Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6-dev, 2016-09-12

| Channels | #perl6-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:04 Zoffix m:  say "!" R~ 10.base: 16
00:04 camelia rakudo-moar 04af57: OUTPUT«A!␤»
00:04 Zoffix Found a good use of `R` metaop :)
00:08 awwaiid I have a list of things and a predicate -- what's a good way to see if they all match the predicate? Would 'so is-prime(<2 5 7 17>.all)' be efficient, or is there a better lazy eval way?
00:09 Zoffix .oO( premature optimization is the root of all evil )
00:10 awwaiid yeah yeah. I switched to a loop already
00:10 awwaiid so nevermind :)
00:15 MasterDuke speaking of premature optimization i just read the transcript of an interview with Knuth, he's a fascinating guy
00:17 Zoffix He also said Perl 6 may have came in too late... Screw him, he's a traitor :P
00:28 tadzik a heathen! :P
01:06 awwaiid Knuth addressed Perl 6?
01:08 FROGGS_ joined #perl6-dev
01:15 Zoffix awwaiid, not really. It was a talk about what languages make it or break it or something and he said the time when the languages comes around is important and said something along the lines of "Perl 6... is it too late? I don't know"
01:16 awwaiid ah
01:17 awwaiid Zoffix: why you not in #perl6? check this out, https://gist.github.com/awwaiid/d55434447e5503d48a252bf2eff53313
01:18 Zoffix Too much traffic.
01:18 awwaiid ah
01:19 Zoffix So, I'm looking at 40 lines of code that replace 4 characters (.all) ?
01:19 Zoffix :)
01:20 Zoffix Also, you asked your original question here, in #perl6-dev, and now your comment in #perl6 is without context :)
01:20 awwaiid not that. The fun of a very core proof-of-concept of QuickCheck -- I'm looking into building something inspired by quickcheck and/or clojure/spec
01:20 Zoffix Oh never mind that comment is not about your thing
01:20 awwaiid yeah ... I didn't actually mean my earlier question in here :)
01:21 awwaiid well it is technically about my thing in that loop, but that's not important overall :)
01:21 Zoffix I've no idea what quickcheck is
01:21 awwaiid Just thought you might like this idea
01:21 awwaiid ahh
01:22 awwaiid it is predicate / random testing. The idea is if you can generate inputs to a function at random and can assert some result of the function (like 'sort' outputs a list where the first element is the smallest), then you do a reasonable sample of inputs, that you can get some easy and declarative testing
01:22 awwaiid generating inputs is fun in perl6 when you can inspect the signature and param types to start making them
01:23 awwaiid quickcheck is a haskell lib that has been ported to a bunch of other langs
01:23 awwaiid So after this basic probabilistic predicate assertion stuff, next is that WHEN you find a bad case you can then minimize it -- try to find the simplest input that defies the predicate
01:24 awwaiid which I haven't gotten to yet of course
01:33 MasterDuke sounds like fuzzing
01:34 awwaiid It is like fuzzing yes, with expected outcomes. Very popular in Haskellville because they get a lot of free input generation through the types
01:37 awwaiid for example, here in the xmonad test suite the authors declare that if you swap a window and then swap it back you get the original configuration. They leave it up to the quickcheck harness setup to generate some inputs to try that against. https://github.com/xmonad/xmonad/blob/master/tests/Properties/Swap.hs#L13
01:37 awwaiid Ends up magical is my understanding. But I haven't used it much myself, just read about it
01:40 geekosaur it's not really magical. for any type you can declare an instance of the typeclass Arbitrary that generates random values; you can use the typeclass machinery to help with that so in many cases the instance is just boilerplate (and there are packages that will auto-create instances using generics, iirc)
01:41 geekosaur the StackSet is parameterized specifically so we can use QuickCheck with non-X11 types to test its behavior, then use the X11 types (which are actually the same but it can't tell) to run it
01:45 geekosaur so you can have QC generate random StackSets, apply W.swap twice, and verify the result is the original StackSet
01:45 awwaiid indeed. In my current experiment I'm making a multi sub 'gen' that will dispatch based on the types, which will take the place of the typeclass Arbitrary
01:45 geekosaur note that you get to reuse the machinery for fields, so gen for something like StackSet just means calling gen on its elements and building a StackSet from them, and the elements in turn use gen on their elements, etc.
