»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.
psch m: my $rx = / foobar /; say "foobar".match($rx) 00:03
camelia rakudo-moar 46874f: OUTPUT«｢foobar｣␤»
psch m: Str.^can('submatch').perl.say
camelia rakudo-moar 46874f: OUTPUT«()␤»
psch timotimo: the Str is what allows the matching to happen (which is another notch on my "something is the wrong way around with '$str ~~ $rx' log), and i don't know if a Str wants to support a submatch 00:04
it doesn't seem particularly valuable, from a "what do we use these constructs for" point of view 00:05
there's probably more than one angle to argue from here. something like "Grammars should have an ACCEPTS" on one hand, "it should be $rx ~~ $str" on the other hand 00:06
m: say "foo" ~~ /foo/
camelia rakudo-moar 46874f: OUTPUT«｢foo｣␤»
psch m: my $rx = /foo/; say $rx.ACCEPTS("foo")
camelia rakudo-moar 46874f: OUTPUT«｢foo｣␤»
psch m: say /foo/.match("foo") 00:07
camelia rakudo-moar 46874f: OUTPUT«Method 'match' not found for invocant of class 'Regex'␤ in block <unit> at /tmp/x1rhiS9nfZ:1␤␤»
psch it's just odd
psch i mean, from a "what can objects do" perspective it seems fitting enough - a string can match against a regex or not 00:07
but a regex ACCEPTS a string feels kind of off there
psch probably starts sounding a bit ranty there... 00:08
ShimmerFairy The problem with the idea of Regex.match is that there's already a Str.match, and I don't think it's a good idea to mix meanings like that. If anything, it should be called Regex.parse (and thus also Regex.subparse) 00:10
softmoth I'm not finding the right way to try using a module, and recover if it doesn't exist 03:45
like try { use Foo; CATCH { default { say "no Foo?" } } } 03:46
llfourn m: BEGIN { require Foo; CATCH { default { say 'no Foo' } } } 04:01
camelia rakudo-moar 46874f: OUTPUT«no Foo␤»
llfourn softmoth: that works I'm not sure how well it will import from Foo if it exists
softmoth ok. I found some possible workaround at github.com/supernovus/flower/blob/...ES.pm6#L36 04:06
softmoth I hope it is is something that is on the Rakudo roadmap for the near future 04:07
llfourn softmoth: that comment is old. it does work now 04:08
you have to do require ::($plugname)
softmoth OK. It doesn't do imports right though, I guess. I'm trying to make crustup (from p6-Crust) work with more Handler types, and not have to add 'use HTTP::Server::Tiny' at the top. If I take out 'use HTTP::Server::Tiny;' from the top of crustup, the require of the handler works, but code that it needs doesn't get loaded. It works if I replace 'use HTTP::Server::Tiny;' with 'use IO::Blob;', so that's the bit that 04:13
isn't loaded.
softmoth But when I try to use a different Handler, then some other dependency isn't loaded right 04:13
softmoth I thought maybe if I could do a conditional 'use' at the top of crustup, I could get things to work with a wider range of installed (or not) handler modules. 04:14
llfourn softmoth: I could probably help you easier if you put it into a gist
softmoth: are you trying to load + import or just load dynamically? 04:15
softmoth: I'm not really familiar with those modules unfortunetly :/
softmoth I'm not sure what I'm trying to do, either, really :-/ 04:16
llfourn hehe I know that feels
softmoth I'll try to get something simplified to ask a proper question.
llfourn sure there is some docs here if that helps: docs.perl6.org/language/modules#Loa..._Importing 04:17
softmoth Although, it appears that the Crust devs have changed the code I was hacking on, now that I look at it.
softmoth So I'll look into that; they've made it require ::($_) for @!modules, and maybe have some way of getting the needed modules in there. 04:18
llfourn good luck :)
softmoth thanks! :-)
moznion Some modules looks like they don't have `updated` column modules.perl6.org/ 04:22
Is this why?
softmoth The solution they've used is to be able to pass needed modules (e.g., -MFastCGI::NativeCall -MFastCGI::NativeCall::PSGI in my case) to crustup, and it does require ::($_) on those. 04:37
It would be nice for 'require ::(Module::That::Uses::FastCGI::NativeCall)' to handle that dependency automatically, but maybe there's something else going on that I'm not aware of 04:38
Anyways, the current solution seems to work as long as I know what dependencies need to be added 04:39
softmoth I'm curious if someone who actually knows how this should work would say if it's correct or not. 04:39
ugexe personally i think that type of stuff should go into some sort of compunitrepo specific to the application. then it can autoload all the modules (middleware) itself 05:21
ugexe as far as whats going on it sounds like its due to lexical scoping 05:28
softmoth ugexe: any pointers to code that does this the right way? 06:06
softmoth I'm not sure where the compunitrepo would live. Defining it in Crust would be brittle; every time FastCGI::NativeCall (for example) changes what modules it uses, Crust would have to make a change. Defining it in FastCGI::NativeCall doesn't make sense, since its author doesn't know that someone may want to require it at runtime. 06:15
softmoth Explicitly specifying modules (what looks like Crust::Runner's current approach) puts the burden on the end user, which is also brittle and tedious, but it works and puts the solution at the same level as the error will show up. 06:17
[Tux] test 50000 38.519 38.405 06:42
test-t 50000 39.674 39.560
azawawi hi #perl6 06:54
DrForr Morning. or UGT. 07:20
ugexe softmoth: the CUR can live anywhere. we have a hacky CUR for $work that autoloads modules this way but its definately a hack due to the lagging CUR situation in general 07:24
ugexe the idea appears to be you will define it from the command line like -M, but you'd only have to define a single repo instead of a path/name for each plugin 07:27
azawawi is there a Perl 6 syntax to return an Array[SomeObject] or Hash[Something]? 07:31
lizmat_ m: say Array[my Int $ = 42]
camelia rakudo-moar 46874f: OUTPUT«Parameter 't' requires a type object of type Int, but an object instance was passed. Did you forget a 'multi'?␤ in block <unit> at /tmp/0jfDGAwMgT:1␤␤»
lizmat_ azawawi: is that what you want ? 07:32
m: say Array[(my Int $ = 42).WHAT]
camelia rakudo-moar 46874f: OUTPUT«(Array[Int])␤»
azawawi i meant like sub foo() returns Array[Int]? 07:32
azawawi m: sub foo { [1] } returns Array[Int] 07:35
camelia rakudo-moar 46874f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/or9bKha4L5␤Strange text after block (missing semicolon or comma?)␤at /tmp/or9bKha4L5:1␤------> 3sub foo { [1] }7⏏5 returns Array[Int]␤ expecting any of:␤ infix␤ infix stopper…»
azawawi m: sub foo() { [1] } returns Array[Int]
camelia rakudo-moar 46874f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/SXH_cejOdU␤Strange text after block (missing semicolon or comma?)␤at /tmp/SXH_cejOdU:1␤------> 3sub foo() { [1] }7⏏5 returns Array[Int]␤ expecting any of:␤ infix␤ infix stoppe…»
azawawi m: sub foo() { [1] } returns Array
camelia rakudo-moar 46874f: OUTPUT«5===SORRY!5=== Error while compiling /tmp/2kzi3TV2PN␤Strange text after block (missing semicolon or comma?)␤at /tmp/2kzi3TV2PN:1␤------> 3sub foo() { [1] }7⏏5 returns Array␤ expecting any of:␤ infix␤ infix stopper␤ …»
azawawi lizmat_: got it. thanks :) 07:36
dalek kudo/curli: 4b1df7b | FROGGS++ | src/Perl6/Grammar.nqp:
allow whitestape in ::( "&foo"), RT #126385
 07:38
rakudo/curli: 42e55ec | lizmat++ | src/Perl6/World.nqp:
rakudo/curli: Make sure use invocant/parameters throw an NYI
synbot6 Link: rt.perl.org/rt3/Public/Bug/Display...?id=126385 07:38
lizmat sorry dalek
azawawi lizmat: is there a reason why perl6 process startup is a bit slow. I remember it was faster before. 07:46
lizmat I see it at 110 msecs at the moment, I've seen it below 100 in the past yes 07:47
lizmat azawawi: are those the numbers you're referring to ? 07:47
FROGGS we don't know exactly why it is slower right now, right?
lizmat or do you see much larger delays ?
FROGGS: my guess would be that we're missing a lot of optimizations post GLR 07:48
azawawi FROGGS: i will benchmark and report back
FROGGS that's also my guess, but it is hard to guess the things
azawawi mainly with the Selenium::WebDriver test suite 01-load.t (which does basically a use XYZ;) 07:49
FROGGS azawawi: well, keep in mind that also precomp is gone atm
azawawi FROGGS: aha 07:50
FROGGS that probably has the biggest impact there
azawawi FROGGS: so basically we're faster on the panda module installation but slower on the runtime?
FROGGS yes 07:51
though it was not about speeding up panda, it was about getting that fragile thing out of panda 07:52
azawawi interesting... time perl6 -MHTTP::UserAgent -e 'say "Hello world"' => real 0m5.916s
tadzik not that bad :)
FROGGS heh
azawawi tadzik: true, 5 seconds times 20 to 30 runs each day is not that bad :) 07:55
nine tadzik: indeed, not that bad if you're used to loading programs from floppy disk 07:56
azawawi nine: the old days :) 07:57
nine: the floppy loading sound made it bearable though lol
nine azawawi: yes, you always knew when it was doing sometihng :)
azawawi tadzik: what do you think is slowing down HTTP::UserAgent loading? 07:58
FROGGS I get there will be SSDs soon with a sound module, so you get more feedback when working... 08:02
like you've got force feedback in your steering wheel nowadays
tadzik azawawi: the sheer amount of code that has to be precompiled
FROGGS (in your car)
tadzik nine: or if you remember rakudo pre-nom :P 08:03
dalek p: 750f859 | (Pawel Murias)++ | src/vm/js/QAST/Compiler.nqp:
[js] Implement nqp::mod_i.
 08:32
p: cdf1f6b | (Pawel Murias)++ | t/nqp/59-nqpop.t:
[js] Test nqp::mod_i.
p: 2dfdbc1 | (Pawel Murias)++ | docs/HACKING-js:
[js] update test list path in docs/HACKING-js.
p: 0b106de | (Pawel Murias)++ | src/vm/js/nqp-runtime/reprs.js:
[js] Set native int attributes to zero when constructing a P6opaque object.
p: f49d614 | (Pawel Murias)++ | src/vm/js/ (2 files):
[js] Fix a bug in nqp::bindpos, make it work for negative indexes.
p: fc21e43 | (Pawel Murias)++ | tools/build/ (2 files):
[js] correct the autogenerated message on top of Makefile-JS.in
p: 5a5939a | (Pawel Murias)++ | src/vm/js/ (2 files):
[js] Stub nqp::objprimspec.
tokuhiro_ how do i include pod in synopsis section? 08:41
Ah, i should use '=begin code'. 08:42
ShimmerFairy tokuhiro_: if you meant a =SYNOPSIS block, it'll have to be =begin SYNOPSIS ... =end SYNOPSIS for multiple paragraph (unlike in POD from Perl 5, Perl 6's Pod doesn't run until a =cut appears) 08:43
tokuhiro_ o 08:44
dalek osystem: 1ebb389 | (Tokuhiro Matsuno)++ | META.list:
Added Test::Base
 08:51
llfourn m: say Date.new 10:11
camelia rakudo-moar 46874f: OUTPUT«2015-12-24␤»
llfourn ....wut
moritz you didn't tell it which day you wanted, so it gave you Christmas 10:12
gfldex m: say now; 10:13
camelia rakudo-moar 46874f: OUTPUT«Instant:1445336024.609831␤»
llfourn moritz: cool so will date give today after the release date?
gfldex m: say Date.new(now);
camelia rakudo-moar 46874f: OUTPUT«2015-10-20␤»
moritz m: say Date.today
camelia rakudo-moar 46874f: OUTPUT«2015-10-20␤»
FROGGS llfourn: no, it will probably give you the Christmas after that 10:14
llfourn FROGGS: ok well I guess at least that will make my christmas involved coding a bit easier 10:15
FROGGS :D
llfourn s/christmas/season's greetings/ 10:15
dalek kudo/nom: be14c49 | lizmat++ | src/core/Exception.pm:
X::Pragma::CannotNo -> CannotWhat

