )=============================================ld: warning: object file (12-sizeof.o) was built for newer OSX version (10.10) than being linked (10.4)
ShimmerFairy perlawhirl: have you tried just an :s (that is, remove the :i and see what happens). Also try switching around the :s and :i
there you go :)
perlawhirl thanks
flussence m: say " matchthis please" ~~ rx:ss { ^ match this\s+please$ } # this might work too
camelia rakudo-moar c0e0c9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Y7s7AZyArI␤Adverb ss not allowed on rx␤at /tmp/Y7s7AZyArI:1␤------> 3ase" ~~ rx:ss { ^ match this\s+please$ }7⏏5 # this might work too␤»
[Coke] sorry, not failing. warning.
flussence bah, never mind
ShimmerFairy flussence: :ss is for subst (samespace)
flussence I know a rule is just syntactic sugar for *something* you can tack on a plain regex... 04:06
(and it works for that purpose, cause I can remember it :)
perlawhirl also... just a suggestion, the error msg advises that it failed "at EVAL_0:1", which is LTA
flussence I'm guessing EVAL_0 would be the <$pattern> thing... agreed that that could be better. 04:07
ShimmerFairy must be because EVAL is used somewhere internally
flussence: token is regex with implied :r (:ratchet, disable backtracking), rule is regex with implied :r and :s (:sigspace) 04:09
flussence ShimmerFairy++ # thanks for reminding me :)
that EVAL looks like it comes from src/core/control.pm:211, I wonder if there's enough information there to say “oh btw, this is a regex” 04:10
well I guess it'd have to, to know to interpolate the string as regex code... 04:11
perlawhirl none of the suggestions seem to work. I'm ok with using subst as a work around, but it could trip others up as it doesn't quite behave like qr//. 04:12
04:14 vendethiel left 04:17 telex left 04:18 telex joined, khw left
nucp6 racudo: say ( ( 5 * 8 + 2 ) ) ~~ 42; 04:18
[Coke] rakudo tagged, signed uploaded. 04:19
flussence m: say 'a' ~~ rx{ { note &?CALLER::ROUTINE ~~ Regex } . };
camelia rakudo-moar c0e0c9: OUTPUT«5===SORRY!5=== Error while compiling /tmp/8IOOTnSZjG␤Undeclared routine:␤ &?ROUTINE used at line 1␤␤»
flussence m: say 'a' ~~ regex { { note &?CALLER::ROUTINE ~~ Regex } . };
camelia rakudo-moar c0e0c9: OUTPUT«True␤｢a｣␤»
[Coke] rakudo.org/downloads/rakudo/rakudo-....09.tar.gz 04:20
email away.
skids perlawhirl: it's probably best not to just throw user input into a pattern, anyway. The Grammar::BNF package is a bit hard to understand but shows how to build up a grammar from input using the metamodel.
That is, assuming you're expressions are simple and youre not trying to bring P6 regex syntax into the cli. 04:21
perlawhirl yes just basic patterns, no p6 syntax... hmmm, i wonder if it would work differently if i make the patterns P5? what is the suffix again? is it just :P5 ? 04:23
[Coke] the possum on my deck seemed awful friendly. :P 04:24
sorry, "opossum".
TEttinger [Coke]: they don't fear people because they know they wouldn't eat them. a possum's defense of playing possum also involves smelling like a dead rotting carcass 04:25
flussence perlawhirl: yeah, that'll work. You'll still be eval'ing user input in either case though...
TEttinger sorry, "virginia opossum"
[Coke] My concern is that he and my cat seemed chummy.
TEttinger they have a nasty bite but they don't know how to use it 04:26
[Coke] GAH and my other cat just snuck in and rubbed up on my leg, scaring the crap out of me. :)
04:26 rmgk_ joined, rmgk left, rmgk_ is now known as rmgk
TEttinger haha 04:26
[Coke] good thing the release is done.
skids perlawhirl: are you actually doing an IOS emu, or was that just an example, out of interest?
04:27 BenGoldberg left
perlawhirl i am actually doing one... it's not exactly an "emulator"... we do weekly captures of our managed devices of a whole bunch of "show" commands... about 20-30 commands 04:27
rather than searching the text file manually
TEttinger [Coke]: what sucks is that australia gets stuff like brush possums, which pollinate flowers and would never hurt a fly, and we get... opossums. which are nasty and proud of it. 04:28
perlawhirl i build a tool that loads the logged file and responds to cli command, and spits out the relevant output
i built it in perl5... already... trying it out in perl6 as a way to learn the syntax
[Coke] TEttinger: australia had to get -one- nice animal.
skids perlawhirl: ah, that sounds like an interesting way to do things. 04:29
perlawhirl i built a pipeline and everything... splits at the rindex of ' | ' and then calls recursively calls the subroutine again
[Coke] heads offline.
skids
.oO(oh good /S32-num/power isn't my local patch... cuz that would have been bizarre)
 04:30
perlawhirl skids: yeah it's a lot easier... if i want to check the output of "show ip route" from 8 weeks ago, can load it into my tool and just run the command... can filter output with include, exclude, begin or section 04:31
skids: resolves partial commands (eg, 'sh int desc'), has command completion... i purposely colour the prompt so people don't confuse it for a live device, lol 04:33
skids The command completion is something I've had trouble finding a good solution for, even with perl6 rx. 04:34
04:34 uvtc left
skids (there are solutions, but none of them are very clean) 04:34
perlawhirl yeah... tell me about it... i tried a few of the readline modules (perl5) and they work well enough... but i wanted MORE control... so i wrote a (sortof) readline using Term::ReadKey... that way i can control what every key does 04:35
04:36 tokuhiro_ left
perlawhirl but even so... Term::Readline allows you to write a completion function. 04:36
the function receives the ($line, $word, $lineindex) and then you build a list of completions from that and return it. 04:37
well... this is on linux... so once you also install Term::Readline::GNU things work well enough... but from what i've read, getting a decent readline on windows is a bit harder 04:40
04:52 telex left 04:54 telex joined 04:55 skids left 05:03 perlawhirl left 05:11 rurban joined
nine ShimmerFairy: Inline::Perl5 is already Rakudo's handler for use Foo:from<Perl5>; and EVAL "say 'hello'", :lang<Perl5>; The user doesn't even have to explicitly load it. 05:15
ShimmerFairy nine: hm. I think that serves as evidence for my "rakudo too happy to require modules" position :P . Are there plans for I::P5 to be integrated as a part of CORE eventually? 05:16
nine ShimmerFairy: the question is: should backwards compatiblity with Perl 5 be part of the Perl 6 specification (i.e. roast)? And I'd strongly argue for it. A Perl 6 compiler has to be backwards compatible and indeed, they all can be, as they have to support NativeCall already. And Inline::Perl5 is not Rakudo specific at all. 05:18
ShimmerFairy: I'd be all for including Inline::Perl5 in CORE. It's just years of moving away from Perl 5 attachment ("let's rewrite these perfectly working Perl 5 build scripts in Perl 6 just because") that I have to work against. 05:19
ShimmerFairy nine: of course it's not rakudo specific, I mean that the rakudo compiler shouldn't require you install certain modules beyond the compiler itself. (Linenoise is another example of what I mean.) If it's meant to be a core part of P6, include it in the damn compiler.
"let's rewrite these perfectly working Perl 5 build scripts in Perl 6 just because" ?
nine So I'm not pushing too much.
ShimmerFairy: just so they are written in Perl 6 instead of Perl 5. 05:20
The "this is a completely separate language, let's forget we ever had something to do with Perl 5" mindset
ShimmerFairy nine: any specific examples? And why not? I'd write a P6 version of something in a heartbeat over using P5, since I don't use P5 :)
nine ShimmerFairy: of course writing new things in Perl 6 is perfectly sane. I'm talking about rewriting perfectly working code that doesn't need changing. 05:21
ShimmerFairy nine: That's close to why I'd argue against P5 interop being core. P6 is no longer the direct successor to P5, there's no reason why it has to provide P5 support standard. It made more sense in the beginning, but not so much anymore.
(And besides, wasn't backwards-compatibility something we specifically avoided in designing P6? :P)
nine It's still required by S01 and tested in the S01 spec tests. 05:22
ShimmerFairy Also, so what if someone rewrites their code in a new language? I don't see how that's a problem.
nine It's simply a waste of time when there's still so much else to do.
ShimmerFairy nine: and so I'd argue that it shouldn't be mentioned in S01 anymore. Nobody's going to accidentally use P6 when expect P5, not with how history has turned out now. 05:23
Again, so what if someone "wastes" their time to rewrite their code?
nine: to be clear, I don't mind seeing P5 and P6 work a bit more closely together; I won't mind efforts towards that. From my viewpoint, however, I'm not familiar with P5, I entered into P6 long since the old "this will succeed P5" days, so I'm usually against preserving vestiges of that older mindset :) 05:26
That whole "Acts like Perl 5 _until_" list that is (or at least was) in S01 is an example of the vestiges I'm referring to.
(In other words, I don't think P5 deserves to be treated more specially than any other foreign language is in Perl 6) 05:27
mst ShimmerFairy: as soon as you convince everybody to stop trying to share CPAN because perl5 shouldn't treat perl6 specially, I'll consider that argument 05:28
but since we're being asked to treat you specially, I think quid pro quo applies
ShimmerFairy ? "we're being asked to treat you specially" ? 05:29
mst we are. lots of perl5 infrastructure is being patched to make perl6 a first class citizen
05:29 aborazmeh joined, aborazmeh left, aborazmeh joined
mst personally, I'm totally happy with this 05:29
I like the fact that the two perl communities have decided to co-operate like this 05:30
I think it's going to be really good fun
you don't have to care, but I don't see why you should actively oppose volunteers doing things they think is fun in the name of working together 05:31
ShimmerFairy mst: sure, like I said I won't get the way of working more on interop. I think in the case of Inline::Perl5, it either should be a part of CORE, or not be referenced by rakudo at all. This inbetween is quite a design smell to me.
mst well, what about making rakduo default to loading Inline::X for from<X> ?
anyway, any time you say 'should', there's an implied expectation other people will do the work 05:32
figure out which you like best and write the patches, and then we can discuss it
otherwise you're basically just urinating into the gale
ShimmerFairy mst: hmm... I think a generic system like that would be nicer than special-casing Perl 5 currently. (We do the same thing with the --doc CLI argument, at least)
mst and sapping other people's time and motivation via futile arguments
then you're welcome to implement it
if you do it well, I'll help convince people wiser than me to review and consider it
but otherwise, gale, urination, WOFTAM, etc. 05:33
ShimmerFairy I don't think it'd be that hard, if Inline::* had a standardized API (Don't know if it does already). You'd just have to construct the module name (which, again, is already done with --doc=HTML -> Pod::To::HTML)
mst then less arguing with people trying to make things better and more making things bettere 05:34
if it won't be that hard, I hope to see your patch soon :)
05:35 telex left
ShimmerFairy mst: I think the reason why I tend to argue hard against P5-related stuff is simply because I've no interest in it, and I don't like when the old "direct sequel" thing creeps back up. I've been very annoyed every time the design docs left something unexplained, because it was implied to be just like P5. 05:35
05:36 telex joined
mst it comes across as "I don't like what you're doing in your free time, please do what I want instead" 05:36
ShimmerFairy (Modernizing the specs is something I've thought about doing, btw. Not to abruptly remove any mention of P5 or anything like that, but just to make it read less like it was written in the early 2000s ☺)
mst I suggest instead that you just -do- the thing you want to see
rather than being mean to the people doing other things
your constant negativity isn't achieving anybody's goals, yours included 05:37
05:37 adu left
ShimmerFairy mst: I _never_ mean it like that, though I will admit that voicing my opinions strongly is not something I'm well practiced in, so I'm likely making a lot of mis-implications in doing so. 05:37
mst well, you pretty much always sound like that 05:38
hence why I've taken over so nine can go back to doing something useful
rather than listen to you tell him how to spend his free time
ShimmerFairy And I think stuff like "If not mentioned, it's just like it was in P5" is what kinda burned me on Perl 6 stuff being unavoidably linked in various cases.
mst yes, and if you wrote patches to fix that, I would absolutely support it 05:39
in spite of being a perl5 person with fairly minimal direct interest in perl6
but attacking people for making interop better doesn't help fix that at all
ShimmerFairy mst: I'm sorry, but I have to disagree that I "always sound like that". I was never putting down nine++'s excellent work on Inline::Perl5, and Inline::Perl6 IIRC, though if I came across as that way then I apologize.
mst it just makes them feel sad and you look like the rear end of a horse
yopu did come across that way 05:40
every time I see you talking in here you come across that way
let me rephrase: to me, every time I've seen you speak on this topic, you've sounded like that
less tellingg other people what they 'should' do, more doing, and you'll end up with a better result
ShimmerFairy Yes, I'm aware that on the subject of interop I've tended to state my opinions quite strongly. And as a note, when I use the word 'should', I'm not using that word to tell other people what to do. I see that word as a not-hypothetical "would" or "could", or a not-requiring "must" 05:42
mst given the combativeness of the rest of your tone, that isn't how it comes across 05:43
05:44 telex left
ShimmerFairy mst: I don't argue how it came across to others, I'm just saying that since I didn't use the word "should" in reference to anyone in particular (the only place I can see above is "the rakudo compiler shouldn't"), I don't quite see how that would turn into a "you should" 05:46
05:46 telex joined
mst and I'm just saying I don't care what you're nitpicking your meaning to after the fact, I care that you're saying things that will demotivate people who're contributing 05:46
and I'd like you to stop doing that, and instead contribute something
05:46 Alina-malina left
mst driving people who care about interop away from perl6 won't help anybody achieve anything 05:47
your original motivation doesn't make your words any less cutting
I don't really care whether it's actual malice or just terrible social skills, I just want you to not do it anymore :)
ShimmerFairy mst: Like I said, I won't argue how it comes across, I'm just hoping to explain what I meant better, to further understanding :) 05:48
mst I already guessed what you thought you meant
but that wasn't the problem
and doesn't actually help anything
ShimmerFairy And just so you know, I do contribute, and regardless I'm not a fan of this kind of "if you won't light a candle, then you better not make a peep about how dark it is" thinking. (That's why, from what I've heard of OpenBSD, I'm not inclined to ever be involved in that community.) 05:49
mst funny, you remind me of the OpenBSD community 05:50
that's rather why I was annoyed
TimToady you guys are repeating yourselves...
mst ShimmerFairy: that's ... not what I was saying. I'm saying "don't complain about other people lighting candles just because you don't like the colour of the light" 05:52
but, yeah, we're going in circles now
on the upside, nine's now gone to do something more useful than having this conversation
mst stops
ShimmerFairy In any case, sorry nine if I upset you. I never meant to attack your work (which as I understand has been impressive so far), I was voicing my perspectives on P5 interop (being someone who doesn't really need it) too strongly there. 05:53
mst ShimmerFairy++ # that's all I was trying to point out
oh!
ShimmerFairy: actually, I have a blog post I wrote when I realised I'd done exactly the same thing somewhere else 05:54
ShimmerFairy: shadow.cat/blog/matt-s-trout/passion-anguish/
please don't take that as an attack, you'll find I'm actually one of the examples of the failure mode :)
ShimmerFairy mst: and with my candle metaphor earlier, I was trying to mention that your "write a patch" suggestions remind me of how some people tend to be "well then, why don't YOU go do it?" to every suggestion (which is what I heard OpenBSD's programmers were like, from elsewhere) 05:56
05:57 rurban left
mst yes, I understood that 05:57
05:57 tping left
mst that's why I gave you an alternative metaphor 05:57
05:58 domidumont joined
ShimmerFairy I think that's a nice alternative, and you could describe the clashing of opinions as people constantly changing the colors back and forth. (For advanced candles, naturally.) 05:58
mst I appreciate your wanting certain colours of candle lit 05:59
but as the passion/anguish thing mentions
it's really easy to slip from there into accidentally criticising people for lighting their own candle of a colour you're not as fond of
or at least having them feel like you did, even if you weren't wanting to go that way
I would like you to retain all of that passion, but put it into encouraging people to write more candles of the colours you like 06:00
if you see what I mean
(I still get this wrong often enough, if I'm good at spotting this failure mode, it's because I've provided myself with lots of practice ;)
ShimmerFairy Yes, and I can definitely agree that P5 interop is (for whatever reason) a weak spot in my usually calmer self :) . I can't even recall _why_ it's like that for me, perhaps that "implied to be like in P5" got me more than I think? 06:01
06:02 domidumont left
ShimmerFairy should probably do something towards that spec modernization soon, so it's not something she can complain about anymore :P 06:02
06:03 domidumont joined, telex left 06:04 telex joined
mst ShimmerFairy: on the other hand, it's a grand historical tradition 06:05
I'm pretty sure I could find you bits of the perl5 docs that say "and the rest of the behaviour is obvious because it works just like awk" ;)
flussence don't worry too much about the authority of the specs, those got semi-demoted to design docs a few months ago. P6's official definition is the test suite now (...like P5, oh the irony!) 06:07
ShimmerFairy mst: sure, the modernization I have in mind isn't to wipe traces of Perl 5, it's just to make things sound less like a perl future-changelog written in 2003 ("X [from P5] is now Y [in P6]", for instance), and to explain things that were left unexplained before.
flussence: really? When was this decided?
mst I dunno, I get really annoyed by those bits of the perl5 docs too
but I always get distracted by something else before I get round to tweaking them 06:08
flussence I think it was one of the earlier YAPCs...