01:48 ilbot3 joined #perl6-dev
01:48 Topic for #perl6-dev is now Perl 6 language and compiler development 2.0 | Logs at http://irclog.perlgeek.de/perl6-dev/today
01:48 geekosaur no. data StackSet i l a sid sd = StackSet { current :: !(Screen i l a sid sd); visible :: [Screen i l a sid sd]; hidden :: [Workspace i l a]; floating :: Map a RationalRect }
01:49 geekosaur hm, I gues sin p6 that would be instance vars
01:49 geekosaur `has` vars
01:49 awwaiid ya
01:50 awwaiid I'm also looking at clojure/spec as a way to declare these things, kinda interesting
01:51 geekosaur (i = tag/workspace name, l = layout, a = window, sid = screen id, sd = screen detail. for testing you can make sd and l (), sid and a Int, i String)
02:19 Zoffix m: say +"70\x[308]93"
02:19 camelia rakudo-moar 04af57: OUTPUT«Cannot convert string to number: trailing characters after number in '7⏏0̈93' (indicated by ⏏)␤  in block <unit> at <tmp> line 1␤␤Actually thrown at:␤  in block <unit> at <tmp> line 1␤␤»
02:19 Zoffix m: say "70\x[308]93"
02:19 camelia rakudo-moar 04af57: OUTPUT«70̈93␤»
02:20 Zoffix Design question: Should the coercion to numeric ignore the diarhesis and give 7093? There's a couple of tickets that look easy to fix, once that decision is made.
02:20 Zoffix jnthn, pmichaud, TimToady ^ if you have time
02:22 Zoffix One thing we already ignore is surrounding whitespace; I'd think ignoring diahresis would make sense.
02:22 Zoffix m: say +"    7093    "
02:22 camelia rakudo-moar 04af57: OUTPUT«7093␤»
02:22 * Zoffix goes to bed; will read replies in the morning, if any
04:06 * TimToady can argue it both ways, but thinks the situation won't arise all that often, and votes for whichever is faster
04:21 ShimmerFairy I'd lean towards no; as an example, what would someone expect +"0.3̅" to do? 3/10 or 1/3 ? In that example, I'd think failing would be better than choosing a behavior silently that may not be desired.
04:27 TimToady that would be a slightly stronger argument if .base-repeating put out that format...
04:28 TimToady but it's a point
04:29 ShimmerFairy yeah, I'm not suggesting it's very reasonable to expect core Perl 6 to make that into 1/3, just that it could theoretically happen (and that it's not all that farfetched).
04:30 ShimmerFairy I think what makes diacritics different from surrounding whitespace is that the presence of a diacritic likely indicates an extra meaning we can't possibly guess (on average), whereas surrounding whitespace doesn't.
04:33 geekosaur base should fail, an ecosystem module could extend it
04:33 ShimmerFairy I suppose that to me, diacritics in the middle of what should be a numeric value indicates either an additional meaning you have to handle yourself, or that your data is (paritally) garbage :)
04:34 geekosaur likewise fwiw
04:40 * TimToady just wonders about someone trying to copy/paste something with incidental formatting, like strikeout
04:41 TimToady or underlining, because it was a fill-in-the-blank form
04:42 TimToady not that is likely to end up with composing underbars
04:42 TimToady *that that
04:43 TimToady but I could imagine an entry form that did it that way
04:45 TimToady otoh, .ords.chrs is not that hard for stripping composers
04:45 TimToady just that the programmer might not anticipate the need for such
04:47 ShimmerFairy Is there any guarantee that we could distinguish a purely "decorative" or "meaningless" diacritic from a "meaningful" one? That is, how can CORE be sure that an underlined number is never really some kind of super-rare notational thing?
04:48 TimToady ya pays yer money and takes yer choice...
04:48 TimToady sometimes you just lose...
04:48 TimToady so like I said, I can argue it both ways
04:49 TimToady the question here is whether the dwim is worth more on average than the wat
04:49 ShimmerFairy Right, I can see the pro argument being that at least our number system almost never uses any sort of "digit modifier" to indicate different meanings.