For more general applicability		 10:31
kudo/nom: 56dbb06 | lizmat++ | src/Perl6/World.nqp:
Make pragma handling a little more manageable
dalek ast: f331fd6 | lizmat++ | S (4 files):
Fix X::Pragma::CannotNo -> CannotWhat breakage
 10:41
kudo/curli: be14c49 | lizmat++ | src/core/Exception.pm:
X::Pragma::CannotNo -> CannotWhat

For more general applicability		 10:42
kudo/curli: 56dbb06 | lizmat++ | src/Perl6/World.nqp:
Make pragma handling a little more manageable
kudo/curli: 546cf76 | lizmat++ | src/ (2 files):
Merge branch 'nom' into curli
mrf ugexe: Is there any merit to moving the various RFCs for HTTP/MIME/URI ect into a seperate repo (Grammar::RFC::*) and just use them in the Grammar::HTTP. Would be nice to have an Email::Address parse for example without using the HTTP repo....Maybe I am overthinking things. Any thoughts. 10:47
ugexe: also I have forked you repo and am going to add RFC5234 which replaces 4234 FYI. Will throw back a PR when I am done. 10:48
nine mrf: +1 on splitting independent parts 10:49
ShimmerFairy mrf: I think it'd be cool to have a dedicated space for various IETF RFCs :) If you wanted to be a bit more specific, then I'd suggest Grammar::IETF::RFC (e.g. Grammar::IETF::RFC::_5234). 10:51
Or IETF::RFC::Grammar , depending on how you want to organize things :)
(you could also do RFC_5234 instead of RFC::_5234 if you'd prefer, of course) 10:52
pmurias moritz: (Date.new defaulting to christmas)-- 10:54
mrf Either is fine. I like the idea of having top level Grammar namespace for things like this and other standardised grammars. (CSS, XML, Email, Date) etc
ShimmerFairy pmurias: why is it bad? You didn't say what date you wanted? Could be February 32nd for all the info you didn't give :P
s:2nd/\?/./ 10:55
mrf ShimmerFairy: Not sure really how to namespace that much. Several RFCs like 5234 are around how to write RFCs which seems like they should have a separate namespace. but not sure
moritz pmurias: I know, we're becoming Enterprise and all, no place for fun anymore
ShimmerFairy mrf: the IETF::RFC::Grammar idea I gave is if your intention was more towards supporting a variety of RFCs in general. As always, TIMTOWTDI :) 10:57
mrf ShimmerFairy: Are there any RFCs outside of the IETF? 10:57
ShimmerFairy mrf: sure, notably the RFC process that started Perl 6 :) 10:58
nine moritz: it fun when you do it and the rare times one discovers it, but negative fun for those who are bitten by the surprising behavior... I guess the latter group is larger. 10:59
mrf ShimmerFairy: that makes sense.
ShimmerFairy mrf: even though I suspect most would first think of the IETF when it comes to 'RFC', I personally would clarify that it's about IETF RFCs somewhere, just to minimize any potential confusion :)
nine: what surprising behavior? I don't understand what people expect an unspecified date to turn out as.
nine ShimmerFairy: it's the worst kind of error. The one that doesn't explode loudly but seems to work at first glance but silently returns wrong results because someone forgot to type the argument. 11:00
mrf ShimmerFairy: Yeah. I was initially leaning towards Grammar::IETF::RFC::* then you can have Grammar::IETF::* for the various actual top level uses of the RFCs like HTTP/MIME/Email/Date. Especially if the RFC's are just roles.
which is what ugexe has done at the mo. Just need spliting 11:01
hoelzro o/ #perl6
ShimmerFairy mrf: the only thing is, if you plan on doing more than just grammars, you'd probably want the IETF::RFC::Grammar ordering (thus IETF::RFC_1923::Grammar or IETF::RFC::_1923::Grammar)
dalek kudo/curli: b154c49 | lizmat++ | src/ (2 files):
Implement 'no precompilation'
nine m: my @ymd; say Date.new(|@ymd) 11:02
camelia rakudo-moar 56dbb0: OUTPUT«2015-12-24␤»
nine ShimmerFairy: ^^^ this kind of surprise
ShimmerFairy and then you can do things like IETF::RFC_1923::Client, IETF::RFC_1923::Generator, and so on
lizmat m: say 'Life, the Universe, and Everything'.WHY # another surprise, but less dangerous :-) 11:03
camelia rakudo-moar 56dbb0: OUTPUT«42␤»
ShimmerFairy nine: I don't think that surprise is more special than any other time you'll try to slip an empty array
lizmat fwiw, I think a date should at least be given a year
nine ShimmerFairy: without the little joke it would explode loudly, would therefore be easy to discover and easy to debug. 11:04
ShimmerFairy nine: not necessarily. If Date defaulted to something more boring like 'today', would you still argue against it? 11:04
mrf ShimmerFairy: hmm. I am not sure what none grammar things would exist in the namespace.
hoelzro I was looking at S17-procasync/kill.t last night after fixing the bug that prevents Proc::Async.kill from working, and it seems that that test enumerates over all signals defined by the OS and throws each at a separate child process
since not all signals are equal (ex. SIGTERM vs SIGTSTP vs SIGKILL), I am wondering if that's such a good idea 11:05
llfourn m: wait 1 { say "worked" } # how I get this to work
camelia ( no output )
llfourn or what is the equiv of Mojo::IOLoop->timer( 1 => sub { say "worked" }) in p6 11:07
dalek ast/curli: 1451c65 | lizmat++ | S02-types/WHICH.t:
Add test for new X::Pragma::CannotPrecomp class
nine ShimmerFairy: would it be documented? The christmas joke isn't. In any case I'd still be against it, even if not as strong. But we do have a Date.today method that makes it explicit, so there's no need for Date.new to do the same. So Date.new can be more strict and catch more bugs before they hurt anyone.
llfourn and how do I make it wait...?
mrf So you might have a IETF::HTTP module that uses Grammar::IETF::HTTP which does a whole slew of Grammar::IETF::RFC::*. The name space of the Action object seems sane to place separate from the Grammars. Having Grammars separate seems sane in my brain. But I am usually wrong
lizmat hoelzro: please adapt test as you see fit... it was just a brute force approach to test what *could* work 11:08
hoelzro alright, will do!
ShimmerFairy nine: The only reason I'm a bit against your example is that it feels a bit unlikely. I don't feel like you'd often pass an array that's supposed to contain a year, month, and date, and have it accidentally be empty. 11:09
lizmat m: my $i = Supply.interval(1); $i.tap: { .say }; $i.wait # llfourn 11:10
camelia rakudo-moar 56dbb0: OUTPUT«(timeout)0␤1␤2␤3␤4␤5␤6␤7␤8␤9␤10␤11␤12␤13␤14␤15␤16␤17␤18␤19␤20␤21␤» 11:11
nine ShimmerFairy: parse some strange date format using a regex, pass the capture groups to Date.new. Something goes wrong and the regex doesn't match -> empty result -> no arguments for Date.new. Something along those lines.
lizmat hmmm... wonder how long camelia will run this
llfourn lizmat: thanks!
ShimmerFairy nine: hm. I'd of course say "check if your match was successful", but that does seem possible.
nine ShimmerFairy: of course! But people write imperfect code. And error paths are notoriously badly tested. I'd expect a bug like that when I write some one-off tool to get some information for a report out of a log file. And due to the wrongly parsed dates our average reaction time is an ok 3 months while in reality it isn't or something like that. 11:13
llfourn lizmat: so there's no once off?
nine ShimmerFairy: torture the implementors for the sake of the users, not implementors have fun at the expense of the users. 11:14
lizmat llfourn: ah, you want to execute it once a second from now?
llfourn lizmat: yep
lizmat m: await start { sleep 1; say "hello world" }
camelia rakudo-moar 56dbb0: OUTPUT«hello world␤»
ShimmerFairy nine: I think we can live with Date dying in the same way DateTime does without a year. I just don't like the "I don't like fun" argument, that's not a very good one by itself :)
llfourn lizmat: interesting... is sleep blocking the thread (or am I not meant to think about that as a p6 programmer)? 11:15
nine ShimmerFairy: oh I love fun! Just not when it's gonna hurt users more than it's fun for us.
lizmat llfourn: sleep is not taking up any CPU or a thread 11:16
I think it is implemented as a timed cue internally
llfourn lizmat: so p6 is not necessarily creating a new thread with "start"? 11:17
lizmat indeed: it is just scheduling the task to be done
llfourn lizmat: that's cool!
lizmat as soon as a thread is available, it will execute it
ShimmerFairy nine: I still don't think it's that problematic (esp. since it doesn't come up all that often), but like I said I'd be fine if it went otherwise :) 11:19
lizmat hmmm... it looks rather trivial to add :at and/or :in parameters to Promise.start (the underlying logic of start {}) 11:27
masak evenin', #perl6 11:27
lizmat masak o/ # cn time
sergot hi masak \o
masak .oO( I cn haz time )
BenGoldberg There already exists a method Promise.start 11:31
err
llfourn m: Promise.in(2).then: { say "win" }
BenGoldberg Promise.in
camelia ( no output )
llfourn m: await Promise.in(2).then: { say "win" } 11:32
camelia rakudo-moar 56dbb0: OUTPUT«win␤»
llfourn lizmat: that what you mean?
lizmat BenGoldberg: Promise.in returns a promise that is kept in X seconds 11:33
it does not execute any given code
ah, but with the then, you're in.... :-)
yup
llfourn :D
BenGoldberg m: Promise.in(1).then: { say "win" }; sleep 2 11:34
lizmat TIMTOWTDI :-)
camelia rakudo-moar 56dbb0: OUTPUT«win␤»
lizmat m: await Promise.in(1).then: { say "win" }
camelia rakudo-moar 56dbb0: OUTPUT«win␤»
lizmat no need for sleep, just await the Promise
pink_mist lizmat: I think he wanted to see if the promise would have started without needing to await 11:35
BenGoldberg Yup :)
lizmat ah, ok... yes :-)
llfourn mm yes just shows that sleep is not like perl5 sleep
lizmat perl 5 sleep doesn't take any CPU either 11:36
BenGoldberg As for a :at or .at, most event systems have either absolute clock timing or relative timing, not both.
llfourn lizmat: but it will block your process
pmurias moritz: I find the christmas joke about as funny as digging a hole and hoping someone falls in ;)
BenGoldberg There's a complicated reason why, but I forget what it was ;)
lizmat BenGoldberg: the :at would just be a front-end fro calculating an :in, really 11:37
*for
pmurias moritz: plus there is a 'new Date()' in javascript 11:38
llfourn lizmat: if you were to say use Mojo::IOLoop->timer(1, sub { say "hello" } ); sleep 10; you would not see your "hello" until after 10 secs I think. 11:39
lizmat llfourn: well, that's sub-optimal :-) 11:39
masak pmurias: re hiding private attributes in Perl 6 -- note that I was arguing for *not* hiding them (from .new/.perl) 11:40
llfourn lizmat: yes which is why you would always have to write a lot of messy callback type code to DWIM. but in p6 you can just sleep 10 the main block and still see your "hello" after 2 secs. 11:41
llfourn \o/ 11:42
lizmat llfourn: yes, that works in P6 :-)
masak pmurias: re "destroyed CPAN" -- I think you should consider taking me way less seriously :)
masak pmurias: I don't *really* think that there would be any utility in destroying CPAN. I was mostly sick of people who took the stance that promoting Inline::Perl5 too much would hinder Perl 6 development, and I wanted to satirize that. 11:44
masak I've heard next to no reactions on strangelyconsistent.org/blog/macros...ut-on-hold , btw -- except I seem to have gotten szabgab++'s head spinning: "It's unclear to me what a quasi block is, but it seems it can somehow hold its breath" 11:47
pmurias masak: re .new setting private attributes wouldn't that break encapsulation? 11:51
moritz it would; which is why it doesn't, right now 11:52
moritz (it used to) 11:52
pmurias masak: for debugging purposes .perl could just have them commented out
ShimmerFairy masak: despite the fact that you don't take it seriously, I still think it's possible for a too-eager promotion of I::P5 to slow down the development of pure-P6 modules. 11:53
moritz ShimmerFairy: my experience is that people port Perl 5 modules to Perl 6 because they want to improve them in the progress 11:54
masak ShimmerFairy: I think the opposite would happen. Inline::Perl5 would make Perl 6 more attractive today, which would pull in more people, create more contributors, a bigger gravitational pull around Perl 6, and more reason to build pure Perl 6 modules 11:55
ShimmerFairy To be fair, I don't think it would be a very significant slowdown, but I still worry that being too quick to suggest it would temporarily divert the attention of people who're more than happy to write the P6 version :) 11:56
masak ShimmerFairy: I dislike almost all kinds of zero-sum either/or arguing around programming languages, and I would much more like to ride nine++'s happy have-cake/eat-cake train.
