Perl 6 - the future is here, just unevenly distributed

IRC log for #perl11, 2015-09-13

| Channels | #perl11 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
04:19 bulk88_3 joined #perl11
04:20 bulk88_2 joined #perl11
05:20 travis-ci RPerl build passed. Will Braswell says 'To Do List, Misc Update'
05:20 travis-ci https://travis-ci.org/wbraswell/rperl/builds/80059914 https://github.com/wbraswell/rperl/compare/ebc37598770a...e01e83bb5c2c
06:40 rurban joined #perl11
07:04 travis-ci perl11/cperl#12 (master - f455765 : Reini Urban): The build passed. https://travis-ci.org/perl11/cperl/builds/80067608
07:43 travis-ci perl11/cperl#13 (master - f1f7872 : Reini Urban): The build was broken. https://travis-ci.org/perl11/cperl/builds/80069019
08:16 travis-ci perl11/cperl#14 (master - 89a57d4 : Reini Urban): The build is still failing. https://travis-ci.org/perl11/cperl/builds/80070640
08:49 travis-ci perl11/cperl#15 (bugfix/gh28-config-boot - c49116a : Reini Urban): The build failed. https://travis-ci.org/perl11/cperl/builds/80072278
09:55 travis-ci perl11/cperl#16 (bugfix/gh29-stash-destruct - 62d3822 : Reini Urban): The build failed. https://travis-ci.org/perl11/cperl/builds/80076105
10:15 travis-ci perl11/cperl#17 (bugfix/gh28-config-boot - e4188ec : Reini Urban): The build was fixed. https://travis-ci.org/perl11/cperl/builds/80077222
10:38 basiliscos joined #perl11
11:23 travis-ci perl11/cperl#18 (bugfix/gh29-stash-destruct - 32ad817 : Reini Urban): The build is still failing. https://travis-ci.org/perl11/cperl/builds/80081995
11:36 travis-ci perl11/cperl#19 (master - d98e9ea : Reini Urban): The build is still failing. https://travis-ci.org/perl11/cperl/builds/80082649
11:48 travis-ci perl11/cperl#20 (master - d9a0aa7 : Reini Urban): The build was fixed. https://travis-ci.org/perl11/cperl/builds/80083128
13:03 basiliscos1 joined #perl11
14:07 rurban joined #perl11
14:40 travis-ci perl11/cperl#21 (bugfix/gh29-stash-destruct - 163cdd3 : Reini Urban): The build was canceled. https://travis-ci.org/perl11/cperl/builds/80099549
14:57 travis-ci perl11/cperl#22 (bugfix/gh29-stash-destruct - 0abfcd6 : Reini Urban): The build passed. https://travis-ci.org/perl11/cperl/builds/80099664
16:28 travis-ci perl11/cperl#23 (master - 4962ec7 : Reini Urban): The build was broken. https://travis-ci.org/perl11/cperl/builds/80108627
16:43 travis-ci perl11/cperl#24 (bugfix/gh31-threads+ro-stashes - 26cb973 : Reini Urban): The build passed. https://travis-ci.org/perl11/cperl/builds/80109196
17:20 travis-ci perl11/cperl#26 (master - 685d776 : Reini Urban): The build failed. https://travis-ci.org/perl11/cperl/builds/80115410
17:46 travis-ci perl11/cperl#27 (bugfix/gh30-thread-tests - 66835fe : Reini Urban): The build failed. https://travis-ci.org/perl11/cperl/builds/80118016
17:57 travis-ci perl11/cperl#28 (master - 1cf89c7 : Reini Urban): The build is still failing. https://travis-ci.org/perl11/cperl/builds/80119066
18:18 travis-ci perl11/cperl#29 (master - 636f245 : Reini Urban): The build is still failing. https://travis-ci.org/perl11/cperl/builds/80119799
18:30 willthechill joined #perl11
18:36 travis-ci perl11/cperl#30 (master - f118019 : Reini Urban): The build was fixed. https://travis-ci.org/perl11/cperl/builds/80121985
18:36 willthechill rurban: why do many commits?
18:37 willthechill too many too many!
18:37 rurban release tomorrow
18:37 rurban seems to be fixed now
18:38 willthechill oh good release!
18:42 rurban I've also got macosx tests on travis
18:43 willthechill how?
19:00 travis-ci perl11/cperl#31 (travis - c53363e : Reini Urban): The build is still failing. https://travis-ci.org/perl11/cperl/builds/80125927
19:03 rurban http://docs.travis-ci.com/user/multi-os/ https://github.com/perl11/cperl/blob/master/.travis.yml
19:04 j0e joined #perl11
19:25 travis-ci perl11/cperl#32 (travis - 5156bd4 : Reini Urban): The build is still failing. https://travis-ci.org/perl11/cperl/builds/80126321
19:29 rurban 276 secs for the whole cperl testsuite! perl5 needs 436 wallclock secs with less tests. almost 2x faster even without sigs.
19:31 rurban oh no, perl5 was with DEBUGGING
19:33 rurban It's just 298 vs 426
19:34 rurban just 43% faster with DEBUGGING
19:46 willthechill how much faster with debugging turned off in both normal perl and cperl?
19:47 travis-ci perl11/cperl#34 (master - d5fd21f : Reini Urban): The build was broken. https://travis-ci.org/perl11/cperl/builds/80127911
19:48 rurban no numbers
19:49 rurban 276 for cperl, but nothing for perl5
19:49 willthechill why nothing for perl?
19:49 rurban on my machine it's usually 520
19:49 rurban no comparable tests yet
19:50 rurban i took the numbers from tonyc's tests and he did no non-DEBUGGING builds
19:50 willthechill okay well the bottom line is, how much faster is non-debugging cperl vs non-debugging normal perl?