06:08 diana_olhovik joined
arnsholt pmichaud has been wanting to make the test suite the authoritative source for a while, if memory serves 06:08
ShimmerFairy mst: basically, I'd like to make explicit stuff that's left out with just a tiny little "if not mentioned, like P5" note in S01, and to make comparsions between P5 and P6 not so temporal, or "direct successor" sounding :) 06:09
mst OTOH fixing the synopses would be good anyway since they're likely to form more of a basis for documentation going forwards
flussence oh yeah, that was it. pmichaud announced the "specs" repo was going to be renamed at some point without warning :)
ShimmerFairy (e.g. "X is now Y" -> "Perl 6's version of X is called Y", or "X is gone" -> "", since stuff that doesn't exist in P6 don't need to be mentioned in the P6 design docs, right? :P) 06:10
I think the design docs, even if demoted, are worth updating. And it's easier to read documentation for how P6 works than source code :) 06:11
mst exactly
also, from the p5 side of aisle, removing 'direct successor' sounding stuff will make us happy too
and from the "I have lots of friends working on both" 'side', removing 'direct successor' stuff means significantly less likelihood of having to stamp down silly misunderstanding fueled drama yet again 06:12
and I think every community cat herder on either project is real bored of that now :D
06:14 telex left
ShimmerFairy It's still worth explaining stuff in comparison to P5, it would just sound less crufty if the comparison wasn't so temporal, as if Perl 6 is still the direct successor :) 06:14
06:15 telex joined, telex left 06:16 telex joined
mst yeah, well, it's only since ... 06:18
yeah, well, it's only since ... '09, IIRC, that we finally got general consensus to stop pretending it was the direct successor and instead start collaborating again rather than having a bunch of unproductive fights
(which is one of the reasons I'm highly in favour of the interop stuff, just because I think human factor wise it's going to be really important to avoiding a bunch of silly drama, again) 06:19
ShimmerFairy Heh, I think 2009 or 2010 is when I joined Perl 6, so apparently I just missed the end of that way of thinking :)
mst yeah, so, basically, it required masak to write strangelyconsistent.org/blog/how-pe...ill-us-all and me to write shadow.cat/blog/matt-s-trout/f_ck-perl-6/ 06:20
06:20 cdc left
mst and then a bunch of running around glaring at people who still weren't on message 06:20
and then it all started to calm down somewhat
but it was all exceedingly silly
people's front of judea versus people's judean front grade silly ;) 06:21
ShimmerFairy mst: I'm sure the Judean People's Front had the best ideas, though :P 06:22
mst ShimmerFairy: split()er!
flussence has "fond memories" of hacking on Pod::To::HTML around 2010ish, and the half hour or so it took to spit out S26.html... 06:23
(I'm glad that didn't put me off :) 06:25
ShimmerFairy I remember when I couldn't compile rakudo on my old Powerbook anymore, Parrot was so bad. Had to work with it remotely 'til I finally got a better computer :)
06:27 laouji left
flussence I went the insane route and added more swap until my netbook would build it... usually overnight 06:28
(I also flipped between rakudo and niecza a lot back then... p6 felt a lot like the web dev landscape from 10 years before) 06:30
mst remembers taking a little over 24h to build KDE base on his first thinkpad 06:35
flussence pretty sure I got on people's bad side with some *terrible* web browser analogies back then too... :)
06:36 _28_ria left 06:37 telex left
ShimmerFairy mst: wow, I'm glad I don't have it that bad anymore. (there's only webkit, which has a horrendous build time, and until I cleared out as much webkit-using stuff I could I needed four(!) different versions of it too) 06:37
06:38 telex joined
flussence tried a few different browsers last month, so got to enjoy compiling webkit. over and over and over. 06:40
06:41 laouji joined, zacts left
ShimmerFairy flussence: thanks to its horrific memory usage, it's one of the few things in portage I've demoted to less than the usual -j3 . And I wish I could replace qtwebkit with some sort of webkit-dummy, but AFAIK that's NYI :) 06:42
dalek osystem: 3751672 | (Tokuhiro Matsuno)++ | META.list:
Added Raw::Socket
 06:45
osystem: 4a3d940 | moritz++ | META.list:
Merge pull request #58 from tokuhirom/master

Added Raw::Socket
moritz \o 06:46
mst ShimmerFairy: note that if you read those two, and imagine what things must've been like to make those two posts seem sensible to write, you can probably get a fair insight into why "give each other special treatment because we're friends and we want to" is a meme I'm going to defend to the death, because the alternative situation as I saw it was definitely not an improvement
ShimmerFairy: I'm not saying that to make a particular argument, only to give you a clearer picture of where -my- particular foibles here come from, since you were kind enough to make an effort to unpack yours :)
ShimmerFairy mst: thanks :) . And if I understand correctly, I'm guessing I perhaps took "P6 doesn't succeed P5" a bit too much to heart, and came to the opposite of the old way of thinking ("P6 and P5 have nothing to do with each other") 06:48
06:50 perlawhirl joined
mst yes. which, though I was mostly trying to get us to leave each other alone at that point, because oh my did we all need a time out and nice cold class of calm the heck down at the time 06:50
I think stronger diplomatic relations as sister communities than with other communities is still a good thing 06:51
perlawhirl hi all... and hi again to those who were here 3 hours ago when i was asking questions 06:52
mst and will give us a much better position to avoid friction from than taking it quite so far as you seem to, though that doesn't mean you should change, I'm sure there are still far more people on the other end of the spectrum and they worry me a lot more
06:53 erxeto joined
perlawhirl i have another quick regex questions... inverse matching / not matching... is there a more idiomatic way to do it other than { $_ !~~ /$pattern/ } ? doing { ! /$pattern/ } does not work 06:54
06:55 telex left
mst ShimmerFairy: anyway. maybe I was falling a bit into the passion/anguish error in the process of diagnosing you as doing the same thing; I think that probably fits, and, y'know, I did say I was a past master at that particular failure mode ;) 06:55
moritz perlawhirl: huh, I expected !/pattern/ to work 06:56
06:56 telex joined, telex left
moritz perlawhirl: does it work with ! m/../? 06:56
perlawhirl hrm... maybe... one sec
ShimmerFairy mst: I think my opinions are also because I came to this language a few years ago (non-P5-knowing newcomers today will probably end up more familiar with P5 than me), and because I came here after trying P5 for two seconds and failing at that weird -> thing
mst: Basically, I ended up going "well, Perl 6 may not have the libraries I want to use yet, but at the syntax makes sense" :P 06:57
perlawhirl moritz: doesn't seem to
Method 'match' not found for invocant of class 'Any'
flussence m: say (^10).grep(!m/<[12345]>/) # same error locally
camelia rakudo-moar c0e0c9: OUTPUT«Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/mSEOJm0k04:1␤␤»
moritz perlawhirl: example?
flussence: needs to be inside a closure 06:58
m: say (^10).grep({ !/<[12345]>/})
camelia rakudo-moar c0e0c9: OUTPUT«Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/2MJcbJvlAq:1␤␤»
06:58 telex joined
moritz m: say (^10).grep({ !m/<[12345]>/}) 06:58
camelia rakudo-moar c0e0c9: OUTPUT«(0 6 7 8 9)␤»
ShimmerFairy perlawhirl: I don't think it's so easy to negate a regex generally. You'd either have to manually negate it, or use !~~ to check for "didn't match"
perlawhirl i'm trying to inverse grep a list, ie. @arr.grep($_ !~~ /$pattern/)
moritz perlawhirl: see above; needs a closure
or maybe
m: say (^10).grep(none(/<[12345]>/)) 06:59
camelia rakudo-moar c0e0c9: OUTPUT«(0 6 7 8 9)␤»
perlawhirl ok, cool
ShimmerFairy fwiw, I would've been quite surprised if !/rx/ did what you expected :)
moritz ShimmerFairy: me too. But I'm surprised that { !/<[12345]>/} doesn't work
perlawhirl moritz: closure works as expected... thanks!
mst ShimmerFairy: so, -> for nested hash derefs etc. was where I firsst used that 07:00
ShimmerFairy: then had hashes of subrefs, so $thing{foo}->(...)
ShimmerFairy moritz: not me, but then again I always use bare regexes with ~~ , since that's in one sense the "regex match" operator to me :)
07:00 rindolf joined
mst ShimmerFairy: so then I finally grokked OO as 'wait a class is jhust a special sort of hash of subrefs', so $thing->foo(...) as ref($thing)->{foo}($thing. ...) didn't seem that odd 07:00
ShimmerFairy: if I'd known OO -first- I'd probably've hated it too
... must remember that progression for teaching perl5, actually 07:01
ShimmerFairy mst: to clarify a bit, I was a teenager, wanted to write an emulator in a scripting language (long story), and when I gave up on Python 3, I tried deciding between the two Perls. I found a P5 Qt tutorial, couldn't make what I wanted to do work (those -> were to blame), and impatiently switched to P6. 07:02
mst heh 07:03
moritz ShimmerFairy: and now you're stuck here, patching the compiler :-)
mst that sounds like a completely reasonable train of thought, at least given my choice of calibration for 'reasonable' ;)
ShimmerFairy moritz: yep, can't say I'm disappointed in how things turned out :)
mst: about Py3, it didn't have @a[0] := @a[5] (for simulating shadow memory), which I got around, but it was when I got to working on opcodes and discovered that Py3 didn't have case statements that I gave up on it :P 07:05
mst what happens to me in python is I get ten minutes in and then I want lexical scope and lambdas and then I go WHAT IS THIS TOYTOWN BULLCRAP and then I throw a chair and use something else 07:08
I am really looking forward to playing with ES6, I think javascript and I will be much better friends with 'let'
moritz if it reduces the amount of (function () { ... })(), I'm all for it. 07:11
ShimmerFairy I haven't touched python since, but nowadays I like TIMTOWTDI and not having an entire language with :sigspace forcibly on :P
mst moritz: that basically goes away. this makes me happy.
babeljs is well worth a look 07:12
perlawhirl ShimmerFairy: have you used powershell... it's dynamic scoping is intolerable
ShimmerFairy I have not. I've been using bash for years now :)
mst dynamic scoping is the one true scoping, that's why everybody loves elisp unconditionally
07:12 abraxxa joined
perlawhirl ha! 07:12
moritz and perl < 5
ShimmerFairy mst: unless you turn on the fancy new lexical scoping in elisp :P
TimToady m: for (1,2,3).pairs X (4,5,6).pairs -> ($x,$y) { say "$x $y" } 07:13
camelia rakudo-moar c0e0c9: OUTPUT«This type cannot unbox to a native string␤ in block <unit> at /tmp/YnA_LhZbRO:1␤␤»
mst damnit, somebody spilled reality over my joke :(
TimToady bug ^^^
moritz TimToady: I tend to agree :-)
mst TimToady: on the upside, while you were away we stopped going round in circles and started tracing mandelbrot sets :D 07:14
ShimmerFairy mst: it's a 24.??? feature, and it only really applies to (let) and things like that, things like (defvar) now guarantee a dynamic var.
TimToady \o/
mst ShimmerFairy: I have no idea at all, I maintain my own personal fork of ex-vi 07:15
it goes well with the lawn defence cannon
however, it's now half an hour since I said I was going to be in 20 minutes, so I'm going to vanish at this point
ShimmerFairy mst: personally, I would love a Perl6-y form of emacs ("Like emacs, but s:g/elisp/perl6/"), if for no other reason than emacs regexps are awful, awful, awful 07:16
mst that was fun once we untangled the confusion, though
ShimmerFairy see'ya later :) o/
mst may all your candles be appropriately dribbly, whatever, their colour, and in the meantime, good night
flussence
.oO( those arrow functions in JS are a bit *too* nice, I got carried away and accidentally wrote a userscript in pseudo-haskell... )
 07:17
07:18 FROGGS joined 07:22 perlawhirl left, laouji left 07:23 laouji joined 07:29 darutoko joined 07:33 Alina-malina joined, Alina-malina left, Alina-malina joined 07:34 telex left
mathw (interesting backlog) wait Python doesn't have case??? 07:34
07:35 cognominal joined 07:36 telex joined
ShimmerFairy mathw: it didn't back in '09 or so, at least. 07:37
mathw wow
ShimmerFairy (and it's entirely possible that I was too much of a dumb teenager to properly find it :P) 07:38
mathw I'm trying to remember if I saw any in the Python codebase I was recently spelunking through at work
I can't recall doing so
But I might have got too lost amidst the copious misuse of XPath :)
07:38 brrt joined
mathw Fortunately I'm not expected to do Python regularly 07:40
I shall now return to my regularly scheduled attempts to convince anybody else at work that F# is possibly worth considering
ShimmerFairy a quick scan shows that I correctly didn't find it, and perhaps it's still not there O_o
[Tux] test 50000 37.837 37.753 07:42
test-t 50000 41.997 41.913
07:45 telex left
mathw ShimmerFairy: I just did some hunting and found that apparently the official reason it doesn't have anything like switch/case is that you can simulate it using either if... elif... or a dictionary 07:45
and that switch/case is hard to debug
that argument seems to come from the idea that all forms of switch/case use fallthrough-by-default like C's bad choices. 07:46
07:46 aborazmeh left
mathw So I don't really buy that. But I don't really buy a lot of Python's arguments. Really the only thing I like about it is the syntax. 07:46
07:46 telex joined
ShimmerFairy mathw: that's what I find too. Which is silly, since at the simplest case you could at least say "well, people are using associative types to do this, lets make the syntax do something like that" 07:46
mathw yeah 07:47
well at least over here we've got given/when :D
arnsholt Speaking of Python, it does have lexical scoping (and closures, if not proper lambdas). What's completely bananas is the declaration rules, not the scoping =)
ShimmerFairy www.pydanny.com/static/aliens.png 07:48
07:55 telex left 07:56 espadrine joined, telex joined 08:14 telex left 08:16 telex joined 08:20 vendethiel joined 08:24 telex left 08:26 telex joined
brrt mathw: i actually use fallthrough a lot in C :-) 08:28
FROGGS brrt: compiler writer do that, yes :o) 08:30
brrt i feel like i've suddenly developed a reputation for being a mad evil person just for mucking around with bits :-P 08:31
although hate of switch/case is pretty pervasive
i knew one guy who rewrote a 10-case switch statement into if/elses
FROGGS :/ 08:32
08:32 advwp joined, woolfy joined
FROGGS to me switch statements are way more readable 08:32
08:32 woolfy left
brrt dunno. i suppose if you read it in a fully literal left-to-right top-to-bottom order, if/else is ok 08:33
but personally i skim
FROGGS though, my mind "expects" then to not fall through I think, so I like Perl 6's default here
it is easy to miss a n 'if' though when it follows an 'if/elsif/elsif/ directly 08:34
that's where I like it when the body of a switch statement is indented... 08:35
fells more block like
feels*
brrt yeah, i agree
ok, in that sense, fallthrough is probably evil 08:36
although *i* know a guy who uses goto within the switch statements
and fallthrough
psch moin #perl6 o/ 08:38
r: for <0 1 2> { .say when 0; .say when 1; .say when 2 } # allomorphs! *shakes fist*
camelia rakudo-moar c0e0c9: OUTPUT«0␤1␤2␤»
( no output )
mathw brrt: I love switch-like constructs. Fallthrough is useful, but I think there are better options. The trouble with C's is that it's designed to compile down to something super-efficient and it comes from an era when your compiler wasn't smart enough to do that when it can but also give you something fancier when it can't 08:39
brrt uhuh 08:40
that's kind of true still 08:41
compilers are still pretty dumb
08:42 telex left 08:44 telex joined
ShimmerFairy brrt: surely not Perl 6 compilers though! :) 08:45
brrt hmm. i can't speak for that :-)
what i do think is that 'sufficiently smart' usually means 'very expensive algorithms' and that's not pretty rarely a good choice 08:46
euh
08:46 pflanze left
brrt s/not// 08:46
lizmat good *, #perl6! 08:47
psch hi lizmat o/ 08:48
lizmat psch o/
brrt hi lizmat 08:49
lizmat brrt o/
08:52 telex left
psch of course it's not actually the allomorphs that are to blame, my getBI hack just didn't quite work as expected in some cases.. :/ 08:52
08:54 telex joined 09:01 laouji left
ShimmerFairy psch: IIRC bigints in java aren't that great in general 09:03
09:05 laouji joined
vendethiel well, the API is a bit... tiring to use 09:06
FROGGS that's quite Javaish, no?
lizmat m: say (-1) ** 4553535345364535345634543534 09:08
camelia rakudo-moar c0e0c9: OUTPUT«0␤»
lizmat there's a spectest expecting 1 there
psch $ ./perl6-j -e'for <0 1 2> { .print when 0; .print when 1; .print when 2 }'
012
fixed it anyway :P
lizmat m: say -1 ** 4553535345364535345634543534
camelia rakudo-moar c0e0c9: OUTPUT«-1␤»
lizmat hmmm...
psch ShimmerFairy: the problem was just that i didn't actually check for the right field type, but relied on get_attribute_native to throw if it didn't find the right one
ShimmerFairy ah 09:09
lizmat $ git describe
warning: tag 'Zürich' is really 'Zu􏿽xCC􏿽x88rich' here
Zürich
huh?