04:50 TimToady and I think at least you can say the wat is easy to explain
04:51 TimToady and we could conceivably special-case overbar
04:52 TimToady but again, I'd like to know whether one way was slower than the other
04:52 TimToady programs spend a lot of time converting strings to numbers and back
04:52 ShimmerFairy My inclination is that this situation isn't very common at all, so I'd need to see some concrete examples of why auto-diacritic-stripping would be helpful to have in CORE.
04:53 TimToady well, it might just be free, given how nqp::ordat works
04:54 ShimmerFairy I imagine the extra steps would _technically_ lead to more execution time, but I'm not certain it would be significantly so.
04:55 TimToady what extra steps?
04:56 ShimmerFairy TimToady: just the extra statements to ordat and so on.
05:00 TimToady it might well be that rejecting synthetics outright could be faster
05:59 brrt joined #perl6-dev
06:10 [Tux] This is Rakudo version 2016.08.1-163-g04af57c built on MoarVM version 2016.08-43-g3d04391
06:10 [Tux] csv-ip5xs        8.988
06:10 [Tux] test            16.534
06:10 [Tux] test-t           7.089
06:10 [Tux] csv-parser      17.076
06:57 cognominal Vendredi les décodeurs du monde nous parlent d'intox sur la santé de Clinton et le lundi de sa pneumonie :)  Malade, peut-etre à cause d'un baril de chlore laché par Assad, ca expliquerait tout :)
07:00 nine ??
07:02 cognominal oops; wrong channel
07:05 lizmat it does explain everything  :-)
07:11 dalek roast: c7dd8f0 | niner++ | S10-packages/precompilation.t:
07:11 dalek roast: Test if we pick up changes of a precompiled dependency's source file.
07:11 dalek roast: review: https://github.com/perl6/roast/commit/c7dd8f03ee
07:15 nine I can bring back the speedup reverted in commit 376b5f4ee210f36d1ec3d683592901705955604b.
07:15 nine But it would still not pick up changes done by --force installation of a module if the precomp file lives in a precomp store further up the chain.
07:16 nine Thing is, that we intended a short-name:ver:auth triplet to be unique and distros to be immutable from the very beginning. The optimization would rely on that.
07:18 nine I'm inclined to just give it a try and force people to adapt by increasing version numbers instead of --force installing.
07:18 nine What do you think?
07:22 lizmat I agree with you
07:25 dalek rakudo/nom: c170ebe | niner++ | src/core/CompUnit/PrecompilationRepository.pm:
07:25 dalek rakudo/nom: Speed up loading deps by only running .modified checks when needed
07:25 dalek rakudo/nom:
07:25 dalek rakudo/nom: $*REPO.id represents the contents of all repositories. We already check if
07:25 dalek rakudo/nom: that changed since we precompiled a module. If it is still the same, there
07:25 dalek rakudo/nom: should be no reason to check the .modified of the source file. If it changed,
07:25 dalek rakudo/nom: we need to check it as a change in our source file could be exactly what
07:25 dalek rakudo/nom: caused the different checksum.
07:25 dalek rakudo/nom: Note that this relies on the immutability of installed distros which we
07:25 dalek rakudo/nom: haven't enforced so far.
07:25 dalek rakudo/nom: Brings GTK::Simple's load time back down to 227ms.