ShimmerFairy: you're arguing for limiting the options of people so that they are forced to make Perl 6 better. 11:57
lizmat fwiw, I was first apprehensive of a fully functional Inline::Perl5 as well...
lizmat but if we can PR it like any of the other Inline:: modules, like Inline::Python, it only becomes more interesting, I think 11:58
aka: not the best of Both worlds, but the best of All worlds :-)
(especially with NativeCall now also supporting C++ :-)
ShimmerFairy masak: thinking about it, I suppose at least part of my concern is just because I::P5 would never be something I'd use, so it's just weird for me to see it recommended so much.
nine I have sometimes wondered if Inline::Python (the Perl 5 version) may actually harm Perl by making it easier for people to migrate from Perl to Python. I've come to the conclusion, that if people want to migrate to Python, they are no happy and productive members of the community anyway. They have their reasons to go and if I help them, so be it. At the end of the day I'd rather help people than force them. 11:59
jdv79 it would be neat if .perl could optionally include privates maybe
ShimmerFairy On .perl, I 12:00
jdv79 otherwise there would have to be some other dumper that poked at hte mop to do it
masak I'm more concerned with .new than with .perl, to be honest
ShimmerFairy On .perl, I'll point out that .gist has been conjectured to have various adverbs on it, so adverbs on .perl wouldn't be an issue methinks :)
jdv79 and not sure keeping dumping in .perl totally or having it split it up...
lizmat jdv79: simple, write your own .perl in a role, and does that in your class
masak it feels weird to talk about encapsulation from the creator of the object
lizmat ?
jdv79 yes. i know.
lizmat: but sometimes i'm using something else and i want to see it 12:01
pmurias masak: creator of the object - you mean the guy calling .new?
ShimmerFairy masak: that's what I pointed out last time, construction doesn't feel like it "counts" as normal object usage.
masak pmurias: right
jdv79 so, idk, i'm more about whether .perl should be the uber does it all dumper or not
which includes pretty, privates, code, etc...
ShimmerFairy (after all, you have no real 'self' in .new, at least usually)
masak ShimmerFairy: arguably when constructing an object, it's still "open" and hence not encapsulated 12:02
nine Rather than Perl having a reputation of "the legacy language that's hard to migrate from", I'd like it to be "The language, that never lets you down. Even if you have to move away from it."
masak I feel we're upholding the privacy barrier during .new/.bless for academic reasons that end up being of no use and just obstacles in practice
nine: never gonna give you up
jdv79 all the hub bub around I::P5 seems silly and petty esp since its a done deal;) 12:03
nine jdv79: oh I'm not done yet!
ShimmerFairy jdv79: there's a spot in the specs that mentions .gist having things like :oneline and such. Whether those should be there, or the fact that it's still NYI is a sign .gist and .perl shouldn't be "do-everything", who can say? :)
masak nine: never gonna run around and desert you
pmurias masak: if we thing about it in terms of protecting the data it's still "open"
nine Oh, speaking of it. I have a bootleg video of my YAPC::EU talk! niner.name/talks/YAPC2015-Granada-N...-HD720.m4v 12:04
pmurias masak: but if we thing it in terms of having a private/public API, .new is still part of the public API
masak re .perl -- maybe an `is perlable` on private attributes could work?
nine For those who missed the early Friday morning slot
masak pmurias: yeah, but that argument does nothing for me when I feel annoyed that I can't init my private attrs through .new/.bless 12:05
tadzik Oh, bootleg:)
masak I can heartily recommend nine's talk. it was one of the best on the conference.
nine blushes
thank you masak
llfourn masak: ....and `is paramable` to allow the private attribute to be set by bless? 12:06
ShimmerFairy pmurias: I'm not sure if you saw, but my contention was that all attributes are private really, and has $.foo; just is nice sugar for creating a getter (and maybe setter) method for you. So the restrictions in .bless and so on feel more arbitrary (or, as masak++ put it, academic) than useful.
llfourn masak: or you think it should be default accessible?
pmurias ShimmerFairy: you repeated it multiple times and was wrong all of them
masak llfourn: the latter. 12:08
pmurias masak: the question is do you want to set private attributes through .new/.bless or do you want public attributes without an accessor? 12:09
llfourn masak: but what if you really don't want it to be able to be set?
pmurias masak: if you can access them using a public method (new) it feels like they are really public attributes 12:10
nine Why should a class' method not be able to access the object's properties? 12:11
pmurias nine: the user external to the class is accessing them through .new 12:12
nine: so they are visible to him
ShimmerFairy pmurias: I don't think .new setting attributes can really be called "access", at least not in the intuitive sense of "access" meaning "get something"
moritz ShimmerFairy: writing is just a valid an access as reading 12:13
llfourn as far as I understand the proposal is to give `bless` special access to privates not `new`..right?
moritz llfourn: but .new delegates to .bless
llfourn: so the point is moot
llfourn lizmat: 12:14
moritz: very well :)
llfourn Moose let you opt out of having your attribute (private or not) be able to be set by the constructor: metacpan.org/pod/distribution/Moos...s-init_arg 12:19
init_arg => undef did that... (of course privates were only adhoc with _attr) 12:20
it was a useful feature from time to time when you really wanted to have your default value be the starting value for your attribute
llfourn and didn't want to person .new -ing it to be able to change that 12:21
azawawi hi 12:24
for some reason I cant use webchat.freenode.net/
llfourn tries to go to webchat.freenode.net/ but it doesn't load 12:25
moritz azawawi: for some reason, this isn't even a useful bug report :-)
azawawi :)
dalek kudo/curli: 4ac23d1 | lizmat++ | src/core/Compiler.pm:
Implement $*PERL.compiler.id
moritz maybe ask #freenode?
azawawi github.com/azawawi/perl6-selenium-...r/issues/2 # Selenium WebDriver protocol client implemented in Perl 6 :) 12:26
azawawi now off to writing doc.perl6.org ui tests :) 12:29
what sort of functionality in our documentation website needs to be tested all the time? 12:30
moritz searching for stuff 12:33
llfourn azawawi: I guess the search bar is the most dynamic part that would benefit from selenium?
azawawi maybe also testing sample links for content etc 12:34
azawawi the thing is that with selenium you could test X browser types against your website 12:35
phantomjs is the headless one ofcourse 12:36
the others will launch a browser process and automate it
dj_goku azawawi: I saw in p6 weekly about Selenium::Webdriver in P6. thanks! 12:37
azawawi dj_goku: thanks. Really? I did not see it :) 12:40
llfourn realises that nine++'s talk is 2.6gb in size :S
moritz quite a lot of talking, eh? :-) 12:41
azawawi compress it :)
llfourn moritz: it must be intense high fidelity talking
azawawi coming to you in 4K and Dolby Surround 7.1 :) 12:43
nine Feel free to re-encode the video and maybe upload somewhere else! 12:44
llfourn azawawi: it's probably 3D
llfourn nine: ok! 12:45
azawawi selenium has a firefox IDE addon to record UI actions/assertion. Now with Parse::Selenese grammar (DONE) we could parse selenese commands and convert them into Perl 6 Selenium::WebDriver tests :)
masak I guess the argument that can be made for allowing .new/.bless write access to private attrs is this: the object is not formed yet, or is under construction, so normal expectations on invariants and encapsulation are not in force yet. 12:46
azawawi so you record them, save them as HTML (via Firefox addon) and then convert and run them via Perl 6. 12:47
ShimmerFairy masak: first analogy that pops up in my head is electrical wiring in a house; very much something you need to mess with during construction, but after you're done you do not want to mess with that, least of all by accident :)
pmurias masak: the counter argument would be that if you create the object with stuff that's not part of the part of the public API the expectations will never be in force 12:50
moritz masak: that feels like an excuse. The "can I rename a private attribute without anything breaking?"-test still fails if .new is allowed to write to rpivate attributes
llfourn feels that both behaviours are useful but you need to be able to opt in or out of them 12:52
masak moritz: good point. 12:54
azawawi letsencrypt.org/ # cool
masak moritz: though that could equally well be called the fault of *%_ 12:55
nine moritz: why is the .new method seen as something external that has to go through the public API? It is part of the Class, isn't it?
llfourn moritz: why is it better that new is not allowed and submethod BUILD is allowed?
pmurias nine: it's not
the .new method is part of the public API
ShimmerFairy fwiw, .bless is part of the public API, but I'm sure just about everyone doesn't see it as such :) 12:56
ShimmerFairy guesses .bless would be a 'protected' method, if we had that level of access in P6
pmurias nine: so if you create an object using Foo.new(private_attr=>234) you are exposed to the fact that private_attr exists 12:57
moritz nine: I don't care which API .new uses
nine: I care about whether the default behavior of a public constructor is to allow setting private data 12:58
nine: if we end up with something where that isn't the case, I'm fine with custom .new methods being allowed
... to do whatever they please
pmurias masak: would having a convinient way to just mark some attributes as settable by the constructor solve your problem? 12:59
moritz llfourn: .new is inherited from Mu; submethods aren't inherited
llfourn moritz: ok :)
llfourn will think over that one 13:00
moritz www.reddit.com/r/perl/comments/3ph...nt_to_the/ 13:01
"Is there / will there be a Perl6 equivalent to the Camel Book?"
nine moritz: just this morning I tried to turn Instant.from-posix into a constructor that set 2 private attributes. I had to write a very boring private method because the constructor couldn't just initialize them with bless. 13:02
moritz nine: I'm fine with .bless initialzing private attributes, as long as the public API doesn't 13:03
ShimmerFairy moritz: but .bless is part of the public API
m: class A { has $.b }; say A.bless(:b(42));
camelia rakudo-moar 56dbb0: OUTPUT«A.new(b => 42)␤»
llfourn thinks mortiz means default constructor by public API
nine moritz: if we make the default .new set private attributes using named arguments, those named arguments are part of the method's API. Not necessarily internals of your class. One can still change the private attribute's names for example while keeping method new's argument names the same.
masak pmurias: I dunno -- I feel slightly swayed by moritz++' argument right now. 13:04
moritz ShimmerFairy: maybe it shouldn't be
azawawi &
moritz or maybe it should have a switch to initialize private attributes that requires trust (that is, either called from within the same class, or one that's explicitly trusted) 13:05
ShimmerFairy moritz: the problem is, if .bless was in fact method !bless , then how could subclasses have a .bless to use? We'd need a 'protected' level of access for that, methinks. 13:06
moritz ShimmerFairy: I'm not sure; I'll think a bit about it
ShimmerFairy moritz: I do agree that I shouldn't be able to call .bless on a object, in case that wasn't clear. Just pointing out the surprising fact that you can right now :) 13:07
masak in unrelated but happy news, 007 doesn't have a public/private distinction on the object level ;)
ShimmerFairy nor does NQP :P 13:08
llfourn the way things are now make sense to me but: class A { has $!a; submethod BUILD(:$!a) { }; } # is quite ugly for such a simple task 13:12
ShimmerFairy llfourn: especially since you'd be required to handle all the public attributes you have too, unless I'm mistaken. 13:13
llfourn ShimmerFairy: let's see 13:14
ShimmerFairy: you're right o.O
m: class A { has $!a; has $.b; submethod BUILD(:$!a,|) { }; }; say A.new(a => 42,b => 42); 13:15
camelia rakudo-moar 56dbb0: OUTPUT«A.new(b => Any)␤»
dalek kudo/curli: 2daba3e | lizmat++ | src/core/Exception.pm:
X::Pragma::CannotPrecomp more generally useful
 13:25
dalek kudo/curli: b14e925 | lizmat++ | src/Perl6/World.nqp:
Disallow precompilation of "use lib"

