»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'p6: say 3;' or rakudo:, std:, or /msg camelia p6: ... | irclog: irc.perl6.org | UTF-8 is our friend!
Set by masak on 12 May 2015.
00:04 colomon joined 00:24 mr_ron left 00:26 mr_ron joined
mr_ron m: sub md5-block(@H is rw, @X) {} 00:26
camelia rakudo-moar 0132b6: OUTPUT«5===SORRY!5=== Error while compiling /tmp/JG3vifxubI␤Can only use 'is rw' on a scalar ('$' sigil) parameter␤at /tmp/JG3vifxubI:1␤»
00:27 colomon left
Timbus_ mr_ron: you might find it's already 'rw' enough for you 00:43
mr_ron Timbus: its an error I'm getting from building Task::Star when building Digest::MD5 which does that on line 41 of MD5.pm breaking my Task::Star build ... 00:54
Timbus_ oh 00:55
time to bug cosimo ? 00:56
oh snap haha you're already on the case 00:58
mr_ron I cloned the repository, took out the rw, and pointed projects.json at my clone. It got me around the problem. I may bug moritz++ if the Task::Star build finishes.
s/cloned/forked/ 00:59
01:02 skids joined
mr_ron panda completed installing the revised Task::Star 01:04
.tell moritz I tried to install rakudo star with rakudobrew and panda install Task::Star. It failed to install Digest::Md5 because line 41 of MD5.pm reads "sub md5-block(@H is rw, @X) {" and the current rakudo does not like the "is rw". I forked and took out the "is rw" and pointed projects.json at my fork and Task::Star built OK. Could you take out the "is rw". 01:12
yoleaux mr_ron: I'll pass your message to moritz.
mr_ron .tell moritz s/Could you takeout .../Could you please take out the "is rw"?/ 01:13
yoleaux mr_ron: I'll pass your message to moritz.
01:28 mattprelude left 01:33 colomon joined 01:38 xinming_ joined 01:40 xinming left 01:43 vendethiel joined 02:06 vendethiel left
dalek q: 8964bde | ShimmerFairy++ | answers.md:
Add answer for needing unspace with method calls

It's a bit long, but hopefully comprehensive enough to be a satisfactory answer. Improvements of course welcome :) .		 02:09
q: 6b3cecc | ShimmerFairy++ | answers.md:
A couple grammar/wording improvements.
 02:12
02:21 aborazmeh joined, aborazmeh left, aborazmeh joined 02:39 noganex joined 02:42 noganex_ left 02:43 mr_ron left
dalek kudo-star-daily: cce58f2 | coke++ | log/ (9 files):
today (automated commit)
 02:43
02:45 mr_ron joined 02:48 nightfrog left
dalek kudo/nom: 73836e1 | TimToady++ | src/core/ (7 files):
infixes that should likely be pure
 02:50
02:51 nightfrog joined 02:58 xinming_ left, xinming joined
TimToady .tell awwaiid --> belong in the signature because it enables return typing on lambdas 03:20
yoleaux TimToady: I'll pass your message to awwaiid.
03:25 uberbaud left 03:44 adu joined
zostay m: my $x = 10; $x max= [max] 4, 6, 8; $x.say; 03:48
camelia rakudo-moar 73836e: OUTPUT«10␤»
03:51 mr_ron left 03:52 mr_ron joined 04:01 AlexDaniel left 04:02 khw left 04:26 rmgk_ joined, rmgk is now known as Guest42982, rmgk_ is now known as rmgk 04:29 tokuhiro_ joined 04:30 Guest42982 left 04:34 tokuhiro_ left 04:35 mr_ron left 05:13 RabidGravy joined
awwaiid interesting reason TimToady. I will noodle that. 05:16
yoleaux 03:20Z <TimToady> awwaiid: --> belong in the signature because it enables return typing on lambdas
05:24 mr_ron joined 05:35 tokuhiro_ joined, rurban joined 05:38 kaare__ joined 05:39 tokuhiro_ left
adu noodle? 05:45
05:50 aborazmeh left
moritz .tell mr_ron that's what the R* releases are for: they are snapshots where the version of rakudo and the modules match 05:57
yoleaux 01:12Z <mr_ron> moritz: I tried to install rakudo star with rakudobrew and panda install Task::Star. It failed to install Digest::Md5 because line 41 of MD5.pm reads "sub md5-block(@H is rw, @X) {" and the current rakudo does not like the "is rw". I forked and took out the "is rw" and pointed projects.json at my fork and Task::Star built OK. Could you take out the "is rw".
moritz: I'll pass your message to mr_ron.
01:13Z <mr_ron> moritz: s/Could you takeout .../Could you please take out the "is rw"?/
dalek kudo/nom: cb25b2f | TimToady++ | src/Perl6/ (2 files):
Allow infix:<.> if next thing is alphabetic

As much as I mistrust "fluent" programming, we already have an infix:<.=> operator, so it's not a big stretch to have infix:<.>. We limit it to actual method names so that we can detect (most) accidental use of P5's use of . for concatenation. Note that this is a term/infix distinction now, so .meth where a term is expected still defaults to $_. Community standards for method cascades will likely ask for whitespace around infix:<.> to make a visual disinction from the term form, but we don't enforce that, if for no other reason than to refrain from giving the fluent programmers something to yammer on about. Basically, I've decided that the slight possibility for greater confusion is outweighed by consistency with .= and utility of have a . that can also serve as a precedence modifier. (Since both .= and . parse leftward with item assignment precedence, but rightward as a term.)		 06:04
ShimmerFairy TimToady: ironically the "fluent programming" page on wikipedia isn't very good at saying what it actually is, so I still don't quite understand "fluent" programming :P 06:06
TimToady Anyway, we can now do cascaded methods to the next line without a \ as long as it's a normal looking method 06:09
ShimmerFairy Sounds fair to me, though I personally have never actually lined up method calls, come to think of it :) . It's always been either method arguments or nested sub calls (e.g. nqp::op(...) "trees") 06:10
06:11 rurban left
TimToady though I don't like the style, I decided it was allowable without too much pain to the rest of us :) 06:12
ShimmerFairy I suppose you could think of the dot as a strange, twisted kind of concatenation, in that you bring together a bunch of things to produce one singular object, which is how read things like +$foo.bar.baz.quux to remember the prefix is after method calls.
TimToady but . will let you run the + before the method now, due to precedence, and the spacing will be intuitive 06:13
ShimmerFairy I also think P5-isms are less of a concern now, enough that I don't worry about this change interfering with the worryobs :)
TimToady: so +$foo.bar means +($foo.bar), and +$foo . bar means (+$foo).bar ? Sounds almost intuitive :P 06:14
TimToady pretty nearly almost :)
JimmyZ heard that many guys around the Community are waiting for the 1.0 beta release news .. 06:15
ShimmerFairy TimToady: also, thanks for invalidating my carefully constructed FAQ answer mere hours after pushing it :P . Though while writing it I was thinking "could someone write an infix:<.> ? But that would probably require a slang, being a 'special' operator" 06:16
TimToady JimmyZ: well, it's a beta of an official release, it's not an official beta release :) 06:19
but we're certainly clamping down on language changes after my birthday, and if we liked, we could label a version of the test suite as "first beta"
but the beta is the test suite, not an actual release, so October's rakudo will be the first "beta" rakudo really, in my mind 06:20
ShimmerFairy TimToady: is my proposed S32::Stringy too much of a change this close, then? Honestly, part of me would be happy to shelve it for 6.1, and focus again on improving Pod6 instead. :) 06:21
TimToady unless we want to do an extra release next week just so we can say "supports 6.birthday"
I need to do more than just skim it 06:22
ShimmerFairy TimToady: I'm alright with saving it for a future Perl 6. It's not a revolutionary change to S32::Str, but I could easily see it being too close for comfort by now. 06:23
JimmyZ TimToady: many chinese friends ask me about it(the news) :P
ShimmerFairy
.oO(Test Suite 6.🎂 )
06:24 mr_ron left
dalek kudo/nom: db0c46e | moritz++ | lib/NativeCall.pm:
NativeCall: explicitly-manage does not need its string arg "is rw"

it modifies the string by mixing something in, not via the container. This should fix a test failure in DBIish, which calls explicitly-manage on a non-container		 06:35
TimToady m: say -42 . abs + 5 06:38
camelia rakudo-moar cb25b2: OUTPUT«47␤»
TimToady m: -42 . abs . say
camelia rakudo-moar cb25b2: OUTPUT«42␤»
ShimmerFairy reminds me of lisp or haskell all of a sudden :) 06:39
06:41 Fleurety joined
ShimmerFairy m: say -42.abs + 5 06:41
camelia rakudo-moar cb25b2: OUTPUT«-37␤»
ShimmerFairy And I suppose it's one way to solve the -42 issue that came up a while back :)
06:44 mr_ron joined
TimToady JimmyZ: otoh, the September release is pretty close to beta quality 06:50
JimmyZ TimToady: yeah, they are waiting for the news(about 1.0 beta), not the star relases, haha 06:52
06:53 Loren joined
Loren afternoon, #perl6 06:53
JimmyZ TimToady: and happy birthday! 06:56
06:57 adu left
TimToady yup, now it's my birthday here :) 07:03
We are now officially in beta. :) 07:04
JimmyZ ;)
ShimmerFairy 🎂 ! :) 07:06
dalek kudo/nom: 96c2ec9 | TimToady++ | src/core/Perl.pm:
change to 6.Birthday
 07:08
TimToady heads off to sleep the first part of his birthday... 07:10
JimmyZ good night 07:11
Loren good night .. :) 07:12
07:24 _itz_ joined 07:32 mr_ron left 07:36 skids left 07:37 tokuhiro_ joined 07:41 tokuhiro_ left 07:49 araujo_ left, vike left 08:12 darutoko joined
masak good antenoon, #perl6 08:18
happy b-day, TimToady. wishing you peace, and long life. 08:19
DrForr Morning.