07:25 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/c170ebe503
07:26 nine lizmat: ok, so let's give it a try :)
07:50 dalek rakudo/nom: 512b5e2 | lizmat++ | src/core/Rakudo/Internals.pm:
07:50 dalek rakudo/nom: Introducing Rakudo::Internals.SeqFromSeqs
07:50 dalek rakudo/nom:
07:50 dalek rakudo/nom: Create a Seq from a Seq of Seqs.  Also tune ListsFromSeq so it can
07:50 dalek rakudo/nom: handle empty List objects without $!reified
07:50 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/512b5e2542
07:50 dalek rakudo/nom: 1379175 | lizmat++ | src/core/List.pm:
07:50 dalek rakudo/nom: Make List.combinations about 4x faster
07:50 dalek rakudo/nom:
07:50 dalek rakudo/nom: - based on "my @a = <a b c d>; for ^500 { my @b = @a.combinations }"
07:50 dalek rakudo/nom: - using the new SeqFromSeqs functionality
07:50 dalek rakudo/nom: - create separate candidate for the combinations() case
07:50 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/137917541f
07:56 RabidGravy joined #perl6-dev
08:46 [Coke] Zoffix: I had a todo from ages ago that may be of interest to you: - update rakudo release announcement to solicit donations to the Perl 6 Core Dev Fund
08:48 DrForr You too can emulate Jimmy Wales :)
09:25 lizmat nine: t/spec/S10-packages/precompilation.t has become flappy for me
09:25 lizmat either test 47 fails, or 48, or all tests, or 4 other tests  :-(
09:25 lizmat (so far)
09:26 lizmat hmmm... seems 47 and 48 are alternating on subsequent runs
09:26 lizmat some cleanup issue ?
09:29 stmuk_ [Coke]: putting a Begging Letter in R* could be a good idea as well
09:30 stmuk_ does anyone know of any copy which could be reused? (just looking on the Perl Foundation site)
09:54 |Tux| joined #perl6-dev
10:04 arnsholt Do we have anyone working semi-regularly on NQP or Rakudo on JVM?
10:06 Zoffix psch used to, but I haven't seen them in a while.
10:06 Zoffix .seen psch
10:06 yoleaux2 I saw psch 4 Aug 2016 16:20Z in #perl6-dev: <psch> i mean, we probably could have infix:sym<.=> do a similar check to methodop, the latter being where we throw that error
10:06 Zoffix NeuralAnomaly, stats
10:06 NeuralAnomaly Zoffix, [✘] Next release will be in 4 days and 18 hours. Since last release, there are 48 new still-open tickets (0 unreviewed and 0 blockers) and 10 unreviewed commits. See http://perl6.fail/release/stats for details
10:07 Zoffix 🎺🎺🎺 REMINDER: The next Rakudo release will be this Saturday. Please review the ChangeLog to ensure the changes you made have been logged properly.
10:07 Zoffix New blog post "Perl 6 Core Hacking: Grammatical Babble": http://perl6.party/post/Perl-6-Core-Hacking-Grammatical-Babble
10:19 nine lizmat: right now precompilation.t fails 2 tests consistently on my laptop. But reverting my optimization doesn't change that :/
10:20 lizmat hmmm... :-(
10:26 nine lizmat: oh wait a sec. Those two failing tests should be TODOed. The test harness doesn't seem to understand that.
10:29 nine Or rather the Test module doesn't output the TODO markers
10:29 nine Despite the line reading: todo('no 6model parametrics interning yet'); is_run
10:31 Zoffix works fine here
10:36 Zoffix Just built from HEAD. Passes fine with and without PERL6_TEST_DIE_ON_FAIL turned on: https://gist.github.com/zoffixznet/fedf091fa93b8d7a725ef90642a29e67
10:42 nine Well maybe I should have run precompilation.rakudo.moar instead of precompilation.t...
10:45 Zoffix :)
10:47 nine lizmat: so far I've only once managed to get precompilation.rakudo.moar to fail due to "duplicate definition of symbol Test"
10:47 nine lizmat: RAKUDO_MODULE_DEBUG output may help
10:54 lizmat nine: will try in a mo
11:13 dalek rakudo/nom: 3c2451c | lizmat++ | src/core/List.pm:
11:13 dalek rakudo/nom: Make combinations(n,k) about 1.8x faster
11:13 dalek rakudo/nom:
11:13 dalek rakudo/nom: - rewrite using nqp ops
11:13 dalek rakudo/nom: - no JVM dependent code necessary anymore
11:13 dalek rakudo/nom: - don't use return for successful find
11:13 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/3c2451c57b
11:15 dalek rakudo/nom: 9d1a086 | lizmat++ | docs/ChangeLog:
11:15 dalek rakudo/nom: Mention the combinations improvements
11:15 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/9d1a086339
11:16 lizmat Method 'version' not found for invocant of class 'Needed'
11:16 lizmat in block <unit> at -e line 3
11:17 lizmat nine  ^^^
11:17 lizmat getting that consistently now
11:18 lizmat siesta&
11:19 nine lizmat: t/spec/S10-packages/precompilation.t:260 should overwrite t/spec/packages/RT128156/Needed.pm6 with a version that contains the "version" method
11:20 nine lizmat: can you confirm that it does so (it gets reverted in line 265)? Does by any chance adding a sleep 2; after that spurt help?
