| [Coke] | dogbert17: how did you run htmlify? | 00:52 | |
| ah, I can duplicate it. | 00:59 | ||
| samcv, dogbert17: the htmlify failure was a result of samcv's commit: 0b005062fb63f7b62f79f318a1bfe450c3d49010 which I just reverted. | 01:50 | ||
| ugexe | m: Nil.?starts-with("foo"); # bug? | 04:23 | |
| camelia | Invocant of method 'starts-with' must be an object instance of type 'Cool', not a type object of type 'Nil'. Did you forget a '.new'? in block <unit> at <tmp> line 1 |
||
| llfourn | m: Nil.^find_method("starts-with").package.say | 04:35 | |
| camelia | (Cool) | ||
| llfourn | wat | ||
| m: Nil ~~ Cool | 04:36 | ||
| camelia | ( no output ) | ||
| llfourn | m: say Nil ~~ Cool | ||
| camelia | True | ||
| llfourn | huh | ||
| BenGoldberg | m: dd Nil | 05:00 | |
| camelia | Nil | ||
| BenGoldberg | m: say Nil.WHAT | ||
| camelia | Nil | ||
| BenGoldberg | m: say Nil.HOW | ||
| camelia | Perl6::Metamodel::ClassHOW.new | ||
| BenGoldberg | m: say Nil.^parents | 05:01 | |
| camelia | () | ||
| BenGoldberg | m: say Mu.^parents | ||
| camelia | () | ||
| BenGoldberg | m: say Cool.^parents | ||
| camelia | () | ||
| samcv | [Coke], is travis at least finishing now? | 05:45 | |
| hopefully. i set it to only do bigpage. is annoying to do without feedback from travis | |||
| thanks for the revert though | |||
| [Tux] | This is Rakudo version 2017.06-209-gc6cc1a7a3 built on MoarVM version 2017.06-56-g161ec639 | 06:50 | |
| csv-ip5xs 2.545 | |||
| test 12.904 | |||
| test-t 4.144 - 4.482 | |||
| csv-parser 12.174 | |||
| Error encoding UTF-8 string: could not encode codepoint 1805894 (0x1B8E46), codepoint out of bounds. Cannot encode higher than 1114111 (0x10FFFF) | |||
| in block <unit> at 88.t line 21 | |||
| Geth | roast: 291b557e0d | (Elizabeth Mattijsen)++ | S03-operators/mix.t Fix faulty tests |
09:19 | |
| roast/6.c-errata: 84e08cabdd | (Elizabeth Mattijsen)++ | S03-operators/mix.t Fix faulty tests, same as in master |
09:23 | ||
| lizmat | wonder if this would affect Rakudo Perl 6 on Win10 as well: randomascii.wordpress.com/2017/07/...-my-mouse/ | 09:41 | |
| yoleaux | 00:27Z <perlawhirl> lizmat: I'm not sure I'm an authority :D The docs say a Mix is like a Bag, so my opinion is it should behave similar to bags, ie: False. | ||
| timotimo | hm. it's about starting and quitting a whole bunch of short-lived processes; i think even roast wouldn't qualify for the amount of processes dying at the same time you have to have to get this bad behaviour | 09:55 | |
| nine | 3 decades of developing Windows and they still haven't figured out the most basic tasks of an OS - creating and destroying processes? | 09:56 | |
| timotimo | nobody could have known that multi-processor computers would be so popular in the future | 09:57 | |
| and also: the problem is on windows 10, but not on windows 7 | |||
| so at some point they have forgotten how to process | |||
| arnsholt | Isn't that one of those fundamental differences between Windows and Unix too though? | 10:02 | |
| Unix processes are cheap, but Windows processes are more heavyweight | 10:03 | ||
| timotimo | hmm. one major difference is the absence of fork in windows | ||
| for whatever that's worth | |||
| Zoffix | heh, I have to put execution cap at 96% on VirtualBox running in Win10 host because otherwise I can't move my mouse either when I'm spectesting :) | 10:19 | |
| though haven't noticed the issue when spectesting natively on Win10 | 10:21 | ||
| ugexe: IMO yeah. We should add (or auto-generate) all the Cool methods into Nil to make them propagate the Nil. Otherwise the safecall operator is useless in many cases where you make use of Cool methods. | 10:26 | ||