TEttinger I hope TimToady gets a gift of hundreds of pull requests fixing every issue in Perl6 for the next year 08:20
08:20 FROGGS joined 08:24 haroldwu left 08:25 rurban joined
Loren That's a pretty gift. 08:31
TEttinger it might take a year to arrive 08:33
08:38 tokuhiro_ joined 08:42 tokuhiro_ left 08:44 adu joined 08:50 vendethiel joined 08:51 araujo joined 09:14 nbd_ joined 09:16 xinming left 09:17 xinming joined 09:24 xinming left 09:25 xinming joined
[Tux] test 50000 35.068 34.967 09:26
test-t 50000 35.727 35.626
09:27 adu left
FROGGS o/ 09:36
yoleaux 26 Sep 2015 20:16Z <lizmat> FROGGS: I commented on your comment on gist.github.com/lizmat/f3807956c354c14902a3
09:38 espadrine left
FROGGS .tell lizmat I commented on your comment on my comment :o) - gist.github.com/lizmat/f3807956c35...nt-1582671 09:44
yoleaux FROGGS: I'll pass your message to lizmat.
09:49 spider-mario joined 09:56 yqt joined 09:58 brrt joined 10:00 pmurias joined 10:14 abaugher left, abaugher joined 10:19 rindolf joined 10:20 mattprelude joined 10:26 Khisanth left
brrt good * #perl6 10:28
10:29 TEttinger left 10:30 leont joined, Akagi201 joined 10:32 Khisanth joined 10:33 brrt left 10:36 luiz_lha left 10:40 tokuhiro_ joined
lizmat wishes TimToady a Happy Birthday 10:42
yoleaux 09:44Z <FROGGS> lizmat: I commented on your comment on my comment :o) - gist.github.com/lizmat/f3807956c35...nt-1582671
lizmat and thanks him for the give that is Foo\n.method :-)
[Tux] (and many other people) will be very happy with that 10:43
*gift
10:44 tokuhiro_ left, emilbarton joined
lizmat m: dd Int.new(42); Int .new(42); Int. new(42) 10:45
camelia rakudo-moar 96c2ec: OUTPUT«42␤»
lizmat m: dd Int.new(42); dd Int .new(42); dd Int. new(42)
camelia rakudo-moar 96c2ec: OUTPUT«42␤42␤42␤»
lizmat :-)
10:46 yqt left
emilbarton hello, I'm puzzled because I have a class with 'has $.id' but somewhere I get "cannot assign to a readonly value or variable" for a regex like 'if $_.id ~~ /^<$user>$/', in other places in my script the same regex works normally. Can you help me solve this? 10:48
10:48 mr_ron joined
dalek ast: a2e9472 | lizmat++ | S02-lexical-conventions/minimal-whitespace.t:
Adapt tests to new reality of cb25b2f475f31335d77
 10:48
lizmat emilbarton: could you create a gist of that ? 10:49
10:49 Khisanth left
psch m: class A { has $.b }; my A $a .= new( :b("foo") ); say $_.b ~~ /foo/ given $a; # must be something more involved 10:50
camelia rakudo-moar 96c2ec: OUTPUT«｢foo｣␤»
emilbarton lizmat: no I'm sorry
psch m: class A { has $.b }; my A $a .= new( :b("foo") ); my $rx = "foo"; say $_.b ~~ /<$rx>/ given $a; # must be something more involved
camelia rakudo-moar 96c2ec: OUTPUT«｢foo｣␤»
moritz emilbarton: it seems you are trying a writing access on it, like with .id ~~ s/a/b/ or so 10:51
emilbarton moritz: but precisely I aint 10:52
10:52 Alina-malina left
emilbarton psch: thanks, I'll try 10:52
moritz emilbarton: then please boil it down to a runnable test case 10:53
10:53 Alina-malina joined
moritz emilbarton: and submit it as a bug report 10:53
psch emilbarton: ...i didn't suggest anything? i just demonstrated that what you describe should work :)
emilbarton ok, it happens just after instanciation
FROGGS .tell TimToady Happy Birthday! <oȏ> <-- should look like a hug if it is not obvious :o) 10:54
yoleaux FROGGS: I'll pass your message to TimToady.
dalek Iish: a2e3ca9 | moritz++ | t/01-ConnectConfig-PG.t:
Avoid uinitialized warning in test
 10:56
moritz .tell [Tux] Tux/CSV/NLPW-2015/090-hooks.txt uses "@f is rw" in a signature (multiple times); please drop the "is rw"; newest rakudo disallows it on array params 11:01
yoleaux moritz: I'll pass your message to [Tux].
11:01 rurban left
[Tux] moritz, why? 11:02
yoleaux 11:01Z <moritz> [Tux]: Tux/CSV/NLPW-2015/090-hooks.txt uses "@f is rw" in a signature (multiple times); please drop the "is rw"; newest rakudo disallows it on array params
11:02 vendethiel- joined, vendethiel left
[Tux] as in "why is it not allowed on arrays? 11:02
moritz [Tux]: because arrays and hashes aren't read-only anyway
[Tux] afk
ok
moritz FROGGS: rakudo-p5/v5/src/Perl5/Terms.pm uses 'is rw' on array params (chop and chomp); please drop the "is rw" 11:06
FROGGS moritz: does it even compile these days? 11:09
moritz FROGGS: I'm sure it does not compile, at least due to the extra "is rw" :-) 11:10
FROGGS :o)
jnthn m: say -42.abs; say -42 .abs; # I love how TimToady++ takes a continued complaint about whitespace not being allowed and makes it mean something in the process of making it significant :) 11:13
yoleaux 26 Sep 2015 18:03Z <FROGGS> jnthn: I try to implement 'my $a; "a" ~~ /$a=[.]/' and fail to do that in P6::A... is that something I should be able to do in the actions?
camelia rakudo-moar 96c2ec: OUTPUT«-42␤42␤»
yoleaux 26 Sep 2015 18:11Z <brrt> jnthn: (poor jnthn for all the spam) i've created a gist with dynvar log results: gist.github.com/bdw/992dfd5b69b7d42f2239
jnthn Uh, in the process of making it allowed 11:14
That visual-pills nicely, especially in the multi-line case.
Happy Birthday, TimToady++ :) 11:15
FROGGS ohh, nice example
jnthn FROGGS: Well, you may need a little effort to make the parse work out right for the $a=... thing
FROGGS jnthn: it is parsing it correctly 11:16
jnthn: but emitting a p6store to that variable in a qast::stmts in a qast::nodelist doesnt work...
lizmat FROGGS: commented on gist.github.com/lizmat/f3807956c354c14902a3 again :-)
FROGGS jnthn: the match result will be empty then, and I dunno what to do
lizmat: that still renders v1.2.+ totally useless 11:18
lizmat well, I can't say I disagree with that...
11:19 rmgk left
lizmat but what you're suggesting, is that we can *only* have semantic versioning, and nothing else 11:19
FROGGS lizmat: I dont
I say that we let the users give more than one option to chose from, and also a way to be less strict / more flexible
lizmat ok, so maybe we need a pragma? Allow semantic versioning ? 11:20
FROGGS and I also say that you cannot predict what will work out and what does not, so let use give the users a chance to evolve to the optimal solution
no
lizmat how would you handle difference in auth ? 11:21
would *that* be an error ?
FROGGS no, if the user decided to not filter on auth, or to use a regex that is satisfied by two auths (github:FROGGS and pause:FROGGS) then we are not allowed to stand in the users way 11:22
that's not snake language here, you know :o)
lizmat so it would load both ? 11:23
nine .tell masak the visitor macros are something that I incidentally though about before falling asleep yesterday. Seems like something really useful for implementing something like LINQ and even more, for making Rakudo's optimizer readable.
yoleaux nine: I'll pass your message to masak.
jnthn FROGGS: OK, I'll have to take a closer look later :) 11:24
jnthn bbl
FROGGS lizmat: no, it will sort the candidates by version, and pick these with the highest version; if the result is not a signle CU --> fail
single* 11:25
lizmat so even major version bump ?
FROGGS lizmat: if the user did not specify the version at all, sure
because then you don't know there is a bump
lizmat well, that *really* feels like the noose that is Perl 5 module packaging :-( 11:26
FROGGS lizmat: it is like duck typing, you can/want to add type information, but you are not forced to
lizmat and we make it way too easy for people to step into that noose
well, I think we agree to disagree on this :-)
FROGGS yeah, sadly :/
lizmat personally, I could live with patch level changes, and take the highest 11:27
anytihing else, I think is asking for trouble in the long run
so I guess we will have to take jnthn up on his offer to put on his pumpkin hat 11:28
FROGGS lizmat: yes, but as I said, it is not up to you to decide... the user is in charge to make clear what exactly (s)he wants
lizmat we've put several "traps" in Perl 6 code to prevent people coming from Perl 5 to continue with the same "bad" habits
use Foo;
is one of them, in my book
nine lizmat: there's usually not much difference in API between 0.* and 1.* versions. Even when people stick to semver. 11:29
lizmat the thing is this: one of the bad reps of Perl in the world, is the nightmare that is dependency hell
with :auth and :ver , Larry has at least pointed to a way out of that 11:30
if we don't lead people that way, then all of that work has really been in vain
and we could as well just drop the whole CUR idea, and just go for Perl 5 semantics
it has brought us this far, why won't it take us further ?? 11:31
sorry, but I feel really deeply about this, having had to deal with the dependency nightmare at former $work
and seen the way it stifles moving forward
yes, it is more strict in the beginning 11:32
but developers need to know the needs of the sysadmins a bit as well
I see it as being a responsible developer
ShimmerFairy I think versioning is such a free-form thing in general that adding any kind of enforcement will inevitably make a number of people unhappy.