For Perl 6.Christmas, we're going to disallow changing repo's from inside a module. Since we don't have that concept yet, but generally only precomp modules (and no scripts), disallowing precomping is a way to catch this at least at this point in the module development cycle.		 13:32
FROGGS lizmat: so one is not allowed to precompile something that contains a 'use lib'? 13:34
lizmat well, it's more general, really 13:35
FROGGS it is how I read it
lizmat modules may not have a "use lib" in it, at least not for 6.c
and modules get precomped
FROGGS ohh 13:36
lizmat even better would be if "use lib" would blow up when inside a module
at parse time, but I haven't gotten to that yet
nine lizmat, FROGGS: what do you think about jnthn++'s gist regarding module installation? Anything we can start to implement? 13:38
lizmat nine: that's what I'm doing :-) 13:38
FROGGS ohh, I did not even know there is one! 13:39
link?
lizmat FROGGS does not read P6W ?
gist.github.com/jnthn/47a42b2e86e7e552b2e2 13:40
FROGGS I did...
hh, under blog posts
ahh*
azawawi github.com/SeleniumHQ/www.selenium...g/pull/48/ # How Perl 6 is initially perceived...
lizmat :-) 13:41
dalek ast/kill-t-fixes: fe064dc | hoelzro++ | S17-procasync/kill.t:
kill.t: Only test SIGINT

SIGINT is available on all of our target operating systems, and it has consistent semantics among them. We shouldn't try every signal available to us, because some of them are unblocked (ex. SIGKILL, SIGSTOP), and others have different semantics than "normal" (ex. SIGTSTP)		 13:45
ast/kill-t-fixes: cc5583c | hoelzro++ | S17-procasync/kill.t:
kill.t: Exit after handling signal in child

The default behavior after handling a signal is to carry on; we want our child to exit early
ast/kill-t-fixes: 9a2db9a | hoelzro++ | S17-procasync/kill.t:
kill.t: Give the child some time to start up

Currently (on MoarVM), calls to kill() are immediate and not enqueued for running after the child has had a chance to start up. So we introduce an artificial delay to make sure the child has had a chance to start
ast/kill-t-fixes: 051ca2b | hoelzro++ | S17-procasync/kill.t:
kill.t: reenable actual sending of signals to child
azawawi my $username-element = $driver.find-element-by-id("username") # any ideas how to make it more Perl6ish? 13:49
dj_goku azawawi: p6weekly.wordpress.com under Ecosystem Additions
azawawi dj_goku: thx
[Coke] azawawi: that line seems fine to me. 13:50
dalek kudo/curli: f8e7eb1 | lizmat++ | src/core/Exception.pm:
Add X::Package::UseLib exception
kudo/curli: 9986c52 | lizmat++ | src/Perl6/World.nqp:
Disallow use lib inside any package declaration
ast/curli: 05a31b0 | lizmat++ | S02-types/WHICH.t:
Add X::Package::UseLib
PerlJam azawawi: $driver.find-element("username", :by<id>); # maybe more 6y along a certain dimension 13:51
azawawi i was also thinking $driver.element( :id('username') ) # more jQuery-like
PerlJam aye, that one is nice too 13:52
azawawi or .element-by(...) that is
nine .element-by-id would be my choice, since it not only finds the element, it also returns it. And by-id is a bit of a special case, because there may only be one element with a given id. 13:53
Other find-* methods have to deal with multiple results.
PerlJam 1 is just a special case of many :) 13:54
nine Good rule of thumb: a method should do one and only one thing
azawawi nine: also auto-completion friendly
FROGGS lizmat: you need to change some packages in t/spec/packages... some of them contain a 'use lib'
nine azawawi: jQuery is the last place where I would look for API design inspiration :)
azawawi nine: true but it is easy to remember lol 13:55
lizmat FROGGS: indeed... but fortunately (?) the use is not inside the package, and the module is not precomped 13:56
so, atm, they're just considered script with a package in it, right ?
dj_goku azawawi: is the idea to add tests here: github.com/perl6/doc ?
lizmat I had some fallout in the spectest related to precomp, but I thought that was related to some other work I'd done earlier 13:57
based on my CURLI gist
azawawi nine: github.com/azawawi/perl6-selenium-...r/issues/4 :)
azawawi dj_goku: im tracking it here github.com/azawawi/perl6-selenium-...r/issues/1 13:58
nine azawawi: shouldn't it be the other way round?
lizmat nine FROGGS I intend to work on jnthn's gist in the coming days, see how far I can get before he returns 13:59
azawawi nine: fixed :)
lizmat first thing I'm going to focus on is CompUnit::Loader 13:59
BTW, all work is done in the "curli" branch 14:00
PerlJam checks to see if we also have mo and larri branches 14:01
lizmat :-) 14:02
azawawi home &
nine lizmat: maybe I can contribute something, too :) 14:06
lizmat that would be excellent! 14:07
do you have questions about jnthn's gist ? 14:08
nine lizmat: none that I remember. I read it at the weekend and I'm gonna read it again this evening 14:09
lizmat ok
this evening I'll be at the Niederrhein PM meeting, so probably won't be able to do much 14:10
I'll try to be on channel though, to answer any questions
itz_stmuk is a heisenbug in rakudo likely to be a bug at a lower level like nqp? 14:11
nine itz_stmuk: it's at least not unlikely 14:12
dalek kudo/curli: 062e778 | lizmat++ | / (4 files):
Add stubs for CompUnit::Handle/Loader

copy-pasted from jnthn's gist + added to setting creation spec		 14:15
itz_stmuk didn't [ptc] run coverity? 14:33
lizmat commute to NR.pm
FROGGS lizmat: I'll also read the gist this evening... ohh well :o) 14:33
itz_stmuk: I think so, yes 14:34
jlb333333 join 14:35
masak jlb333333: join in! welcome! :D 14:36
Ven oi, oy 14:39
ugexe mrf: you can use any of the individual rfcs in grammar::http, you just create your grammar and `does Grammar::HTTP::RFCxxxx` to it
the reason the rfcs aren't individual distros is because 7230-7235 all need each other, and all the other grammars are of trivial length that they aren't worth adding external dependency (4647 is 2 lines for instance)
ely-se Ven: vei, vey
Ven ely-se: well met 14:39
Ven ely-se: i thought you didn't like perl6, yet you're still around :P 14:39
.oO( by the power of the nice community, I compel you! )
 14:40
ugexe i dont like brushing my teeth but i still do it 14:40
Ven ugexe: do you go to #toothbruth on IRC? :P
ugexe just #toothbruh 14:40
ely-se #toothbrush is my favourite channel
n0tjack I like #payingtaxes 14:41
mrf ugexe: I was thinking more of placing all IETF RFCs into a single repo as they are used in more than HTTP.
|Tux| www.perlmonks.org/?node_id=1145436 <= whose fault? Can someone else help?
mrf ugexe: I spent time writing them myself becuase I didn't think of looking in Grammer::HTTP for them.
I would likely have looked in Grammer::RFC or the like 14:42
n0tjack |Tux|: I believe that means your panda (and probably Rakudo) are out of date 14:42
huh, panda doesn't have a --version flag. that's a bit silly. 14:43
|Tux| it is not me: my panda workes like a charm
|Tux| feeds it enough bamboo to keep it satisfied 14:44
n0tjack Well, the guy who has the problem then.
You might ask him to do perl6 --version, and/or panda update
ugexe mrf: right, believe me i considered it enough that the repo was originall called Perl6-IETC--RFC_Grammar--HTTP
mrf ugexe: though I accept that this would add additional dependencies that might be undesirable.
ugexe: why the link to HTTP
n0tjack once saw a panda poop on another panda in the san diego zoo. to all appearances, panda #2 was unfazed. 14:45
ugexe well when i search for something to parse http, the first thing i look for isnt some cryptic rfc
mrf ugexe: I think teh idea I have is more to seperate the Grammars that implement the RFC from the Grammars that use them to implement something specific
ugexe you can't piece together grammars like that very well yet 14:46
thats why all the rfcs are roles in grammar::http
mrf ugexe: indeed. So I would expect a HTTP::Message module that uses a HTTP::Message grammar which does a whole bunch of Grammars.
ugexe and there is a single centralized grammar that doesnt do anything but `does RFCxxx RFCyyy`
i think a better module search would take `provides` into consideration, so searching for rfc7230 would turn up Grammar::HTTP's lib/Grammar/HTTP/RFC7230.pm6 14:47
dalek ecs: 93d6e5f | (Nova Patch)++ | S05-regex.pod:
tcuc() was removed two years ago