| Last time such a proposal was met with resistance as being messy | 10:27 | ||
| Similar issue exists with Failure and Any methods that make 1-item list | |||
| m: say Failure.new.elems; # look mah, no failures! | 10:28 | ||
| camelia | 1 | ||
| timotimo | has Nil $!foo handles Cool :) | 10:31 | |
| DrForr_ | Egads. String matching in Python is amazingly clunky. | 10:35 | |
| timotimo | how so? just because regexes are second-class citizen? | ||
| DrForr_ | Mostly. | 10:36 | |
| I'm properly layering stuff for TensorFlow, and the hoops I need to go through in order to unpack a type declaration semi-cleanly aren't quite on fire but they're close. | 10:37 | ||
| Geth | rakudo/nom: 0cf3487160 | (Elizabeth Mattijsen)++ | src/core/Rakudo/QuantHash.pm Any (<=) Any should always be True, not False Unlike Any (<) Any is always False. Caught while overhauling subset tests. This e.g. fixes set() (<=) SetHash.new. This is *not* the same cod path as set() (<=) set(), as that will return True based on the fact that the nqp::eqaddr() test fires. |
10:40 | |
| lizmat | .oO( fishy flows ) |
10:41 | |
| *sigh* :-) | |||
| DrForr_ | o/ I'm gonna eat you little fishie o/ | ||
| At least I've cured the segfault. | 10:42 | ||
| lizmat | m: dd Nil (<=) Nil # wonders if this should be True or not | 10:59 | |
| camelia | Bool::True | ||
| lizmat | currently it is the same as: | 11:00 | |
| m: dd Nil.Set (<=) Nil.Set | |||
| camelia | Bool::True | ||
| lizmat | but oddly enough: | ||
| m: dd Nil.Set | |||
| camelia | set(Any) | ||
| lizmat | which implies: | 11:01 | |
| dd Any.Set eqv Nil.Set # :-( | |||
| m: dd Any.Set eqv Nil.Set # :-( | |||
| camelia | Bool::True | ||
| lizmat feels that Nil.Set should be Nil | 11:02 | ||
| m: dd Set.new(Nil) # subtle difference | 11:03 | ||
| camelia | set(Nil) | ||
| lizmat | ok, so either Nil.Set should be Nil, or should be set(Nil) | 11:04 | |
| m: dd Nil.Set # clearly wrong | 11:05 | ||
| camelia | set(Any) | ||
| DrForr_ | I'd go with set(Nil), arguing simply that the null set and null are different things. | 11:09 | |
| samcv | or it could be set().. though idk really. | 11:11 | |
| is Nil nothing or is Nil something | |||
| lizmat | well, Nil should indicate the absence of a value where there should be one | 11:12 | |
| samcv | and with that i'm gonig to bed. night all. hopefully going to work more on making sure nothing concatted has to ever re-normalize the entire length of both strings tomorrow! | ||
| i will have a party when i achieve that i think | |||
| lizmat | good night, samcv | 11:13 | |
| samcv | at least i made it so that renormalization gets triggered like < 15% of the time compared to before | 11:14 | |
| only triggers now when renormalization will actually result in a different result than just bonding the two strings | 11:15 | ||
| dogbert17 | .tell [Coke] htmlify works again, many thanks for fixing | 11:17 | |
| yoleaux | dogbert17: I'll pass your message to [Coke]. | ||
| Geth | rakudo/nom: f42e53cbee | (Elizabeth Mattijsen)++ | src/core/Nil.pm Make QuantHash.new(Nil) and Nil.QuantHash consistent Although I'm not 100% sure this is correct, it is at least better than generating a QuantHash.new(Any). |
11:22 | |
| nqp/master: 6 commits pushed by pmurias++ | 11:59 | ||
| rakudo/nom: 5074158712 | (Elizabeth Mattijsen)++ | src/core/Rakudo/QuantHash.pm Don't compare yourself, duh! Fixes <a b>.Mix (<=) <a>.Mix. Caught when overhauling subset tests. |
12:16 | ||
| rakudo/nom: 1ce8071e5e | (Elizabeth Mattijsen)++ | src/core/set_subset.pm Need two additional similar candidates To prevent Ambiguous call to 'infix:«(<=)» for mix() (<=) set() . Spotted while overhauling subset tests. |