lizmat if you say "use Foo"
it just means you don't care
ShimmerFairy cf. the fact that you can construct a Version object with anything you want
lizmat and you should in a production environment 11:33
11:35 mr_ron left, mr_ron joined
ShimmerFairy FROGGS: I think highest version picking should only happen with packages from the same auth (that is, sort by auths and _then_ versions), because otherwise you can't be sure the versioning scheme is the same. 11:36
btyler the node ecosystem does a pretty good job of this, I think 11:37
ShimmerFairy I don't see how 'supersede' is related to module loading at all, btw. 11:39
btyler when you declare dependencies you specify whether you care about version, and if you do, you can pick "X, + or - minor patch level", "anything newer than X", "precisely X", or "X.Y or Y.Z, but nothing else"
and best practice depends if you're writing an application or a library
ShimmerFairy lizmat: I think having multiple candidates should only die if they have different auths, whereas multiple versions of one auth can be sorted and chosen as appropriate. 11:41
leont would agree p5's take on versioning is suboptimal
In theory, p5 meta supports relatively complex behavior, in practice no one has implemented it in a useful way
btyler oh, and the other bit about node/npm -- every single library/all has its own local::lib-alike, so different dependencies relying on different versions of the same lib doesn't matter 11:42
*library/app
at the expense of disk space, of course 11:43
lizmat btyler: I don't quite follow you there
leont As for numeric versus semantic versioning: I don't think it's unreasonable to pick one of them and force users to use it. It's hard-to-impossible to make both work at the same time.
lizmat disk space is cheap, generally _0
:-)
leont Encouraging people to specify more about their versions sounds like a good idea 11:44
Subtly related to that: having a mechanism to install tests, and rerun them (for example after the upgrade of a dependency) would be a killer feature too
We've been talking about that in p5 land for years, but it's very NYI so far
lizmat in P6 land that would be possible, as the tests get installed as well 11:45
leont I was hoping you were going to say that
lizmat yeah, the thinking on that has changed :-)
btyler lizmat: sorry, I might be a bit off topic; I don't totally understand the p6 versioning/dependency thread. in short: I have an app Foo. Foo depends on library libbar and libbaz. the way node/npm handles things is to give both libbar and libbaz their own local::lib (called node_modules) with the precise version that it needs
so if both libbar and libbaz need libawesome, but one needs libawesome 0.3 and the other needs libawesome 1.5, there's no problem 11:46
leont I suspect that testing if things still work is often more reliable than specifying "0.3.x should work, no idea about 0.4.x"
btyler you just end up with two copies of libawesome installed; one in libbar's local::lib and the other in libbaz's local::lib
lizmat btyler: that problem will not exist in P6, as 2 versions of the same compunit can be installed in the same CUR
ShimmerFairy Gentoo, for example, will complain about you not specifying the package category when multiple packages have the same name, but you're free to leave the version unspecified and get the latest possible. I think I'd like that to be how 'use' decls work 11:47
btyler goes to the glossary
lizmat *and* they can be loaded simulteanously as well (even in the same lexical scope, if you can rename one of them)
btyler very nice
ShimmerFairy I also don't think enforcing one particular versioning scheme would be a good idea. People like their conventions :) 11:49
lizmat well, that is true :-( 11:50
:-)
but conventions can change
11:51 vendethiel- left
lizmat and I'm thinking in 5 years from now: people will think: how did we *ever* build large applications without specifying which part needs which version of a library 11:51
and it being a USP for Perl
ShimmerFairy And I don't think restricting the version scheme would suddenly make cross-auth version sorting reliable; one auth's v0.2.7 is another's v42.0.0 11:52
lizmat in my experience, the use of Perl in a lot of corporations, is blocked by sysadmins, not by developers
ShimmerFairy: I'm not sure what you're saying :-) 11:54
btyler that's partially a product of the deployment artifact too, right? e.g. golang sysadmins don't need to care about what's running on the production box, because all the libs are wrapped up in the thing you deploy
jdv79 fwiw im with FROGGS on this
ShimmerFairy lizmat: I think someone made the implication (FROGGS?) that :auth<Foo> and :auth<Bar> candidates wouldn't die, and instead pick the highest version
but I don't recall if he stated that explicitly, or it was just the vagueness of what he said 11:55
lizmat ah, yes, that should *never* happen :p)
jdv79 no need to be overly strict out of the gate methinks
ShimmerFairy lizmat: the one reason I disagree with forcing a strict :auth<> and :ver<> requirement is that not every program endeavors to be stable, or left unmaintained. 11:56
lizmat ShimmerFairy: I'm not saying that either
but *if* CURLI.candidates finds more than one CompUnit for -use Foo-, it should die 11:57
jdv79 id be interested to how the laxer approaches would work out in complex codebases
ShimmerFairy lizmat: my thoughts boil down to: require an :auth<> only to disambiguate, and default to highest version for multiple same-auth matches. Without a disambiguating :auth<> where it's needed, it should die
lizmat well, that still leaves us with most of the shit that has happened in P5, because most modules *are* maintained by the one developer 11:58
*and* they've been known to cause problems between versions even if they didn't intend to 11:59
lizmat being one of them
ShimmerFairy I don't think you can reasonably require a :ver<> specifier without causing a lot of grief. An unqualified 'use Foo' is allowed for a reason :)
lizmat in my eyes, an unqualified -use Foo- is a one liner, or an historical artefact 12:00
12:00 AlexDaniel joined
ShimmerFairy I disagree. And I predict people would just start using :ver<*> anyway, at which point requiring it doesn't seem to add much value :) 12:01
lizmat well, yeah, :ver<*> is just as badd 12:02
well, I can't seem to convince
guess it will need to become a PBP :-(
ShimmerFairy I think lack of :ver could reasonably be considered a synonym for :ver<*> 12:03
jdv79 pbp?
lizmat Perl Best Practice
ShimmerFairy You could have a pragma that enforces :ver, of course, I just don't think it's a good default for all cases.
jdv79 kinda like optional typing as FROGGS suggested? 12:04
lizmat well, I suggested a pragma before
and we were so glad that -use strict- is now the default :-(
ShimmerFairy I don't think that's quite the same as 'use strict', though :) 12:05
emilbarton lizmat, moritz, psch: ok I managed to make a gist of my problem. It's my first script in perl6 so be kind! It's a client-server app for sqlite3. The gist is at gist.github.com/emilbarton/41d3e70...c0d6115c00 . I'v tried to remove as much as possible of my spaghetti...
lizmat emilbarton: I get a 404 on that link ? 12:06
jdv79 the strict has to be tamed though:) this is still Perl afterall.
emilbarton gist.github.com/emilbarton/41d3e7094dc0d6115c00
lizmat :-)
emilbarton there is an example usage to reproduce the error at the end 12:07
12:07 kaare__ left
lizmat hmmm... the gist is 142 lines, but the line number refers to line 187 ? 12:09
jdv79 couldnt a versioning pbp be hoisted into core later anywsy? 12:11
*anyways
lizmat well, yes, but not a default setting change, as that could break code
jdv79 so dont upgrsde?
lizmat not upgrading is not an option in most cases in real life, I'm afraid 12:12
ShimmerFairy That's one thing I'd like to see about P6's design process in future versions: don't upgrade if you're not willing to make the necessary changes.
jdv79 what. $work was on 5.8.5 until 2 years ago 12:13
lizmat anyways, to summarize: on what we *do* agree on (I think):
1. if two CompUnits are found with different auths, it should die 12:14
12:14 _mg_ joined
lizmat hmm.. that's not much 12:14
gfldex emilbarton: try changing 172: my @auth =(1,2,3); to my @auth = [1,2,3];
emilbarton lizmat: the problem happens at line 178
lizmat: in the server file 12:15
lizmat my proposal would be to allow auto-select to the highest version of a CompUnit if: 1. name / auth are equal and there's only a patch level difference
emilbarton gfldex: ok but it doesn't matter for the error 12:16
jdv79 could thatvlast bit be dropped for now?
lizmat m: my @a = (1,2,3); my @b = [1,2,3]; dd @a, @b # there is a difference ???
camelia rakudo-moar 96c2ec: OUTPUT«Array @a = [1, 2, 3]␤Array @b = [1, 2, 3]␤»
ShimmerFairy lizmat: that would require settling on a specific versioning scheme, but I do think something specifying a general range of acceptable versions would be a good idea 12:17
btyler lizmat: is there a middle ground of a 'warn' if it's the same name/auth and there's a > patch level diff, or it isn't clear how to calculate the version diff (if someone is using a goofy naming scheme)?
12:17 zed__ joined
ShimmerFairy *cough* gentoo's SLOTs *cough* 12:17
btyler or maybe we need pluggable version comparison :D
use Version::SemVer in my library 12:18
ShimmerFairy ponders a separate :slot<> specifier :P
lizmat: i.e. I think it'd be best if we had some way of distinguishing groups of compatible versions, that preferably didn't rely on a certain portion of the version number 12:19
lizmat ShimmerFairy: using slots may be an idea, but at the META6.json level, I would think 12:20
12:20 zed__ left
ShimmerFairy e.g. use Qt:slot<4> for getting a Qt4-based package, latest existing version 12:20
lizmat not at that level
I see it more as a tie-breaker
so the slot info lives in the meta-info 12:21
if you specify a -use- command that selects more than one CompUnit, but they are in the same slot, *then* use the one with the highest version
ShimmerFairy lizmat: you mean in terms of the 'depends' section, perhaps? (Which would be analogous with ebuild files, I _think_?) 12:22
lizmat I think "slot" is a bad name, though, API version would be better, I think
maybe an example works better:
ShimmerFairy Yeah, it of course doesn't have to be "slot", that's just what gentoo calls it :P
lizmat I have Foo 1.1 and Foo 2.0 installed
they are by the same auth 12:23
and they have the same "slot"
when I do "use Foo 1+", it would load Foo 2.0 without any issue
if Foo 2.0 would be in a different "slot"
12:23 _mg_ left
lizmat the same "use Foo 1+" would die, because there are 2 CompUnits that match that specification, and they are in different slots 12:24
semantically versioned modules would auto-slot for patch levels
ShimmerFairy Gentoo's SLOTs are designed to let you have multiple versions of a package installed at once, and choose which slot you need something from. I don't think the "multiple installed" is a problem for us, it'd just be the "choose a slot" thing, in my mind 12:25
lizmat (unless differently specified in the META6.jsnon)
emilbarton lizmat: I've updated the error output at the end of the 2 files
moritz: should I submit a bug report or is there something obviously wrong in my code?