reference: d1f9930373d9aa92dd24df24212cb3cbd74f4de6
n0tjack m: o.WHAT; 14:49
camelia rakudo-moar 56dbb0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/VIl41OkcxJ␤Undeclared routine:␤ o used at line 1␤␤»
n0tjack m: sqrt o +;
camelia rakudo-moar 56dbb0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/44mBXFRF22␤Prefix + requires an argument, but no valid term found␤at /tmp/44mBXFRF22:1␤------> 3sqrt o +7⏏5;␤ expecting any of:␤ prefix␤»
mrf ugexe: so for context I came at this wanting an Email::Address style parser that could validate etc email address. I didn't even think of looking in *HTTP* for the Grammars.
modules like Email::Address and HTTP::Message would likely share a whole bunch of grammars yes? 14:50
ugexe: Am I making any sense. Or am I just talking bollocks? 14:52
ilmari m: &sqrt o &infix:<+> 14:53
camelia ( no output )
ilmari m: say &sqrt o &infix:<+>
camelia rakudo-moar 56dbb0: OUTPUT«-> |args is raw { #`(Block|51403952) ... }␤»
ilmari m: say &infix:<o>.WHAT 14:54
camelia rakudo-moar 56dbb0: OUTPUT«(Sub)␤»
n0tjack what are the semantics for %hash1 ~~ %hash2 ? Identical keys and match-y values?
ilmari n0tjack: operators have funny names
n0tjack ilmari: thanks
I do overlook the use/mention distinction for subs in P6 a lot
that's one place p6 has traded off DWIMminess for consistency 14:55
or, rather, workability :)
ilmari s/subs/operators/
ugexe mrf: it makes sense, but pulling it off leads to its own problems. that is how i originally tried to design it but with things like 7230-7235 you would have circular dependencies
n0tjack well, for subs too
grondilu m: say { foo => "bar" } ~~ { foo => "bar" }
camelia rakudo-moar 56dbb0: OUTPUT«False␤»
n0tjack yikes 14:56
me no likey
ilmari m: say { foo => "bar" }.WHAT
camelia rakudo-moar 56dbb0: OUTPUT«(Hash)␤»
grondilu looks LTA indeed 14:56
n0tjack LTA?
ah 14:57
grondilu "Less Then Awesome". See S99
*Than
n0tjack yeah, just did
thanks
mrf ugexe: could you not split *all* the roles into one repo and leave Grammer::HTTP.pm6 in the current repo (I realise this leaves one tiny repo but it makes semantic sense) 14:58
ugexe not really, because it makes testing almost impossible 14:58
n0tjack m: say 1000.WHAT; say 1e3.WHAT; say (10**3).WHAT;
camelia rakudo-moar 56dbb0: OUTPUT«(Int)␤(Num)␤(Int)␤»
mrf ugexe: Why?
grondilu m: say { foo => "bar" } === { foo => "bar" } 14:59
camelia rakudo-moar 56dbb0: OUTPUT«False␤»
mrf SUrely just a dummy class that tests a specific grammar.
ugexe because then you have dependencies, and you will have circular dependencies
ilmari there is no Map.ACCEPTS(Map $)
docs.perl6.org/routine/ACCEPTS#class_Map
ilmari m: say { foo => "bar" } == { foo => "bar" } 15:00
camelia rakudo-moar 56dbb0: OUTPUT«True␤»
n0tjack m: say { foo => 12, bar = 13 } == {bar => , foo => 12};
camelia rakudo-moar 56dbb0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/nE0ceogkVP␤Preceding context expects a term, but found infix = instead␤at /tmp/nE0ceogkVP:1␤------> 3say { foo => 12, bar =7⏏5 13 } == {bar => , foo => 12};␤» 15:01
n0tjack m: say { foo => 12, bar = 13 } == {bar => 13, foo => 12};
camelia rakudo-moar 56dbb0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/oZ3ZxSLNkq␤Preceding context expects a term, but found infix = instead␤at /tmp/oZ3ZxSLNkq:1␤------> 3say { foo => 12, bar =7⏏5 13 } == {bar => 13, foo => 12};␤»
n0tjack m: say { foo => 12, bar => 13 } == {bar => 13, foo => 12};
camelia rakudo-moar 56dbb0: OUTPUT«True␤»
ugexe mrf: once a grammar can `does OtherGrammar` it might be possible 15:02
mrf ugexe: I thought that was possible. That kind of sucks. 15:03
ShimmerFairy m: role G { token foo { foo } }; grammar H does G { }; say H.parse("foo", :rule<foo>) # you can certainly compose a Grammar-ish role, however 15:04
camelia rakudo-moar 56dbb0: OUTPUT«｢foo｣␤»
ugexe yes, thats how http::grammar is setup
ShimmerFairy you can also 'is' a grammar, I don't know if that's acceptable in this case though. 15:05
ugexe not usually, because rfcs have tokens that will conflict 15:06
mrf ugexe: because all rfcs use localised rules? 15:07
ShimmerFairy any reason role G is Grammar isn't acceptable as a composeable Grammar? Unless I've forgotten something grammars are just classes with a default parent of Grammar instead of the more usual one. 15:07
ugexe because the token `time` might mean a different format of timestamp depending on the rfc
mrf ugexe: but shouldn't the tokens be localised to the RFC anyway. 15:08
ugexe no, because the rfcs say stuff like `<.FWS>` that other rfcs use 15:09
mrf ugexe: There are only a few special common rules that I have seen. Can you give an example of any outside those declared in RFC4234.5234 15:10
ugexe 5322 has like 30 rules 15:11
rarara_ Hello, with hashes it is possible to do things like %hash<first second>.join(" ") .... is there a similar construct for calsses? something like $class{.first .second}.join(" ") ? 15:12
ugexe about half are localized, the others no (because they are used by other rfcs)
n0tjack rarara_: you could try something like [.first, .second].join(" ") given Foo.new("one", "two"); 15:13
might have to fiddle the syntax a bit
mrf ugexe: hmm ok. That kinda sucks 15:15
rarara_ n0tjack : Thank you
rarara_ works also with normal parenthesis (vs squared ones) 15:16
But I don't know which one would be more performant
n0tjack rarara_: Round parens () make an Array, Square bracket [] make a List 15:17
rarara_ n0tjack it is the opposite here 15:18
n0tjack_ I'd say at this point in p6's development, learning the language will be a better use of your time than worrying about performance 15:19
there's a lot of work to do yet in the compiler that will have a bigger impact on performance than stuff you do in your own code
n0tjack_ rarara_: I might have gotten that mixed up. But one is a list, the other an array. 15:19
ugexe i dunno, i can write some pretty bad code
rarara_ I agree, but I was thinking that list might make more sense since we use the data immediately in join 15:20
while Array seems to me more suited for storing modifiable things
n0tjack_ that's the general idea, yes
but it's a small piece of code and I don't imagine the choice will have much impact on the overall application 15:21
alright, have to go visit a customer
rarara_ n0tjack good luck!
ugexe mrf: if you find a way to split them apart where they are still testable i would be open to it. its just when i tried this like a year ago i had problems (and maybe they are resolved these days) 15:22
i have no problem dropping 20-some modules and making dalek earn his bones lol 15:23
mrf ugexe: I will see what I can do.
ugexe: however I suspect that if you couldn't manage it I won't either, but you never know 15:24
ugexe well, i couldnt manage it *6+ months ago*. lots has changed 15:25
mrf ugexe: that sounds like something a go developer would say
ugexe it took me this long to split the grammar out from zef
golang changes have nothing on perl6 changes heh 15:26
ugexe personally i'd prefer to drop the pointless namespace and just call them `RFCxxx` instead of IETF_Grammar::RFC[\d+ | "::RFC" \d+] 15:28
that way you can mix in a lot of rfcs without huge swaths of `IETF_Grammar::` 15:30
mrf ugexe: how would you then distinguish between IETF RFCs from not IETF ones. 15:31
PerlJam mrf: call the non-IETF ones something else :) 15:32
mrf PerlJam: true. 15:33
ugexe as long as they dont share the same names i dont think it will confuse anyone
PerlJam What non-IETF RFCs are there in potential module space anyway? 15:34
mrf ugexe: so as a (top of the head) example Grammar::Perl6::RFC111
PerlJam: That was the question I asked and the Perl6 rfcs were the example ShimmerFairy gave 15:35
PerlJam ah, those should clearly live in a Perl6 space of some sort. 15:35
mrf PerlJam: I suspect that its not a large use case but I like to be specifc.
masak 'night, #perl6 15:36
PerlJam btw, why do people put "Grammar" somewhere in the namespace of a grammar? 15:37
masak: good night. sleep well!
I mean, is "Grammar" meant to give a hint as to how the name should be used?
mrf PerlJam: example?
PerlJam you just gave one :) Grammar::PErl6::RFC111 15:38
Foo::Grammar.parse($string)
mrf Grammar gives a namespace for modules that are just Grammars.
PerlJam Why mention that it's a grammar?
mrf: sure, but ... why?
mrf PerlJam: So I can find them. 15:39
ugexe well, grammars will be slower than some string index type stuff
mrf If I want to find a grammar for parsing XML. I would want to look for something like Grammar::XML or some such
ugexe it also suggests you can pass in an action
PerlJam ugexe: could you expand on that please? 15:40
you mean that other classes may have a .parse() method, but they may not take :action ? But if you know it's a Grammar, you know that .parse will take an :aciton ? 15:41
ugexe well to me something with Grammar:: suggests you are meant to work with the match object or can pass in $actions to the parse call. I would expect something like HTTP::Parser to just give me a perl6 object, not a match from a grammar
PerlJam er, I guess it's :actions (I always get that mixed up) 15:42
ugexe take for instance this (not in the ecosystem): Perl6-Grammar--XQuery It is useless without any :actions, but it can (used to?) kinda parse it. calling it XQuery would not be correct here 15:45
mrf PerlJam: A good example for me is the difference between JSON::Tiny and JSON::Tiny::Grammar. As an end user I would not expect to go anywhere near the ::Grammar module and would only use JSON::Tiny as the interface. In fact the ::Grammar module could be then extracted out for others who wanted to implement JSON parsers 15:47
moritz: as a point is there any worth putting JSON::Tiny::Grammar into the Grammar:: namespace. Assuming its a complete JSON grammar (which it appears to be) other JSON::* implementations could make use of it (With say different action objects) 15:50
ugexe URI bothers me like that. URI.new() doesnt seem like it should be returning a big ol' match object 15:53
which just led me to this bug 15:54
star: use URI; my $x = URI.new("www.google.com"); say $x.WHAT
camelia star-m 2015.09: OUTPUT«Cannot look up attributes in a type object␤ in method scheme at /home/camelia/star-2015.09/share/perl6/lib/URI.pm:137␤ in method gist at /home/camelia/star-2015.09/share/perl6/lib/URI.pm:204␤ in block <unit> at /tmp/Cz8uGTbsuh:1␤␤»
jdv79 yeah, i did a rewrite of URI but never polished and pushed it
it could be a lot better
moritz mrf: fwiw I want JSON::Tiny to stay without dependencies 15:55
mrf: but afaict you can already use the grammar from elsewhere
mrf moritz: yeah sure. Just my pigeonholing brain wanting a separation from the Grammars (the definition of something) from the thing that uses that Grammar (the parser) 15:56
moritz mrf: but isn't it separated already? 15:57
mrf I am aware that I could 'panda JSON::Tiny' and then 'use JSON::Tiny::Grammar' but that just seems weird in my brain. Its not a Grammar for JSON Tiny. Its a JSON Grammar.
I would have liked to install Grammar::JSON (or likewise) without installing JSON::Tiny.
RabidGravy well it's the JSON grammar that JSON::Tiny uses
other grammars are possible I guess 15:58
mrf true enough. I think I am just being pedantic 15:59
PerlJam I wouldn't say "pedantic" 16:00
I think you're trying to fit your brain to the world and vice versa.
moritz mrf: you're welcome to write a Grammar::JSON that does a "use JSON::Tiny::Grammar; constant Grammar::JSON = JSON::Tiny::Grammar" :-)
RabidGravy I could see there being some virtue in there being a standalone JSON grammar, nothing precludes taking the best one around and forking it to a separate distribution
PerlJam much the same reason I was asking why people put Grammar in the namespace
mrf See that the URI module has its own Grammar for IETF when it could use Grammar::HTTP to get that
moritz: :D 16:01
moritz: I had one problem then I linked one Grammar to another now I have a recursive problem
RabidGravy PerlJam, I could see a top-level Grammar:: for pure grammars (with no actions or other code) for people to use for their own purposes, but yup 16:03
moritz I think you're idealizing the usefulness of a grammar
mrf RabidGravy++ # Yes this
moritz actions are rather tightly coupled to the structure of a grammar
PerlJam indeed
moritz so code reuse is far from trivial
RabidGravy oh for sure 16:04
PerlJam even with no actions, if you're going to use the grammar for anything, you need to be highly familiar with the rules used within the grammar and their relationships 16:05
mrf PerlJam: true but that can be mitigated by documenting the rules/tokens ect 16:08
RabidGravy mrf, I think at least part of the thing here is, if you want things in the ecosystem at this stage in the game, just do it 16:10
:)
mrf RabidGravy: true. 16:14
moritz mrf: the problem is that changes / bug fixes quickly become backwards incompataible, because the distinction between implementation and API basically doesn't exist 16:15
kmel m: say 'hello' 16:16
camelia rakudo-moar 56dbb0: OUTPUT«hello␤»
mrf moritz: Because action object methods are tied to rules in the grammar? 16:20
RabidGravy yeah 16:22
RabidGravy to be honest the only example I have found where the grammar could be used in a different way than the author intended is in Router::Boost where the basic grammar could be alternatively used in a URI template thingy 16:23
PerlJam Because grammars need to be designed with sharing in mind for that to happen (usually) 16:25
RabidGravy yeah, and I think in that case an eventual design would bend it out of shape from the original purpose 16:26
PerlJam and then readability/maintainability/understandability probably suffer 16:27
mrf I can certainly see pitfalls in sharing of grammars but really want to avoid a situation were every single Email/HTTP/URI module defines the same RFC grammar. Which seems like wasted effort.
RabidGravy on the other hand, and I am rambling here, there may well be things for which a standalone grammar can answer "is this this a valid $foo" when you don't want to do a full parse
mrf, only if it's your effort or effort that would otherwise be doing something else and I think the reasons people make the modules the latter is unlikely 16:30
ugexe grammar::http can be passed actions that turn the parse into a deep data structure, or handle things like closing a socket when it hits certain parts of a header. with rfc grammars the definitions and shareability are mostly solved 16:31
mrf maybe that is where my blind spot is coming from . The kind of things I am focusing on right now have very stong specifications (RFC's) and define syntaxt used by a huge swath of things I use daily which makes their Utility greater. 16:35
I think in the end I will just need to implement something and see where it breaks.
Thanks people you have all been awesomely helpful
-> Home time 16:36
RabidGravy :)
lizmat waves from Cologne 16:58
moritz }o Cologne 16:59
[ptc] o/ 17:02
jdv79 one random thing on the grammar discussion - it might be hard when people are optimizing grammars 17:03
what i mean is, for instance, shortcutting or not capturing in the name of perf
[ptc]: hey 17:04
[ptc] jdv79: hi!
[Coke] is tied to a work project where "/" is a special character. wtf. 17:28
dalek c: 9320442 | (Nova Patch)++ | doc/Type/Str.pod:
remove Str.tcuc