13:15 | ||
| jnthn | Man, such a lot of backlog for such a short vacation :) | 13:17 | |
| lizmat | jnthn o/ | 13:18 | |
| you could wait for the next P6W :-) | |||
| llfourn | wb jnthn | 13:19 | |
| Zoffix | jnthn: someone in #perl6 was asking last night about something you likely know about: irclog.perlgeek.de/perl6/2017-07-11#i_14858173 | ||
| jnthn | lizmat: heh, that's not a bad idea :) | 13:22 | |
| At least just a handful of RTs :) | |||
| Zoffix: Hm, I get DNS resolution issue on the logs again right now :S | |||
| It's odd, seems transient but repreating | 13:23 | ||
| hah, yup, now they finally loaded | 13:25 | ||
| But scumbag browser doesn't scroll to the anchor | 13:26 | ||
| Ah, an RT showed up about that | |||
| m: await start { say $*THREAD.id } | 13:27 | ||
| camelia | 3 | ||
| jnthn | I'm wondering if it's some issue caused by us tracking the dynamic scope of the place we do the start | 13:28 | |
| But clearly it doesn't happen in the above example | |||
| m: start { say $*THREAD.id; await Promise.in(2); say $*THREAD.id }; sleep 1; start { sleep 3 } | 13:29 | ||
| m: start { say $*THREAD.id; await Promise.in(2); say $*THREAD.id }; sleep 1; await start { sleep 3 } # d'oh | |||
| Hm, but why no output | |||
| camelia | Resource temporarily unavailable | ||
| 3 3 |
|||
| jnthn | m: use nqp; start { say $*THREAD.id; say nqp::threadid; await Promise.in(2); say $*THREAD.id; say nqp::threadid }; sleep 1; await start { sleep 3 } # d'oh | 13:30 | |
| camelia | ===SORRY!=== Arg count 0 doesn't equal required operand count 2 for op 'threadid' |
||
| jnthn | ah, darn | ||
| m: use nqp; start { say $*THREAD.id; say nqp::threadid(nqp::currentthread); await Promise.in(2); say $*THREAD.id; say nqp::threadid(nqp::currentthread) }; sleep 1; await start { sleep 3 } # d'oh | |||
| camelia | 3 3 3 3 |
||
| jnthn | hah, no, it apparently doesn't like | 13:31 | |
| *lie | |||
| m: await do for ^1000 -> $d { start { my $before = $*THREAD.id; await Promise.in(0.002 * $d); say $before != $*THREAD.id ?? 'D' !! 'S' } } | 13:33 | ||
| camelia | (timeout)S S S |
||
| jnthn | oh damn | ||
| I forgot the 6.d.PREVIEW on all of the above! | |||
| m: use v6.d.PREVIEW; await do for ^1000 -> $d { start { my $before = $*THREAD.id; await Promise.in(0.002 * $d); say $before != $*THREAD.id ?? 'D' !! 'S' } } | |||
| camelia | S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S S… |
13:34 | |
| jnthn | m: use v6.d.PREVIEW; await do for ^1000 -> $d { start { my $before = $*THREAD.id; await Promise.in(0.002 * $d); say 'D' if $before != $*THREAD.id } } | ||
| camelia | ( no output ) | ||
| jnthn | OK, that's...a bit odd | ||
| m: use v6.d.PREVIEW; use nqp; await do for ^1000 -> $d { start { await Promise.in(0.002 * $d); say 'oops' if nqp::threadid(nqp::currentthread) != $*THREAD.id } } | 13:35 | ||
| camelia | ( no output ) | ||
| jnthn | m: use v6.d.PREVIEW; use nqp; await do for (^1000).roll(1000) -> $d { start { await Promise.in(0.002 * $d); say 'oops' if nqp::threadid(nqp::currentthread) != $*THREAD.id } } | 13:36 | |
| camelia | ( no output ) | 13:37 | |
| jnthn | m: use v6.d.PREVIEW; await do for (^1000).roll(1000) -> $d { start { my $before = $*THREAD.id; await Promise.in(0.002 * $d); say 'D' if $before != $*THREAD.id } } | ||
| camelia | ( no output ) | ||
| jnthn | Yeah, something feels suspect there. Clearly it's working right in that it's doing the continuation trick, but the resumption always being on the same thread feels a bit odd | 13:38 | |
| m: use v6.d.PREVIEW; use NativeCall; sub gettid() returns int is native {*}; await do for (^1000).roll(1000) -> $d { start { my $before = gettid(); await Promise.in(0.002 * $d); say 'D' if $before != gettid() } } | 13:39 | ||