12:25 z8 joined
lizmat ShimmerFairy: slots in this context, would only be for selection, not for installing, indeed 12:26
ShimmerFairy lizmat: calling it an API version instead, I think you should have to select a particular version of the API if there are multiple, but then within that highest-version by default
so if someone has Qt modules installed for both Qt4 and Qt5, you'd have to specify somewhere which API you want, e.g. use Qt:api<5>
(because there isn't necessarily a clean relationship between "slot"s and versions, at least in Gentoo's implementation of the concept) 12:27
lizmat well, I would sorta expect a Qt4 and Qt5 module then, if they are that radically different API's ?
12:27 spebern joined
dalek ast: 450b919 | TimToady++ | S (2 files):
adjust some exceptions
 12:27
ast: 4321092 | TimToady++ | S02-lexical-conventions/minimal-whitespace.t:
merge the changed tests I forgot to check in
kudo/nom: be141b8 | TimToady++ | src/Perl6/Grammar.nqp:
. and .= should parse tighter

To be intuitive, . and .= need to parse at about the level of unaries, such that "-3**2 . abs" treats the left as a single pill, but nothing looser than that, lest it eat up the multiplication of "-3 * 2 . abs".
  (Changing .= away from item assignment is okay because initializer .=
was effectively tighter precedence anyway, and nobody ever noticed. And .= was always syntactically privileged anyway.)
TimToady woke up early realizing two things: 1) he forgot to check in his tests for . and 2) the precedence of . and .= is too loose, and they should just eat things that would ordinarily be kept together visually, like unaries and exponentiation, but nothing looser
yoleaux 10:54Z <FROGGS> TimToady: Happy Birthday! <oȏ> <-- should look like a hug if it is not obvious :o)
lizmat good morning, TimToady, and what a beautiful day this is (I hope :-) 12:28
ShimmerFairy lizmat: true, it's just that thinking of concrete examples from gentoo, Qt was the first thing that came to mind (there are two SLOTs for 4 and 5 for Qt's packages, qtcore etc.) 12:29
12:30 telex left
TimToady heads back to bed maybe... 12:31
lizmat 5:30 is a bit early, right ?
ShimmerFairy lizmat: I think we both agree that there needs to be a distinction between "API versions", and the other, regular kind. I just happen to think that you should be able to disambiguate on API without specifying the regular version if you want :) . 12:32
12:32 diana_olhovik_ joined
ShimmerFairy ♞ TimToady maybe o/ 12:32
12:32 telex joined
[Tux] moritz, removed from .txt there were not present in the actual code anymore 12:37
I also removed the remaining is rw's that were unneeded 12:38
dalek ecs: 2fa83b9 | lizmat++ | S22-package-format.pod:
Tests *will* probably be installed
 12:40
12:41 spebern left, tokuhiro_ joined 12:44 FROGGS left 12:46 tokuhiro_ left 12:47 cognominal joined 12:48 mr_ron left 12:50 yqt joined
lizmat ShimmerFairy: thinking more about slots / API version: isn't that in the end just the same as semantic versioning really ? 12:50
ShimmerFairy lizmat: I suppose, but having a separate concept of API version would allow more freedom in how someone can actually version their stuff. 12:51
lizmat true 12:52
12:52 vendethiel joined
ShimmerFairy (which is something Gentoo has to allow for, but P6 is perhaps able to be more influential on its packages right now ☺) 12:52
12:53 mattprelude left
[ptc] TimToady: Happy Birthday! 12:54
lizmat if we would add an :api, I think it should be a name, rather than a number
ShimmerFairy: use QT:api<QT4>
use Qt:api<Qt4> rather :-) 12:55
12:55 z8 left 12:57 FROGGS joined 13:00 abaugher left 13:01 abaugher joined
timotimo o/ 13:02
lizmat: i expect you might know a bit about this: should parameters like args in "sub foo(|args)" be treated the same was as "is raw"?
because right now we allocate a scalar container for those and i'm sure there's no need at all 13:03
lizmat feels to me they should be treated the same... but that's just a feeling, I would defer to jnthn for a authoritative answer
timotimo also, there's a whole bunch of subs that have a $¢ lexical registered, but not used, so i think the optimizer wants to learn about making unnecessary dollarcent lexicals disappear as well 13:04
lizmat yup, feels like :-)
timotimo i suppose i'll just cook up a patch for the "is raw by default" thing and run spec tests 13:05
if they agree i'll go the "forgiveness > permission" route; can always revert it and put a good explanation in the commit message if the idea was bad 13:06
ShimmerFairy timotimo: ah, I made $¢ work in grammars (but not bare /.../ regexes, grr), but I didn't touch the optimizer :)
timotimo oh i see :) 13:07
ShimmerFairy lizmat: I can agree on :api<> . There's no reason we should limit ourselves to mere numbers :) .oO(use webkit-gtk:api<random-api-change-yaaay>) 13:08
timotimo updates all the components 13:11
lizmat goes cycling for some fresh air, ideas and energy 13:12
timotimo i noticed the |args thing when CALL-ME from inside NativeCall was the #1 allocator of Scalar objects in my nativecall-heavy program 13:13
the only scalar it allocated was the one for |args in that case
so removing that was quite worth it :)
13:13 emilbarton left, kaare__ joined
timotimo sadly, that program isn't a good benchmark yet, as its workload depends on random numbers 13:13
but we do have srand, right? maybe that'll help 13:14
13:15 espadrine joined
ShimmerFairy timotimo: btw, if you happen to figure out how to make $¢ work in places like $a ~~ /<stuff> { say $¢ } <stuff>/ , that'd be cool :P 13:16
[Tux] $ panda list --installed --verbose 13:17
resolve stage failed for list: Project list not found in the ecosystem
in method throw at /pro/3gl/CPAN/rakudobrew/moar-nom/install/share/perl6/runtime/CORE.setting.moarvm:1
timotimo ShimmerFairy: i don't have a clue yet 13:18
ShimmerFairy timotimo: don't worry about it, I couldn't figure it out either :) (closest I got, IIRC, was inserting my own $¢ to bypass qbuildsub's insert-if-nonexistent, and then removing what I added, but then EVALs of regexes died on 'missing lexical'(?)) 13:19
timotimo :| 13:21
13:25 BenGoldberg joined
timotimo at least i think i found the right peg to hang the "is raw by default" thing on 13:25
spec test time 13:26
13:26 BenGoldberg left 13:27 BenGoldberg joined
pmurias lizmat: isn't pinning down exact version deps something that Carton (and similiar such things) try to fix in Perl 5 land? 13:29
lizmat I don't know Carton well enough to answer that question 13:30
i *think* Carton is more about having a single CUR in @?INC with only the modules you need for a specific application 13:31
and then start perl6 with just that one CUR
if I would transpose this from P5 to P6
13:31 leont left 13:32 khw joined 13:33 Loren left
timotimo only t/spec/S02-types/WHICH.t, t/spec/S05-transliteration/trans.rakudo.moar and t/spec/S32-num/power.rakudo.moar have failures 13:34
i expect that's not because of my change
lizmat no, it's not 13:35
timotimo yays
lizmat m: StrDistance.new
camelia rakudo-moar be141b: OUTPUT«Cannot make a Block object using .new␤ in block <unit> at /tmp/O9_9bpmVfG:1␤␤»
lizmat now what is that about???
13:35 Loren joined
lizmat basically the issue in WHICH 13:35
timotimo huh, well, that's interesting
lizmat m: StrDIstance.WHICH
camelia rakudo-moar be141b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/_hTGl0TykO␤Undeclared name:␤ StrDIstance used at line 1␤␤»
lizmat m: StrDistance.WHICH
camelia ( no output )
lizmat
.oO( still not cycling :-)
 13:37
timotimo m: say StrDistance.WHAT 13:38
camelia rakudo-moar be141b: OUTPUT«(Block)␤»
timotimo what the flying foot is up with that? 13:39
m: my class FooBar is Cool { }; say FooBar.WHAT
camelia rakudo-moar be141b: OUTPUT«(FooBar)␤»
ShimmerFairy m: say StrDistance.^mro 13:40
camelia rakudo-moar be141b: OUTPUT«((Block) (Code) (Any) (Mu))␤»
timotimo StrDistance only occurs in the generated m-CORE.setting in the parts included from StrDistance.pm
ShimmerFairy m: class FooBar is Cool { }; say FooBar.^mro
camelia rakudo-moar be141b: OUTPUT«((FooBar) (Cool) (Any) (Mu))␤»
timotimo so nothing overrides that after the fact
and running that exact file with a "say StrDistance" afterwards gives me "(StrDistance)" instead 13:41
lizmat my class StrDistance is Cool 13:42
timotimo yeah
lizmat so not sure where the Block come from...
the ACCEPTS looks a bit wonky?
timotimo me neither
ShimmerFairy m: my class Foo is StrDistance { }; say Foo.WHAT 13:43
camelia rakudo-moar be141b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/irBn76jOdj␤Cannot call trait_mod:<is>(Foo, Block); none of these signatures match:␤ (Mu:U $child, Mu:U $parent)␤ (Mu:U $child, :$DEPRECATED!)␤ (Mu:U $type, :$rw!)␤ (Mu:U $type, :$nativesiz…»
timotimo that looks like an artifact from the days where we had that same thing for "list", too
jnthn Just passing by but: some patch somewhere has caused some *very* odd kind of corruption/mis-compilation
It happened around the time I merged the isrwcont change 13:44
psch $_ = "foo"; if tr/o/u/ > 1 { say "more than one replacement" } # is the reason for StrDistance.ACCEPTS
ShimmerFairy lizmat: both ACCEPTS and BUILD look suspicious to me
psch doesn't work right now though, obviously
jnthn m: say StrDistance
camelia rakudo-moar be141b: OUTPUT«-> { #`(Block|59519208) ... }␤»
lizmat ShimmerFairy: BUILD is fine
jnthn That's one of the very odd consequences
StrDistance should be a type object. 13:45
ShimmerFairy lizmat: isn't it just the same as standard BUILD, though? (I agree nothing is wrong with it, just odd that it's there ☺)
lizmat jnthn: agree :-)
ShimmerFairy: actually, you're right
psch ShimmerFairy: :$!before is coerced to Str
lizmat ah, yes
psch ShimmerFairy: standard BUILD doesn't coerce
ShimmerFairy psch: and after isn't ?