ShimmerFairy lizmat: the second quoted Zürich has a decomposed umlaut after the u
(and my irc client puts it over the wrong letter. nice.)
lizmat aha! 09:10
welcome to the world of unicode :-)
dalek p: cdd57fe | FROGGS++ | tools/build/MOAR_REVISION:
bump moar for fix for RT #125408
ShimmerFairy I'm familiar with it, it's just that a frightening number of programs get the placement of diacritics wrong :/
dalek kudo/nom: 5ddb978 | FROGGS++ | t (2 files):
bump nqp/moar for fix for RT #125408
 09:11
p: 9a53b94 | peschwa++ | src/vm/jvm/runtime/org/perl6/nqp/runtime/Ops.java:
Correct getBI hack.

Apparently we can't rely on get_attribute_native to only get BigIntegers, so we actually look for one.
psch FROGGS: bump again! :P
psch can do that too, though 09:12
FROGGS psch: ohh, then I'll let you doing it :o)
09:12 yakudza left
dalek kudo/nom: d2d975e | peschwa++ | tools/build/NQP_REVISION:
Bump NQP_REVISION (again).

Get proper IntStr handling on jvm.		 09:14
09:23 jack_rabbit_ joined 09:26 telex left 09:28 telex joined 09:32 TEttinger left, andreoss joined, laouji left 09:36 vendethiel left 09:40 yakudza joined
lizmat is still seeing a new spectest error in t/spec/S32-num/power.t 09:41
09:47 telex left 09:48 telex joined 09:49 pdcawley joined 09:54 mprelude joined 10:02 araujo_ joined 10:05 araujo left 10:06 telex left 10:08 telex joined 10:18 [Sno] left 10:20 telex left 10:22 telex joined
psch gist.github.com/peschwa/90c8dcc516ea934a17be # updated j-spectest summary 10:27
down to 43 files mentioned, with 2 being only mentioned as TODO PASS 10:28
actually that's 3, from the looks of it
moritz r: say (-1) ** -i
camelia rakudo-moar d2d975: OUTPUT«NaN+NaN\i␤»
( no output )
psch r: say (-1) ** -i # jvm just sleepy 10:29
camelia rakudo-{moar,jvm} d2d975: OUTPUT«NaN+NaN\i␤»
moritz ah, that one is TODOed
the ones that fails is
m: say (-1) ** 4553535345364535345634543534 10:30
camelia rakudo-moar d2d975: OUTPUT«0␤»
psch m: say <3> ~~ Int 10:31
camelia rakudo-moar d2d975: OUTPUT«True␤»
10:31 telex left
psch m: use Test; isa-ok <3>, Int 10:31
camelia rakudo-moar d2d975: OUTPUT«ok 1 - The object is-a 'Int'␤»
psch r: use Test; isa-ok <3>, Int
camelia rakudo-moar d2d975: OUTPUT«ok 1 - The object is-a 'Int'␤»
..rakudo-jvm d2d975: OUTPUT«not ok 1 - The object is-a 'Int'␤␤# Failed test 'The object is-a 'Int''␤# at /tmp/tmpfile line 1␤# Actual type: Str␤»
10:32 telex joined
psch r: use Test; isa-ok <3>, Str 10:32
camelia rakudo-{moar,jvm} d2d975: OUTPUT«ok 1 - The object is-a 'Str'␤»
psch r: say IntStr.isa(Str) 10:33
camelia rakudo-{moar,jvm} d2d975: OUTPUT«True␤»
psch r: say IntStr.isa(Int)
camelia rakudo-{moar,jvm} d2d975: OUTPUT«True␤»
psch guesses Binder :P
ilmari m: say 1 << 73
camelia rakudo-moar d2d975: OUTPUT«5===SORRY!5=== Error while compiling /tmp/_pfCDbRbAX␤Unsupported use of << to do left shift; in Perl 6 please use +< or ~<␤at /tmp/_pfCDbRbAX:1␤------> 3say 1 <<7⏏5 73␤»
ilmari m: say 1 +< 73
camelia rakudo-moar d2d975: OUTPUT«9444732965739290427392␤»
psch i'll fix hash ordering tests first suppose 10:34
'cause that also helps moar :)
10:37 [Sno] joined 10:41 telex left 10:42 telex joined 10:43 brrt left
psch hm, not sure about S02-types/baghash.t:134 10:45
where exactly is it that we're testing flattening there..?
isn't that test actually just conflating the two tests below 10:46
lizmat psch: that might well be the case, as these tests predate GLR and were quickly fudged 10:47
10:48 telex left, telex joined, telex left 10:50 telex joined, vendethiel joined
dalek ast: 66f9a46 | peschwa++ | S02-types/bag.t:
Make a test hash-elem-ordering proof.
 10:52
ast: 7a58cad | peschwa++ | S02-types/ (3 files):
Remove a duplicate/conflated test.
10:52 colomon left
psch s/ a / a few / 10:52
oh 10:53
i should probably have checked if bag.t has something like &showkv :/
hm, maybe not. i don't immediately see how i can utilize it there 10:54
10:56 colomon joined
psch eh, but that test doesn't feel particularly good in any case :/ 10:56
i mean, neither with nor without my changes
the fact that the glr changed the semantics is part of the reason i guess 11:01
m: my Bag $bag .= new: <foo foo bar>; for $bag.keys X $bag.keys -> $a, $b { say $a, $b }
camelia rakudo-moar d2d975: OUTPUT«(foo foo)(foo bar)␤(bar foo)(bar bar)␤»
psch ...or did they
they didn't i must've done something weird locally :) 11:02
anyway, that's another 4 files clean on j-spectest
11:02 araujo_ left
psch r: use Test; my @a = <foo 3 4.5 5.60 1.2e1 -2+3i>; isa-ok @a[1], Int, '< ... 3 ...> is an Int'; 11:04
camelia rakudo-moar d2d975: OUTPUT«ok 1 - < ... 3 ...> is an Int␤»
..rakudo-jvm d2d975: OUTPUT«not ok 1 - < ... 3 ...> is an Int␤␤# Failed test '< ... 3 ...> is an Int'␤# at /tmp/tmpfile line 1␤# Actual type: Str␤»
psch oh ffs, getBI still wonky? 11:05
r: say $_.WHAT for <0 1 2 3 4 5 6 7 8 9>
camelia rakudo-jvm d2d975: OUTPUT«(Str)␤(Str)␤(Str)␤(Str)␤(Str)␤(Str)␤(Str)␤(Str)␤(Str)␤(Str)␤»
..rakudo-moar d2d975: OUTPUT«(IntStr)␤(IntStr)␤(IntStr)␤(IntStr)␤(IntStr)␤(IntStr)␤(IntStr)␤(IntStr)␤(IntStr)␤(IntStr)␤»
psch agh 11:06
allomorphic.t probably should be in spectest.data... 11:07
dalek kudo/nom: aa34562 | peschwa++ | t/spectest.data:
Test S02-literals/allomorphic.t
 11:08
11:17 brrt joined 11:18 domidumont left
masak hi, #perl6 11:20
lizmat masak o/ 11:21
11:23 araujo joined, araujo left, araujo joined 11:24 domidumont joined 11:25 [Sno] left 11:26 telex left 11:27 skids joined 11:28 telex joined 11:29 brrt left
[ptc] masak: o/ 11:30
masak lizmat, [ptc]: \o 11:31
11:34 xinming joined 11:36 jack_rabbit_ left 11:38 xinming_ left 11:39 kid51 joined 11:40 abraxxa left, telex left 11:42 telex joined 11:48 Skarsnik joined, jack_rabbit_ joined 11:51 FROGGS left 11:52 FROGGS joined 11:53 [Sno] joined 11:58 kid51 left, skids left 12:01 brrt joined, telex left 12:02 telex joined, telex left 12:04 telex joined 12:17 g4 left
dalek pan style="color: #395be5">perl6-examples: 2d4b164 | (Steve Mynott)++ | t/categories/99-problems.t:
[99probs] reorder expected output for P91-edpratomo and strip <> so it passes
 12:18
[Coke] psch: yay, I hoped I'd come back to jvm hacking. :) stresstest was very sad last night. 12:25
lizmat psch: did you actually try to run bag.t after doing github.com/perl6/roast/commit/66f9a46816 ? :-) 12:29
Ulti my rakudobrew build is failing on Mac at the moment even after I completely wipe everything... is this a known issue?
Ulti notices self-upgrade did a lot 12:30
psch lizmat: i... think so? which means, probably not... :/
lizmat it has a compile error in it 12:31
moritz Ulti: if you wiped everything, how can you do a self-upgrade?
lizmat parings vs pairings
psch ugh ._.
Ulti moritz everything but the rakudobrew script :P
still fails in the same place I'll gist it 12:32
psch ...i did have this commit hook for roast once
maybe i should revisit it, ensure that i don't commit bogus tests :/
psch waits until jvm finished building and *runs* S02-types/bag.t before pushing... 12:33
dalek kudo/nom: 96c65bf | lizmat++ | src/core/IO/Handle.pm:
Make "words".IO.lines.elems much faster

For the chomp case, 20%, for the !chomp case, 40%.
Basically, the nqp::eoffh op is useless as it stands now, so just don't use it. If nqp::readlinefh returns 0 chars, we've reached EOF. Also, don't bother updating $!ins if we're about to close the file handle at the end anyway.
12:34 telex left
psch [Coke]: yeah, there's still quite a few pretty bad failure modes around... lowlevel Exceptions without a msg, StackOverflows... 12:34
Ulti gist.github.com/MattOates/6f8c4139a51956bff728
psch the former FROGGS (i think) says is leaking control exceptions from label stuff iirc, the latter is nigh impossible to debug... 12:35
12:35 rurban joined
_itz Ulti: did you run a mac update .. I had to "sudo gcc" (actually clang) and accept an agreement? 12:35
Ulti orly 12:36
I'll try that thanks _itz
12:36 telex joined
Ulti wow.... just wow 12:36
masak lizmat: what about checking $!ins on a closed handle? 12:37
Ulti thanks _itz looks to really be building now
jees
masak JEZU
lizmat masak: well, this is really about the case where the handle is hidden
masak aha.
Ulti sure love this opaque OSX updates where they tell you nothing
lizmat "words".IO.lines
in that case, the handle is never visible 12:38
_itz I was a bit annoyed by that one myself .. although Apple seem to be improving clang on the upside
lizmat and if you're doing $handle.lines(:close) yourself, you get what you indicate
psch hrm, the getBI/IntStr stuff breaks bag.t..? i can't get *to* the test that i wanted to verify as working... :S
on the plus, allomorphic.t runs about half the tests (and fails about half of that...) 12:39
geekosaur seeing email indicating that apple pushed xcode 7 12:41
12:43 telex left 12:44 telex joined
geekosaur ...and apparently it breaks macports toolchains because it uses a whole new linker mechanism. workaround: sudo port upgrade --enforce-variants ld64 +ld64_xcode (if it complains about incompatible variants, add the variant it complains about with leading - instead of +, e.g.port upgrade --enforce-variants ld64 -ld64_236 +ld64_xcode) 12:46
FROGGS geekosaur: sounds fun 12:47
geekosaur pretty much par for apple :/
FROGGS yeah, heard that already
12:50 sufrostico joined
[Coke] I did the xcode upgrade but haven't lost anything in macports yet, thankfully. 12:52
12:53 abaugher joined
psch well, i have a getBI implementation that works with &val 12:57
it doesn't work quite right with angle-quoted lists, though...
[Coke] psch: if you have an N core machine, how high can you push TEST_JOBS on a jvm spectest? N-1? 12:58
psch [Coke]: i'm running TEST_JOBS=4 on hack, usually
[Coke]: so that's a "yes", i suppose, although ISTR that someone suggested 2*N here some (probably "a long") time ago 12:59
[Coke] ok.
maybe on moar, I would expect that to kill jvm stuff.
rindolf Hi all! Running «perl6 prob329-shlomif.p6» gives me these warnings - why? www.shlomifish.org/Files/files/text...rnings.txt
I've recently updated my rakudobrew and then the trouble started. 13:00
[Coke] relink to the script?
psch $ ./perl6-j -e'for "0", "1", "2" { print val($_).Int ~ " " }; print "\t"; for <0 1 2> { print .Int ~ " " }; say()' 13:04
0 1 2 1 1 1
that's just weird :/
rindolf [Coke]: here - github.com/perl6/perl6-examples/bl...shlomif.p6
psch ShimmerFairy: ^^^ any idea where i should look? it seems to be somewhere above nqp-level... 13:05
arnsholt psch: What's the AST for <0 1 2>?
Or just <0>, even
rindolf m: for "0", "1", "2" { print val($_).Int ~ " " }; print "\t"; for <0 1 2> { print .Int ~ " " }; say() 13:06
camelia rakudo-moar 96c65b: OUTPUT«0 1 2 0 1 2 ␤»
arnsholt It's a JVM bug
FROGGS m: say my @init_probab = (0, (map { FatRat.new(1, 500) }, 1 .. 500) , 0);
camelia rakudo-moar 96c65b: OUTPUT«[0 (0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 …»
FROGGS m: say my @init_probab = flat (0, (map { FatRat.new(1, 500) }, 1 .. 500) , 0);
camelia rakudo-moar 96c65b: OUTPUT«[0 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0.002 0…»
ShimmerFairy psch: only thing I can think is that I had to change some of the Q lang to support multiple post-processing adverbs (since <> is just q:w:v, for example). But as has been mentioned, it's a VM-level issue.
FROGGS rindolf: do that ^^
arnsholt psch: Also, does the same thing happen when you do "for <0>, <1>, <2>"? 13:07
psch arnsholt: - QAST::Stmts <0>␤- QAST::Stmt <0>␤- QAST::WVal(IntStr) 0
FROGGS add a 'flat' after the assignment op to line 9, 11, 12 and 30
rindolf FROGGS: is this the correct behaviour?
FROGGS rindolf: yes
rindolf FROGGS: oh.
psch arnsholt: $ ./perl6-j -e'for <0>, <1>, <2> { print .Int ~ " " }; say()' 13:08
1 1 1
rindolf FROGGS: didn't it break the code of all the world and his mother?
FROGGS rindolf: yes
[Coke] rindolf: @p_letter only has 4 elements in it.
psch ShimmerFairy: well, yeah, it's vm-level, but it's apparently not in the ops impl
13:08 telex left
ShimmerFairy psch: since I never had to do anything obviously VM-specific ("obvious" according to one who's not familiar with what VM-differentiated code looks like), I couldn't tell you where the issue could be :/ 13:09
13:10 rarara joined
arnsholt psch: Weird. Is the .WHAT of the val($_) the same as the .WHAT in the <...> case? 13:10
[Coke] sees FROGGS is already on the case and gets more coffee.
ShimmerFairy psch: I assume the weird "1 1 1" thing doesn't happen with RatStrs or anything?
13:10 telex joined
dalek kudo/nom: 328a03a | lizmat++ | src/core/IO/Handle.pm:
Make 'for "words".IO.lines' about 25% faster

Basically the same opt done for '"words".IO.lines.elems'		 13:10
psch $ ./perl6-j -e'print <0>.WHAT.gist, " ", val("0").WHAT.gist, " "; say <0>.WHAT =:= val("0").WHAT' 13:11
(IntStr) (IntStr) True
arnsholt: ^^^
arnsholt Very, very weird 13:12
FROGGS j: for "0", "1", "2" { print val($_).Int ~ " " }; print "\t"; for <0 1 2> { print .Int ~ " " }; say()
camelia rakudo-jvm 96c65b: OUTPUT«java.lang.NullPointerException␤ in block <unit> at /tmp/J6cm3OijYG:1␤␤»
psch FROGGS: i can push what i have, but...
FROGGS psch: what does this give? for "0", "1", "2" { print val($_).Int ~ " " }; print "\t"; for <0 1 2> { print .WHAT; print .Int ~ " " }; say() 13:13
13:13 Peter_R left, Peter__R joined
psch FROGGS: 0 1 2 1 1 1 13:14
FROGGS ohh, print was stupid there...
psch: would be interesting to know what $_ is the second loop...
psch oh
miscopied
FROGGS its type and its .WHICH
no, .WHERE
psch $ ./perl6-j -e'for "0", "1", "2" { print val($_).Int ~ " " }; print "\t"; for <0 1 2> { print .^name, " ", .WHERE, " ", .Int ~ " " }; say()' 13:16
0 1 2 IntStr 1367725438 1 IntStr 696478978 1 IntStr 1936307096 1
that good? :S
ShimmerFairy psch: Here's a silly idea: <1 2 3>.map: { val(~$_) } :)
psch ShimmerFairy: well, that works - <>-quoting gives the right .Str but the wrong .Int here 13:17
$ ./perl6 -e'<1 2 3>.map: { print val(~$_).Int, " " }; say()' 13:18
1 2 3
$ ./perl6 -e'<1 2 3>.map: { print val(~($_.Int)).Int, " " }; say()'
1 1 1
13:18 larion joined
psch ^^^ as seen there :l 13:18
lizmat cycling& 13:19
ShimmerFairy psch: oh! try something like my ($a, $b, $c) = (1, 2, 3); for «$a $b $c» { print .Int ~ " " }; say()
FROGGS hmmm 13:21
psch ShimmerFairy: 1 1 1
ShimmerFairy psch: darn, I was hoping the calling of &val at compile-time was perhaps the issue. («$a» isn't counted as being a compile_time_value for me at least) 13:22
13:26 telex left, khw joined, telex joined, telex left 13:27 colomon left
FROGGS j: use nqp; my $a = 42; say $a.WHERE; say nqp::add_I(nqp::decont($a), 0, IntStr) # ??? 13:28
camelia rakudo-jvm 96c65b: OUTPUT«1008783893␤%_␤»
FROGGS j: use nqp; my $a = 42; say $a.WHERE; say nqp::add_I(nqp::decont($a), 0, IntStr).Int
camelia rakudo-jvm 96c65b: OUTPUT«728034546␤42␤»
FROGGS ahh
13:28 telex joined
FROGGS phew 13:28
j: use nqp; my $a = 42; say $a.WHERE; say nqp::add_I(nqp::decont($a), 0, IntStr).Int.WHERE
camelia rakudo-jvm 96c65b: OUTPUT«455979265␤1826478070␤»
13:29 colomon joined, rarara left
dalek pan style="color: #395be5">perl6-examples: 07cf4eb | (Shlomi Fish)++ | categories/euler/prob329-shlomif.p6:
Fix the code on recent versions of Rakudo.