| camelia | 5===SORRY!5=== Error while compiling <tmp> Cannot resolve caller infix:<==>(NQPMu, Int); none of these signatures match: ($?) (\a, \b) (Real \a, Real \b) (Int:D \a, Int:D \b) (int $a, int $b) (Num:D \a, Num:D \b … |
||
| jnthn | wat, NQPMu?! | ||
| m: use v6.d.PREVIEW; use NativeCall; sub gettid() returns int32 is native {*}; await do for (^1000).roll(1000) -> $d { start { my $before = gettid(); await Promise.in(0.002 * $d); say 'D' if $before != gettid() } } | 13:40 | ||
| camelia | An operation first awaited: in block <unit> at <tmp> line 1 Died with the exception: Cannot locate symbol 'gettid' in native library '' in method setup at /home/camelia/rakudo-m-inst-2/share/perl6/sources/24DD121B5B4774C04A7084827B… |
||
| jnthn | m: use v6.d.PREVIEW; use NativeCall; sub pthread_self() returns int32 is native {*}; await do for (^1000).roll(1000) -> $d { start { my $before = gettid(); await Promise.in(0.002 * $d); say 'D' if $before != gettid() } } | ||
| camelia | 5===SORRY!5=== Error while compiling <tmp> Undeclared routine: gettid used at line 1 |
||
| jnthn | m: use v6.d.PREVIEW; use NativeCall; sub pthread_self() returns int32 is native {*}; await do for (^1000).roll(1000) -> $d { start { my $before = pthread_self(); await Promise.in(0.002 * $d); say 'D' if $before != pthread_self() } } | 13:41 | |
| camelia | D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D D… |
||
| jnthn | oooh | ||
| The plot thickens. | |||
| Also, wat?! | |||
| Geth | roast: 631cea88c3 | (Elizabeth Mattijsen)++ | S03-operators/set_subset.t Completed overhaul of (<=) tests Basically only use "a" and "b" as keys, and values -2,-1,0,1 as weights. Testing other values should be part of te basic functionality of a QuantHash type. |
13:51 | |
| lizmat | this uncovered 4 bugs, which got fixed :-) | 13:52 | |
| now on to overhauling (<) tests | |||
| jnthn | I messed up some of the above examples, it seems. It is about $*THREAD. | 13:54 | |
| Also, I suspect it's going to turn out to be about the fact that we cache dynamic variable lookups | 13:55 | ||
| But those probably all need invalidating in the case of a continuation take/resume | 13:56 | ||
| lizmat | which then begs the question: should we cache at all ? | ||
| jnthn | Yeah, it's woth it | 13:57 | |
| *worth | |||
| In general | |||
| TimToady++ has done lots of measuring on this | |||
| lizmat | but that was before massive multi-threading, right ? | 13:58 | |
| but I guess the context thing in a Promise only made &DYNAMIC more expensive | |||
| should invalidate until the context of the $*PROMISE ? | 13:59 | ||
| jnthn | The easy/cheap point to do it is when taking the continuation | 14:00 | |
| As we are walking the frames anyway to try and find the continuation tag. | |||
| lizmat | ah, yes :-) | 14:01 | |
| jnthn | Which seems to resolve the issue :) | ||
| Guess now I need to write a spectest for this :) | 14:02 | ||
| Which is, um, a little interesting to try and work out how to do reliably. | |||
| Especially because in the future we probably will have a scheduler that gives the thread that did the await first preference on running the resumption. | 14:03 | ||
| jnthn struggles to think of a non-probabalistic test | |||
| Oh! | 14:05 | ||
| I can write one using NQP ops and stick it into the Rakudo tests ratherr than spectest | |||
| lizmat | m: dd {:a} (<) {:a,:b} # another buglet found | 14:11 | |
| camelia | Bool::False | ||
| lizmat | but first afk for a bit& | 14:25 | |
| Geth | nqp: 124754efce | (Jonathan Worthington)++ | tools/build/MOAR_REVISION Bump to latest MoarVM revision. |
