| [Tux] | This is Rakudo version 2017.01-113-g5b3ac83b9 built on MoarVM version 2017.01-14-g357438a9 | 07:34 | |
| csv-ip5xs 2.847 | |||
| test 12.305 | |||
| test-t 5.027 | |||
| csv-parser 14.210 | |||
| second run 5.111 | |||
| gfldex | m: sub f { say Backtrace } | 08:46 | |
| camelia | ( no output ) | ||
| gfldex | m: sub f { say Backtrace }; | ||
| camelia | ( no output ) | ||
| gfldex | m: sub f { say Backtrace }; f | ||
| camelia | rakudo-moar 5b3ac8: OUTPUT«(Backtrace)» | ||
| gfldex | m: sub f { say Backtrace.Str }; f | 08:47 | |
| camelia | rakudo-moar 5b3ac8: OUTPUT«Use of uninitialized value of type Backtrace in string context.Methods .^name, .perl, .gist, or .say can be used to stringify it to something meaningful. in sub f at <tmp> line 1» | ||
| gfldex | m: sub f { say Backtrace.new.Str }; f | ||
| camelia | rakudo-moar 5b3ac8: OUTPUT« in sub f at <tmp> line 1 in block <unit> at <tmp> line 1» | ||
| gfldex | m: sub f { say Backtrace.new }; f | ||
| camelia | rakudo-moar 5b3ac8: OUTPUT«Backtrace.new» | ||
| gfldex | m: sub f { say Backtrace.new.full }; f | ||
| camelia | rakudo-moar 5b3ac8: OUTPUT« in code at SETTING::src/core/Backtrace.pm line 85 in method new at SETTING::src/core/Backtrace.pm line 85 in sub f at <tmp> line 1 in block <unit> at <tmp> line 1» | ||
| Geth | kudo: usev6++ created pull request #1007: [JVM] Set nqp-lib=blib for building CORE.d.setting |
12:55 | |
| bartolin can't commit directly to rakudo, yet | 12:57 | ||
| yoleaux2 | 22 Jan 2017 23:53Z <MasterDuke> bartolin: do you have an opinion on Nd's in sprintf format strings? e.g., irclog.perlgeek.de/perl6-dev/2017-..._13970314. should they be made to work? | ||
| Geth | kudo/nom: bfbf348ade | usev6++ | tools/build/Makefile-JVM.in [JVM] Set nqp-lib=blib for building CORE.d.setting ... as we do for CORE.setting This unbreaks the JVM build. |
||
| kudo/nom: 87d40ab183 | niner++ | tools/build/Makefile-JVM.in Merge pull request #1007 from usev6/jvm_core.d.setting [JVM] Set nqp-lib=blib for building CORE.d.setting |
|||
| brokenchicken | bartolin: ask jnthn++ to give you (and ugexe) commit bit | 15:12 | |
| jnthn | Ah, they sent CLA but didn't get the bits yet? | 15:14 | |
| timotimo | i seem to recall the arrival of the CLA has been confirmedon the channel at some point | ||
| ugexe | Coke confirmed the CLAs. Not sure if I have a bit or not but I dont think so | 15:15 | |
| jnthn | Yeah, it even made it into p6weekly | 15:17 | |
| bartolin | [Coke] confirmed that TPF got our CLAs: irclog.perlgeek.de/perl6-dev/2017-...i_13943097 | 15:18 | |
| jnthn | Invites sent | 15:23 | |
| bartolin | thanks a lot! | ||
| [Coke] | jnthn++ handling the last mile | 15:40 | |
| Geth | ast: bcb8ea57c2 | usev6++ | 3 files Fudge some newly added tests for JVM |
16:28 | |
| ast: 69493573bd | usev6++ | S32-basics/warn.t Unfudge passing test for JVM |
16:35 | ||
| kudo/nom: 3e28b1eb3e | (Elizabeth Mattijsen)++ | src/core/Any-iterable-methods.pm Optimize Any::IterateOneWithoutPhasers.push-all a bit - change the return sig to IterationEnd - add an extra scope for faster JITting and OSRing - makes @b = @a.map: { } about 10% faster - makes @b = do for @a { } about 20% faster Having an additional scope make things faster feels counter-intuitive. Feels to me something's wrong with the way a scope of a method/sub is handled wrt to spesh/JIT, if a bare scope gives us so much extra performance. |
19:51 | ||
| jnthn | lizmat: Smaller things will be spesh'd/JIT-compiled sooner (smaller things get hot faster) | 20:21 | |
| So it may be that | |||
| It may also be one of a whole load of thing things, including a finer-grained deopt | |||
| lizmat | jnthn: the extra scope does not have any code outside of it :-) | ||