 13:34
13:35 nucp6 left
timotimo o/ everyone 13:37
13:37 FROGGS left 13:40 mspo joined
psch hi timotimo o/ 13:40
mspo I'm trying to get the pkgsrc rakudo-star package up to date and change it into the "official" perl6 pkgsrc lang/perl6 13:43
13:43 thou joined
mspo is it possible to pre-fetch everything outside of the build steps? 13:43
Ulti /win/win 2 13:46
:(
13:46 simcop2387 left, zostay joined 13:47 brrt left
masak Ulti: it's one of those /win/win situations... :P 13:49
13:49 telex left 13:50 telex joined
[Coke] mspo: star already prefetches most things. 13:51
the .tar file should have everything needed to install.
13:51 simcop2387 joined, muraiki joined
mspo [Coke]: when I build it manually it seemed to do a bit of downloading 13:52
muraiki morning all. is there a stat() function somewhere in perl6? I looked in IO and couldn't find it... some old docs referenced File::stat but that doesn't seem to exist either
mspo [Coke]: this is not great for pkg systems :)
[Coke] mspo: "build it manually" ... are you building what's in the github repo? 13:53
mspo [Coke]: alternatively I could package nqp, moarvm, and (other) seperately I suppose
[Coke] or are you building from rakudo-star.2015.mumb...
13:55 thou left
mspo let me do a fresh build and clarify what I"m asking; sorry for the noise so far 13:55
Skarsnik Hi, what is the opposite of the .perl method? I have a $str holding a object.perl result and I want to put it back in a object. 13:57
jnthn EVAL 13:58
14:01 Loren joined
Loren m: sprintf("In my perl6, sprintf not work ."); 14:01
camelia ( no output )
muraiki is there a way to convert a decimal to octal? I know that if I my $foo = 0o776 that is turned into an Int; there doesn't seem to be an Octal type
pink_mist sprintf "%o", $num; 14:02
muraiki hah, duh. thanks!
Loren Is 'sprintf' not implement or my code is wrong 14:03
Skarsnik jnthn, oh thx, why EVAL (in caps)? or does $var.eval exist? 14:04
muraiki it worked for me
pink_mist heh, I thought this was #perl when I gave the answer ... so I didn't even check if it was available in perl6 :P
mspo [Coke]: I'm talking about this: 14:05
--gen-moar[=branch]
Download and build a copy of MoarVM
--gen-nqp[=branch]
Download and build a copy of NQP
[Coke]: how do I pre-download those (or give my own paths) so I can validate checksums etc 14:06
pink_mist m: print sprintf("In my perl6, sprintf not work .");
camelia rakudo-moar 328a03: OUTPUT«In my perl6, sprintf not work .»
pink_mist Loren: you forgot to print
Loren ... 14:07
OH ,my bad...
psch m: "50".base(8).say
camelia rakudo-moar 328a03: OUTPUT«Method 'base' not found for invocant of class 'Str'␤ in block <unit> at /tmp/mteoQKwHva:1␤␤»
psch m: 50.base(8).say
camelia rakudo-moar 328a03: OUTPUT«62␤»
pink_mist psch: nice 14:08
psch m: :8("0777").Int.base(8).say # round-trip Str -> Str
camelia rakudo-moar 328a03: OUTPUT«777␤»
14:08 rindolf left
psch oh the .Int is superfluous too i think 14:09
m: :8("0777").base(8).say # round-trip Str -> Str
camelia rakudo-moar 328a03: OUTPUT«777␤»
Loren I just saw the sample, forgot it was 'sprintf'!
14:09 perl6_newbee joined 14:10 telex left
psch Skarsnik: EVAL is caps to stand out - it's usually a sign of code doing something the reader probably wants to pay attention to 14:10
jnthn Skarsnik: We user uppercase for most of the time-shifty things (like BEGIN); also because it's macro-y rather than function-y
psch jnthn++ # for the reason *why* you want to pay attention :) 14:11
14:12 FROGGS joined, telex joined, Peter__R left 14:14 Sqirrel left
FROGGS o/ 14:15
Loren ~_~ 14:19
14:23 lolisa joined 14:25 lolisa left
psch gist.github.com/peschwa/09f512e0a3f95a90dc81 14:26
i'm out of ideas
14:26 telex left 14:27 vendethiel left
moritz psch: what does .Str.say give you? 14:27
psch moritz: the right values, i.e. "0 1 2" in this case 14:28
14:28 telex joined
moritz psch: so that means the Int component is wrong? 14:28
also IMHO it would be better to still emit a QAST::Want
psch moritz: well, that's the thing. it's only wrong through <>
moritz: call &val directly works 14:29
+ing
moritz QAST::Want($IntStr, 'i', $int_value, 's', $str_value)
psch moritz: there's a bit more in the backlog, about an hour ago
moritz ah yes 14:30
14:30 skids joined 14:32 jack_rabbit_ left 14:35 telex left 14:36 telex joined, telex left 14:38 telex joined
erxeto moritz: by any chance are you the author of DBIish ? 14:40
moritz erxeto: one of the maintainers, yes
the original author was mberends, back when it was called MiniDBI
erxeto I'm trying to install Task::Star via panda and and I have an error. I've seen that you already have a Issue at github for this 14:41
Just wanted to know if this is mantained (meaning eventually fixed) or not.
I'm just trying to start with perl6, so is not really an issue. 14:42
moritz erxeto: yes, it's maintained
erxeto this is the issue I'm talking about (just FYI): github.com/perl6/DBIish/issues/19 14:43
moritz erxeto: which version of rakudo are you using? 14:46
erxeto $ perl6 -v 14:47
This is perl6 version 2015.09 built on MoarVM version 2015.09
14:47 lichtkind joined
moritz erxeto: excellent 14:48
erxeto excellent ?
moritz erxeto: you rakudo isn't too old
14:48 g6 left
moritz erxeto: and it's the rakudo I'll test with 14:48
the part that's not excellent is that I fear it's a rakudo bug 14:49
erxeto moritz: I installed it a couple of hours ago with rakudobrew ... hehehe
psch heh, at least allomorphs.t doesn't look that much better on moar :S
Failed tests: 37-38, 40, 42-45
Parse errors: Bad plan. You planned 93 tests but ran 45.
still...
moritz FROGGS: has 'require' stopped working? 14:50
FROGGS moritz: dont think so 14:51
moritz: we have working tests for it
moritz FROGGS: perlpunks.de/paste/show/55fc2506.7e9c.2e3
FROGGS huh 14:52
ohh, I think I know...
_itz I thought panda without args used to do --help 14:53
FROGGS moritz: this might be to blame: github.com/rakudo/rakudo/commit/a9...6fd805b59f
_itz: install is the default no
now*
14:54 Sqirrel joined, telex left
_itz I mean if I type "panda" it doesn't do anything which isn't helpful for a new user 14:54
FROGGS true 14:55
moritz FROGGS: I'm pretty sure this warents a bugfix release; we can't make a working star release on top of 2015.09
FROGGS _itz: easily fixable though
moritz: :S
14:55 telex joined
FROGGS moritz: but yeah, makes sense 14:55
14:55 telex left
erxeto huh, I was thinking in say nothing. Just thought it was my mistake hehehe. 14:56
14:56 telex joined
moritz I'll rakudobug it. 14:56
RT #126096 14:58
[Coke] mspo: again, are you building from a .tar.gz release of star, or from star in the github repo? 15:01
psch: I'm sure you have something already, but here's my current jvm failures: gist.github.com/coke/e590a53cc4c3e2a75cfd
S17-supply/syntax.t never exited, had to kill it, but passed 50 tests regardless. 15:02
15:03 telex left
[Coke] star in github is intended ONLY to generate the release files - it's not something anyone else should be using, even package managers. If you're doing that, you might as well bundle it yourself and not use any of the star infrastructure (which may be what you're ultimately asking here) 15:04
15:04 telex joined, telex left
psch [Coke]: your results look a bit worse, but that could just be because of a "somewhat working but often wrong" instead of "not really working at all" IntStr... 15:05
15:06 telex joined
psch i'm fairly confident that <>-handling of Ints is behind quite a lot of failures outside of S02-types/allomorphs.t 15:06
15:08 synbot6 joined 15:09 diana_olhovik left
[Coke] is hoping that stress test will be clear once spectest is. 15:09
15:09 rurban left
[Coke] mspo: you can absolutely check out things in advance instead of letting Configure.pl do it for you - for example: 15:10
github.com/coke/perl6-roast-data/b...oar-jit.sh
that does all the git clones -first- and then does the configure.pl so that git doesn't make a network connection during build. But again, star releases should already have this covered. 15:11
(the repos/ dir mentioned there is a local bare checkout since I'm building a lot of things that are shared many times and only wanted to do the network fetch once) 15:12
15:14 telex left, telex joined
mspo [Coke]: from the tar.gz, maybe it does actually have everything already 15:14
[Coke]: I'm not noticing any downloads despite the messages 15:15
15:15 DrForr joined
Skarsnik hm, what is the proper method to test if a Str contain only number? 15:16
PerlJam Skarsnik: depends on what you mean by "number" 15:17
DrForr waves weakly.
[Coke] psch: with a java build, does this work for you:
_itz DrForr++
[Coke] prove -v -e t/fudgeandrun t/spec/S05-substitution/subst.t
15:17 Loren left
Skarsnik only digit (no stuff like 3.000) 15:17
moritz Skarsnik: is "0xFF" a valid number for you? 15:18
15:18 Loren joined
moritz if not, probably $str ~~ / ^ <[0..9]>+ $ / 15:18
Skarsnik Not in this case, but I get the prob with this question ~~
moritz if yes, defined $str.Int
[Coke] ahh, need to make install first. bah. 15:19
PerlJam Skarsnik: I might've used $str !~~ / \D /;
ugexe what about 0002
[Coke] psch: doing a verbose run on the tests with failures.
15:21 mr_ron joined, rurban joined, rurban left
Skarsnik $str.Int crash if $str has no number ~~ 15:21
TimToady Morning has broken; anyone have some glue?
mr_ron m: gist.github.com/ronaldxs/2ced79b6b334009af9db 15:22
camelia rakudo-moar 328a03: OUTPUT«Any case that results in a type of (Any) is broken␤(Match)␤(Any)␤(Match)␤(Any)␤(Match)␤(Any)␤»
PerlJam Skarsnik: mine returns true as long as the string does not contain any non-digits. This will also return true for the empty string (I dunno if that's a problem for you)
DrForr No glue, but I can offer a few titanium screws.
ugexe friends dont let friends huff glue first thing in the morning 15:23
mr_ron I was trying to kebab case the URI.pm grammar and ran into trouble with character classes. Could someone take a look at my gist?
Skarsnik for a 'if ($key.Int) {stuff} I get Cannot convert string to number: base-10 number must begin with valid digits or '.' in '⏏bsname' (indicated by ⏏) 15:24
mr_ron the rfc for URI bnf is kebab case
PerlJam mr_ron: I don't think reserved2 is breakage.
TimToady DrForr!!! You can keep your screws!!! Did you make it home?
DrForr Sitting up in bed as I type. 15:25
TimToady at least we can offer you a sedentary occupation... 15:26
mr_ron from the synopsis "Character classes can be combined (additively or subtractively) within a single set of angle brackets. Whitespace is ignored. For example: / <[a..z] - [aeiou] + xdigit> /
should we allow kebab case in character classes for grammars? 15:27
ugexe mr_ron: ive had that problem as well
psch phew
found the culprit for the IntStr breakage. Ops.java:5995 silently binds into a native_str slot 15:28
[Coke]: any specific file you'd prefer me to run?
PerlJam mr_ron: Also read S05:1413
synbot6 Link: design.perl6.org/S05.html#line_1413
[Coke] psch: no, I figured out my damage with the prove line. 15:29
TimToady mr_ron: I think kebab case should be allowed, and we can just require space for disambiguation, as in normal code 15:31
15:31 rurban1 joined
DrForr Indeed, Amalia's rather anxious for me to get back to work, for reasons made obvious by Ovid's keynote. 15:32
jdv79 huh?
DrForr blogs.perl.org. 15:33
jdv79 is it broke?
PerlJam mr_ron: Actually, I don't think your gist is elucidating any problems with regex, but rather with your code.
DrForr Somewhat, yes. If you've been able to log in and use it over the last 6 months without several password resets, you're lucky. 15:34
PerlJam mr_ron: reserved3 matches "d" characters but you're trying to match against "ccc" and reserved suffers the same problem as reserved2
mr_ron working on fixing the example to focus on kebab case ... 15:35
pink_mist ... now I'm getting hungry
15:36 rurban joined 15:37 telex left 15:38 telex joined
mr_ron m: gist.github.com/ronaldxs/227c871493478ce5224e 15:39
camelia rakudo-moar 328a03: OUTPUT«(Match)␤(Match)␤(Match)␤(Match)␤Method 'delims' not found for invocant of class 'TestGrammar'␤ in regex reserved at /tmp/O46UAy0Q0q:17␤ in block <unit> at /tmp/O46UAy0Q0q:40␤␤»
PerlJam aye, that's clearly a bug
15:39 rurban1 left
mr_ron So we do kebab case for simple scenario but combined seems to break. Correct? 15:40
[Coke] psch: gist.github.com/coke/e590a53cc4c3e...erbose-out 15:41
grondilu still gets 'JIT isn't supported on arm-linux-gnueabihf-thread-multi-64int yet. 15:42
' on his Raspberry Pi 2
timotimo hm, would you have to --no-jit for moarvm?
we should probably just warn and build the jitless version regardlesss 15:43
psch [Coke]: what am i looking for..?
[Coke] It's just a list of specific failures, you said I had more than you did, just trying to provide more data. 15:44
Skarsnik why http::ua took me etheir 0-1 to do a http request (a simple get) or 46-47 sec ~~
_itz jit is amd64 only? I guess Configure.pl should know this
timotimo Skarsnik: perhaps it prefers ipv6 and your ipv6 connectivity isn't awesome?
psch [Coke]: oh, thanks. i'm getting closer to fixing IntStr though, so i guess that'll change a bit anyway... :)
geekosaur Skarsnik, that delay makes me think something wonky with DNS 15:45
FROGGS moritz: I think I've got a fix 15:46
Skarsnik well, if I replace with wget call, there is no issue ~~ 15:48
15:49 domidumont left
FROGGS timotimo: we *do* warn an keep building without jit 15:50
ugexe how big is the cookie the site you are fetching trying to get?
it uses a grammar for cookies i think... might be some backtracking going on 15:51
Ulti is there some special magic I need to do to get panda to use a web proxy? other than have the enivronment variables set 15:52
_itz Ulti: it should work 15:53
Skarsnik ugexe, there should be none. I just get on a page that return some json 15:54
ugexe is it an https proxy? 15:55
_itz no 15:56
grondilu just to be clear : it's just a warning message. The builds goes through. Only without JIT.