which was never implemented in rakudo and was removed from the spec in perl6/spec@d1f9930
[Coke] is nova here in channel?
Nova++ 17:29
FROGGS lizmat: would be nice to perhaps do TDD now (I'm just reading jnthn++'s gist) 17:44
lizmat FROGGS: that's why I also created a curli branch in roast
nine FROGGS: where would the tests live?
FROGGS nine: in roast :o) 17:45
nine That's mildly helpful ;)
S11-modules? 17:46
lizmat if they are deemed not to be part of the Perl 6 definition, then we can always move them somewhere else
FROGGS S11-precomp?
nine Precomp is only a part. How about S11-compunit? 17:46
FROGGS I think there are enough precomp tests to create a new dir for it 17:47
lizmat S11-compunit is fine by me
FROGGS okay
one test would be to ask for precompilation on A.pm, but this will trigger precomp of C.pm and and B.pm 17:48
lizmat indeed...
and the idea is that that would happen transparently
lizmat as in: while precompiling A.pm, the "use B" is encountered 17:49
nine I'm adding CompUnit::DependencySpecification
FROGGS another test would be have a lax use statement, do precomp, then add another dist that would have been preferred to a cur, then run the precomped code...
lizmat at that point, precomp of A should be halted, and B should be precomped first
FROGGS lizmat: aye
lizmat then, when B is precomped, precomp of A should continue
FROGGS same should probably happen if the cache of B is invalidated 17:50
FROGGS but, that is another test :o) 17:50
lizmat nine: CU::DependencySpecification probably also needs an api-matcher :-) 17:51
rindolf m: 0b1000 ~& 0b1010
camelia rakudo-moar 56dbb0: OUTPUT«WARNINGS:␤Useless use of "~&" in expression "0b1000 ~& 0b1010" in sink context (line 1)␤»
rindolf m: say (0b1000 ~& 0b1010)
camelia rakudo-moar 56dbb0: OUTPUT«0␤»
rindolf This seems wrong. 17:52
rindolf <rindolf> perlbot: eval: [0b1010 & 0b1000] 17:53
<perlbot> rindolf: [8]
FROGGS rindolf: depends how that op is meant to work :o)
FROGGS m: say (0b1000 +& 0b1010) 17:53
camelia rakudo-moar 56dbb0: OUTPUT«8␤»
rindolf FROGGS: isn't it bitwise or? 17:53
Ah, so it's +& ?
FROGGS rindolf: I think there is some sort of consensus that ~& should work on blobs only
rindolf btw, this search returns junk - duckduckgo.com/?q=perl%206%20bitwi...operations 17:54
FROGGS rindolf: numeric is +&, stringy is ~&
rindolf FROGGS: ah. 17:55
nine Is the has $version-matcher on purpose or should it be has $.version-matcher? 17:56
[Coke] rindolf: try searching on doc.perl6.org ... though you may have no better luck, we actually control that. 17:58
rindolf [Coke]: ok.
[Coke]: Google gave me this - doc.perl6.org/language/operators
[Coke] that works. 17:59
FROGGS ahh, I like the linked list design in jnthn's gist
lizmat nine: not sure
FROGGS use lib "Foo"; will then build down to: PROCESS::<$REPO> = NewCur.new( "Foo", :next(PROCESS::<$REPO>) ) # more or less 18:00
nine FROGGS: yes 18:01
lizmat wouldn't that be: my $*REPO = NewCur.new("Foo", :next($*REPO) ) ? 18:04
nine FROGGS ^^^
FROGGS lizmat: that doesnt work 18:05
nine There's also $?REPO
FROGGS if you declare something, you shadow the old value
lizmat hmmm....
dalek ast/curli: f3eb270 | (Stefan Seifert)++ | S11-compunit/compunit-dependencyspecification.t:
Test CompUnit::DependencySpecification
 18:07
kudo/curli: 771c4cf | (Stefan Seifert)++ | / (7 files):
Add CompUnit::DependencySpecification
rindolf How do I put a reference to a subroutine as a value in a hash? 18:10
moritz %h{$key} = &yoursub; 18:11
gfldex m: sub f(){ say 'i haz a f' }; my %h; %h = f => &f; %h<f>.();
lizmat rindolf: why a reference>
camelia rakudo-moar 56dbb0: OUTPUT«i haz a f␤»
lizmat m: %h<a> = { say "foo" }; %h<a>() # no need to make a sub 18:12
camelia rakudo-moar 56dbb0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/Qbi0B_asYb␤Variable '%h' is not declared␤at /tmp/Qbi0B_asYb:1␤------> 3<BOL>7⏏5%h<a> = { say "foo" }; %h<a>() # no nee␤»
lizmat m: my %h; %h<a> = { say "foo" }; %h<a>() # do need to make a hash :-)
camelia rakudo-moar 56dbb0: OUTPUT«foo␤»
rindolf lizmat: well, it's kinda long.
kmel hello everyone 18:15
nine hello kmel 18:16
kmel if i want to play a bit with Perl 6 oriented towards web development. What options do i have? 18:17
DrForr Bailador comes to mind. 18:18
kmel i found bailador but thought i'd ask before delving in
kmel thanks DrForr 18:19
nine I hear good things about Mojomojo
DrForr I haven't checked into what's in modules.perl6.org lately :/
nine no, not Mojomojo....curse you brain 18:20
I meant Mojolicious
kmel nine isn't mojolicious Perl 5?
FROGGS it is 18:21
it will be usable with Inline::Perl5 of course but its interface won't be Perl 6-ish I fear
kmel so from a Perl 6 point of view there's only bailador? 18:23
rindolf paste.debian.net/317172/ - this code gives me an error - «Missing infix inside []» - how can I fix it and what does it mean?
FROGGS kmel: you can also search for 'server' here: modules.perl6.org/ 18:25
moritz rindolf: I can't parse your code either
((@new_rec[$old_digits_sum + $new_digit]
[($old_mod + $BASE_MOD * $new_digit) % $BASE]
AlexDaniel I feel like there was an attempt to port Mojolicious 18:26
github.com/coke/mojo6
moritz rindolf: what should the second pair of brackets introduce?
AlexDaniel this one
moritz rindolf: is it supposed to be a subscript?
rindolf moritz: an array dereference.
moritz: yes.
moritz: a subscript.
moritz rindolf: can't have whitespace between two subscripts
FROGGS unspace missing? 18:27
rindolf Like $x->[6]->[100] in Perl 5 or arr[5][100] in Python.
moritz: ah.
FROGGS m: my @a; @a[0]␤[0] = 42
camelia rakudo-moar 56dbb0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/d7IKaf95Tk␤Missing infix inside []␤at /tmp/d7IKaf95Tk:2␤------> 3[7⏏050] = 42␤ expecting any of:␤ bracketed infix␤ infix␤ infix stopper␤»
FROGGS m: my @a; @a[0]\␤[0] = 42
camelia ( no output )
moritz rindolf: return (0 .. $DIGITS_SUM).map, { [(0) x $BASE] };
FROGGS there you go
moritz rindolf: that looks like a 5ism
AlexDaniel
.oO( What about using CGI.pm with Inline::Perl5? Ha-ha-ha )
moritz rindolf: in Perl 6, xx is the list repetition operator
FROGGS rindolf: add a backslash before the newline 18:28
moritz and that map call is also weird
maybe you want [0 xx $BASE] xx ($DIGITS_SUM + 1) or so?
itz_stmuk the cool kids don't use CGI.pm anymore :) 18:29
kmel AlexDaniel I am trying to go for a pure 6 module
AlexDaniel kmel: I think that even if you find one, it seems like you will have to become an active contributor to get your stuff going. Which is good, in some sense… 18:31
AlexDaniel kmel: However, Bailador seems to be pretty active 18:31
not sure how usable it is, but try it? 18:32
DrForr There's Perl::ToPerl6 in the Perl5 space that might help :)
kmel AlexDaniel I am not looking for something sophisticated, just something i can test the waters with 18:33
i think i'll try bailador
ugexe grammar::http can be passed actions that turn the parse into a deep data structure, or handle things like closing a socket when it hits certain parts of a header. with rfc grammars the definitions and shareability are mostly solved
oops
[Coke] mojo jojo! ... I need to get back to that 18:44
RabidGravy ugexe, so good you had to send it twice ;-)
rindolf «Method 'eval' not found for invocant of class 'Str'» - how do I do string eval in rakudo-moar? 18:46
pink_mist EVAL "string here" 18:47
rindolf pink_mist: in all caps? 18:48
pink_mist as far as I understand it, yes
rindolf pink_mist: thanks! 18:49
pink_mist: what's up, BTW?
lizmat yes, all caps, because you probably SHOULD NOT DO THAT -)
pink_mist rindolf: np ... and I'm busy busy busy =) in fact gotta go right now :P
AlexDaniel lizmat: are you saying that I should not any phasers at all? 18:50
should not use*
lizmat AlexDaniel: hehe :-)
rindolf pink_mist: bye.
lizmat okokok...
basically, all caps means it draws attention for something special happening
AlexDaniel ok :) 18:51
rindolf OK, my Euler program seems much slower in p6 than it is in p5.
AlexDaniel but I always felt that all caps were used just to distinguish it from other stuff 18:51
lizmat AlexDaniel: yeah, you're right 18:52
originally, eval was just eval and not EVAL :-)
n0tjack back then, it was primeval 18:54
lizmat
.oO( the best of times )
 18:55
AlexDaniel votes for universal prefix for potentially unsafe operations – ☠. That is, ☠EVAL, ☠qqx//, etc…
lizmat use Slang::Unsafe :-) 18:56
AlexDaniel well, I doubt anybody will add a dependency just to protect him from himself. 18:58
lizmat could be a Company Best Practice :-) 18:59
[Coke] rindolf: feel free to post source - we might be able to suggest improvements.
(gist it)
rindolf [Coke]: oh, I see - I used & instead of +& 19:03
dalek osystem: 9de12e6 | (Shoichi Kaji)++ | META.list:
Add Data::Section::Simple to ecosystem

See github.com/shoichikaji/Data-Section-Simple		 19:08
osystem: decc31c | (Zoffix Znet)++ | META.list:
Merge pull request #82 from shoichikaji/Data-Section-Simple