dalek kudo/nom: 7116059 | timotimo++ | src/Perl6/World.nqp:
don't give capture parameters a scalar container

spec tests are fine, suggesting that there's never a need for a capture parameter to have a scalar container to live in.		 13:46
timotimo ^- please notice my commit, senpai :3
psch ShimmerFairy: no, after is the result of Str.trans, which is Str
just stating reasons for why i wrote it as it is, no idea where the Block weirdness comes from either... 13:47
jnthn timotimo: Looks OK
psch although ISTR that &trans had Str as return type..?
timotimo jnthn: i noticed this when a piece of code using SDL2::Raw allocated most of its scalars inside CALL-ME 13:48
and i'm still wondering why CALL-ME doesn't end up getting speshed at all :\
dalek ecs: 3cd3789 | lizmat++ | S11-modules.pod:
Introduce "api" and :api
 13:49
lizmat FROGGS: this implies that if a distribution does not specify any API, it will revert to the behaviour you want 13:50
if a distribution *does* specify an API, and the selected compunits have a different API, it will fail (which would be the behaviour *I* want)
I think this is the best of both worlds, thanks ShimmerFairy for keeping coughing :-) 13:51
ShimmerFairy++ # coughing
ShimmerFairy lizmat: I'm just sharing my portage-addled brain :P
lizmat and we thank you for that! 13:52
ShimmerFairy The other Neat Idea™ I can think of would be "virtuals", that let you select between different satisfactory implementations (e.g. Oracle or IcedTea JVM), but I don't think we need a special concept for that, at least not yet :)
lizmat I think that is covered by "emulates" 13:53
really cycling&
ShimmerFairy ah, yes, I thought that sounded familiar :)
bye lizmat o/
13:54 zakharyas joined
Ulti m: say -42.abs + -42 . abs 13:55
camelia rakudo-moar be141b: OUTPUT«0␤»
Ulti ;__;
ShimmerFairy I think it's cool we have a lower-precedence methodop (though TimToady++ did mention something about infix:<.> and .= having too low a precedence atm) 13:56
Ulti m: say -42.abs + -42 .abs #also 13:57
camelia rakudo-moar be141b: OUTPUT«0␤»
Ulti thats one space different semantically super close but does something different and its the - which is the issue really not the new method magic
ilmari m: say -42.abs + -42.abs 13:58
camelia rakudo-moar be141b: OUTPUT«-84␤»
Zoffix :S
ShimmerFairy Ulti: the issue is that the prefix isn't interpreted as part of a numeric literal (and I think it'd be weird if it was, it'd be the only numeric literal thingy that has an op version too) 13:59
Ulti yes I know
Zoffix m: say (-42).abs
camelia rakudo-moar be141b: OUTPUT«42␤»
Ulti I dont think it would be weird at all that prefix - with no space was always seen as the literal numerically
timotimo but but but custom operators 14:00
Ulti I think that is literally what programmer on the planet would expect, literal pun intended
*every
ShimmerFairy Ulti: I personally wouldn't mind making mathematicians cry over -4 ** 2 -> 16 :)
timotimo ShimmerFairy: before adding $¢ to the optimizer, there were 4229 lines matching \d+: .*$¢, now there are 3465
i have no clue how many of these are "legitimate", but ... 14:01
oh, this is the core setting
pink_mist why would they cry over that? maybe I'm just not a mathematician so I don't get it :>
Zoffix m: say -4 ** 2
camelia rakudo-moar be141b: OUTPUT«-16␤»
Zoffix :)
ShimmerFairy pink_mist: in math, prefix:<-> has looser precedence than exponentiation (makes a tiny bit more sense written as -4², but still)
pink_mist oh I see 14:02
ilmari m: say -4²
camelia rakudo-moar be141b: OUTPUT«5===SORRY!5=== Error while compiling /tmp/srmhsIzfLF␤Bogus postfix␤at /tmp/srmhsIzfLF:1␤------> 3say -47⏏5²␤ expecting any of:␤ infix␤ infix stopper␤ postfix␤ statement end␤ statement modif…»
ilmari is disappoint
timotimo hah
jnthn "There will be modules" :P
timotimo m: sub postfix:<²>($a) { $a ** 2 }; say 100² 14:03
camelia rakudo-moar be141b: OUTPUT«10000␤»
pink_mist ShimmerFairy++ thanks for explaining =)
Zoffix wow, that's so cool.
ShimmerFairy jnthn: that's gonna be the weirdest thing for me, I think. Getting used to the idea that language improvements will come from modules rather than additions to core :)
timotimo you're susprised by that? you need to pay more attention in #perl6 it seems ;) ;)
pmurias lizmat: isn't having an exact set of dependencies how we should deploy our apps to production?
pink_mist m: sub postfix:<²>($a) { $a ** 2 }; say 100²² #I wonder... 14:04
camelia rakudo-moar be141b: OUTPUT«100000000␤»
pink_mist damn, that wasn't interpreted as ^22 :P
ShimmerFairy You'd need a slang of some kind to generalize superscripts :)
timotimo LTM would allow you to build a ²² postfix
14:05 mn080202 joined
Ulti I'm not entirely sure why anyone has a problem with - being part of a numeric literal given we have . already -___- 14:05
. is now all sorts of things!
jnthn ShimmerFairy: I'm really glad of that, in fact. It means whenever people want to propose things for core, we'll be able to say "do it as a module and see how it works out", and bring in the things that see widespread and unproblemtic use
ShimmerFairy Ulti: I certainly don't, I perceive it as literal integer syntax.
timotimo very yes.
ShimmerFairy: the question now becomes: why don't we accept - as literal integer syntax, too?
ShimmerFairy jnthn: sure, I'm not saying it's a bad thing, just weird for me :) 14:06
timotimo: to make the mathemagicians happy, I guess. Also, what about my $a = 4; say -$a ** 2
Ulti I dont want to have to use NegativeNumbers :P
jnthn ShimmerFairy: I guess it's 'cus very few languages are in a place to make that happen :)
mn080202 hey guys if I mod a library how do I recompile the module so it picks up the changes?
with latest rakudo panda
timotimo mn080202: panda doesn't do pre-compilation any more, so you can just re-install the new version and everything will work out immediately 14:07
mn080202 and if I mod it in the filesystem? it should pick it up?
timotimo yes, but you shouldn't do that :)
oh, wait
ShimmerFairy jnthn: I just put up a repo (need to add it to the ecosystem) that offers error reporting within grammars, with a hope that it could someday be core functionality, for example :)
timotimo what exactly do you mean by "in the filesystem"?
mn080202 hm I am trying to debug HTTP::UserAgent it is failing so I found the module where it is installed 14:08
and tried to insert debug statements
and it does not print stuff so I thought it most probably compiles it since I saw that output on install
Compiling lib/HTTP/Header/Field.pm6 to mbc 14:09
timotimo huh? compiling to mbc?
is that really latest panda?
mn080202 I tried to follow tuts online I will check to find something newer 14:10
14:11 FROGGS_ joined
timotimo the easiest thing to do is use "rakudobrew", to be honest 14:11
14:12 emilbarton joined 14:15 FROGGS left
pmurias lizmat: re using a use without a version numer one thing I often want to use when developing code is to just use the latest thing from CPAN 14:15
lizmat: releasing it is the moment I would like to have a tool go and write down the list of things my module requires 14:16
mn080202 well I tried straight from github and before I tried rakudobrew but again it compiles I will try rakudobrew again 14:18
14:19 emilbarton left
Ulti m/win 2 14:22
mn080202 how do you remove modules with panda? 14:27
14:33 emilbarton joined
dalek kudo/nom: 08d9122 | timotimo++ | src/Perl6/Optimizer.nqp:
throw out unused $¢

those were added to subs and methods in a recent-ish patch, but aren't actually used terribly often.		 14:33
timotimo mn080202: that's not yet implemented :\ 14:34
mn080202 huh thanks timotimo I finally managed to get HTTP::UserAgent to scrape a simple web page 14:39
timotimo phew
mn080202 used rakudobrew to install everything
timotimo that's good; rakudobrew can install panda for you, too, but i expect you've noticed
was there a problem inside HTTP::UserAgent?
mn080202 no there does not seem to be one what I did in the begining was I used rakudobrew then I installed rakudo 2015.02 then 2015.07 after removing all rakudo.xy.tar.gz and using rakudobrew again it worked 14:40
timotimo ShimmerFairy: would you be interested in looking at a moar --dump of the core setting with that latest commit applied and seeing if there's any pattern to $¢ being unnecessary, but the optimizer not having a clue?
maybe the moar --dump is less helpful than a --target=optimize, actually 14:41
ShimmerFairy sure, but not now, it's bed for me :) 14:42
14:42 tokuhiro_ joined
timotimo ah 14:43
good night :)
14:44 yqt left 14:47 tokuhiro_ left 14:49 nbd_ left 14:55 Akagi201 left 14:56 Akagi201 joined 15:00 Akagi201 left 15:05 spebern joined 15:15 spebern left 15:17 Akagi201 joined 15:18 xinming left
dalek kudo/nom: 2b04b50 | TimToady++ | src/Perl6/Grammar.nqp:
fake up intermediate precedence for . and .=

until we straighten out the associativity of unaries and infixes sharing the same precedence level....		 15:27
vendethiel TimToady: I wanna say, thanks a lot for this. being able to line up calls nicely is a great thing, imho :) 15:35
15:36 zakharyas left 15:44 tokuhiro_ joined 15:49 Akagi201 left, tokuhiro_ left, xinming joined, Akagi201 joined 15:58 Loren left, steelwilly joined, nwc10 joined
nwc10 good Birthday, TimToady 16:01
yoleaux 25 Sep 2015 13:33Z <tony-o> nwc10: there is a perl6 'prove' module/bin out there called 'Green', can be installed with zef or panda.