11:42 nine lizmat: now that I thought some more, the sleep should probably be before the spurt as we need to see a difference between the outdated precomp file's .modified and the source. Probably safest to add it before and after though.
12:00 [Coke] I could really use a new docker image based on the 2016.09 release (coming soon)
12:01 [Coke] Willing to offer a 2-beer bounty. :)
13:05 AlexDaniel joined #perl6-dev
13:57 [Coke] P6M Merging GLOBAL symbols failed: duplicate definition of symbol Test
13:57 * [Coke] does a realclean
14:08 dalek roast: 3ceb078 | coke++ | S07-iterators/range-iterator.t:
14:08 dalek roast: fudge for moar
14:08 dalek roast: review: https://github.com/perl6/roast/commit/3ceb078400
14:08 dalek rakudo/nom: 15532db | coke++ | t/spectest.data:
14:08 dalek rakudo/nom: run these (fudged) tests
14:08 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/15532dbc19
14:10 nine [Coke]: when exactly did you get that error?
14:13 ugexe should we start pushing for distribution names to use `-` instead of `::` as a separator?
14:13 [Coke] should this test file be deleted? It's not run: S03-operators/shortcuts.t
14:14 [Coke] nine: during a 'prove -v -e t/fudgeandrun /path/to/file' - went away with a realclean/rebuild. (was in a build directory)
14:15 * [Coke] opens a ticket.
14:17 [Coke] https://github.com/perl6/roast/issues/157
14:35 MasterDuke_ joined #perl6-dev
14:38 MasterDuke_ jnthn: hello hello hello. have you had any chance to think about https://github.com/rakudo/rakudo/pull/859? this is the fix that "quietly" does union hyperoperators for hashes
14:59 lizmat back from siesta
14:59 lizmat nine: looks like putting a sleep before/after does fix the issue
14:59 lizmat shall I commit it ?
15:17 tadzik joined #perl6-dev
15:26 MetaZoffix joined #perl6-dev
15:27 dalek roast: 4585a9c | lizmat++ | S10-packages/precompilation.t:
15:27 dalek roast: Add some rest to make for less flappy testing
15:27 dalek roast: review: https://github.com/perl6/roast/commit/4585a9cd1f
15:29 |Tux| This is Rakudo version 2016.08.1-169-g15532db built on MoarVM version 2016.08-43-g3d04391
15:29 |Tux| csv-ip5xs        9.015
15:29 |Tux| test            15.223
15:29 |Tux| test-t           6.787
15:29 |Tux| csv-parser      17.359
15:29 MetaZoffix \o/
15:30 [Coke] nice.
15:32 timotimo very not bad
16:07 SourceBaby joined #perl6-dev
16:22 ilmari[m] joined #perl6-dev
17:30 monada joined #perl6-dev
17:53 monada MasterDuke: I see this comment: http://irclog.perlgeek.de/perl6-dev/2016-09-12#i_13192789  was jnthn++ having concerns about https://github.com/rakudo/rakudo/pull/859 ? I think it's OK to merge, but wanted to check first.
17:57 nine lizmat: yes please. As the test is about time stamps, a sleep does seem like an appropriate fix
18:06 nine Those other spurious errors worry me :/
18:09 nine A kinda ironic detail: the CORE dist is always installed as version 6.c violating our own rule.
18:12 jnthn MasterDuke_: Sorry, was tied up with $other-job and just reading backlog here now. :)
18:12 yoleaux2 11 Sep 2016 03:28Z <Zoffix> jnthn: would you add what 'async-socket-str-fixes' brought in to the ChangeLog? I've no idea.