Add Data::Section::Simple to ecosystem: github.com/shoichikaji/Data-Section-Simple
ast/curli: 4e223c5 | (Stefan Seifert)++ | S11-compunit/compunit-repository.t:
First tests for CompUnit::Repository
 19:09
kudo/curli: 53da925 | (Stefan Seifert)++ | / (4 files):
Add CompUnit::Repository role
vendethiel amazingly enough, nine++ took a liking to core hacking :P
rindolf [Coke]: here is the source - bitbucket.org/shlomif/project-eule...ew-default 19:16
n0tjack I need something to code up in p6 19:21
ZoffixW Hey. Is it valid to have "authors" field in META.info instead of singular "author"? I'm just looking at this: raw.githubusercontent.com/shoichik...META6.json
n0tjack hmm, maybe project euler
ZoffixW n0tjack, go nuts: github.com/perl6/perl6-most-wanted...modules.md
:D
n0tjack .... I'm not sure you want me to write code other ppl are expected to use :)
Ulti n0tjack: www.reddit.com/r/perl6/comments/3p..._show_off/ 19:22
ZoffixW n0tjack, I'd not worry too much about that. Just write good tests and you should be fine :) 19:23
Ulti yeah its more motivational if people do use your code... which reminds me I was meant to finish the histogram stuff in Stats
Ulti maybe if I wait long enough there will be a PR waiting instead.... 19:24
RabidGravy n0tjack, plus if people really want it and they want it better they can pitch in and fix it ;-)
flussence m: my @a; @a[0]␤.[0] = 42; say @a 19:27
camelia rakudo-moar 56dbb0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/X51OdAaiwf␤Unsupported use of . to concatenate strings; in Perl 6 please use ~␤at /tmp/X51OdAaiwf:2␤------> 3.[7⏏050] = 42; say @a␤»
flussence std: my @a; @a[0]␤.[0] = 42; say @a
camelia std 28329a7: OUTPUT«5===SORRY!5===␤Method call found where infix expected (change whitespace?) at /tmp/K8ucLWJlGv line 2:␤------> 3<BOL>7⏏5.[0] = 42; say @a␤ expecting any of:␤ POST␤ infix or meta-infix␤ infixed function␤ postcircumfix␤ postfix␤ postfi…»
FROGGS m: my @a; @a[0] . [0] = 42; say @a 19:32
camelia rakudo-moar 56dbb0: OUTPUT«5===SORRY!5=== Error while compiling /tmp/lC7p29Bb01␤Unsupported use of . to concatenate strings; in Perl 6 please use ~␤at /tmp/lC7p29Bb01:1␤------> 3my @a; @a[0] . [7⏏050] = 42; say @a␤»
FROGGS hmmm, sad
but only a little
n0tjack is roast the general testing framework for p6? 19:36
as in, how do I write my own tests? 19:37
moritz it's not a framework
nine n0tjack: use Test;
moritz it's a test suite
n0tjack: doc.perl6.org/language/testing.html
n0tjack thanks nine, moritz
FROGGS m: say $*KERNEL ~~ 'linux' 19:38
camelia rakudo-moar 56dbb0: OUTPUT«True␤»
FROGGS wow, that documentation is very good 19:39
lizmat FROGGS: re [] . [] I tried exactly the same thing :-) 19:41
vendethiel "is($womble->location" probably should be "." 19:42
the link at the bottom is down
FROGGS :o)
dalek c: 229c4f9 | ven++ | doc/Language/testing.pod:
Fix P5ism '->' in language/testing
 19:43
n0tjack sorry for the silly question, is use some-other-file-in-this-dir.pl6 the way to import a script into a test? 19:49
lizmat there's also EVALFILE ? 19:51
n0tjack :4 19:53
ugh, too many terminals
dalek kudo/nom: 5d83db3 | FROGGS++ | src/core/Regex.pm:
handle junction on LHS of regex correctly
 19:55
ast: 51d1a48 | FROGGS++ | S03-junctions/misc.t:
unfudge/add junction ~~ regex tests
 19:56
FROGGS gnight 20:02
[Coke] .ask rindolf - the description says S(1112), but you're looking at 11**12? 20:12
yoleaux [Coke]: I'll pass your message to rindolf.
[Coke] that code looks some gnarly perl4 code. 20:13
RabidGravy gnarly dude 20:20
[Coke] Does anyone else follow his code? I am having a hard time seeing how it's doing what's asked in the description. 20:21
[Coke] m: say int.max; 20:24
camelia rakudo-moar 5d83db: OUTPUT«-Inf␤»
TimToady m: say int.range 20:26
camelia rakudo-moar 5d83db: OUTPUT«Method 'range' not found for invocant of class 'int'␤ in block <unit> at /tmp/briVW7t5yx:1␤␤»
TimToady m: say int.Range
camelia rakudo-moar 5d83db: OUTPUT«-9223372036854775808..9223372036854775807␤»
[Coke] m: say [+] 12345.split 20:28
camelia rakudo-moar 5d83db: OUTPUT«Cannot call split(Int: ); none of these signatures match:␤ (Cool $: Regex $pat, $limit = { ... };; :$all, *%_)␤ (Cool $: Cool $pat, $limit = { ... };; :$all, *%_)␤ in block <unit> at /tmp/Estzn2AtWw:1␤␤»
[Coke] m: say [+] 12345.split('')
camelia rakudo-moar 5d83db: OUTPUT«15␤»
PerlJam m: say [+] 12345.comb; 20:29
camelia rakudo-moar 5d83db: OUTPUT«15␤»
n0tjack now do it in base 8 :)
moritz base 8 is just like base 10, really, if you're missing two fingers
(a Tom Lehrer quote, iirc) 20:30
PerlJam moritz: nah, you just count the spaces between the fingers instead.
n0tjack all bases are base 10
that came as a huge relevation to me when i was a kid
m: say ox30071; say :8[3,0,0,7,1]; 20:31
camelia rakudo-moar 5d83db: OUTPUT«5===SORRY!5=== Error while compiling /tmp/lu_QdwQ5ja␤Undeclared routine:␤ ox30071 used at line 1␤␤»
n0tjack I forget the notation for octal
TimToady 0o
n0tjack m: say 0o30071; say :8[3,0,0,7,1]; 20:32
camelia rakudo-moar 5d83db: OUTPUT«12345␤12345␤»
n0tjack I wrote an "antibase" function that will take (12345,8) and produce (3,0,0,7,1), but is there a native way in p6? 20:33
PerlJam is having deja vu right now
gfldex m: say 12345.base(8)
camelia rakudo-moar 5d83db: OUTPUT«30071␤» 20:33
n0tjack gfldex++
gfldex doc.perl6.org/routine/base#role_Real ++
n0tjack m: say 12345.base(8).comb; 20:34
camelia rakudo-moar 5d83db: OUTPUT«(3 0 0 7 1)␤»
n0tjack heh!
[Coke] .tell rindolf - oh, because your copy of the problem statement is missing formatting.
yoleaux [Coke]: I'll pass your message to rindolf.
n0tjack m: say :8[3,0,0,7,1].base(8).comb;
camelia rakudo-moar 5d83db: OUTPUT«(3 0 0 7 1)␤»
n0tjack I'm never 100% sure when I need to parenthesize the LHA to . but it doesn't seem mandatory very often 20:35
TimToady well, terms are the only thing tighter than methods and method-like postfixes 20:36
n0tjack terms as in pi, i, e, * ? 20:37
TimToady or :8[3,0,0,7,1]
n0tjack yeah I'm not sure whether :8[3, 0, 0,7, 1] is atomic or composite syntax 20:37
to me it lookes like a weird Pair followed by an Array 20:38
PerlJam it's covalent
n0tjack Can you elaborate on that? In J terms, covalent means "can take one argument, or two arguments"
TimToady even if it were a pair, pair terms include their brackets
I think it was a chemistry joke
n0tjack ah, got it 20:39
PerlJam it was (sorta)
skids m: say 12345.polymod(8 xx *).reverse
camelia rakudo-moar 5d83db: OUTPUT«(3 0 0 7 1)␤»
n0tjack m: say :8[1;2;3]; # multidim indexing 20:40
camelia rakudo-moar 5d83db: OUTPUT«83␤»
n0tjack whoa!
someone needs to explain that one to me
TimToady m: say [1;2;3]
camelia rakudo-moar 5d83db: OUTPUT«[1 2 3]␤»
TimToady it's just taking them as commas there, I suspect
it's not a subscript
n0tjack m: say :8[1;*;3]; #...
camelia rakudo-moar 5d83db: OUTPUT«Inf␤»
TimToady remember :foo[1,2,3] is really short for foo => [1,2,3] 20:41
so it's a circumfix in it's parsing
n0tjack and how does :8 with postcirumfix [1,2,3] produce 83?
TimToady m: say (*).Num 20:42
camelia rakudo-moar 5d83db: OUTPUT«WhateverCode.new␤»
TimToady m: say Numeric(*)
camelia rakudo-moar 5d83db: OUTPUT«Cannot invoke this object␤ in block <unit> at /tmp/rcY42ElJbv:1␤␤»
TimToady m: say Num(*)
camelia rakudo-moar 5d83db: OUTPUT«Cannot find method 'Num'␤ in block <unit> at /tmp/brpvyH0LHf:1␤␤»
TimToady hmm 20:43
TimToady some in there * is numifying to Inf 20:43
*somewhere
haven't decided yet whether I like that...
n0tjack what is being expressed by :number[some array of ints] ?
as opposed to the normal base syntax 20:44
skids m: say 3 * 1 + 2 * 8 + 1 * 64
camelia rakudo-moar 5d83db: OUTPUT«83␤»
PerlJam normal?
TimToady it's just an escape valve for bases larger than 36
n0tjack skids: I see, it's the same as the normal syntax
TimToady m: say :60[24,59,59]
camelia rakudo-moar 5d83db: OUTPUT«89999␤»
TimToady m: say :60[23,59,59]
camelia rakudo-moar 5d83db: OUTPUT«86399␤»
n0tjack TimToady: No, I got that, I was confused by the , means ; thing
uh, ; means , thing
we have a builtin for that kinda calculation too (spelled #. and its inverse #:) 20:45
TimToady we used to turn those into LoLs, but I'm not sure why ; is turning into , these days
n0tjack S99: LoL
TimToady we should probably reserve ; in those situations for more monadish notations 20:46
m: say (1,2;3).perl
camelia rakudo-moar 5d83db: OUTPUT«((1, 2), 3)␤»
n0tjack that notation makes sense to me 20:46
TimToady m: say [1,2; 3].perl
camelia rakudo-moar 5d83db: OUTPUT«[(1, 2), 3]␤»
n0tjack esp. as it pertains to multidim indexing 20:47
TimToady m: say [1;2;3].perl
camelia rakudo-moar 5d83db: OUTPUT«[1, 2, 3]␤»
TimToady it's proably just doing single-arg rule on each dimension there
n0tjack m: say [(1);(2);(3)].perl;
camelia rakudo-moar 5d83db: OUTPUT«[1, 2, 3]␤»
TimToady m: say [1,;2,;3,].perl
camelia rakudo-moar 5d83db: OUTPUT«[(1,), (2,), (3,)]␤»
TimToady but again, that's an array, not a subscript
skids m: say :8[1;1,1;1,1,1] 20:48
camelia rakudo-moar 5d83db: OUTPUT«83␤»
n0tjack yeah but notionally the selection of indices along each dimension is itself an array
so having the notation for LoLs and multidim subscripts be similar is sensible
TimToady they don't really map 1:1 though since slicing adds a dimension in subscripting 20:49
n0tjack [1..2;3..4] can be thought of as a LoL or a 2D array slice
rows 1 and 2, cols 3 and 4 20:50
PerlJam n0tjack: Do you use J in your day to day life?
n0tjack yes
every day
PerlJam for what exactly? 20:50
dalek kudo/curli: f293126 | (Stefan Seifert)++ | src/core/CompUnit (4 files):
Add $.repo attribute to CompUnit