mr_ron m: grammar G { token TOP { <+ kebab-case> }; token kebab-case { <[a]> } }; G.parse('aaa');
camelia rakudo-moar 328a03: OUTPUT«Method 'case' not found for invocant of class 'G'␤ in regex TOP at /tmp/nO8n9KP_TK:1␤ in block <unit> at /tmp/nO8n9KP_TK:1␤␤»
mr_ron PerlJam: a simplified version of the bug - yes? 15:57
skids mr_ron: well golfed. 15:58
PerlJam mr_ron: indeed
16:00 Peter_R joined, Peter_R left 16:01 telex left 16:02 telex joined
TimToady m: grammar G { token TOP { <+ :kebab-case> }; token kebab-case { <[a]> } }; G.parse('aaa'); 16:03
camelia ( no output )
16:03 mullagainn left
TimToady m: grammar G { token TOP { <+ :kebab-case> }; token kebab-case { <[a]> } }; say G.parse('aaa'); 16:04
camelia rakudo-moar 328a03: OUTPUT«(Any)␤»
Skarsnik ugexe, if you are curious pastebin.com/GY9M25Vp 16:05
16:05 captain-adequate joined
timotimo FROGGS: oh, i thought grondilu said his build wasn't working; okay 16:06
grondilu: feel free to try to build an arm-based jit :) 16:07
16:09 telex left, telex joined 16:10 telex left, araujo left
TimToady m: grammar G { token TOP { <+ :kebab-case> }; token kebab-case { <[a]> } }; say G.subparse('aaa'); 16:10
camelia rakudo-moar 328a03: OUTPUT«#<failed match>␤»
TimToady m: grammar G { token TOP { <+ :kebab-case> }; token kebab-case { 'a' } }; say G.subparse('aaa'); 16:11
camelia rakudo-moar 328a03: OUTPUT«#<failed match>␤»
16:12 telex joined 16:13 araujo joined, araujo left
dalek p: 5848638 | TimToady++ | src/QRegex/P6Regex/Grammar.nqp:
allow <+ mr_ron> to be <+ mr-ron> too
 16:14
16:17 travis-ci joined
travis-ci NQP build failed. TimToady 'allow <+ mr_ron> to be <+ mr-ron> too' 16:17
travis-ci.org/perl6/nqp/builds/81043051 github.com/perl6/nqp/compare/9a53b...48638c16ca
16:17 travis-ci left
DrForr I need to get travis running on my builds too... 16:18
16:18 araujo joined, araujo left, araujo joined 16:19 araujo left 16:20 pdcawley left, telex left 16:22 telex joined 16:24 vendethiel joined 16:25 cognominal left
andreoss m: grammar Y { token TOP { <num>+ % ' ' { make $/>>.made } } ; token num { \d+ { make $/.Int.sqrt } } } ; Y.parse("1 2 3").ast.perl.say; 16:25
camelia rakudo-moar 328a03: OUTPUT«()␤»
TimToady >> would work only on positionals? 16:26
16:28 Peter_R joined
TimToady m: grammar Y { token TOP { <num>+ % ' ' { make $<num>>>.made } } ; token num { \d+ { make $/.Int.sqrt } } } ; Y.parse("1 2 3").ast.perl.say; 16:28
camelia rakudo-moar 328a03: OUTPUT«[1e0, 1.4142135623731e0, 1.73205080756888e0]␤»
TimToady m: grammar Y { token TOP { <num>+ % ' ' { make $<num>».made } } ; token num { \d+ { make $/.Int.sqrt } } } ; Y.parse("1 2 3").ast.perl.say; 16:29
camelia rakudo-moar 328a03: OUTPUT«[1e0, 1.4142135623731e0, 1.73205080756888e0]␤»
16:32 araujo joined, araujo left 16:33 domidumont joined, Loren left, telex left 16:34 telex joined, telex left, domidumont left 16:35 domidumont joined 16:36 telex joined, araujo joined 16:37 araujo left 16:40 larion left
andreoss >> breaks the order of elements 16:41
m: grammar Y { token TOP { <num>+ % ' ' { make $/.values>>.made } } ; token num { \d+ { make $/.Int } } } ; Y.parse("1 2 3 4 5 6").ast.perl.say; 16:42
camelia rakudo-moar 328a03: OUTPUT«(1, 2, 3, 4, 5, 6)␤»
andreoss hm 16:43
TimToady why are you using $/.values instead of just $<num>? 16:44
andreoss m: grammar Y { token TOP { [ <n> | <l> ]+ % ' ' { make $/.values>>.made } } ; token n { \d { make +$/ } } ; token l { \w { make ~$/ } }} ; Y.parse("1 a 2 b 3 c 4 d").ast.perl.say;
camelia rakudo-moar 328a03: OUTPUT«(1, 2, 3, 4, "a", "b", "c", "d")␤»
16:45 telex left
TimToady .perl sorts 16:45
16:46 telex joined, telex left 16:48 telex joined 16:50 Peter_R left
andreoss m: grammar Y { token TOP { <norl>+ % \s { make $/.values>>.made } } ; token norl { [ <n> | <l> ] { make ($<n>//$<l>).made } } ; token n { \d { make +$/ } } ; token l { \w { make ~$/ } }} ; Y.parse("1 a 2 b 3 c 4 d").ast.gist.say; 16:50
camelia rakudo-moar 328a03: OUTPUT«(1 a 2 b 3 c 4 d)␤»
16:50 uvtc joined
uvtc Is anyone keeping a running log of ecosystem's `wc -l META.list`? 16:51
timotimo huh, what happened to the nqp build? 16:52
TimToady I thought I proved it wasn't my patch, but let me doublecheck 16:53
timotimo oh, ok
andreoss can it be done without extra token?
PerlJam uvtc: git has a log of it if you know how to ask
TimToady yeah, was already failing before my patch 16:54
well, make test was failing, it was building fine 16:55
timotimo tries
uvtc PerlJam: Ok, thanks. Was thinking it might be nice to have a plot of it over time (especially in time for the Christmas rush).
TimToady does travis run make test?
timotimo ah, yes 16:56
ooooh 16:57
nqp builds with MVM_SPESH_NODELAY=1 by now!
16:57 uvtc left 17:02 cognominal joined 17:04 nwc10 joined
dalek ast: 838b9c1 | TimToady++ | S05-metasyntax/charset.t:
check kebab-case in character classes
 17:06
17:07 larion joined 17:08 araujo joined 17:09 telex left 17:10 telex joined 17:12 Peter_R joined
dalek rl6-roast-data: 7dfda69 | coke++ | / (9 files):
today (automated commit)
 17:13
nwc10 TimToady: for me, commit 5848638c16cabe9d2e4ae288bd23796b103ea9ea breaks the NQP build. 17:14
the compile of gen/moar/stage2/NQPP5QRegex.nqp explodes with a backtrace
Cannot find method 'quote_escape-stopper-starter' at <unknown>:1 (./NQPHLL.moarvm:quote_atom:127)
... 17:15
moritz travis-ci said as much 17:18
17:18 Sqirrel left 17:22 Peter_R left 17:23 espadrine left
TimToady okay, that'll need whitespace now; dunno why it didn't blow up here 17:27
17:28 telex left, telex joined 17:31 Peter_R joined 17:33 Peter_R left
dalek p: a2c0550 | TimToady++ | src/HLL/Grammar.nqp:
remove accidental kebab from cclass
 17:34
japhb *yawn* 17:35
So what remains of the GLR fallout? Just JVM needing work? 17:36
Anyone have a rough idea of how the ecosystem is doing? 17:37
17:37 rindolf joined, telex left
japhb Oh, I guess ++jnthn has more he wanted to add to S07. 17:37
TimToady we're still auditing various listops to see which ones should be more one-arg-y 17:38
17:38 telex joined, travis-ci joined
travis-ci NQP build passed. TimToady 'remove accidental kebab from cclass' 17:38
travis-ci.org/perl6/nqp/builds/81055798 github.com/perl6/nqp/compare/58486...c05500a1f0
17:38 travis-ci left, Peter_R joined
timotimo accident all the kebab! 17:40
nwc10 TimToady: yes, that fixes it for me.
TimToady travis said as much :P 17:41
17:41 Sqirrel joined 17:44 Peter_R left, andreoss left
dalek kudo/nom: 74420e5 | TimToady++ | tools/build/NQP_REVISION:
bump NQP to get <+ mr-ron> fix
 17:45
kudo/nom: 38846d9 | TimToady++ | tools/build/NQP_REVISION:
Revert "bump NQP to get <+ mr-ron> fix"

This reverts commit ad27f74e734ed7bf0e6d89c8b4f335ae035393db.
kudo/nom: ff9c09f | TimToady++ | tools/build/NQP_REVISION:
bump NQP to get kebab-related fixes
[Tux] test 50000 36.936 36.856
test-t 50000 39.453 39.373
TimToady mr_ron++
skids (or + mr-ron as the case may be) 17:48
mr_ron Thank you. rfc3986 for URI's has BNF that is all kebab case but with inexperience I coded the URI grammar with underscores. The fix will hopefully let me fix that problem and make the grammar more like the rfc. Hopefully coming soon. 17:49
17:49 telex left 17:50 telex joined 17:54 Peter_R joined 17:58 Peter_R left 17:59 telex left 18:00 Peter_R joined, telex joined 18:02 raiph left 18:14 Gruber is now known as Grrrr, Sqirrel_ joined, Sqirrel left
mr_ron m: gist.github.com/ronaldxs/df389180dffb7c216131 18:16
camelia rakudo-moar 328a03: OUTPUT«Method 'delims' not found for invocant of class 'TestGrammar'␤ in regex reserved at /tmp/KCUFwvw2Lx:14␤ in regex TOP at /tmp/KCUFwvw2Lx:13␤ in block <unit> at /tmp/KCUFwvw2Lx:20␤␤»
mr_ron worked on my system at home ...
[Coke] j: use Test; is-deeply(<a>, 'a')
camelia ( no output )
[Coke] m: use Test; is-deeply(<a>, 'a') 18:17
camelia rakudo-moar 328a03: OUTPUT«ok 1 - ␤»
18:17 telex left 18:18 perl6_newbee left, telex joined 18:19 leont joined
PerlJam mr_ron: you should make a little parser that translates from the notation used in RFC3986 to Perl 6 regex 18:19
mr_ron: oh, and it mayn't have worked here because camelia hasn't updated yet. 18:22
[Coke] p6: say nqp::p6box_n(nqp::time_n); 18:24
camelia rakudo-moar ff9c09: OUTPUT«1442600680.11175␤»
[Coke] j: say nqp::p6box_n(nqp::time_n)
camelia rakudo-jvm ff9c09: OUTPUT«===SORRY!===␤java.lang.NullPointerException␤»
[Coke] j: say nqp::time_n 18:25
camelia rakudo-jvm ff9c09: OUTPUT«===SORRY!===␤java.lang.NullPointerException␤»
[Coke] j: nqp::time_n
camelia rakudo-jvm ff9c09: OUTPUT«===SORRY!===␤java.lang.NullPointerException␤»
[Coke] "welp, there's yer problem."
dalek ast: 13af7d6 | lizmat++ | S02-types/bag.t:
Make S02-types/bag.t at least compile
 18:26
[Coke] m: nqp::barf; 18:27
camelia rakudo-moar ff9c09: OUTPUT«===SORRY!===␤No registered operation handler for 'barf'␤»
[Coke] j: nqp::barf;
camelia rakudo-jvm ff9c09: OUTPUT«===SORRY!===␤java.lang.NullPointerException␤»
18:30 telex left
_itz is there anyway of putting constraints on slurpy parms *@a since I can't use types and ideally would like to specify "at least one" slurpy parm 18:31
dalek ast: 95a921a | TimToady++ | S02-types/bag.t:
don't need both flat and subsig
18:32 telex joined 18:33 larion left 18:34 rurban left
japhb _itz: *@a where * > 0 18:37
(Can be shorter, but that's one of the clearer methods)
_itz ah yes thanks 18:39
dalek kudo/nom: 72f58f7 | FROGGS++ | src/core/CompUnitRepo (2 files):
let CURLF respect :file, fixes RT #126096
 18:43
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126096
moritz where ?* # shorter, less clear
FROGGS umm 18:45
18:47 telex left
PerlJam can't you just make the slurpy a required parameter? 18:47
TimToady where @a
er, where +@a
er, nevermind
where ?@a is slightly clearer than ?* 18:48
18:48 telex_ joined
TimToady misses his coffee sometimes... 18:48
PerlJam _itz: why do you want "at least one" anyway?
18:49 telex_ is now known as telex
lizmat TimToady: can't handle coffee anymore either, found tea to be a good substitute :-) 18:49
FROGGS m: my $module = "Test"; require ::($module)
camelia ( no output )
FROGGS m: my $module = "Test"; require ::($module); ::("&ok")(1, "...") 18:50
camelia rakudo-moar ff9c09: OUTPUT«No such symbol '&ok'␤ in block <unit> at /tmp/5fceaqZy0W:1␤␤Actually thrown at:␤ in block <unit> at /tmp/5fceaqZy0W:1␤␤»
FROGGS m: my $module = "Test"; require ::($module); import ::($module); ::("&ok")(1, "...")
camelia rakudo-moar ff9c09: OUTPUT«5===SORRY!5=== Error while compiling /tmp/mQrangPm3S␤Could not find module ::($module) to import symbols from␤at /tmp/mQrangPm3S:1␤------> 3 require ::($module); import ::($module)7⏏5; ::("&ok")(1, "...")␤»
FROGGS m: my $module = "Test"; require ::($module); import $module; ::("&ok")(1, "...")
camelia rakudo-moar ff9c09: OUTPUT«5===SORRY!5=== Error while compiling /tmp/8NR1vyaT15␤Undeclared routine:␤ import used at line 1␤␤»
FROGGS moritz: that code here is invalid: github.com/perl6/DBIish/blob/maste...sh.pm6#L14 18:51
moritz: it just happened to work by accident
moritz FROGGS: how so?
FROGGS require $foo, or require "foo" is always about a path
require ::($foo) is about a symbol 18:52
18:52 larion joined
moritz FROGGS: since when? 18:52
FROGGS: also, it makes no sense at all to me
FROGGS design.perl6.org/S11.html#line_345 18:53
moritz FROGGS: ::($thing) looks up a symbol; but how can we look up a symbol that hasn't been loaded yet?
18:53 telex left
dalek kudo/nom: aed3f93 | lizmat++ | src/core/IO/Handle.pm:
We don't need no nqp::not_i

By rearranging the code, we made lines again a few percent faster.		 18:53
ast: 76adbe5 | usev6++ | S32-list/minmax.t:
Add tests for RT #112250
 18:54
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=112250
FROGGS moritz: ::($foo) compiles to longname, like Foo::Bar does...
it is not only for looking up existing symbols
_itz PerlJam: because I want usage with no args and the slurpy matches that case 18:55
18:55 telex joined, telex left
lizmat _itz: merged 18:55
FROGGS moritz: so, seems the correct code also works with the rakudo release version... my patch just makes 'require Foo:file<bar>' work 18:56
_itz lizmat: ty
18:56 telex joined
FROGGS lizmat: btw, when looking at CURLF.candidates we should consider checking for a META6.json, this would allow us to handle distribution directories with non-standard CU file paths 18:58
or with unicode package names and another representation on disk
lizmat probably: atm I'm trying to get into a programming flow again in an area that's not as complicated as loading compunits 18:59
FROGGS: I expect to be in a flow again by next week
FROGGS lizmat: just wanted to tell somebody, in case I forget :o) 19:00
19:00 sufrostico left
FROGGS lizmat: I'm doing it now... 19:01
PerlJam It seems weird to me that require would have a :file colonpair where in other parts of the language we would have split require into 2 parts: require Symbol; require-path $filename;
FROGGS PerlJam: you can also do: require Symbol:file($filename) 19:02
PerlJam er, that didn't quite come out right, but I hope you know what I mean.
moritz PerlJam: you mean "different operations should have different operators" 19:03
PerlJam moritz: ding!
moritz where "loading by file name" and "loading by symbol" are different operations
I always hated that require behavior in p5
FROGGS well... the symbols boils down to a file also
moritz everything boils down to bits and bytes
FROGGS hehe 19:04
moritz and yet we still distinguish Int and Str
FROGGS IntStr /o/
psch grumbls
FROGGS :D
19:04 domidumont1 joined
PerlJam FROGGS: if we had the require/require-path of my imagination, require would just be a wrapper around require-path that knew how to translate symbols into paths. 19:05
19:06 telex left
FROGGS hmmm, I still think that these operations are too similar to not be split into several names 19:07
19:07 domidumont left
moritz I still find it surprising that "require Symbol" and "require $thing" treat their arguments quite differently 19:08
19:08 telex joined
moritz ah well 19:08
moritz lets it rest
[Coke] psch: not being able to run nqp::time_n is part of the problem, did you see that?
FROGGS that is just there to allow 'require "foo/bar.pm"', which might be the common case 19:09
moritz FROGGS: it might? really?
FROGGS I think so
psch [Coke]: part of which problem? 19:10
moritz FROGGS: in p5, everytime I've used the string form of require, I had to translate to the foo/bar.pm form first
FROGGS moritz: we could change it of course, which would mean that "my" common case would either be 'require SomeSymbol:file<...>' or 'require-path ...' or so 19:11
[Coke] j: nqp::time_n
camelia rakudo-jvm ff9c09: OUTPUT«===SORRY!===␤java.lang.NullPointerException␤»
[Coke] ^^ that opcode is used in is-deeply.
so every call to is-deeply is probably dying.
unfortunately, that NPE is happening in such a way I can't get a p6 backtrace. 19:12
psch $ ./perl6-j -e'use nqp; say nqp::time_n()'
1.44260353391E9
psch seems to accidentally have fixed that
i'm runnig spectest right now, for a hopefully-actually-finally-correct getBI (and makeBI) implementation
FROGGS psch: also works on my box... that jamelia here is a bit picky 19:13
psch++
psch FROGGS: well, it's running a pretty wonky getBI implementation right now :/
FROGGS moritz: I have no strong feelings about 'require STRING' btw, I just a feeling about changing things at this time
:S 19:14
[Coke] psch - if you leave off the use nqp;, you get a LTA failure.
... so probably not the problem. golfed it too much. 19:15
psch [Coke]: i don't, i just get the deprecation message and correct say output...