||
| ¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...0-ga3a58c8 | |||
| rakudo/nom: d74ebd8250 | (Jonathan Worthington)++ | tools/build/NQP_REVISION Bump for dynamic cache/continuation bug fix. |
14:36 | ||
| rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....5-g124754e 8aa1b0bc0b | (Jonathan Worthington)++ | t/02-rakudo/09-thread-id-after-await.t |
|||
| synopsebot6 | Link: rt.perl.org/rt3/Public/Bug/Display...?id=131740 | ||
| Geth | ¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....5-g124754e | ||
| pmurias | jnthn: do we use nqp::continuationcontrol(1, ...) for anything? | 15:38 | |
| ugexe | does IO::Path.slurp(:!close) not act the same as IO::Handle.slurp(:!close) because it would just leak handles? | 16:17 | |
| Zoffix | ugexe: IO::Path.slurp doesn't have a :close. It always closes | 16:18 | |
| There's no filehandle in user's hands. | |||
| So there's nothing to not-close from the user's perspective. i.e. the fact that there are any handles involved in teh process of slurping is implementation detail. | 16:19 | ||
| ugexe | right, there could technically be a share interface between the two slurps at the cost of allowing something that is not useful | 16:21 | |
| Zoffix doesn't see the point of deliberatelly leaking handles for interface consistency | 16:22 | ||
| Considering nothing's stopping you for passing :!close to IO::Path.slurp; it's just ignored | |||
| Or mostly-ignored. IIRC there's a fastpath for when there aren't any args | |||
| Ah. Not any more. OK :) | 16:23 | ||
| Zoffix & | 16:24 | ||
| ugexe | yeah, so what it should technically be method slurp(..., :$close where *.not) ignoring any performance penalties | ||
| er, :$close = True where *.so | 16:25 | ||
| Zoffix | Why? | 16:26 | |
| What is $close closing in IO::Path? | 16:29 | ||
| nine | Those methods are related but they do not do the same thing. So it's OK that the interface is not 100 % identical | 16:30 | |
| ugexe | i understand its pointless to do IO::Path.slurp(:!close), and that extra named params are ignored. but since slurp seems to try and mimick behavior across different types it seems like it would be nice for it to at least do *something* with the named args its ignoring that other implementations use | ||
| Zoffix | What do you mean "would be nice"? | 16:31 | |
| There's nothing open for it to close. There are no errors generated if you pass :close parameter. There's nothing for it to do, so it's ignored. | 16:32 | ||
| I don't understand the need for making it leak handles on purpose. The only reason IO::Handle.slurp can take a :close because you're holding the handle and can close it at another, better, time. It's not there to leak handles and it shouldn't be in IO::Path to leak handles either. | 16:33 | ||
| Zoffix & for real | |||
| ugexe | m: $*OUT = class :: { method print (|) {} }; say 1; note 2; warn 3; | 16:46 | |
| camelia | 2 3 in block <unit> at <tmp> line 1 |
||
| ugexe | m: $*ERR = class :: { method print (|) {} }; say 1; note 2; warn 3; | 16:47 | |
| camelia | 1 No exception handler located for catch at <unknown>:1 (/home/camelia/rakudo-m-inst-1/share/perl6/runtime/CORE.setting.moarvm:print_exception) from SETTING::src/core/Exception.pm:452 (/home/camelia/rakudo-m-inst-1/share/perl6/runtime/CORE.se… |
||
| ugexe | m: $*ERR = class :: { method print (|) {} }; note 2; | 16:49 | |
| camelia | ( no output ) | ||
| ugexe | m: $*ERR = class :: { method print (|) {} }; warn 2; | ||
| camelia | No exception handler located for catch at <unknown>:1 (/home/camelia/rakudo-m-inst-1/share/perl6/runtime/CORE.setting.moarvm:print_exception) from SETTING::src/core/Exception.pm:452 (/home/camelia/rakudo-m-inst-1/share/perl6/runtime/CORE.settin… |