Start morphing CURLI and CURLF into CompUnit::Repositorys.
n0tjack everything you use perl for, I imagine
mostly one-off scripts to accomplish a particular task
PerlJam more math oriented or no? 20:51
n0tjack As a concrete example: I'm in sales, and I sell a lot of heavy/complex software to Wall St firms. We often to proofs-of-concept, where the customer sends us some data, adn we get it into the product to show them how they could use it
today, I took a bunch of order and trade files from a customer, parse them, interpolated some series, and converted them into the neccesary format 20:52
wow my spelling is horrible 20:53
vendethiel
.oO( write a Inline::R, or Inline::J, or Inline::APL module :P )
PerlJam yes, that sounds exactly what i would use Perl for. All your mentions of J though remind me of PDL or R
n0tjack vendethiel: I am indeed! Though I'm patiently waiting for shaped arrays before I progress with it. 20:54
vendethiel that makes sense :-)
n0tjack PerlJam: It is more R-like. R's syntax is an abomination, but it's much more popular and you can find an off-the-shelf script for anything 20:55
vendethiel n0tjack: isn't J code terser than R code? 20:55
TimToady Both of them are pretty terse; I can see where doing some of those things in a language like Perl 6 would make the non-J/R people you show the programs to think they know what's going on a little more than J or R would :) 20:56
because we've tried very hard to make different things look different
n0tjack vendethiel: Ɐlang is-terser(J, lang) 20:57
TimToady yes, J usually wins the Shortest Entry Award on rosettacode :) 20:57
vendethiel *g*
the best award of them all ;-)
TimToady and Perl 6 is often 2nd
n0tjack TimToady: Yeah, J is pretty impenetrable from an outsider's perspective. Not too dissimilar to regexen. But that doesn't bother me. Chinese looks impenetrable to me too, but over a billion people seem to get some use out of it :) 20:58
TimToady: Welll... the RC thing is probably my own personal fault :)
vendethiel n0tjack++ # Rosetta Code Driver Development everywhere 20:58
n0tjack: were one to pick up J; where would one start? 20:59
s:2nd/one/they/
n0tjack vendethiel: our answer to #perl6 is the J Forum mailing lists. 20:59
vendethiel: if you know C or C-like languages, Henry Rich's J for C Programmers is a great bootstrap 21:00
www.jsoftware.com/help/jforc/contents.htm
PerlJam n0tjack++
vendethiel n0tjack: thanks :-). will read that
PerlJam me too
PerlJam n0tjack: you may have just instigated some new J programmers :) 21:01
confused_as_usal multi method pairs(Set:D:) 21:01
Could somebody explain what's the colon after Set:D: does 21:02
vendethiel confused_as_usal: defines the invocant (self, or this in other languages)
which means self is a defined Set (as given by Set.new, or set()) 21:03
n0tjack PerlJam: I wonder if that gets me any extra years in Purgatory :)
confused_as_usal I get the defined part which is Set:D
confused_as_usal but why there is another ":" after that tacked on 21:04
vendethiel confused_as_usal: because it's not a parameter -- it's a constraint on self
PerlJam confused_as_usal: to say that the first arg is the invocant
vendethiel m: $a.foo(); <- $a is what's before said colon 21:04
camelia rakudo-moar 5d83db: OUTPUT«5===SORRY!5=== Error while compiling /tmp/bQyUi47YeU␤Variable '$a' is not declared␤at /tmp/bQyUi47YeU:1␤------> 3<BOL>7⏏5$a.foo(); <- $a is what's before said co␤»
confused_as_usal so, every method must start with this invocant? 21:07
PerlJam no
confused_as_usal if not, would it then become a class method? 21:08
PerlJam nope
confused_as_usal appreciate if you point me to any doc I can read.. thx 21:09
PerlJam confused_as_usal: methods /may/ list constraints on the invocant as the first parameter
gfldex m: class C { method foo($c where $c ~~ C:){ say 'I\'m fooing!' } }; C.new.foo;
camelia rakudo-moar 5d83db: OUTPUT«I'm fooing!␤»
confused_as_usal so, if unconstrained, there is no need to mention it?
PerlJam confused_as_usal: if you omit the invocant constraint, you still get an invocant, but it's called "self" instead of $whatever-you-define 21:10
confused_as_usal: usually you list the invocant if you want to rename it or constrain it
s/rename/give it an explicit name/
confused_as_usal cool, that makes sense now.. thanks 21:11
PerlJam confused_as_usal: but you can also end up with some weird stuff like class A { }; class B { method m(A:D $self) { } }
n0tjack m: #`(this is cool comment syntax I'd never noticed before) 21:12
camelia ( no output )
PerlJam er, with the other colon after $self
tony-o grammar tracer borken :-( 21:21
Ulti pull-one gen/moar/m-CORE.setting:2926 is mega slow for the following code my @a = (^1000000).map({("hello")}); 21:31
tony-o .tell jnthn i've submitted a PR for Grammar::Debugger to fix Grammar::Tracer for most recent rakudo build
yoleaux tony-o: I'll pass your message to jnthn.
Ulti takes up 60% of the runtime which on my laptop is 6s
mattoates.co.uk/files/perl6/mapmill...#/routines 21:32
lizmat decommute& 21:34
flussence m: my $c = “{say ‘hi’}”; /<$c>/ # arbitrary string execution with no EVAL in sight, something to pay attention to 22:49
camelia rakudo-moar 5d83db: OUTPUT«hi␤Method 'match' not found for invocant of class 'Any'␤ in block <unit> at /tmp/na66lpgcv7:1␤␤»
RabidGravy m: $_ = 'foo'; my $c = “{say ‘hi’}”; /<$c>/ 22:54
camelia rakudo-moar 5d83db: OUTPUT«hi␤»
flussence (1 char shorter than .EVAL too!) 22:56
RabidGravy m: $_ = 'foo'; my $b = “{say ‘hi’}”; my $c = $b; /<$c>/ 22:58
camelia rakudo-moar 5d83db: OUTPUT«hi␤»
RabidGravy m: $_ = 'foo'; my $b = “{say ‘hi’}”; my $c = $b;
camelia rakudo-moar 5d83db: OUTPUT«hi␤»
RabidGravy m: $_ = 'foo'; my $b = “{say ‘hi’}”;
camelia rakudo-moar 5d83db: OUTPUT«hi␤»
flussence hm, maybe I should've used single outer quotes...
RabidGravy m: my $b = “{say ‘hi’}”; 22:59
camelia rakudo-moar 5d83db: OUTPUT«hi␤»
flussence m: my $b = '{say “hi”}'; 'foo' ~~ /<$b>/
camelia rakudo-moar 5d83db: OUTPUT«hi␤»
RabidGravy ah
flussence maybe rx// needs to borrow some of the q:qcv{} modifiers... or... 23:01
m: my $b = '{say “hi”}'; 'foo' ~~ rx'<$b>'
camelia rakudo-moar 5d83db: OUTPUT«hi␤»
flussence looks like there's no way to do safe-ish interpolation there. 23:02
nick25 Hi! Why are parentheses kept mandatory around the signature (parameter list) of functions definitions? Couldn't this requirement be relaxed to be in line with other places of the language where parentheses are optional, e.g. w.r.t. if-statements, "pointy blocks", function calls? At first glance, this looks syntactically plausible to me as well: (*) sub add $x, $y { $x + $y; }. Are there technical/syntactical reasons behind this 23:03
requirement?
RabidGravy flussence, I'd go bug on that, I can see why one might want to if I squint at it, but it should require some flag 23:04
grondilu what could I put in $bar so that f is called with no argument? f($test ?? $foo !! $bar) 23:05
RabidGravy Empty?
Nil? 23:06
grondilu m: sub f($a?) { say $a.perl }; f(Empty)
camelia rakudo-moar 5d83db: OUTPUT«$()␤»
RabidGravy one of those things
grondilu m: sub f($a?) { say $a.perl }; f(Nil)
camelia rakudo-moar 5d83db: OUTPUT«Nil␤»
grondilu m: class Foo {}; say Foo.new: Empty
camelia rakudo-moar 5d83db: OUTPUT«Default constructor for 'Foo' only takes named arguments␤ in block <unit> at /tmp/QasFhCv99F:1␤␤»
flussence m: multi sub f() { say False }; multi sub f($a) { say True }; f(Empty); f(Nil); 23:07
camelia rakudo-moar 5d83db: OUTPUT«True␤True␤»
flussence m: multi sub f() { say False }; multi sub f($a) { say True }; my $bar = (); f($bar);
camelia rakudo-moar 5d83db: OUTPUT«True␤»
RabidGravy so nothing
flussence m: multi sub f() { say False }; multi sub f($a) { say True }; my $bar = slip(); f($bar); 23:08
camelia rakudo-moar 5d83db: OUTPUT«True␤»
flussence kinda hard... you could check for definedness in the signature instead 23:08
grondilu the thing is I wanted to factorize Foo.new in an expression like $some-arg ?? Foo.new($some-arg) !! Foo.new() 23:09
grondilu and write Foo.new: $some-arg ?? $some-arg !! $something-corresponding-to-no-arg 23:10
flussence m: multi sub f() { say False }; multi sub f($a) { say True }; my $bar = Empty; f($_ ?? 'arg' !! |$bar) for True, False;
camelia rakudo-moar 5d83db: OUTPUT«True␤True␤»
flussence m: multi sub f() { say False }; multi sub f($a) { say True }; my $bar = Empty; f |($_ ?? 'arg' !! $bar) for True, False; 23:11
camelia rakudo-moar 5d83db: OUTPUT«True␤False␤»
flussence there you go, just cheat a little :) 23:11
grondilu m: multi sub f() { say False }; multi sub f($a) { say True }; f |($_ ?? 'arg' !! Empty) for True, False; 23:12
camelia rakudo-moar 5d83db: OUTPUT«True␤False␤»
grondilu flussence++
m: class A {}; A.new: |Empty
camelia ( no output )
grondilu nice
flussence «f |($_ with $some-arg)» may work too 23:13
grondilu I no nothing about this 'with' keyword. Need to get up-to-date I guess. 23:14
flussence s/with/for/
"with" is like "if" but tests .defined instead of .so
grondilu ok
flussence it gets a bit hard to remember all the if/given/for/with details sometimes :) 23:15
but yeah, I'm pretty sure one of those should work there.
m: say $_ if 5
camelia rakudo-moar 5d83db: OUTPUT«(Any)␤»
flussence kinda odd that the others topicalize but that one doesn't. 23:16
grondilu it's in S04 I suppose?
grondilu fount it in S04 indeed 23:19
dalek osystem: 91c6fc1 | (Andy Weidenbaum)++ | META.list:
add Digest::xxHash
 23:21
osystem: 9aeae98 | RabidGravy++ | META.list:
Merge pull request #83 from atweiden/digest-xxhash

add Digest::xxHash
Ben_Goldberg m: multi sub f() { say False }; multi sub f($a) { say True }; my $bar = slip(); f(|$bar); 23:23
camelia rakudo-moar 5d83db: OUTPUT«False␤»
RabidGravy ooh didn't realise it was tomorrow, catch you all later 23:24
Ben_Goldberg m: multi sub f() { say False }; multi sub f($a) { say True }; my $bar = slip(); f(|[]);
camelia rakudo-moar 5d83db: OUTPUT«False␤»
Ben_Goldberg m: multi sub f() { say False }; multi sub f($a) { say True }; f(|Slip);
camelia rakudo-moar 5d83db: OUTPUT«Cannot look up attributes in a type object␤ in block <unit> at /tmp/DM5HBD7QhR:1␤␤»
Ben_Goldberg m: multi sub f() { say False }; multi sub f($a) { say True }; f(|Slip.new);
camelia rakudo-moar 5d83db: OUTPUT«False␤»
Ben_Goldberg m: multi sub f() { say False }; multi sub f($a) { say True }; f(|{});
camelia rakudo-moar 5d83db: OUTPUT«False␤»

Powered by Cro and the Raku® Programming Language.

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

Thank you!