18:13 jnthn MasterDuke_: The main concern is if there's any lazy evaluation taking place and we could swallow the warnings associated with that. But since it's Associative, I guess not.
18:14 jnthn MasterDuke_: The safe fix would be `my &quiet-op = -> |c { quietly op(|c) }`
18:15 jnthn MasterDuke_: Which of course comes at quite a cost because of the extra invocation and args shuffling
18:15 jnthn otoh I guess if we were to assume it's a binary op we can do it much cheaper
18:15 jnthn But that'd make sure only the evaluation of the op is quieted
18:16 jnthn To show up in this case we'd need somebody to have a custom Associative implementation that could warn when indexed though, which is kinda marginal
18:16 jnthn So this is more of an issue for a similar patch for hyper on Positional, where laziness (and so suppressing warnings at a distance) really is a possibility.
18:17 FROGGS joined #perl6-dev
18:27 dalek roast: eedf53d | usev6++ | S17-supply/interval.t:
18:27 dalek roast: Fudge (skip) test using Proc::Async for JVM
18:27 dalek roast: review: https://github.com/perl6/roast/commit/eedf53d6b4
18:29 dalek roast: 2a5e743 | usev6++ | S17-supply/interval.t:
18:29 dalek roast: Fix fudging for JVM
18:29 dalek roast: review: https://github.com/perl6/roast/commit/2a5e743d97
18:40 AlexDaniel joined #perl6-dev
18:41 dalek rakudo/nom: a9ed671 | TimToady++ | src/Perl6/Grammar.nqp:
18:41 dalek rakudo/nom: desigilname var checked too soon in quotes
18:41 dalek rakudo/nom:
18:41 dalek rakudo/nom: When parsing quotes, interpolated variables with multiple sigils
18:41 dalek rakudo/nom: were being checked immediately, but that violates the constraint
18:41 dalek rakudo/nom: that backtrackable interpolations should not have side effects.
18:41 dalek rakudo/nom: We now delay that variable check out to where normal variables are
18:41 dalek rakudo/nom: checked, after any such backtracking would be triggered.
18:41 dalek rakudo/nom:
18:41 dalek rakudo/nom: Fixes RT #129257.
18:41 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/a9ed67168e
18:41 synopsebot6 Link:  https://rt.perl.org/rt3//Public/Bug/Display.html?id=129257
19:13 TheLemonMan joined #perl6-dev
19:18 Ven joined #perl6-dev
19:36 MasterDuke_ jnthn: yeah, Positional seems a thornier problem. and like Zoffix said in his comment on th PR, you can't really know what to replace an undefined with. so the impression i'm getting is that this is good to merge?
19:49 dalek rakudo/nom: 804bf67 | (Zoffix Znet)++ | / (3 files):
19:49 dalek rakudo/nom: Fix NativeCall CArray hanging when created from empty list
19:49 dalek rakudo/nom:
19:49 dalek rakudo/nom: The issue was we were calling nextsame when @values was empty,
19:49 dalek rakudo/nom: but the nextsame candidate was the one with a slurpy, that
19:49 dalek rakudo/nom: called the @values candidate again, resulting in a loop.
19:49 dalek rakudo/nom:
19:49 dalek rakudo/nom: Instead of having a separate multi for empty argument case, we
19:49 dalek rakudo/nom: return its value when the @values is empty, resolving the infiniloop.
19:49 dalek rakudo/nom:
19:49 dalek rakudo/nom: Fixes RT#129256: https://rt.perl.org/Ticket/Display.html?id=129256
19:49 synopsebot6 Link:  https://rt.perl.org/rt3//Public/Bug/Display.html?id=129256
19:49 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/804bf671f8
19:58 Ven joined #perl6-dev
20:00 Ven joined #perl6-dev
20:03 dogbert17 joined #perl6-dev
20:05 dogbert17 Zoffix: I've made the change locally in a branch, done some cmdline testing and then 'make spectest'. However, there were failures ...
20:06 dogbert17 failures in t/spec/S32-io/socket-accept-and-working-threads.t   and   t/spec/S32-num/power.rakudo.moar
20:06 Ven joined #perl6-dev
20:07 dogbert17 t/spec/S32-num/power.rakudo.moar .................................. Dubious, test returned 1 (wstat 256, 0x100)
20:09 ZoffixMobile joined #perl6-dev
20:10 ZoffixMobile dogbert17, that sounds like you aren't roast HEAD.