||
| ugexe | [Coke]: looks related to RT 131670 | 16:50 | |
| stmuk_ | URL doesn't install - Cannot do 'say' on a handle in binary mode in block <unit> at t/01.t line 33 | 16:54 | |
| err URI I mean | |||
| Zoffix | /me knows a fix | 16:55 | |
| buggable: eco URI | |||
| buggable | Zoffix, URI 'A URI implementation using Perl 6 grammars to implement RFC 3986 BNF': github.com/perl6-community-modules/uri 9 other matching results: modules.perl6.org/#q=URI | ||
| Zoffix | hm. can't repro on 2017.05-389-g326faed | 16:57 | |
| c: 2017.05-389-g326faed $*OUT.encoding: Nil; $*OUT.say: 42 | 16:58 | ||
| committable6 | Zoffix, ¦2017.05-389-g326faed: «Cannot do 'say' on a handle in binary mode in block <unit> at /tmp/QxWoPHUVTT line 1 «exit code = 1»» | ||
| stmuk_ | This is using 2017.06-215-g8aa1b0b | ||
| Zoffix | It's also a Succ on -193 toast: toaster.perl6.party/module?module=...3-gcb43086 | 17:01 | |
| It's this module, right? | 17:03 | ||
| github.com/perl6-community-modules/uri | |||
| can't repro on 2017.06-197-g83e1570 | 17:05 | ||
| Zoffix builds HEAD | |||
| stmuk_: what're you running for test? zef? does it pass with prove? | 17:06 | ||
| stmuk_ | odd I did zef update and I'm getting 0.1.3 | ||
| deleting dirs now | 17:07 | ||
| ah my issue just outdated version | 17:09 | ||
| Zoffix | Oh *phew* :) | ||
| stmuk_ | I'm wondering if "rakudobrew nuke" should blow away ~/.zef and ~/.perl6 too? or maybe new "thermonuclear" option | 17:15 | |
| ugexe | zef nuke RootDir | ||
| or `zef nuke RootDir site home` for "blow up everything except base rakudo install" | |||
| if you were using a zef that was older than 2 weeks old the problem probably got fixed. file locks on most things in ~/.zef (so multi copies of zef and future concurrent fetches shouldn't stomp on each other) and refactored the cache indexer | 17:26 | ||
| Zoffix | ZOFVM: Files=1259, Tests=143197, 123 wallclock secs (21.65 usr 3.48 sys + 2416.67 cusr 186.69 csys = 2628.49 CPU) | 17:27 | |
| jnthn | pmurias: Perhaps nothing does in Rakudo; I think maybe some NQP tests do, but then I'm not sure about that either :) | 17:28 | |
| Zoffix | 6.c-errata has failures: t/spec/S03-operators/mix. test 91 | 17:42 | |
| t/spec/S06-multi/type-based.t test 56 | |||
| I think that's all (the passing TODOs really dirty up the output; 6.d should remove all the fudges on release) | 17:43 | ||
| type-based: "not ok 56 - native types and where clauses do not cause spurious circularities" | |||
| Geth | roast/6.c-errata: 7775417330 | (Jonathan Worthington)++ (committed by Zoffix Znet) | S06-multi/type-based.t Update test for multi literal semantics. |
17:45 | |
| Zoffix | .tell lizmat S03-operators/mix.t test 91 fails: "not ok 91 - ⊅ - mix(d(2.2), e(4.4), n(2.2), y(2.2)) is not a strict supermix of mix(d(2.2), e(2.2), n(2.2))". Would you know how to resolve this? | 17:47 | |
| yoleaux | Zoffix: I'll pass your message to lizmat. | ||
| ugexe | when is it ok to change something in roast? specifically github.com/perl6/roast/blob/7851e7...tall.t#L19 - the test itself is not specifically about class Distribution, just *installing -a- Distribution*. if that line can be changed, can it be changed to use code that only works on 2016.04+ ? | 17:48 | |
| Zoffix | I think such questions is what we'd need to nail down more thoroughly for 6.d. What exactly constitutes a wrong test and what constitutes just supporting code that itself isn't being tested. | 17:52 | |
| ¯\_(ツ)_/¯ | 17:53 | ||
| ugexe | well, Distribution has a 40 line compatability shim for that single line. and the shim is technically broken anyway, just not in a tested way | ||