FROGGS [Coke]: still works here
19:20 colomon left, Possum joined 19:24 rurban joined 19:27 sufrostico joined, telex left, zakharyas joined
dalek ast: f5a77ba | usev6++ | S06-other/main-usage.t:
Add test for RT #119001
 19:27
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=119001
19:28 telex joined 19:35 Andrie joined
bartolin m: say 0.9999999999999999999999.perl # from RT #126016; haven't seen this one before 19:38
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126016
camelia rakudo-moar aed3f9: OUTPUT«0.A␤»
pink_mist 0_o A? 19:39
19:39 rurban left
geekosaur hex? 19:39
and o.O indeed
bartolin star-m: say 0.9999999999999999999999.perl 19:40
camelia star-m 2015.03: OUTPUT«0.10␤»
19:40 telex left 19:41 FROGGS left, Andrie left, FROGGS joined 19:42 telex joined
ugexe FROGGS: assuming Distribution.content was implemented, it would seem this would be a CUR::* universal way to copy the data to its new location. To me it seems like a nicer way of handling it, but at the same time CURLI and CURLF would probably remain much faster just copying the files. So would a new CURLI use Distribution.content? 19:42
pink_mist is it a specific number of 9s? does it do the same with one more or one less?
psch hrm, spectest started hanging somewhere 19:44
FROGGS ugexe: dunno, different files need different treatments... thinking of bin/ lib/ and resoures
ugexe: how would that work with .content?
psch S02-literals/allomorphics.t also dies at the second test...
ugexe yea i ran into that last night as well
psch apparently <0> and val("0") are somehow very different..?
i'm gonna let this sit for the night o/
ugexe although anything in lib/ that was in `provides` would be accessable 19:45
FROGGS sleep val, psch++
19:45 oskie joined
FROGGS hides 19:45
oskie when using rakudo, are "intermediate" files stored on disk somewhere? or just in memory when you "run" a perl6 program?
dalek Iish: 08d218e | moritz++ | lib/DBIish.pm6:
Fix "require" syntax.

  FROGGS++.
Closes #19		 19:46
flussence there aren't intermediate files, at least currently
moritz oskie: just in memory, unless you you explicitly precompile
PerlJam oskie: why do you ask?
bartolin pink_mist: looks like rounding starts with twenty 9's after decimal dot. 19:47
flussence (if it generated disk files on the fly, the first place I'd look is in $XDG_CACHE_DIR...)
oskie i ask only to try to learn stuff. it's confusing, compilers, intermediate formats, VMs, etc
flussence you can ignore all that stuff for the time being, the stuff you can view in a text editor is WYSIWYG 19:48
ugexe also RESOURCES would be available to Distribution.content as they are supposed to be in the meta
flussence (it used to precompile, but that was removed for a rewrite and hasn't been put back)
ugexe bin/ and hooks/ were the 2 i could not determine how to handle
oskie ok. also i'm curious what it takes to distribute and run perl6 programs to a set of servers (at the moment all my server admin code at work is written in perl5 or bash) 19:49
19:49 nwc10 left 19:50 zakharyas left
flussence that's easy, same as p5. -I and "use lib" work exactly the same way if you need to run unpackaged modules 19:50
[Coke] depending on your server OS, there might be a packaged version. more likely for that to happen after christmas. 19:51
oskie sounds good, and the only prerequisitve will be 'apt-get install rakudo' i guess (debian or ubuntu)
19:51 telex left 19:52 telex joined
jdv79 so is the release coke did the beta? 19:52
dalek kudo/nom: 719a90f | FROGGS++ | src/core/CompUnitRepo/Local/File.pm:
handle META6.json files in CURLF

That means we finally can have full unicode package names put into the META6.json in the provides section, but let that entry point to a file containing e.g. ASCII characters only. If one does -I/directory where the META6.json is in there, the mapping of package names to file paths will work out. This does not affect the current way of including -I/directory/lib directly, as this will not know about a META6.json.
PerlJam Why is is_approx() still in Test.pm? I would have thought it deprecated in favor of is-approx().
[Coke] probably missed. 19:53
FROGGS ugexe: we need to install the wrappers of the stuff in bin/ as we currently do in CURLI
[Coke] everything else was kebab'd.
flussence FROGGS: that commit... does that mean I can prove -e perl6 instead of prove -e 'perl6 -Ilib' now?
PerlJam There is an is-approx too, but there's also is_approx (with no deprecation notice) 19:54
jdv79 [Coke]: ?
FROGGS flussence: if there is an implicit -I.
flussence: but there is not, is there? 19:55
ugexe FROGGS: yea but the origin bin/ scripts are not required to be mapped in the meta file is what i mean, so .content would not be able to access those
PerlJam jdv79: The release Coke did is just another monthly release but with the GLR done. I don't know what you mean by "beta"
flussence I'll try it now, just to see what happens :)
FROGGS ugexe: what if you want to shell out to the scripts of your dist? then you'd need its actual path
jdv79 its near TimToady's bday is it not? 19:56
FROGGS jdv79: the RC0 you are talking about is about Perl 6 the language... not rakudo the compiler
mst jdv79: hard to tell - we know what date it is but unlike perl6 he hasn't yet announced which year 19:57
FROGGS jdv79: so if you see a release of roast, then that's The Release™
though, both go kinda hand in hand
mst: he said, "2015, most likely, maybe", or some such :o) 19:58
unless, you know, stuff like buses happen
mst FROGGS: that was about perl6
FROGGS: it's TimToady's birthday I'm talking about :)
(helps if you read the line I'm replying to)
FROGGS mst: well, I know as much as that his birthay will also happen this year :o) 19:59
ugexe i mean for installation. right now it copies the files, but .content would let you steam it to a new file (but the end result is the same). So some cloud based CUR that can't use copy() could still access the files of bin/ or whatever
mst FROGGS: suuure, people keep saying that, but until he confirms it I don't believe you
FROGGS mst: and I still think it can only be about both, you can't release a language without an implementation and vice versa
mst what?
larry's birthday is neither a language nor an implementation 20:00
I don't understand what you're trying to tell me
20:00 domidumont1 left
jdv79 anyway... thanks. 20:00
bartolin PerlJam: (is-approx) I don't know about details, but I vaguely remembered this commit: github.com/rakudo/rakudo/commit/e83b6020ca One day ago lizmat++ made a note about a better version of is_approx: irclog.perlgeek.de/perl6/2015-06-10#i_10731692
FROGGS mst: ohh, both meant 'compiler and language spec'
mst FROGGS: yeah, but I wasn't addressing either of those :D
FROGGS mst: ahh :o) 20:01
PerlJam FROGGS: Perl 5 was released without a language spec many many many times :)
[Coke] so, this compiler release might be considered as 1/2 of that. I don't think we've chatted about how we're going to "cut" a spec.
mst FROGGS: all I was doing was making a joke in response to < jdv79> its near TimToady's bday is it not?
[Coke] not taking the spec into account, we hopefully will just be improving and not making any huge incompatible changes for the remainder of the year. 20:02
20:02 telex left
mst FROGGS: that's why I suggested you went back and read the line I was replying to :D 20:02
PerlJam mst: confusion is hard to excise once it digs in.
20:02 telex joined
masak way I understand it, "releasing" the spec would mean making a snapshot of the spectests we believe are solid enough to deserve to specify the language. 20:02
20:02 telex left
mst PerlJam: I'm probably going to need to take off and nuke the joke from orbit 20:03
FROGGS the way I understand it, mst is just joking about TimToady's bday :o)
[Coke] masak: the problem with that is that there are on the order of 100K individual tests, and we have no plan on how we're going to mark them.
PerlJam [Coke]++ I was just typing something like that 20:04
[Coke] (well, there's some syntax for that. but then do we do a prune based on that... in repo? out of repo? still unanswered questions there.)
20:04 telex joined
masak oh, I agree there are unanswered questions and that this hasn't been discussed/planned enough. 20:04
[Coke] can we just say "this whole file as of right now is 6.Acme"? test by test... 20:05
FROGGS [Coke]: we already can add fudge markers telling that tests meet a certain Perl 6 version or are conjectural
[Coke]: yes
[Coke]: test by test at leastrr
-rr
[Coke]: github.com/perl6/roast/blob/master/fudge#L86 20:06
20:08 araujo_ joined
PerlJam There's still a piece missing where we can link features to Perl6 langauge versions in the compiler for those inevitable future "use v6.christmas;" 20:08
FROGGS PerlJam: let's just not add features that can ever break existing code :o) 20:09
20:09 darutoko left
PerlJam mumbles something about history and doom 20:09
[Coke] FROGGS: right, we have the syntax and can mark things... but no plan around it, is what i'm saying. 20:10
FROGGS [Coke]: and no attempt at developing a plan, sadly
20:10 araujo left
[Coke] well, plan A: Make larry do it, it's his language. 20:11
... that won't fly, so let's add jnthn, since he's clever.
after that my plan kind of peters out. What's the 3rd approximation? ;)
FROGGS [Coke]: I was hoping that either pmichaud or TimToady would give a statement about these markers, or like, play with it in a roest file
roast*
Ulti I cant tell if its real or not but panda feels a *lot* faster than it used to be
FROGGS Ulti: also feels faster to me, fwiw 20:12
Ulti tadzik++ and everyone doing the GLR and IO optimisations I guess
flussence it got a faster JSON thing, that probably helped a bit
PerlJam &
FROGGS Ulti: ohh, maybe that also because it does not precompile anymore?
Ulti ahhh yeah JSON::Fast
hah so actually its just a big cheater 20:13
doing less
FROGGS aye
:o)
that's how you do it
my wife won't let me though :o(
20:13 telex left
Ulti what cheat or do less :P 20:14
20:14 telex joined, rurban joined
Ulti either way its nice to be able to suck down some modules quickly 20:14
Linenoise is definately doing some compiling though and that was still a lot faster 20:15
20:15 rurban left
Ulti like the compiling bit was the slow part for a change 20:15
FROGGS Ulti: doing less :o)
[Coke] ponders pursuing a CS/IT related masters, and wonders how to tie that to p6. 20:17
Ulti [Coke] do you mean into language dev or just using P6? 20:19
jdv79 [Coke]: why? 20:20
20:24 telex left, telex joined 20:25 telex left, sufrostico left 20:26 telex joined
flussence does anyone know the correct procedure to manually precomp installed modules? I tried `perl6 --target=mbc --output=Linenoise.moarvm Linenoise.pm` and get a 1.5MB file from it, but the repl still takes 4 seconds to start up. 20:30
timotimo oh hey Ulti 20:31
were you the one who recently-ish built some code to find prime numbers?
20:32 lea joined
prammer Hello all 20:33
20:33 sufrostico joined
prammer I'm reading S07, about the single argument rule 20:33
and how "for @a, { }" is a single iteration
Part me wants the next line to say that "for @a {}" is @a.elems iterations
masak m: my @a = 1, 2, 3; for @a, { .say }
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5===␤Expression needs parens to avoid gobbling block␤at /tmp/VZkh18RdPj:1␤------> 3my @a = 1, 2, 3; for @a, { .say }7⏏5<EOL>␤Missing block (apparently taken by expression)␤at /tmp/VZkh18RdPj:1␤------> 3my @a = 1, 2, 3; for …»
masak prammer: that seems to need parens 20:34
m: my @a = 1, 2, 3; for (@a,) { .say }
camelia rakudo-moar 719a90: OUTPUT«[1 2 3]␤»
masak m: my @a = 1, 2, 3; for @a { .say }
camelia rakudo-moar 719a90: OUTPUT«1␤2␤3␤»
prammer masak: so it does. it's in S07 though
20:35 telex left
prammer perhaps S07 should say for (@a,) and right after that show for (@a) and/or for @a 20:36
20:36 telex joined, eternaleye joined
prammer just what came to mind when reading it 20:36
masak yes, if that's in S07, then it should be changed. 20:39
flussence oh and fwiw, "prove -e perl6" in a properly packaged module's dir without -Ilib does indeed work now 20:41
masak wow, cool 20:43
what does the "properly packaged module" expand to in terms of individual steps?
pink_mist presumably a properly populated META6.json? 20:44
flussence yep, just that.
(mv'ing it makes the command stop working right away)
(editing it to have wrong paths does the same too, so it's definitely being read) 20:45
20:45 telex left 20:46 telex joined
TimToady rosettacode.org/wiki/Mad_Libs#Perl_6 is kind of amazingly short 20:47
20:50 timbunce joined
[Coke] jdv79: why what? 20:50
20:52 telex left, telex joined, telex left
jdv79 just curious why do you want to do a masters. 20:53
_itz www.youtube.com/watch?v=07KHwjebf7k
20:54 telex joined
[Coke] wants to learn, which he can do on his own, but also wants to further his career. 20:54
masak [Coke]: I'm pretty sure there is some masters topic that would benefit both you, academia, and Perl 6 20:55
[Coke]: Perl 6's particular brand of grammars comes to mind 20:56
vendethiel TimToady: why %. (the dot)?
Skarsnik Hm, I try to get the attribute list of a class. How I can do that without creating an object of the class? Metamodel::className::attributes?
dalek ast: 1cd6d48 | skids++ | S06- (3 files):
Add tests for anonymous capture with subsig and/or post-constraint
vendethiel m: class A { has $.a; };say A.^attributes;
camelia rakudo-moar 719a90: OUTPUT«(Mu $!a)␤»
timotimo you could do a great contribution to macadamia! 20:57
vendethiel Skarsnik: ^^ ?
20:57 skids left
timotimo Skarsnik: perl 6 has a pretty interesting thing called "the type object" 20:57
Skarsnik Oooh thanks
timotimo where the type itself is an object that stands in for the type
vendethiel m: say %{5}
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/AJvlsilohq␤Unsupported use of %{5}; in Perl 6 please use %4␤at /tmp/AJvlsilohq:1␤------> 3say %{5}7⏏5<EOL>␤»
jdv79 i was looking for some gem of a reason that might help motivate me to finish a bachelors;) as time goes on it seems less intetesting. mostly.
vendethiel m: say %.{5}
camelia rakudo-moar 719a90: OUTPUT«(Any)␤»
vendethiel ok, I see why now. TimToady++ 20:58
masak m: say %{$0}
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/idb1uR0MAV␤Unsupported use of %{$0}; in Perl 6 please use %($0)␤at /tmp/idb1uR0MAV:1␤------> 3say %{$0}7⏏5<EOL>␤»
masak vendethiel: there's basically an over-eager warning in the way :)
vendethiel doesn't like those that much, as a non-fiver
masak s/warning/parse error/
vendethiel but that solution is *really* cute 20:59
TimToady especially compared to all the others
vendethiel TimToady: just a question, why doesn't the Perl5 version use <>?
TimToady don't ask me 21:00
21:00 telex left
vendethiel ..but it could, right? I'm just trying to figure that out 21:00
21:01 telex joined, telex left 21:02 telex joined
vendethiel TimToady: I'm now using this one-liner so scare people in :D 21:11
Skarsnik ok, another probably dumb question, how I get the type of an attribute? (when defined) 21:14
vendethiel m: class A { has $.a; };say A.^attributes[0].WHAT; 21:17
camelia rakudo-moar 719a90: OUTPUT«(Attribute)␤»
vendethiel Skarsnik: when I see that, I go look at github.com/rakudo/rakudo/blob/nom/...tribute.pm
and see there's a $!type
Skarsnik Oh, there is a WHAT method. I was looking at doc.perl6.org/type/Attribute 21:18
ugexe and .^name, but i forget what the slight difference is between that and .WHAT
Skarsnik Oh yeah, could have use ^name on it ~~ 21:19
21:21 muraiki left, telex left 21:22 telex joined
dalek ast: d375959 | usev6++ | S32-array/ (2 files):
Replace eval-dies-ok with throws-like (EVAL form) for some tests
 21:26
21:27 CurtisPoe joined
CurtisPoe p6: my Rat $foo = 3 21:28
camelia rakudo-moar 719a90: OUTPUT«Type check failed in assignment to '$foo'; expected 'Rat' but got 'Int'␤ in block <unit> at /tmp/tmpfile:1␤␤»
CurtisPoe Can someone remind me why Ints aren't automatically promoted to Rats here?
TimToady m: my Rat() $foo = 3 21:30
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/aI7lcTiuMT␤Coercion Rat(Any) is insufficiently type-like to qualify a variable␤at /tmp/aI7lcTiuMT:1␤------> 3my Rat() $foo7⏏5 = 3␤ expecting any of:␤ constraint␤»
Skarsnik I wonder if there is already a module that do what I am writing (creating a table in a sqlthing from the class definition)
pink_mist you mean ... an ORM?
bartolin m: my @pop = 1 .. 5; @pop.pop = 7; say "alive" # should this die? S32-array/pop.t assumes that 21:31
camelia rakudo-moar 719a90: OUTPUT«alive␤»
21:32 mr_ron left 21:33 mr_ron joined 21:35 telex left 21:36 telex joined
masak CurtisPoe: because when you opt into typechecking, that means something. if you don't want that, you don't want to declare a type. 21:37
a Rat is not a kind of Int in any meaningful way. 21:38
CurtisPoe masak: 3/1 is a Rat.
masak right. and 3 isn't.
m: my Rat $foo = 3/1; say "alive" 21:39
camelia rakudo-moar 719a90: OUTPUT«alive␤»
CurtisPoe p6: say 3 == 3/1
camelia rakudo-moar 719a90: OUTPUT«True␤»
masak numeric equality has little to do with types.
Skarsnik pink_mist, yes, well I am just writing the thing to create the db. but a quick look on the modules list did not give something like a orm
masak m: say 3 eq "3" 21:40
camelia rakudo-moar 719a90: OUTPUT«True␤»
masak you wouldn't expect `my Int $foo = "3"` to work....
Skarsnik is there a pragma to force this to fail? 21:41
bartolin m: my Rat $foo = 3.Rat; say $foo.nude
camelia rakudo-moar 719a90: OUTPUT«(3 1)␤»
CurtisPoe masak: mathematically, all of us know that .1 + .2 - .3 equals 0 and Perl 6 agrees that it is. Mathematically, all of us know that 3 equals 3/1, and perl 6 agrees that it is. Perl 6 tries hard to do the right thing by the developer, but here it fails hard.
masak CurtisPoe: it fails because you promised to give it one type, and gave it another.