20:10 ZoffixMobile did you start from a fresh checkout of rakudo?
20:10 dogbert17 yes
20:11 ZoffixMobile Weird. Did you already commit the change? Can I see it?
20:11 dogbert17 when I wrote make spectest, roast was automatically downloaded
20:12 dogbert17 haven't committed yet, wanted to ask about the failures first (otoh, I did the change you suggested)
20:12 ZoffixMobile can you pastebin output of git diff
20:14 dogbert17 https://gist.github.com/dogbert17/f0bac4e05f46b87327006ebccec05a26
20:17 ZoffixMobile Yeah, that's good. Submit that as PR, I spectested that before I headed out and it was clean so the failures aren't due to this change. I'm unsure why you have spectest failures. Some tests flop sometimes, so you can try running the spectest again. Both of those recently had failures (but were then fixed), which is why I thought you may have been running an older spectest
20:18 dogbert17 ok, I'll commit and send a PR
20:20 jnthn MasterDuke: I think so, but we should consider the Positional case also
20:23 dalek rakudo/nom: 6fc1d9b | jnthn++ | docs/ChangeLog:
20:23 dalek rakudo/nom: Explain IO::Socket::Async fixes in ChangeLog.
20:23 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/6fc1d9b5be
20:27 dogbert17 Zoffix: done (I think)
20:32 dalek rakudo/nom: 59ce05e | (Jan-Olof Hendig)++ | src/core/Cool.pm:
20:32 dalek rakudo/nom: Make Cool.trans work the same as Str.trans
20:32 dalek rakudo/nom:
20:32 dalek rakudo/nom: Cool.trans coerces the invovant to Str and then directly calls
20:32 dalek rakudo/nom: Str.trans. The way this is done causes adverbs, if specified, to
20:33 dalek rakudo/nom: be thrown away. We must ensure that the these parameters are sent
20:33 dalek rakudo/nom: to Str.trans and not being thrown away. A couple of changes
20:33 dalek rakudo/nom: suggested by Zoffix++ takes care of this problem.
20:33 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/59ce05ed9b
20:33 dalek rakudo/nom: 22946ed | lizmat++ | src/core/Cool.pm:
20:33 dalek rakudo/nom: Merge pull request #877 from dogbert17/fix-rt-129258
20:33 dalek rakudo/nom:
20:33 dalek rakudo/nom: Make Cool.trans work the same as Str.trans
20:33 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/22946ed9bb
20:34 dogbert17 wohoo
20:35 nine dogbert17: your branch was probably not up to date with rakudo nom or you missed running config.status. Those failures are because of an out of date nqp/moarvm
20:37 dogbert17 nine: I probably missed something. Both nqp and moarvm were downloaded when I typed 'perl Configure.pl --gen-moar --gen-nqp --backends=moar'
20:41 ZoffixMobile dogbert17, ah, but your rakudo fork may have been out of date
20:41 ZoffixMobile git pull https//github.org/rakudo/rakudo
20:42 ZoffixMobile Anyway, dogbert17++ Achievement Unlocked: Fix Rakudo Bug :)
20:42 nine dogbert17: rakudo will build the nqp and moar that was required at the time of that rakudo version. roast may already test fixes in newer version.
20:44 dogbert17 When I forked and cloned, the last fix included was 'desigilname var checked too soon in quotes' by TimToady
20:45 nine odd
20:46 dogbert17 not all socket async tests failed, only five of them
20:46 dogbert17 Also, I'm running a 32 bit vm if that makes a difference
20:48 nine That could actually be a deciding factor
20:48 nine Few people run rakudo on 32 bit systems. It's much less well tested there.
20:49 dogbert17 is 32 bit officially supported?
20:52 ZoffixMobile I think our "officially supported" is constrained by whether or not there are volunteers to provide support on the platform in question :)
20:52 ZoffixMobile I use it on a 32bit box at $work. Seems to work fine
21:07 Zoffix dogbert17, will you submit roast tests for that bug as well?