| it is literally sub a { { ... } } | 20:22 | ||
| jnthn | Hm, yes, wat | ||
| I wonder if it's spesh at all or something with the static optimizer... | |||
| lizmat | my @a = ^1000; for ^1000 { my @b = do for @a { 42 } } # my benchmark | 20:23 | |
| jnthn | Note that it's still smaller in so far as the signature is compiled in the outer block, but even then that'd not add a ton | ||
| lizmat | my @a = ^1000; for ^1000 { my @b = @a.map: { 42 } } # other benchmark | ||
| jnthn | Feel free to file it as a MoarVM oddity :) | ||
| lizmat | ok, will do :-) | ||
| after I finish the P6W :-) | 20:24 | ||
| jdv79 | another one? seems every other day there's a new one! | 20:25 | |
| jnthn | D'oh, another week when I didn't get a blog post done. I have like 2 blog post in my queue of "stuff to write about"... | 20:26 | |
| lizmat | well, if did feel like a week of blue mondays the past week | ||
| stmuk_ | rakudo.org/2017/01/30/announce-raku...e-2017-01/ | 20:28 | |
| lizmat | stmuk_++ | 20:29 | |
| jnthn | \o/ | 20:30 | |
| TimToady | throwing extras scopes in sounds offhand like it could bite us on closure cloning policy | 20:49 | |
| though it doesn't seem to in this case | 20:55 | ||
| and I was worried about inserting scopes into the user's AST, and I see this is just in iterator implementation, so nevermind :) | 20:57 | ||
| lizmat | yeah, it's pretty harmless I think, but weird nonetheless | 21:04 | |
| twitter.com/perl6org/status/825392308606611458 # seem to miss something essential here | 21:10 | ||
| RabidGravy | bit 10% for a common pattern isn't to be sneezed at | ||
| lizmat | m: say "a" ~~ <:Letter> | ||
| camelia | rakudo-moar 3e28b1: OUTPUT«False» | ||
| RabidGravy | but | 21:12 | |
| m: say "a" ~~ /<:Letter>/ | |||
| camelia | rakudo-moar 3e28b1: OUTPUT«｢a｣» | ||
| lizmat | arg | 21:13 | |
| brokenchicken | m: say ":Letter" ~~ <:Letter> | ||
| camelia | rakudo-moar 3e28b1: OUTPUT«True» | ||
| lizmat | flu has muddled my brain | ||
| stmuk_ | get better for FOSDEM! | ||
| RabidGravy | my brain is always mouddled :) | ||
| lizmat | m: say "a" ~~ /<:Digit>/ # huh ? | ||
| camelia | rakudo-moar 3e28b1: OUTPUT«｢a｣» | ||
| lizmat | m: say "g" ~~ /<:Digit>/ # not just hex ? | 21:14 | |
| camelia | rakudo-moar 3e28b1: OUTPUT«｢g｣» | ||
| jnthn | Think there's already an RT about that one :) | ||
| lizmat | ah, ok | ||
| jnthn | Digit apparently matches anything | ||
| lizmat | I guess it all depends on your base :-) | ||
| jnthn | All your base are belong to us! | 21:15 | |
| lizmat | and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2017/01/30/...e-mondays/ | 22:11 | |
| jnthn | lizmat++ # thanks for doing these! | 22:17 | |
| Wow, I totally missed the couple of percent TimToady++ parse speedup | 22:18 | ||
| lizmat | well, I will be on holiday for the next two weeks | ||
| jnthn | Ah, found the patch | 22:19 | |
| TimToady++ | |||
| [Coke] | lizmat: OH NOES HAVE FUN | 22:20 | |
| lizmat | [Coke]: we will, if we make it past immigration in time | 22:22 | |
| [Coke] | you coming to the US? | 22:23 | |
| if so, best of luck. :| | |||
| lizmat | [Coke]: yeah, we have a window of about 14 hour to make it from Orlando to Tampa | 22:24 | |
| fortunately, I'm from a country aspiring to be #2 | 22:25 | ||
| www.youtube.com/watch?v=j-xxis7hDOE # in case you didn't get the #2 reference | 22:26 | ||
| good night, #perl6-dev! | 22:36 | ||
| brokenchicken | night | ||
| b2gills | m: #`(It seems like <:Digit> is reversed) .print for (^0x30000)».chr.grep: /<:!Digit>/ # note the ｢!｣ | 23:38 | |
| camelia | rakudo-moar 3e28b1: OUTPUT«0123456789²³¹¼½¾٠١٢٣٤٥٦٧٨٩۰۱۲۳۴۵۶۷۸۹߀߁߂߃߄߅߆߇߈߉०१२३४५६७८९০১২৩৪৫৬৭৮৯৴৵৶৷৸৹੦੧੨੩੪੫੬੭੮੯૦૧૨૩૪૫૬૭૮૯୦୧୨୩୪୫୬୭୮୯୲୳…» | 23:39 |