nwc10 good UGT, #perl6 16:02
16:06 jasanj joined 16:07 steelwilly left 16:09 cognominal left, Akagi201 left
moritz m: my @a = 1, 2, 3; for @a.item { .say } 16:10
camelia rakudo-moar 2b04b5: OUTPUT«[1 2 3]␤»
nwc10 it looks like if one creates tickets directly in RT, they don't hit the list 16:12
moritz m: say <a b c>.kv.perl 16:19
camelia rakudo-moar 2b04b5: OUTPUT«(0, "a", 1, "b", 2, "c").Seq␤»
moritz m: say <a b c>.pairs.perl
camelia rakudo-moar 2b04b5: OUTPUT«(0 => "a", 1 => "b", 2 => "c").Seq␤»
moritz m: say join '|', <a b>, <d e> 16:20
camelia rakudo-moar 2b04b5: OUTPUT«a|b|d|e␤»
moritz m: say join '|', <a b>, $<d e>
camelia rakudo-moar 2b04b5: OUTPUT«Use of uninitialized value <element> of type Any in string context␤Any of .^name, .perl, .gist, or .say can stringify undefined things, if needed. in block <unit> at /tmp/fVx5bYkx37:1␤Use of uninitialized value <element> of type Any in string context…»
moritz m: say join '|', <a b>, <d e>.item
camelia rakudo-moar 2b04b5: OUTPUT«a|b|d e␤»
moritz m: sub fe(*@flat) { @flat.elems }; say fe(<a b>, <d e>.item); 16:22
camelia rakudo-moar 2b04b5: OUTPUT«3␤»
vendethiel mh, maybe someone will expect $<foo> to itemize :P
moritz it's a rather obscure use case 16:23
interesting for explaining list mechanics, but seldomly useful in real-world code
timotimo nwc10: how do you actually create a ticket in rt? 16:25
moritz m: say <a b c d e>.pick: 3; 16:26
camelia rakudo-moar 2b04b5: OUTPUT«(c a b)␤»
timotimo i was under the impression that's impossible
moritz timotimo: mailto:rakudobug@perl.org
timotimo yeah
that's not "directly in rt", that's "in your mailclient"
moritz no, you write your mail in your MUA, but it becomes a ticket inside RT 16:28
sorry for being a nuisance; I maintain a fork of a ticket system for $work :/
m: say <a b c>.rotate(1).^name 16:29
camelia rakudo-moar 2b04b5: OUTPUT«List␤»
moritz m: say <a b c>.reverse.^name
camelia rakudo-moar 2b04b5: OUTPUT«List␤»
nine TimToady: Happy birthday! 16:32
moritz TimToady++ # one day a year the aniversary brings free karma 16:34
m: my @a = 1, 2, 3; say @a.^splice(0, 2).^name 16:35
camelia rakudo-moar 2b04b5: OUTPUT«Method 'splice' not found for invocant of class 'Perl6::Metamodel::ClassHOW+{<anon>}'␤ in block <unit> at /tmp/cDO9K5yu_Z:1␤␤»
moritz m: my @a = 1, 2, 3; say @a.splice(0, 2).^name
camelia rakudo-moar 2b04b5: OUTPUT«Array␤»
moritz m: my @a = 1, 2, 3; say @a.rotor(2 => -1).^name 16:37
camelia rakudo-moar 2b04b5: OUTPUT«Seq␤»
moritz m: say (<a b c> Z <d e> Z (1, 2)).perl 16:40
camelia rakudo-moar 2b04b5: OUTPUT«(("a", "d", 1), ("b", "e", 2)).Seq␤»
moritz m: say .join for zip <a b c>, <d e f> 16:41
camelia rakudo-moar 2b04b5: OUTPUT«ad␤be␤cf␤»
moritz m: say .join for zip <a b c>, ('d', *) 16:43
camelia rakudo-moar 2b04b5: OUTPUT«ad␤bWhatever<139937550545840>␤»
timotimo bwhatever 16:47
b whatever u want
dalek c: 645466d | moritz++ | lib/Type/List.pod:
GLRify List.pod
 16:48
moritz why is zef no longer in the modules list? 16:50
raydiak good morning, #perl6 16:52
sergot m: my @a = <1 2>; my @b = <3 4>; @a.push: @a,;
camelia ( no output )
sergot m: my @a = <1 2>; my @b = <3 4>; @a.push: @b,; say @a; 16:53
camelia rakudo-moar 2b04b5: OUTPUT«[1 2 3 4]␤»
sergot m: my @a = <1 2>; my @b = <3 4>; @a.push: @b,; say @a.perl;
camelia rakudo-moar 2b04b5: OUTPUT«[IntStr.new(1, "1"), IntStr.new(2, "2"), IntStr.new(3, "3"), IntStr.new(4, "4")]␤»
sergot m: my @a = <1 2>; my @b = <3 4>; @a.push: @b, ; say @a.perl;
camelia rakudo-moar 2b04b5: OUTPUT«[IntStr.new(1, "1"), IntStr.new(2, "2"), IntStr.new(3, "3"), IntStr.new(4, "4")]␤»
moritz hopes this is a bug
sergot hmm 16:54
m: my @a = <1 2>; my @b = <3 4>; @a.push: @b; say @a;
camelia rakudo-moar 2b04b5: OUTPUT«[1 2 3 4]␤»
sergot , doesn't matter here
16:56 emilbarton left
jasanj m: say 1.WAT 16:56
camelia rakudo-moar 2b04b5: OUTPUT«Method 'WAT' not found for invocant of class 'Int'␤ in block <unit> at /tmp/xXai9V6HAq:1␤␤»
jasanj m: say 1.WHAT 16:57
camelia rakudo-moar 2b04b5: OUTPUT«(Int)␤»
jasanj m: say 1.WHICH
camelia rakudo-moar 2b04b5: OUTPUT«Int|1␤»
jasanj m: say 1.WHO
camelia rakudo-moar 2b04b5: OUTPUT«␤»
jasanj m: say 1.WHERE 16:58
camelia rakudo-moar 2b04b5: OUTPUT«139948576800912␤»
jasanj m: say 1.HOW
camelia rakudo-moar 2b04b5: OUTPUT«Perl6::Metamodel::ClassHOW.new␤»
jasanj m: say 1.WHY
camelia rakudo-moar 2b04b5: OUTPUT«(Any)␤»
TimToady maybe .WHY should default to "Mu" :) 17:01
moritz TimToady: or to class Dunno is Mu { } 17:02
17:03 skids joined
TimToady or Cool :) 17:03
17:05 vendethiel left
moritz m: my @a; my @b = <a b >; @a.push: @b,; say @a.elems 17:06
camelia rakudo-moar 2b04b5: OUTPUT«2␤»
moritz TimToady: would you agree this is a bug?
m: my @a; my @b = <a b >; @a = @b,; say @a.elems
camelia rakudo-moar 2b04b5: OUTPUT«1␤»
moritz seems inconsistent with array assignment, at least
17:06 MadcapJake joined
[Tux] TimToady, belated congrats from the Netherlands 17:14
TimToady moritz: as jnthn pointed out a day or so ago, arglist commas are different, and we don't really want to encourage that notation anyway 17:24
that's why we have $@b and $[1,2,3] 17:26
m: my @a = <1 2>; my @b = <3 4>; @a.push: item @b; say @a;
camelia rakudo-moar 2b04b5: OUTPUT«[1 2 [3 4]]␤»
TimToady or that
17:29 rindolf left
nwc10 timotimo: one needs to be logged in, and have suffienct queue privs 17:31
timotimo oh 17:32
interesting
hey TimToady, happy birthday!
17:36 rindolf joined 17:45 tokuhiro_ joined 17:47 mr_ron joined 17:50 tokuhiro_ left 17:59 thou joined 18:13 cognominal joined
nwc10 anyway, the ticket in question is rt.perl.org/Ticket/Display.html?id=126195 18:15
It has a patch, and all :-)
dalek kudo/nom: 3813a26 | nicholas++ | t/harness:
Only enable 'trap' for new enough harnesses.

The 'trap' option was added to TAP::Harness version 3.22. Attempting to set it on earlier versions is a fatal error, and aborts testing. Hence pass argument conditionally, based on $TAP::Harness::VERSION.
Signed-off-by: Moritz Lenz [email@hidden.address]		 18:19
nwc10 moritz++ 18:20
thanks
jnthn
.oO( If only things ignored named options they didn't understand, we'd not have this problem :P )
 18:22
FROGGS_ heh 18:23
nwc10 Thermal::Detonator.explode(dry_run => 1); 18:25
dry_run was added in version 3.
there are arguments both ways :-)
pink_mist . o O ( and then there are those who refer to code that aborts as 'exploding' ) 18:28
lizmat is back
18:32 yqt joined 18:33 patrickz joined 18:36 spider-mario left
dalek p: e1cc288 | (Pawel Murias)++ | src/vm/js/nqp-runtime/bootstrap.js:
[js] fix bootstraping bug
 18:37
p: 3f2ad11 | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (2 files):
[js] serialization of simple STables
18:39 tokuhiro_ joined 18:44 tokuhiro_ left, Ven joined
pmurias lizmat: re silently allowing version differences up to patch level, aren't a lot of the breakages caused by accident instead of concious api changes? 18:47
lizmat yes, they are 18:48
that is why *I* think it is best to be as strict in your -use- statements as possible
jdv79 or desired 18:49
pmurias for deploying applications having an exact list of tested dependencies seems the sanest option 18:50
having cpan modules have exact deps would cause some problems 18:51
18:52 Ven left
pmurias like having your up load a bunch of old Moose (replace with any common dependency) versions 18:52
lizmat pmurias: that's why we have "supersedes" 18:53
pmurias isn't having "supersedes" everywhere the same thing as assuming that higher versions supersed older ones unless marked as breaking 18:55
lizmat you can indicate that a compunit is superseding another (range of) compunits: this is very specific 18:56
if Foo 1.2 has a supersedes Foo <1.2, then any attempt to select a Foo < 1,2 will result in selecting Foo 1.2 18:58
jdv79 isn't emulates teh same thing as this new api tag thing?
pmurias api tag thing?
lizmat emulates is when you ask for Foo 1.2 and that is not available, but there is a Bar 3.4 that emulates Foo 1.2, *then* Bar will be selected 18:59
if there *is* a Foo 1.2, then that will be used
jdv79 k
pmurias: i mean the, as of a few hours ago, addition of an api specifier 19:00
see S11
pmurias lizmat: marking a module as using semantic versioning and specifing a version would work nearly the same as supersedes 19:03
lizmat accept with supersedes you can actually list which compunits you're superseding 19:04
19:04 Ven joined
lizmat if Foo 1.2, 1.5 and 1.7 had serious issues, you can make Foo 1.8 supersede only *those* versions 19:04
Ven is there a "code index" for the advent calendar? to update all the code samples for the post-GLR world, say. 19:10
moritz Ven: well, some of them were made into tests in roast 19:15
masak Ven: the only thing like that (that I know about) is the spectests that have been adapted from the advent calendar.
yoleaux 11:23Z <nine> masak: the visitor macros are something that I incidentally though about before falling asleep yesterday. Seems like something really useful for implementing something like LINQ and even more, for making Rakudo's optimizer readable.