CurtisPoe: what's funny is that I argued your side about 6 years ago :)
CurtisPoe Since we *know* that 3/1 and 3 are the same thing, why can't Perl 6 apply this evenly? What's the downside?
masak CurtisPoe: and moritz++ argued my side. he eventually convinced me. 21:42
CurtisPoe: if you don't want the strict typechecking, why are you declaring the strict type?
you're being inconsistent, and blaming Perl 6 for a stricture you *asked for*.
CurtisPoe I'm hitting cases in my new Perl 6 for Mere Mortals talk where I tell people how wonderful Rats are and then fall back to Reals in my slides because otherwise I have to keep telling people "the computer doesn't understand intergers here/
s/intergers/integers/
It's an embarrassment. 21:43
masak it's not an embarrassment to have a consistent type system.
CurtisPoe Actually, like in many languages when programmers argue that 7/2 must equal 3 because integer/op/integer must give an integer when we clearly know that's not true.
masak this feels like the opposite of that. 21:44
CurtisPoe Please tell me the downside of saying "my Rat $foo = 3" so I can explain it to a skeptical audience of people who aren't Perl devs.
masak you asked for a Rat, and gave yourself an Int. Perl 6 complained. now you're unsatisfied with that.
lizmat m: my Int $foo = "3"; say $foo
camelia rakudo-moar 719a90: OUTPUT«Type check failed in assignment to '$foo'; expected 'Int' but got 'Str'␤ in block <unit> at /tmp/9B0CbFg0_p:1␤␤»
21:44 telex left
lizmat m: my Int $foo = <3>; say $foo 21:44
camelia rakudo-moar 719a90: OUTPUT«3␤»
masak we've been down the road of implicit coercion, and rejected it every time. 21:45
it sounds like a great idea, but it's a bad idea.
CurtisPoe Why?
Here we
masak it introduces laxity which directly or indirectly causes logical errors in the code.
lizmat m: my Rat $a = <3>
camelia rakudo-moar 719a90: OUTPUT«Type check failed in assignment to '$a'; expected 'Rat' but got 'IntStr'␤ in block <unit> at /tmp/Suokno2cOD:1␤␤»
CurtisPoe Here we're now exposing computer/compiler limitations when much of the talk is about how Perl 6 solves human problems, not computer ones. And then there's this ...
Skarsnik IntStr fun type x) 21:46
21:46 telex joined
masak CurtisPoe: 3 is an Int. Int !~~ Rat. you're getting the type error because you're trying to feed the 3 value to the Rat type. 21:46
21:46 sufrostico left
masak CurtisPoe: if you don't want the type error, you need to either (a) coerce, or (b) fix the value to be a Rat, like 3/1. 21:47
CurtisPoe You're exposing a compiler limitation? You keep saying this is good, but you haven't explained why.
masak I don't call it a compiler limitation. I call it a consistent type system.
vendethiel m: say Int ~~ Rat; # *g* 21:48
camelia rakudo-moar 719a90: OUTPUT«False␤»
masak you're asking to make the type system guess more, thus making it less good at its job: catching type errors.
CurtisPoe It's counter-intuitive. You said yourself that you argued against it. In this case, we know mathematically that every Int in existence is equivalent to the Rat $int/1. There is no problem with promoting the type. 21:49
masak from a design perspective, yes, I've raged against the way things are factored just as you do now. but I'm basically not unhappy about it anymore.
it really does come down to having the right expecations of the type system.
if you don't want it to give you an error there, you should take a step back and think about what it is you want to check for. 21:50
CurtisPoe Again, please explain how promoting something to a GUARANTEED equivalent value is a problem.
Int/1 == Int (every time).
masak you're arguing from mathematical intuition, which is fine. 21:51
but in Perl 6, for better or worse, Int !~~ Rat
CurtisPoe At OSCON, I'm going to be dealing with an already skeptical (possibly hostile) audience and I don't want to say "because #perl6 said type theory"
vendethiel I just noticed how much c# is cheating there: rosettacode.org/wiki/Mad_Libs#C.23
masak first off, I don't speak for #perl6 :)
vendethiel on another news, recent chrome update broke # -_-""
CurtisPoe That's fair enough, but Perl 6 seems to be very deeply wrong on this one :) 21:52
vendethiel (i mean, i'm sure it's only for my french keyboard and alt gre, but seriously, grrrr)
masak second, am I understaing you correct that you are asking for a special case of Int literal -> Rat type promotion in Rakudo?
Skarsnik m: my Int a = 2; my Str b = "5"; say $a +$b;
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/bbpYxr53St␤Malformed my (did you mean to declare a sigilless \a or $a?)␤at /tmp/bbpYxr53St:1␤------> 3my Int a7⏏5 = 2; my Str b = "5"; say $a +$b;␤»
Skarsnik m: my Int a = 2; my Str $b = "5"; say $a +$b;
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/CtXaQU6T76␤Malformed my (did you mean to declare a sigilless \a or $a?)␤at /tmp/CtXaQU6T76:1␤------> 3my Int a7⏏5 = 2; my Str $b = "5"; say $a +$b;␤»
vendethiel has not much against implicit conversions that are totally guaranteed to be non-lossy, and that'd require stuff like "integer suffixes/prefixes", etc 21:53
masak CurtisPoe: what about Int variable -> Rat type promotion?
pink_mist Skarsnik: $a, not a
Skarsnik I sucks
m: my Int $a = 2; my Str $b = "5"; say $a +$b;
camelia rakudo-moar 719a90: OUTPUT«7␤»
CurtisPoe I'm not going to argue that strings and numbers should share operators.
Skarsnik why does that work?
masak CurtisPoe: that was a bit of a non sequitur answer, but OK.
CurtisPoe: I'm trying to map out the structure of your feature request here. 21:54
CurtisPoe Skarsnik: well, that's awkward. It looks like Perl 6 is coercing a Str to an Int, which isn't safe, but won't promote an Int to a Rat, which is.
vendethiel does it?
m: my Int $a = "123";
camelia rakudo-moar 719a90: OUTPUT«Type check failed in assignment to '$a'; expected 'Int' but got 'Str'␤ in block <unit> at /tmp/24KrD7ej1E:1␤␤»
vendethiel doesn't seem like it does..
masak of course infix:<+> coerces to numeric. 21:55
CurtisPoe vendethial: look at Skarnik's example.
Skarsnik why of course?
masak but it's safe, in that it *dies* if the string is not a number.
Skarsnik: because Perl.
Skarsnik: that's how Perl operators work. always did.
vendethiel CurtisPoe: you know, it's that much easier with <tab>, and it removes the possibility of mistakes :P
(i'm talking about name completion
21:55 telex left
vendethiel ) 21:55
masak Skarsnik: there's infix:<+> for numeric addition, and infix:<~> for string concat.
CurtisPoe masak: Ah, I see. That still appears troublesome to me. So Perl 6 is trying to DWIM on a possibly unsafe operations, but not on the safe ones.
masak Skarsnik: Perl operators are defined by working on a particular type, or range of types. 21:56
CurtisPoe: there's a crucial difference.
21:56 telex joined
Skarsnik yes, but how to force perl6 to not allow to write this? 21:56
vendethiel doesn't believe that much in DWIM, and thinks "$a == @b" will surprise and bite a lot of people, but at least when you read it, you know it smells weird
masak CurtisPoe: assignment isn't magical in the way infix:<+> is magical. assignment just tries to assign.
CurtisPoe: one reason assignment wants/needs to be very non-magical is performance.
Skarsnik for me the type system is pretty pointless if you have to know where there is so 'magic' that occur at run time 21:57
flussence it's a lot easier to add typecasts where needed than to parse user input.
masak CurtisPoe: if we overload it with all kinds of coercing magic, then that might affect the performance of all of Perl 6. (maybe. I'm no expect.)
expert*
CurtisPoe masak: ok, I won't argue that. I'm just asking for ANY counter-example which shows why assuming the known mathematical fact that Int/1 == Int would be a problem.
I'm only asking for the one, obvious change that is going to byte (sic) developers: the common case where Int/1 == Int.
masak CurtisPoe: I could see it happening. unlikely but possible. that's why I started asking about what exactly it is you're expecting. 21:58
CurtisPoe: but I do think it sets a bad precedent.
vendethiel also, Perl6 cheats WRT its type system in a few places already :)
masak CurtisPoe: the problem is still that you have an expectation about the relation between Rat and Int which doesn't hold in Perl 6.
CurtisPoe And then I could avoid a very awkward line of questioning in my talk, or start rewriting my slides to (dishonestly) avoid awkward questions.
vendethiel and int vs Int or num vs Num is very hard to argue about when you value "mathematical purity"
masak CurtisPoe: I could imagine TimToady popping in at any moment and saying "sure, we could give you that kind of magic, no prob!" 21:59
CurtisPoe: but I could also imagine jnthn popping in and saying "that would *kill* performance"
CurtisPoe masak: I realize my expectation is different, I'm merely asking for ANY counter-example, anywhere, for anything, which would show why a known mathematical fact is problematic.
First make it right, then make it fast :)
masak CurtisPoe: I think it's a bad precedent to introduce exceptions in the type system.
leont CurtisPoe: my real question would be: why declare types explicitly there? 22:00
masak CurtisPoe: the fact remains that Int !~~ Rat.
leont IME, it's rarely (though occasionally) useful for lexicals to be types, but not that often
vendethiel masak: the type system already has exceptions
that'd not be a precedent
leont s/types/typed/
lizmat I wonder if the typecheck exception couldn't be catchable, and allow for ad-hoc coercion on the fly
masak vendethiel: and, IIRC, jnthn has argued against most of them.
CurtisPoe Int !~~ Rat only in the case of how the type system is defined. In reality, Int/1 always equals Int.
vendethiel m: my int $a = 1; my Int $a = 1; # that's the kind of thing that Just Works, expectedly
camelia rakudo-moar 719a90: OUTPUT«Potential difficulties:␤ Redeclaration of symbol $a␤ at /tmp/uzcuXa3nmX:1␤ ------> 3my int $a = 1; my Int $a7⏏5 = 1; # that's the kind of thing that J␤Cannot assign to an immutable value␤ in block <unit> at /tmp/uzcuXa3nmX:1␤␤»
vendethiel masak: then he must've argued against Failure, etc, right? 22:01
(I honestly don't know)
masak vendethiel: yes, I think so.
leont CurtisPoe: you could define your own subtype that DWIMs, I think
vendethiel nice. I'm happy to know I'm not the only one :-)
masak m: my Rat() $foo = 3 # CurtisPoe: why don't you like this solution?
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/LaopcHHRyd␤Coercion Rat(Any) is insufficiently type-like to qualify a variable␤at /tmp/LaopcHHRyd:1␤------> 3my Rat() $foo7⏏5 = 3 # CurtisPoe: why don't you like thi␤ expecting any of:␤ …»
masak er.
CurtisPoe leont: not a solution when I'm trying to tell non-Perl devs about how easy and natural Perl 6 is :)
masak what...?
vendethiel also had to introduce () in his talk about MAIN, and felt somewhat "dirty" (er..) about it 22:02
feels like it's a feature you shouldn't be introduced to there, for that
CurtisPoe masak: even if that worked, that would be a special case of gradual typing that would look just weird.
Skarsnik For me it's really sad to have type when most of the time it feel like it something pointless ~~
masak CurtisPoe: it's saying "I know this isn't a Rat, but please do the last step of coercing it to one for me" 22:03
CurtisPoe: which is *exactly* what you're asking for here, but explicitly in code instead of magic in the implementation.
CurtisPoe I'm trying to show people that Perl 6 is *easy* and my Rat $foo = 3 defeats my argument.
vendethiel Skarsnik: it doesn't to some of us :)
leont 􏿽xABmy Rat $foo = 3􏿽xBB still doesn't look like sensible code to me 22:04
22:04 synbot6 left
vendethiel m: say 3.Rat.perl 22:04
camelia rakudo-moar 719a90: OUTPUT«3.0␤»
leont I mean, what does that type give you?
vendethiel m: my Rat $a = 3.0; # o//
camelia ( no output )
vendethiel CurtisPoe: apparently, you can just add .0 as a workaround =P
CurtisPoe vendethiel: that's the problem. I don't want a workaround. I want it to work :) 22:05
masak CurtisPoe: "I'm trying to show people that Perl 6 is *easy* and <misunderstanding of the type system> defeats my argument" is not convincing from my point of view
vendethiel CurtisPoe: sorry, they actually managed to convince me that 3 is an Int, not a Rat 22:06
CurtisPoe Again, can anyone point to ANY counter-example, ANY, which shows why that's a problem?
masak CurtisPoe: the more you argue your position, the more I want to *protect* the Perl 6 type system from random tampering.
vendethiel well, if the solution were to be "integer {pre,suf}fix", I'd take auto-promotion anyday
masak CurtisPoe: you have an itch to scratch, and you want to go so far as to weaken the type system for it, not accepting any compromise along the way.
CurtisPoe Especially since Int/1 is guaranteed to be correct. 22:07
lizmat my $a Rat() = 3 # CurtisPoe: would that work for you ?
CurtisPoe How is that weakening the type system?
masak you're wanting to introduce an exception to the strict type error you're getting.
an error which is *correct*.
CurtisPoe Again, I'm willing to accept ANY counter-example of why that would be bad.
masak because it would set a bad precedent to weakening the type system.
22:08 telex left
lizmat CurtisPoe: type checking is done at a *very* low level, it is actually written in C for each backend 22:08
22:08 telex joined
lizmat because it is a very hot path 22:08
CurtisPoe It's not weakening it. 3/1 == 3 mathematically. There are no exceptions.
22:08 telex left
lizmat this is why at least jnthn would be against fixing it at *that* level 22:08
CurtisPoe If it's a performance hack, I could maybe present that, but that opens up another can of worms.
OK, I have to hit the hay. Sorry for being a pain :) 22:09
lizmat m: say Int ~~ Real; say Int ~~ Rat # the latter shows the problem
camelia rakudo-moar 719a90: OUTPUT«True␤False␤»
vendethiel masak: I strongly disagree with "precedent" though. How isn't Failure the precedent of weakened type system? The question isn't about whether it's good or not, it's just there
(at least for now)
masak no problem. I'm happy to have this discussion again, six years later, from the other side ;)
flussence I'm not sure what one could usefully do with an Int/1 that can't also be done with an Int. 22:10
22:10 telex joined
flussence += ? 22:10
masak I think the failure mode/surprise with that would be when people did `my Int $foo = 3/1` and *that* failed.
lizmat m: use nqp; say nqp::istype(Int,Rat); say nqp::istype(Int,Real) # the lower level version 22:11
camelia rakudo-moar 719a90: OUTPUT«0␤1␤»
masak lizmat++ # arguing from hot paths and low-level implementation
vendethiel: you're right, of course.
vendethiel: still, it makes sense to not worsen the damage and introduce more exceptions and corner cases.
lizmat FWIW, I agree with CurtisPoe that it is a WAT
masak oh, me too
it is. 22:12
but it's at least an instructive one.
vendethiel masak: that's all I was arguing about! maybe I'm gonna argue about removing Failure one day :P
masak "you said you wanted a Rat there, but then you provided an Int. what's up with that? do you want types or not?"
lizmat m: my Rat() $a = 42 # why doesn't this work, is this an LTA error for a NYI feature ?
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/yhDNrNuVUL␤Coercion Rat(Any) is insufficiently type-like to qualify a variable␤at /tmp/yhDNrNuVUL:1␤------> 3my Rat() $a7⏏5 = 42 # why doesn't this work, is this␤ expecting any of:␤ …»
masak lizmat: I hope it's an NYI 22:13
flussence m: my $a is Rat = 42; # this is one!
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/vUXS7P2Wxq␤Variable trait 'is TypeObject' not yet implemented. Sorry. ␤at /tmp/vUXS7P2Wxq:1␤------> 3my $a is Rat7⏏5 = 42; # this is one!␤ expecting any of:␤ constraint␤»
masak lizmat: unfortunately, implementing it will not be sufficient for CurtisPoe's needs.
lichtkind hej there will be big article here on christmas :)
vendethiel
.oO( Failure is to Either what Effects are to Monad Transforms )
ers*
masak lichtkind: yay 22:14
lichtkind yes
flussence m: my $a is Rat() = 3; # this causes a bizarre error on my repl 22:15
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/nzXV5FoGzL␤Can't use unknown trait 'is ' in a variable declaration.␤at /tmp/nzXV5FoGzL:1␤------> 3my $a is Rat()7⏏5 = 3; # this causes a bizarre error on m␤ expecting any of:␤ TypeObje…»
lichtkind and granada article will be published tomorrow
lizmat m: sub a(Rat() $a) { say $a.nude }; a 42 # so why does this work then ?
camelia rakudo-moar 719a90: OUTPUT«(42 1)␤»
flussence (doesn't show up in evalbot)
lichtkind celebrating whenever in third last paragraph
becuase i felt it was a thing in granada
so i will be askeing more questions :) 22:17
good night
masak I do think it's an interesting discussion with Int !~~ Rat. but the whole argument above, which went in circles a few times, can be summed up as "why doesn't it typecheck?" -- "because you gave it the wrong type" -- "oh, don't give me that type theory crap!"
lizmat masak: why wouldn't my Rat() $a = 42 work for CurtisPoe?
22:18 TEttinger joined
masak lizmat: because he thinks that `()` is unintuitive and not easy. 22:18
lizmat: he thinks it should Just Work without the `()`.