21:13 Zoffix dogbert17++ # making me more aware of what Core Hacking blogs need to be written :)
21:26 dalek roast: a2e54a7 | (Zoffix Znet)++ | S02-literals/string-interpolation.t:
21:26 dalek roast: %%, @@, and && do not attempt to interpolate any variables
21:26 dalek roast:
21:26 dalek roast: RT#129257: https://rt.perl.org/Ticket/Display.html?id=129257
21:26 dalek roast: review: https://github.com/perl6/roast/commit/a2e54a7e09
21:26 synopsebot6 Link:  https://rt.perl.org/rt3//Public/Bug/Display.html?id=129257
21:27 timotimo m: my $foo; say "$foo.say().notamethod"
21:27 camelia rakudo-moar 22946e: OUTPUT«(Any)␤True.notamethod␤»
21:27 timotimo m: my $foo; say "$foo.say().notamethod()"
21:27 camelia rakudo-moar 22946e: OUTPUT«(Any)␤Method 'notamethod' not found for invocant of class 'Bool'␤  in block <unit> at <tmp> line 1␤␤»
21:27 timotimo hm
21:27 Zoffix m: my $foo; say "$foo.say.notamethod()"
21:27 camelia rakudo-moar 22946e: OUTPUT«(Any)␤Method 'notamethod' not found for invocant of class 'Bool'␤  in block <unit> at <tmp> line 1␤␤»
21:27 Zoffix Looks good to me
21:28 timotimo i seem to recall i had some issue at some point
21:31 Zoffix The Unguided is such a good band to hack while listening to \o/ https://www.youtube.com/watch?v=1ZNhfeIcArE
21:43 dalek roast: 0cbc5bf | (Zoffix Znet)++ | S05-transliteration/trans.t:
21:43 dalek roast: Remove trailing whitespace
21:43 dalek roast: review: https://github.com/perl6/roast/commit/0cbc5bf24f
21:45 dalek roast: 54ad7fd | (Zoffix Znet)++ | S05-transliteration/trans.t:
21:45 dalek roast: Adverbs on Cool.trans work the same as on Str.trans
21:45 dalek roast:
21:45 dalek roast: RT#129258: https://rt.perl.org/Ticket/Display.html?id=129258
21:45 dalek roast: review: https://github.com/perl6/roast/commit/54ad7fda46
21:45 synopsebot6 Link:  https://rt.perl.org/rt3//Public/Bug/Display.html?id=129258
21:45 Zoffix dogbert17, I added the tests in https://github.com/perl6/roast/commit/54ad7fda46    Teamwork ✋
22:12 lizmat and another Perl 6 Weekly hits the Net: https://p6weekly.wordpress.com/2016/09/12/warmen-up-das-youtuben/
22:16 jnthn lizmat++
22:17 jnthn A nice bedtime read
22:17 jnthn 'night, #perl6
22:17 jnthn *-dev :)
22:17 lizmat gnight jnthn
22:17 * lizmat also calls it a night
22:18 timotimo gnite lizmat and jnthn
22:23 Zoffix lizmat++ # good weekly
22:24 timotimo Zoffix++ # good week
22:41 Zoffix m: say ^∞ ~~ ∞
22:41 camelia rakudo-moar 22946e: OUTPUT«True␤»
22:41 Zoffix m: say ^4 ~~ 4
22:41 camelia rakudo-moar 22946e: OUTPUT«True␤»
22:41 Zoffix huh
22:41 Zoffix m: say ^4 ~~ 3
22:41 camelia rakudo-moar 22946e: OUTPUT«False␤»
22:42 Zoffix oh, it's using .elems ... I thought it'd say whether it's in range
22:42 timotimo yeah, that would be the other way around
22:42 Zoffix m: say 3 ~~ ^4
22:42 camelia rakudo-moar 22946e: OUTPUT«True␤»
22:42 Zoffix m: say ∞ ~~ ^∞
22:42 camelia rakudo-moar 22946e: OUTPUT«False␤»
22:42 Zoffix timotimo++ cheers

| Channels | #perl6-dev index | Today | | Search | Google Search | Plain-Text | summary