Ven right, those I'm aware of
masak nine: nice :)
nine: if you have time to talk about them at some point, I'd be happy to 19:16
19:16 nwc10 left
Ven Linq, for me, falls into the other scope. Vmacros would only be useful for analysis and modifying-slang-ish things. 19:17
mmh, or should it... well, depends which kind of stuff we can catch
masak yeah, LINQ feels more slang-y to me as well. 19:18
basically visitor macros do not change the syntax of the language, though they can both analyze and modify the resulting Qtrees if they want. 19:19
Ven masak: which category does timotimo++'s ADT module fall into? 19:22
AlexDaniel “New slurpy parameter, +args or +@args, to allow for one-argument style binding” 19:23
ooooooooh
If I got that right, there are some of early RTs waiting for that 19:25
in fact, some of them were fixed and should now be reopened to use that new thing
some of my early reports* 19:26
m: say 2 + Ⅵ;
camelia rakudo-moar 3813a2: OUTPUT«5===SORRY!5=== Error while compiling /tmp/bPb40erFJ9␤Missing required term after infix␤at /tmp/bPb40erFJ9:1␤------> 3say 2 +7⏏5 Ⅵ;␤ expecting any of:␤ prefix␤ term␤»
masak Ven: you mean the thing inside the q{} in github.com/timo/ADT/blob/master/README.md 19:27
...?
that's clearly a slang to me.
Ven masak: well, the idea would be to have something akin to "ADT foo = ..."
19:27 philh joined
masak well, then the `ADT` bit would be the slang "opener". 19:27
philh p6: say ((1,2),3).flat 19:28
camelia rakudo-moar 3813a2: OUTPUT«(1 2 3)␤»
philh p6: say [(1,2),3].flat
camelia rakudo-moar 3813a2: OUTPUT«((1 2) 3)␤»
pmurias lizmat: I'm not sure I see a use for the api modifier, if we have sister versions like Perl 6 and Perl 5 wouldn't it be better to have the number as part of the name?
philh Should .flat on an Array be doing that? 19:29
19:29 apotheon left
lizmat pmurias: for that we have :from<perl5> 19:29
19:29 apotheon joined
lizmat which, BTW, already works if you have Inline::Perl5 installed 19:29
19:30 cognominal left, diana_olhovik_ left, cognominal joined 19:31 cognominal left
masak m: say [[1, 2], 3].flat 19:31
camelia rakudo-moar 3813a2: OUTPUT«([1 2] 3)␤»
pmurias lizmat: what's the intended use of the :api modifier?
19:32 mr_ron left
masak philh: seems suspect to me. does Array.flat only convert the outermost [] to a () ? seems a bit surprising. 19:32
jnthn Array.flat is a no-op 'cus an Array is by definition already flat 19:33
philh masak: well, it seems to return a Seq rather than a list. But yeah, it doesn't seem to be actually flattening.
lizmat if a module author indicates that a certain version of a compunit adheres to a certain api, then a CURLI can automatically allow/disallow selecting a higher numbered version of a module in the case where multiple compunit match a give -use- statement specification
jnthn (Because an Array is a bunch of Scalar containers)
So, working as designed.
lizmat jnthn: perhaps Array.flat should warn then ?
jnthn lizmat: But flat is recursive, so then for flat @a, @b { } would warn. 19:34
Which surely isn't desirable.
lizmat that would be the sub case, I was talking about the Array.flat method
philh p6: say [(1,2),3].list.flat
camelia rakudo-moar 3813a2: OUTPUT«((1 2) 3)␤»
jnthn lizmat: The sub only delegates to the .flat method 19:35
philh So is there a simple way to get the flattening behaviour on an array?
lizmat ah, eh, ok
jnthn m: say flat [[1, 2], 3]>>[] 19:36
camelia rakudo-moar 3813a2: OUTPUT«(1 2 3)␤»
pmurias lizmat: isn't that something that can be done with supersedes?
jdv79 pmurias: the api mod (in backlog) is a solution to bridging the gap between the strict and lax camps.
jnthn (That is, apply the zen slice over the data structure)
TimToady m: say gather [[1,2],3].deepmap: *.take
camelia rakudo-moar 3813a2: OUTPUT«(1 2 3)␤»
jnthn This works any number of levels deep
m: say flat [[1, 2], [3, [4, 5]]]>>[] 19:37
camelia rakudo-moar 3813a2: OUTPUT«(1 2 3 [4 5])␤»
jnthn oh? :)
Oh, nodal...not deep
The TimToady one works any number of levels deep, I guess.
timotimo Ven: i'd call my ADT module just "code gen" 19:39
Ven timotimo: hey, better syntax is always nice ;-)
TimToady lunch & 19:41
pmurias jdv79: I have read through todays backlog but it seems to be just an attempt to port stuff from gentoo 19:42
19:42 dha joined
masak "an array is already flat" -- interesting. I can see how that is true, but it's something I've yet to internalize. 19:43
philh I'm trying to look up the >>[] thing but can't find >>, could I get a pointer?
19:44 japhb joined
jnthn philh: Hyper operator 19:44
pmurias jdv79: with the use case being binding to different incompatible versions of libraries
jnthn m: my @a = 1, 2, 3; @a>>++; say @a;
camelia rakudo-moar 3813a2: OUTPUT«[2 3 4]␤»
dha You may have to search for 􏿽xAB
japhb .tell TimToady Happy Birthday!
yoleaux japhb: I'll pass your message to TimToady.
jnthn That's another example
19:44 Ven left
dha sorry, 􏿽xBB 19:44
japhb .tell masak three-types-of-macros.md looks very promising. How much of that do you have working in 007 at this point? 19:45
yoleaux japhb: I'll pass your message to masak.
19:45 dha left 19:46 mr_ron joined
philh jnthn: Ah, thanks. So the [...]>>[] is like ( [1,2][], 3[] ) i.e. ( [1,2], 3 ) which flattens accordingly. Got it. 19:47
japhb What's the proper post-GLR way to map over an array, producing a Pair for each item in the original array in which the Pair's key is a (computed) string and the Pair's value is an array of two values, and assigning that to another array, so that the end result is [ key1 => [ a, b ], key2 => [ c, d], ]? 19:48
I keep trying to fix this pre-GLR code in front of me with cargo culting, and (no surprise) that isn't actually working for me.
masak m: my @a = 1..6; say [@a.map(-> $x, $y { "key{ ++$ }" => [$x, $y] })].perl 19:49
yoleaux 19:45Z <japhb> masak: three-types-of-macros.md looks very promising. How much of that do you have working in 007 at this point?
camelia rakudo-moar 3813a2: OUTPUT«[:key1([1, 2]), :key2([3, 4]), :key3([5, 6])]␤»
moritz m: say <a b c>.map({; ($_ x 2) => [ $_ xx 2] }).perl
camelia rakudo-moar 3813a2: OUTPUT«(:aa(["a", "a"]), :bb(["b", "b"]), :cc(["c", "c"])).Seq␤»
masak japhb: ^
japhb: in 007 (just as in Rakudo), the first of the three is implemented. just about to dive into the latter two. I don't think 007 does hygiene properly yet, but working on that. 19:51
jnthn masak: superstitious ; ?
japhb moritz was closer to what I meant, but I'll file both of those in my head for later. Thanks masak++, moritz++
19:51 spider-mario joined
jnthn Oh, moritz wrote the superstitious ; 19:51
masak jnthn: did you mean "moritz"? :P
jnthn :)
japhb jnthn: You mean moritz?
jnthn You know, guys, I might just have meant moritz! :P
japhb Heh
masak perhaps you meant that other guy.
jnthn m: say <a b c>.map({ ($_ x 2) => [ $_ xx 2] }).perl 19:52
camelia rakudo-moar 3813a2: OUTPUT«(:aa(["a", "a"]), :bb(["b", "b"]), :cc(["c", "c"])).Seq␤»
masak .oO( superstitious "masak:" ) :P
japhb Although now I'm really wondering what I didn't get right before.
masak japhb: yah, I got it in one! :P
jnthn The use of $_ implies it's a Block (though we hadn't used to get that one right)
moritz is stuck in his "superstitious" past 19:54
japhb moritz: Yeah, I have a lot of them to unlearn after all these years.