22:18 araujo_ left
masak basically, because mathematical intuition tells him that 3/1 == 1, he is disappointed that `my Rat $foo = 3` doesn't DTRT. 22:20
and that's the WAT to him.
to me, it's a case of "you promised you'd do Rat, and then you did Int, and Int !~~ Rat, so you broke that promise" 22:21
in a way, the `Int !~~ Rat` part is where Perl 6 diverges from mathematical intuition.
they're both Real, but Int is not a kind of Rat in Perl 6.
(which it kind of is in mathematical intuition, although mathematicians seem to properly talk about "embeddings", which is a bit more careful than just equating things) 22:22
22:22 TEttinger left
sjn is that something worth "fixing", if for no other reason than to follow the principle of least astonishmen? 22:22
masak sjn: that's exactly what we were discussing. 22:23
and disagreeing on.
22:23 lichtkind left
masak sjn: the thing is, Least Astonishment is up here against "do you want a proper type system or not?" 22:23
sjn well, I'm a fan of the WTF/min metric. which option would give fewer?
masak which option would ruin the type system less? that's also worth asking. 22:24
sjn mm
masak because in the long run, having a predictable/simple/consistent type system will also provide benefits. 22:25
geekosaur I'm not convinced that numeric literals shouldn't be polymorphic though
masak there's the Real type, which subsumes both Int and Rat. if you declare things as `Real`, you're essentially saying that you don't care if you get a Rat or an Int back.
TimToady or a Num, which Rat can mutate into 22:26
masak TimToady: come again?
TimToady if you exceed 64 bits in the denom, you get a Num, not a FatRat
masak ah, yes.
TimToady which is also "anti-mathematical" 22:27
cognominal If we go that way and stay consistent, one should be able to detect/know when a type is a subset (or. more exactly equivalent of to the subset of another). If we go to that path, we get something like idris, not Perl 6
masak maybe the real question here is "do people benefit from the type distinction between Int, Num, and Rat?"
I kind of think they do.
cognominal And it does not dwim.
* detect/know/declare
masak because real type modeling is usually concerned with the exact type of things. sometimes those types are numeric types. 22:28
flussence well it does map neatly to non-programming maths (natural numbers, fractions, abstract symbolic things)
masak saying "I want this to actually be a rational, not an integer or a floating-point number" is *useful*.
which is why the strict type checking is ultimately worth defending.
and people who want looser type checking because they don't really care can do `Real`. 22:29
Skarsnik If you need to distinct them, you probably want something strict behind. otherwise you will just use like Int/Float
masak cognominal: yes, I was a little worried about that aspect, too.
cognominal: but I didn't want to say "slippery slope" to CurtisPoe.
leont Sounds to me like we should have answered "you want to use Real there"
masak leont: we did tell him that. 22:30
cognominal: if we give in to people's expectation of auto-coercing Int to Rat, then what about when two user-defined types don't behave like that?
sjn well, fwiw, I'm a fan of "easy to understand" > "correct" > "efficient", but I'm not sure how that translates from the implementation sphere into the usage sphare.. Haven't really thought about it much, since I usually assume that questions of usage (WTF/h or PoLA) usually are more focused on expectations and conventions than on implementation...
leont Vaguely, but not really why (or at least I'm only now understanding why, I was expecting Real to mean something closer to Num)
masak leont: yes, that's another common one. :) 22:31
leont To my mind, Num and Real should have swapped names
lizmat m: say Num ~~ Rat # isn't this a problem if the denom exceeds 64bit ??? TimToady
camelia rakudo-moar 719a90: OUTPUT«False␤»
22:31 skids joined
masak lizmat: only if you typed the variable as Rat and assumed it would stay that way forever. 22:31
cognominal masak: yea, slippery slope indeed.
masak leont: well, Real gets its name because these things are all on the real line. 22:32
lizmat masak: feels like a WAT that would typically occur after a *long* run of computations :-(
leont It's probably because real has floating point connotations to me
masak leont: and Num I cannot really provide a rationale for, except that I like it better than Float :)
lizmat: and then the user goes "awrgh, I shoulda typed it as Real!" 22:33
lizmat: and hopefully learns a lesson
leont Num is awfully generic, which is perfect for "this covers everything", which is what Real currently does
masak leont: no, you're thinking of Numeric :P
lizmat masak: this could be in a library :-(
masak m: say Real ~~ Numeric; say Complex ~~ Numeric 22:34
camelia rakudo-moar 719a90: OUTPUT«True␤True␤»
masak lizmat: we need to teach people how to use the numeric types right.
lizmat: there's just no way around that.
sjn would be very happy if we don't scare away math geeks (or Mathematicians) because of Perl 6 having "our kind of" types
cognominal When we see the underlying complexity of the conversion to boolean, we can imagine what it would be to play a similar game with numeric types.
masak intuition only goes so far.
sjn: nod
sjn: I think the thing that bothered me the most was that Perl 6 opted out of any kind of "numeric tower" business 22:35
22:35 captain-adequate left
masak sjn: which would essentially grant CurtisPoe his wish 22:35
but another point: I think it would wreak havoc with people's expectations in *signature* matching if we made Int ~~ Rat
or given/when matching, for that matter 22:36
users are inconsistent as hell
"I passed it an Int, and it matched on Rat" -- "well, you shoulda known that" -- "this sucks!"
22:37 telex left
sjn if a "numeric tower" would be implemented later at some time, what would break when compared with the current system? 22:37
masak sjn: some things that failed at compile time would start working. no big deal.
sjn: we could even try it out safely in a module/pragma.
sjn: people could opt into it if they wanted that semantics.
lizmat m: sub a(Rat() $a) { say $a.nude }; a 42 # masak: should this break ?? 22:38
camelia rakudo-moar 719a90: OUTPUT«(42 1)␤»
22:38 telex joined
masak lizmat: no, seems 100% legit to me. 22:38
lizmat: (because the coercion is explicitly asked for)
sjn well, basic mathematic concepts isn't really something that is ok to just make up as you go (even considering user's whims)
lizmat ok
masak sjn: I... agree...? :P 22:39
"don't make up math as you go!" -- probably most mathematicians
TimToady isn't that what mathematicians are for?
sjn and quite a lot of things out there that might provoke a "you should have know that" response 22:40
masak: sure, but the basics are the same :)
an integer in India is the same as an integer in Greece :)
or a Rat
mst sjn: that generally depends whether they open the till 22:41
22:41 rindolf left
masak to be clear, the surprise here is exactly that people expect `my Rat $foo = 3` to work from a *mathematical intuition* perspective (just like CurtisPoe did), because we know from math that Int can be embedded in Rat, and thus "ought to" smartmatch as Int ~~ Rat 22:42
but it doesn't in Perl 6, and we kinda consider that a feature, even though it goes clearly against mathematical intuition
and so we're keeping that WAT as a chance to tell you about how Ints are not Rats
cognominal Just like people must learn that (often doubtful) intuitions about reals don't match the behavior of floating point. 22:44
masak I was thinking about that earlier today.
sjn will this tradeoff make life harder for mathematicians considering using Perl 6? 22:45
masak the reason the reals are uncountable is that when we get to the point of defining them, we have to go all the way to *relations* with the rationals to do so. that's a heavy dependency.
whereas the data of a rational is literally just a sign and two non-negative integers.
sjn: no, I think it's all about habit. and getting the right answers after the WAT happens to you. 22:46
cognominal s/doubtful/dubious/
sjn ok, so this can be "solved" with documentation and sane warnings
masak sjn: what's interesting is that CurtisPoe didn't buy the "no, you have the wrong expectations" argument. basically because he's a very picky shopper who writes presentations that are meant to be easy for neophytes. 22:47
sjn: and `my Rat $foo = 3` not working fails his "it's easy" lithmus test.
sjn mm
well, it does
masak yes. 22:48
ShimmerFairy masak: I'm sure this is the same problem behind our persistent Inf issue, fwiw :)
masak ShimmerFairy: I don't think so...
well, they both have to do with the type system.
if you remove the type system, both of the problems go away :P 22:49
ShimmerFairy masak: Yeah, not quite the same, but it's still that same problem of wanting "type transcendence" in numerics
.oO(not to be confused with mathematical transcendentals :P)
 22:50
masak I guess it's a bit annoying for people arguing the "mathematical intuition" side that a number of languages out there *do* effortlessly implement the numeric tower, and then get on with life.
and Perl 6 choses not to, for whatever reasons.
and the WATs occur because of that.
ShimmerFairy I would ask what rearranging our type system would impact, but we have 'Num' and 'Rat' covering the same space, for example, so we have a bit of snag in cleanly mathematical organization there. 22:51
masak ShimmerFairy: I still see them as different things. CurtisPoe wants a thing to basically "upgrade" itself to a more derived type when it's completely safe to do so. Inf kind of wants to "blend in" as a closed set ({Int, Num, Str}) of types, and pass for either of them. 22:52
ShimmerFairy Perhaps other langs can organize things mathematically because they only have Nums, not Rats :)
masak ShimmerFairy: 'Int', 'Rat', and 'Num' all cover the "same space". and you could throw 'Complex' in there, too.
sjn 'Imaginary' :) 22:53
ShimmerFairy masak: sure, but I wasn't counting proper subsets :P
(FatRat and Num are better coupled as covering the same exact space, btw) 22:54
sjn Well, maybe this could be solved with a module? (use Mathematical::Types)
masak I think it could.
but I would call it `use Math::Typing::Lax` :)
ShimmerFairy Mathematicians wouldn't call it "lax mode" though :P 22:55
masak or `use Math::Typing::Sloppy`
well, the point is, it kind of *is* "lax mode" from Perl 6's point of view
sjn yeah, better name it so it goes well with mathematician's expecations
22:55 Peter_R left
masak it gave you this nice distinction between Int/Rat/Num, which you can actually use for things 22:55
and then you go "nah, I don't want that" 22:56
sjn 'use Proper::Math::Types' :)
ShimmerFairy masak: maybe if the numerics had some sort of "pseudo-C<subset>" feature, where Ints are considered a subset of Rats for example (even though they aren't really, in a P6 sense)
leont ShimmerFairy: thing is, sometimes they're not
masak ShimmerFairy: yes, cognominal was thinking about the same earlier. (and warned about becoming Idris.)
leont E.g. Complex is a superset of Int, except you can't order the former like you can with the latter 22:57
masak that's fine with supersets, though
you may lose properties that were true of the subset
22:57 Peter_R joined
masak (because you're zooming out and becoming more general, so specific properties may no longer hold) 22:58
if you zoom out further to the quaternions, you lose commutativity of multiplication
if you zoom out even further to the octonions, you lose associativity of multiplication
and after that, everything nice breaks down, so there isn't really a "hexadeconion" algebraic system 22:59
ShimmerFairy masak: nope, just sedenions :) 23:00
masak ShimmerFairy: oh, you're right 23:02
they lose yet another property, of being "alternative", whatever that is
ShimmerFairy apparently, that's x*(x*y) = (x*x)*y
masak we're following the same trail here :)
yeah, that sounds like... something I'd generally want to be true :) 23:03
ok, and it's weaker than associativity's x(yz) = (xy)z, which requires all three to vary
here, two are the same 23:04
apparently, the Cayley-Dickson construction goes as far up 2**N as you care to follow it. 23:05
leont quaternions at least are used in the non-academic world, I don't think octonions are, let alone sedenions.
masak but no-one seems to pay any interest past 16.
leont: that's my impression too. 23:06
leont Though even complex numbers aren't used all that often, IME 23:07
ShimmerFairy Without getting too lost in the math, it sounds like sedenions end up destroying what little faith you had left in order of operations (for multiplication at least), by making the direction you evaluate important :P
masak ShimmerFairy: well, it forces you to put in the parentheses in basically every multiplication. 23:08
(something I'd do just in case after the loss of associativity)
leont: complex numbers seem to have found their place in various branches of physics and engineering. 23:09
basically, I think of complex numbers as being almost as nice as the reals. (unexpectedly) 23:10
sure, you can't order them, but if that's not a problem, then they can basically be expected to behave as nicely as the reals.
23:10 telex left 23:12 telex joined
ShimmerFairy Anyway, I think that my Rat $a = 4 should either "promote" the literal there, or at least warn about assigning an Int to a Rat variable. 23:13
I think perhaps having assignment require a Rat() instead of Rat (or Rat(Numeric), or whatever) might work out better 23:15
(should assignment be coercive in general? I'd be surprised if such a change didn't cause problems) 23:18
23:19 telex left
skids Personally I think we have to diversify our vocabulary around "coerce" 23:19
dalek kudo/nom: 6252fc4 | lizmat++ | src/core/IO/Handle.pm:
Make "words".IO.lines($limit) lazy when needed

Basically, remove the old pre-GLR candidate and replace it with a class that does the right magic. In the process, also allow for Ranges to be specified as a limit.		 23:20
23:20 telex joined
lizmat which concludes my commits for today 23:20
sleep&
ShimmerFairy skids: I personally would prefer having better words for describing it. FWIW I don't use "convert" because to me that's something the 'owner' of the data does (i.e. my $a = $b.Rat is "conversion" to me, and foo($a) with an arg of Rat() is "coercion" to me) 23:25
23:27 mr_ron left 23:28 cognominal left
masak ShimmerFairy: if you think `my Rat $a = 4` should promote, what about `my Rat $a = $x`, where $x happens to be a Rat? 23:28
23:28 telex left 23:30 telex joined 23:39 telex left 23:40 telex joined, telex left 23:42 telex joined
ShimmerFairy masak: I thought about that too, and I'm not sure. (That "Rat() assignment" thing would solve it, but like I said probably cause more problems than solves) 23:44
m: my Rat() $a = 4 # aw :P 23:45
camelia rakudo-moar 719a90: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Ste2iwsGPf␤Coercion Rat(Any) is insufficiently type-like to qualify a variable␤at /tmp/Ste2iwsGPf:1␤------> 3my Rat() $a7⏏5 = 4 # aw :P␤ expecting any of:␤ constraint␤»
TimToady basically NYI
ShimmerFairy TimToady: the coercive assignment thing or the my Rat() $a thing? 23:46
TimToady or? they're much the same thing
there's no coercive assignment yet because there's no way to declare a coercive lvalue 23:47
ShimmerFairy oh, so my Rat $a = 4 wouldn't work, you'd have to do Rat() as the type decl ?
TimToady that's the plan 23:48
ShimmerFairy Sounds like the perfect solution to those silly mathematical perspectives on numeric types :) 23:49
TimToady we make a lot of concessions to mathematic intuition, but at some point mathematics must give way to computation
.oO(Computation is hard, lets go mathing...)
 23:50
23:50 leont left
masak ok, good. I'm gonna go out on a limb and take that as a Ruling. 23:51
TimToady maybe we'll let people say: subset Ratty of Rat(); 23:52
masak then the question becomes "how do we best explain to people how best to deal with math, numeric (non-)tower, types, and (non-)coercion?"
TimToady 'course, that's the same number of chars...
ShimmerFairy wonders what mathematician came up with -1 ** 2 == -1 , and if that counts as a mathematical concession in P6 :P
masak TimToady: I think it's just as elegant, or more so, to teach people the general thing with `()`
TimToady all of our types are implementation types, just some are closer to the metal than others 23:53
ShimmerFairy: yes, that's one of 'em
masak ShimmerFairy: that one doesn't even register as wrong or weird to me.
ShimmerFairy: but that's probably because I'm damaged by too much knowledge of precedence. 23:54
TimToady well, looks a lot better if you write it -1²
masak that's how I read the infix:<**> version too, basically
ShimmerFairy masak: whenever a language treats prefix:<-> as anything but part of a number literal (or more generally, part of the closest term), it feels like something's off.
masak ShimmerFairy: indeed.
23:54 TEttinger joined
ShimmerFairy (This counts when I read an introductory thing on Haskell, and it mentioned the weirdness on negation) 23:54
masak ShimmerFairy: there are parts of Perl 6 where it bothers me a lot more, though.
m: multi foo(1) { say "one" }; multi foo(-1) { say "minus one" }; foo(-1) 23:55
camelia rakudo-moar 6252fc: OUTPUT«5===SORRY!5=== Error while compiling /tmp/s5ZQYbhi1P␤Missing block␤at /tmp/s5ZQYbhi1P:1␤------> 3multi foo(1) { say "one" }; multi foo(7⏏5-1) { say "minus one" }; foo(-1)␤»
masak that's the most bothersome one, IMO.
23:55 mr_ron joined
ShimmerFairy maybe we should have <-1> literal Ints :) 23:55
masak no, I'd just like -1 to Just Work in that case. no questions asked. 23:56
TimToady we haven't used - in siggies yet
masak it's a place where I'd prefer the implementors to be tormented a wee bit more than they currently are.
ShimmerFairy masak: sure. While it's Cool™ that P6 does the right mathy thing with -1 ** 2 , I'd prefer if it was more aligned to how it looks
masak I don't have a corresponding wish with either prefix:<!> or prefix:<~>
TimToady ShimmerFairy: it's all tradeoffs, and there's no perfect
masak what TimToady said. 23:57
I have no sympathy for people who don't want to learn about precedence :)
ShimmerFairy TimToady: but of course :) That's why I thought of <-1> as a compromise of sorts, so that people who really want that behavior can get it
ShimmerFairy afk
masak I don't want a compromise. I want -1 to Just Work. 23:58
TimToady yes, well, the same instinct would make -1 ** 2 wrong
but we might cheat this one 23:59
masak ...in a roundabout way, it would, yes.
but I'm strangely not worried at all about that discontinuity.
for the purposes of a siggie, a "-1" is a single integer literal.

Powered by Cro and the Raku® Programming Language.

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

Thank you!