| m: class Foo is Distribution {}; Distribution.new(:name<Foo>).name.say; Foo.new(:name<Foo>).name.say | 17:54 | ||
| Geth | 6.d-prep: f82fb006e3 | (Zoffix Znet)++ (committed using GitHub Web editor) | TODO/README.md List consideration of what's 'supporting code' in roast changes |
||
| camelia | Foo No such method 'name' for invocant of type 'Foo'. Did you mean any of these? none note take in block <unit> at <tmp> line 1 |
||
| [Tux] | This is Rakudo version 2017.06-215-g8aa1b0bc0 built on MoarVM version 2017.06-75-g8aa657d6 | 18:52 | |
| csv-ip5xs 2.443 | |||
| test 12.504 | |||
| test-t 4.104 - 4.110 | |||
| csv-parser 11.790 | |||
| ugexe | nine: looks like some segfaults when doing -Ilib with code that does `use lib <lib>` were fixed by the fix to up-to-date/handles | 18:55 | |
| Geth | roast: 28efdda0e8 | (Elizabeth Mattijsen)++ | S03-operators/set_proper_subset.t Completed overhaul of (<) tests Basically only use "a" and "b" as keys, and values -2,-1,0,1 as weights. Testing other values should be part of the basic functionality of a QuantHash type. |
19:21 | |
| rakudo/nom: 103869be17 | (Elizabeth Mattijsen)++ | src/core/set_proper_subset.pm Remove Map (<) Map candidate for now While overhauling the (<) tests, it became clear to me that it is not as simple as I thought, simply because the number of keys on either side, does not have any bearing on actual number of keys that would be coerced into a Set: think {:a} (<) {:a,:0b} |
19:27 | ||
| ugexe | `run("ls", :merge).out.slurp.say` # regression from 2017.06 | 19:35 | |
| Zoffix | What happens? | 19:36 | |
| ugexe | its deadlocking | ||
| Zoffix | :o | ||
| ugexe | mind you before 2017.06 using :merge threw an exception telling you it was broken | 19:37 | |
| github.com/perl6/roast/blob/master...stem.t#L83 this never got unfudged, but it works in 2017.06 but not head | 19:42 | ||
| Geth | roast: c518b59457 | (Elizabeth Mattijsen)++ | S03-operators/set_difference.t Some more (-) tests and some reorg |
19:46 | |
| AlexDaniel | I don't think :merge ever worked | 19:50 | |
| rt.perl.org/Ticket/Display.html?id=128594 | |||
| geekosaur | I think the complaint is that either it should work or it should throw, just not deadlock? | 19:52 | |
| AlexDaniel | uh! Would be lovely if it started to work | 20:02 | |
| I'd be able to throw away the workaround I'm using everywhere, plus it will probably work faster… | |||
| japhb | Is there a channel for perl6 game dev? ISTR something about that, but I can't find it now. Maybe it was a mailing list and I'm just forgetting .... | 20:09 | |
| Zoffix | japhb: yeah, it's #perl6-gaming | 20:14 | |
| lizmat | Zoffix: I don't see that in my #perl6- channel list ? | 20:24 | |
| yoleaux | 17:47Z <Zoffix> lizmat: S03-operators/mix.t test 91 fails: "not ok 91 - ⊅ - mix(d(2.2), e(4.4), n(2.2), y(2.2)) is not a strict supermix of mix(d(2.2), e(2.2), n(2.2))". Would you know how to resolve this? | ||
| lizmat | Zoffix: I don't see that? is that in 6.c-errata ? | 20:25 | |
| jnthn | AlexDaniel: It was certainly busted, and made to throw an error 'cus it SEGV'd; now Proc::Async can do it (you just .Supply to get it) and that's used to implement it for Proc too | 20:31 | |
| AlexDaniel | oh, so it is supposed to work now? | ||
| mmmm nice | |||
| I'll try to play with it soon, hopefully | |||
| jnthn | Uh, supposed to, but I see in backlog apparently not? :s | 20:32 | |
| Ah well, I'll repeat my regular refrain: use Proc::Async, at least when that one's wrong I know what it's meant to do :P | |||
| But yeah, instead of .stdout or .stderr your just .Supply | |||