19:50 rurban we will see when I get the numbers
19:50 rurban estimated 45% faster
19:51 rurban with sigs it's ~100% faster, but I cannot yet parse \n within sigs
19:54 willthechill and how much faster is perlcc than normal perl?
19:57 rurban I got core testsuite results for perlcc in my yapc talk
19:57 rurban https://rurban.github.io/perl-compiler/yapceu2015.html#21
19:58 rurban that's unoptimized, only startup and destruction time improvements
19:58 rurban no times for CC
19:58 willthechill these numbers don't give me much info
19:58 willthechill I just want to know how fast perlcc is compared to normal perl!
19:59 rurban https://rurban.github.io/perl-compiler/yapceu2015.html#11
19:59 rurban those numbers are the same since years
20:00 willthechill I am only talking about runtime
20:00 willthechill not memory
20:00 willthechill not startup time
20:00 willthechill not destruction time
20:00 willthechill only RUNTIME
20:00 willthechill how much faster is perlcc RUNTIME than normal perl RUNTIME?
20:00 willthechill oh dang sorry I was ambigous
20:00 rurban we are talking about B::C which has the same runtime
20:00 willthechill *ambiguous
20:01 rurban B::CC times are here: https://github.com/rurban/perl-compiler/blob/master/ramblings/blogs-optimizing-3.md
20:01 willthechill I am talking about B::CC not B::C
20:01 willthechill my bad!
20:01 willthechill how much faster is perlcc B::CC runtime versus normal perl runtime?
20:01 willthechill (not B::C)
20:01 rurban 22m14s => 3m58.249s
20:02 willthechill okay so about 5x - 6x runtime speedup from normal Perl to B::CC
20:02 willthechill yes?
20:03 rurban yes
20:03 rurban with nbody
20:03 rurban with regex almost nothing
20:04 willthechill sure regex is exempt
20:05 willthechill it is a different computational engine than Perl
20:05 willthechill so B::CC is only slightly less speedup than RPerl in CPPOPS_PERLTYPES mode
20:05 willthechill which gives 7x - 10x speedup
20:05 willthechill good to know
20:05 rurban yes
20:06 rurban and cperl only about 2x, with native types I hope 4x
20:06 rurban php 7 is much faster
20:06 rurban but cperl can run 99.999% of CPAN
20:07 willthechill yes that is good it can run CPAN
20:07 willthechill very very good
20:08 rurban and with the object system it will be ~10x faster than Moose and perl6
20:08 willthechill that's good
20:08 willthechill of course RPerl's object system is hundreds to thousands of times faster
20:09 willthechill but still some people want crazy MOP crap
20:09 willthechill so that's fine for them!
20:09 rurban of course. but mine is dynamic
20:09 rurban I will not do MOP, just reflection and mixins
20:09 willthechill okay that's good I guess
20:10 rurban People have no idea what a MOP is good for. We don't need one.
20:10 willthechill some people will like it, I'm sure!
20:10 willthechill MOP is crap
20:10 willthechill nice theory for CS grad students
20:10 willthechill that's it
20:10 rurban metaclasses are fine, but I'm not sure if I will add them. later probably to slow everything down
20:11 rurban types are much more important
20:11 willthechill uh yeah there is no comparison between the importance of types and metaclasses
20:11 willthechill types == 1, metaclasses == 0
20:13 rurban I'm still not sure if I should make classes finalized as default. One can always derive from it to extend it for Test mockers
20:13 rurban damian went for final, larry for loose
20:13 willthechill the only answer is: which one is faster?
20:14 rurban final is using a compile-time vtable, loose a dynamic hash
20:14 rurban about 20x faster I guess
20:14 rurban fixed offsets for fields and methods, for classes and mixins.
20:15 willthechill well C++ uses a compile-time vtable, so that's what RPerl uses
20:16 rurban parent lookup is slow of course. same problem as with C++. dynamic dispatch
20:16 rurban esp with multiple dispatch on typed arguments
20:16 rurban namemangling or not is the question
20:17 rurban I'll do that next month
21:04 bulk88_2 joined #perl11
22:36 rurban joined #perl11
22:58 travis-ci perl11/cperl#35 (feature/CM-712-cperl-types-proto - feeb260 : Reini Urban): The build has errored. https://travis-ci.org/perl11/cperl/builds/80148548
23:19 travis-ci perl11/cperl#36 (feature/CM-626-cperl-use-dots - f551010 : Reini Urban): The build failed. https://travis-ci.org/perl11/cperl/builds/80148570
23:41 travis-ci perl11/cperl#37 (bugfix/CM-891-rt123878-goto-die - c3b0e3b : Reini Urban): The build failed. https://travis-ci.org/perl11/cperl/builds/80148589
23:53 travis-ci perl11/cperl#38 (feature/gh16-multi - 9a979fd : Reini Urban): The build failed. https://travis-ci.org/perl11/cperl/builds/80148614

| Channels | #perl11 index | Today | | Search | Google Search | Plain-Text | summary