19:55 Ven joined
japhb Especially since some of the bugs still exist, and some only got fixed in the last week. 19:55
dalek ast: 9d54b21 | usev6++ | S12- (14 files):
Replace some eval-dies-ok in S12-* with throws-like
 20:00
20:01 colomon left 20:03 darutoko left 20:04 rurban joined 20:05 Ben_Goldberg joined, _dolmen_ joined 20:08 BenGoldberg left
jnthn back tomorrow, with tuits :) 20:08
'night
timotimo gnite jnthn :) 20:09
20:10 AlexDani` joined, colomon joined 20:11 AlexDaniel left
masak I have that "stuck in the past" thing, too. 20:13
I solve it by jealously observing newcomers quickly arrive at clean, un-superstitious solutions, and then telling them how horrible things used to be :P 20:14
Ven so, how should I call an external program with output capturing *and* parameter escaping? seems `run` is not the one droid I'm looking for 20:16
sergot #perl6: don't forget about the lunar eclipse this night! :)
japhb Ven: What kind of parameter escaping are you looking for? run doesn't need shell escaping, since it skips the shell ... 20:18
(Well, at least on OSen that have sane spawn semantics)
masak saw "parameter escaping", assumed the wrong context (escape analysis) and went "wuh, parameters trivially escape, always!" :P 20:19
Ven japhb: true. I'm never clear enough... I want to invoke a program with an escaped strng 20:20
20:21 patrickz left
timotimo so you need to unescape the escaping? 20:22
20:23 perler joined 20:25 colomon left 20:26 perler left 20:27 cognominal joined, colomon joined 20:28 pmurias left 20:30 dha joined 20:32 Ven left 20:33 Ven joined, mr_ron left 20:36 kaare__ left 20:37 vendethiel joined 20:38 mr_ron joined
lizmat has been working on the next issue of P6W: gist.github.com/lizmat/28d6d9b8f3a968e6c92e 20:40
please let me know if I missed any videos, blogs, slides :-)
20:41 tokuhiro_ joined, _dolmen_ left
timotimo good work, lizmat! 20:43
20:45 tokuhiro_ left
timotimo lizmat: we don't have any clue how to make .min and .max faster, right? 20:47
masak lizmat++ # working on p6w
dha lizmat++ # just generally.
I probably won't have a blog post up until I get home. Sorry. 20:48
lizmat dha: I won't publish until tomorrow evening, in about 24 hours
I've been only combing the net for videos, blogs, slides, ecosystem additions so far
dha Ok. I'm flying home now, so maybe...
*nod* 20:49
lizmat timotimo: we could actually make it faster, I think
timotimo good flight, dha!
lizmat I mean, why do we need the map at all ? 20:50
timotimo yeah
dha I'm considering attempting to corral the documentation efforts into something more coherent while we still have a couple of months before the release.
timotimo well, maybe that's the cleverest way to pull out all values from the iterator?
dha thanks timotimo!
Ven ++lizmat++ # YAY!
lizmat timotimo: no, because map needs to take care of the case where it returns more than one value in an iteration
timotimo oooh 20:51
of course
lizmat also, it's generating a list there
when it is immediately discarded
timotimo so, what is the other way we can do it? get the .iterator and .shift off of it?
timotimo not yet very good at the new GLR internals api
lizmat lemme see if I can whip up a proto type 20:52
need some time of from blogging :-)
timotimo oh, glad to be supplying something easy-ish for you :)
lizmat: min accounts for 20% inclusive time on this one task i have :) 20:53
in the mean time i'll work towards making this thing a more proper benchmark 20:55
with reproducible decisions whether or not to do work at certain places - which are currently probabilistic
20:57 rurban left, espadrine left
timotimo very good, it's reproducible now 20:59
dha Hm. Is p6doc broken at all? 21:02
"===SORRY!===\nCannot invoke this object (REPR: Uninstantiable)"
lizmat dha: I have no idea :-(
wow
dha That's when I try to read docs for an installed module. Maybe it's not documented... 21:03
21:03 apotheon left
lizmat no, that's a deep compile error 21:03
21:04 renormalist left, renormalist joined
lizmat timotimoL would 2x as fast do ? 21:04
21:04 Ven left
lizmat timotimo: would 2x as fast do ? 21:04
timotimo hm 21:05
dha Ironically, I'm looking to read the docs on Benchmark. :-)
timotimo well, i'd prefer 20x faster, but that's probably beyond the point of diminishing returns ;)
just kidding, i'd love 2x as fast!
lizmat lemme check some more..
timotimo my min was called 213668 times and took 21739.4ms inclusive time 21:06
172267.72ms is the complete run time of the program
(an amazing 37% gc time) 21:07
21:08 apotheon joined
lizmat timotimo: do you need a "by" ? 21:09
dha Ah. regardless of p6doc's status, Benchmark isn't documented.
lizmat it's an ecosystem module, is it not ?
dha Dunno. I'll look.
timotimo i do not
21:10 Ven joined
timotimo this is a bunch of ints and a couple of Inf 21:10
Ven Also -- before it's too late, right? – Happy birthday, TimToady++
21:10 cognominal left
timotimo sorry, lizmat, i meant to highlight you. no :&by needed 21:11
lizmat 2.6 times is what I can do for you... :-) 21:13
checking spectest result before committing
21:14 rindolf left
timotimo hooray 21:14
lizmat if the spectest comes out ok, indeed :-)
masak .oO( conditional hooray! )
lizmat hmmm... some breakage... :-( looking at why in a mo 21:15
timotimo there's like three test files that b0rk 21:16
lizmat I have a few more now 21:17
but I also got a lot of roast updates...
Ven .u ⍵ 21:18
yoleaux U+2375 APL FUNCTIONAL SYMBOL OMEGA [So] (⍵)
Ven .u apl alpha 21:19
yoleaux No characters found
Ven .u apl functional symbol alpha 21:20
yoleaux U+2376 APL FUNCTIONAL SYMBOL ALPHA UNDERBAR [So] (⍶)
U+237A APL FUNCTIONAL SYMBOL ALPHA [So] (⍺)
masak APL: the first and the last programming language that got its own set of Unicode symbols. 21:21
Ben_Goldberg m: gist.github.com/BenGoldberg1/bc39a...aeb733012a 21:23
camelia rakudo-moar 3813a2: OUTPUT«Cannot send a message on a closed channel␤ in block <unit> at /tmp/T9gWZ7YfGJ:33␤␤»
lizmat timotimo: found an error in my logic, fixing now and re-testing
Ben_Goldberg Can anyone tell me why the error seems to come from line 33, which doesn't contain a .send ? 21:25
21:25 mr_ron left
masak Ben_Goldberg: maybe the channel closes before the logic in `start { ... }` has time to finish? 21:28
21:28 Ven left
masak confesses to not knowing any details of this 21:28
lizmat timotimo: my opt breaks (^10).pick(3).min :-( (This Seq has been consumed) 21:31
timotimo: gist.github.com/lizmat/44a6d97adf6ab4588e97 21:32
I thought I could get away by just using AT-POS 21:33
21:33 bakedb joined
lizmat probably need a full-fledged iterator class.. :-) 21:33
gfldex Ben_Goldberg: you may be running out of threads from the pool, unless that bug was fixed already. I am guessing as hard as masak tho. 21:34
21:35 colomon left
Ben_Goldberg Running out of threads from the pool should cause a deadlock, no? 21:35
gfldex you may be able to test is by limiting the number of threads to 10 21:36
21:36 colomon joined
Ben_Goldberg How do I do that? 21:37
gfldex change 1..20 to something much smaller
Ben_Goldberg m: gist.github.com/BenGoldberg1/bc39a...aeb733012a 21:38
camelia rakudo-moar 3813a2: OUTPUT«Attempt to return outside of any Routine␤ in block <unit> at /tmp/4c1CWudfwj:33␤␤»
Ben_Goldberg Oooh, different error now.
So it's a threads problem.
gfldex there are many bugs to choose from :) 21:39
21:42 dha left
lizmat timotimo: think I got it covered now 21:42
21:42 colomon left
gfldex Ben_Goldberg: it maybe wise to rakudobug your findings 21:42
21:46 mr_ron joined, colomon joined
dalek kudo/nom: 91b8b1e | lizmat++ | src/core/Any-iterable-methods.pm:
Make .min about 2.5x faster
 21:48
lizmat timotimo: ^^^ 21:49
21:51 FROGGS_ left 21:53 TEttinger joined 21:56 makapa joined 21:58 AlexDani` left
masak lizmat++ 22:01
'night, #perl6
timotimo lovely! 22:04
lizmat: i expect i can just copy-paste that into .max, too?
lizmat I'm working on that, but that caused build breakage somehow 22:05
22:05 dnmfarrell joined 22:08 makapa left
lizmat weird: I changed the code just a little bit, no logic changes, and now it compiles like a charm :-( 22:10
running spectest now, if clean, will push 22:11
22:14 telex left 22:15 sftp left 22:16 telex joined
dalek kudo/nom: 1ed1ea0 | lizmat++ | src/core/Any-iterable-methods.pm:
Make .max about 2.5x faster

Plus some refactoring on .min as well to reduce code bloat		 22:17
lizmat timotimo ^^^
22:18 sftp joined
lizmat wonders how timotimo's program runs now 22:21
timotimo right, gimme a bit 22:23
22:25 bjz left
timotimo a bunch fewer GC runs 22:27
22:27 vendethiel left
timotimo noticably less run time in total, too 22:27
m: say "the previous run took { 92149.34 / 82652.85 }x as much time for executing code" 22:28
camelia rakudo-moar 91b8b1: OUTPUT«the previous run took 1.11489610x as much time for executing code␤»
lizmat ? is that backwards ? 22:29
timotimo it was slower before
so it's faster now
lizmat ah, yes
well, I guess we could further optimize the @a.min and @a.max case 22:30
timotimo m: say "min used to take { 13226.42 / 6069.31 }x as much time"
camelia rakudo-moar 91b8b1: OUTPUT«min used to take 2.1792296x as much time␤»
timotimo not quite 2.5x faster, but it's still very welcome :)
lizmat my bench was: (^10000).list.min.say 22:32
*100000
22:33 RabidGravy left
lizmat anyways, glad to have been of service 22:33
timotimo \o/
lizmat gets some sleep before getting up to see a red moon
timotimo gnite lizmat :)
good luck with that moon
i'll be trying to catch it, too
22:34 mr_ron left
timotimo huh, i may have compared it to a version that lacked another patch, too 22:34
so the time difference might be less pronounced actually
22:41 Akagi201 joined 22:45 Akagi201 left 22:46 dnmfarrell left 22:52 leont joined 22:57 mr_ron joined 23:01 philh left 23:28 Khisanth joined 23:29 krunen left 23:43 krunen joined, tokuhiro_ joined 23:47 tokuhiro_ left 23:54 leont left

Powered by Cro and the Raku® Programming Language.

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

Thank you!