| And since whenever will call .Supply for you, then my $p = Proc::Async.new(...); and later whenever $p { } just gets you the merged output in the whenever | 20:33 | ||
| Zoffix | lizmat: oh yeah, sorry, it's in 6.c-errata | 20:40 | |
| lizmat | ok, will check | ||
| Zoffix | lizmat: *shrug* there were a couple of people in it when I tried the channel a few minutes ago. I think there's also #perl6-noisegang (unsure) | 20:41 | |
| Geth | rakudo/nom: 68e1baa35d | (Elizabeth Mattijsen)++ | src/core/set_symmetric_difference.pm Add a number of (^) candidates we need Spotted while developing thorough (^) (symmetric difference) testing |
20:55 | |
| lizmat | did we lose Geth again, or is github just slow? | 21:11 | |
| lizmat just pushed github.com/rakudo/rakudo/commit/72...eaf998d333 | 21:12 | ||
| Geth | rakudo/nom: 72733bcbab | (Elizabeth Mattijsen)++ | src/core/set_symmetric_difference.pm Fix typo in object hash handling of (^) Spotted while working on thorough (^) testing. |
||
| Zoffix | Wow. Looks like we have a very wealthy individual willing to sponsor Rakudo development! | 21:13 | |
| "From: Florence Martinez | |||
| I have a donation of ($5,500,000.00) for you please respond immediately." | |||
| Zoffix responds immediately. | 21:14 | ||
| ZofBot: thank you Florence Martinez! | |||
| ZofBot | Zoffix, All such values pay attention to the current ":ignorecase" and ":ignoremark" settings | ||
| lizmat | Zoffix: it doesn't state what type of $ so be careful! :-) | 21:16 | |
| Zoffix | :) | 21:20 | |
| lizmat | oh, by the way: Zoffix++ # github.com/perl6/marketing/blob/ma...-bleed.pdf | 21:25 | |
| Geth | rakudo/nom: aee4a46c1b | (Elizabeth Mattijsen)++ | src/core/set_symmetric_difference.pm Make ⊖ an alias for (^) This fixes various issues with calling infix:<⊖> with anything other than 2 parameters. Spotted while working on the thorough testing of symmetric difference tests. |
21:37 | |
| roast: e6d8dd11ce | (Elizabeth Mattijsen)++ | S03-operators/set_symmetric_difference.t Initial version of thorough (^) (symmetric difference) tests |
21:39 | ||
| rakudo/nom: 533eacfdb2 | (Elizabeth Mattijsen)++ | t/spectest.data Make sure we run thorough (^) tests |
21:43 | ||
| lizmat | good night, #perl6-dev! | 21:45 | |
| ugexe | :merge stopped working in github.com/rakudo/rakudo/commit/7b...c69c52a4a0 | 21:48 | |
| Geth | rakudo/nom: c86090e35f | (Nick Logan)++ (committed using GitHub Web editor) | src/core/Proc.pm Fix Proc :merge See: github.com/rakudo/rakudo/commit/7b175bd |
21:57 | |
| Zoffix | haw awesome | 22:07 | |
| Geth | roast: ccef1af2fc | (Nick Logan)++ (committed using GitHub Web editor) | S29-os/system.t Unfudge Proc :merge test Fixed in github.com/rakudo/rakudo/commit/c86090e |
22:31 | |
| samcv | don't bump moarvm until i wrate some more tests | 23:03 | |
| just thought of something and want to be certain there are no issues | |||
| travis-ci | Rakudo build failed. Elizabeth Mattijsen 'Add a number of (^) candidates we need | 23:13 | |
| travis-ci.org/rakudo/rakudo/builds/252970439 github.com/rakudo/rakudo/compare/1...e1baa35dd2 | |||
| buggable | [travis build above] ✓ All failures are due to timeout (0), missing build log (0), GitHub connectivity (1), or failed make test (0). | ||
| Geth | roast/ugexe-patch-1: 750b3b0b4e | (Nick Logan)++ (committed using GitHub Web editor) | S29-os/system.t Fix proc test failing due to prior chdir Use `LEAVE { chdir $orig-cwd }` to fix being left in a different directory for remaining tests. Use `$*EXECUTABLE.absolute` so the proper interpreter is always used. |
23:48 | |
| roast: ugexe++ created pull request #281: Fix proc test failing due to prior chdir |
23:49 |