Perl 6 - the future is here, just unevenly distributed

IRC log for #perl11, 2017-02-07

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

All times shown according to UTC.

Time Nick Message
00:16 sten joined #perl11
01:49 bpmedley joined #perl11
01:49 rurban joined #perl11
04:11 kentnl joined #perl11
06:56 rurban @esaym153: with 'package A; use Moose;  has "name" => (is => "ro", isa => "Str", required => 1); package main; my $z = A->new(name => 1);' there's not much to speed optimize. rather than using Mouse (5x faster). You'd need longer examples. and -Accflags=-DFAKE_SIGNATURES
07:00 rurban -DPERL_FAKE_SIGNATURES it is
07:00 willthechill rurban: is there a web page which lists the proper benchmarks for showing off cperl's performance & memory benefits?
07:00 rurban yes.
07:01 rurban http://perl11.org/cperl/STATUS.html
07:02 willthechill okay I'm looking at it now, which part shows the benchmark commands?
07:03 rurban perlbench-run -s --color cperl=./perl
07:03 rurban cpan PerlBench
07:04 rurban But these are only micro benchmarks, not realistic ones. it doesn't catch the method speedup with 5.22
07:05 rurban running testsuites is much better
07:05 willthechill which test suites?
07:05 rurban your
07:06 willthechill well esaym153 is not showing much success running his own test suites, that is why I am asking which test suites do you recommend in order to show off cperl?
07:06 rurban I improved the core testsuite from 14min to 2min
07:06 willthechill okay great
07:06 rurban this was not a good test :)
07:06 willthechill so what is the command for that?
07:06 rurban time make test
07:07 rurban for memory it's massif
07:07 willthechill okay that is a good start
07:09 rurban my mini bench is: #!/usr/bin/perl
07:09 rurban my @a = (0..31);
07:09 rurban my %h = (1..100);
07:09 rurban my $s = '0123456789';
07:09 rurban sub f {
07:09 rurban my ($n) = @_;
07:09 rurban $n % 2 and bless{1..4} and $h{$n} and $a[$n] and $s =~ s/$/../;
07:09 rurban $n<2 and return $n;
07:10 willthechill crazy
07:12 rurban ackermann, lots of functions calls, with a mix of some other stuff in between
07:12 willthechill cool
07:13 rurban only my ($n) = @_; is optimized by cperl
07:16 rurban e.g. there's no add and mult, which are faster in cperl with new gcc/clang
07:16 rurban @a and $n are not typed, which would be much faster
07:17 willthechill okay so I guess you still need a nice micro-benchmark which shows off cperl
07:21 rurban that's unfair
07:21 willthechill I'm just going off of what you just told me
07:21 willthechill "only my ($n) = @_; is optimized by cperl"
07:21 willthechill not unfair at all
07:21 rurban because perl doesn't support shaped arrays and types
07:21 willthechill that's fine, create a cperl-only benchmark
07:22 willthechill all my benchmarks are rperl-only benchmarks, because normal Perl does not support types
07:22 rurban na, I rather fix the big parts. warnings XS right now. unicode next
07:25 willthechill well the fact remains that users such as esaym153 still need a bit more guidance on how to test out the benefits of cperl
07:26 rurban he will try his big samples
07:26 willthechill yes that is good
07:26 rurban A oneliner will only see strict/DynaLoader/Config benefits
07:28 willthechill *nods*
07:47 rurban need to reboot…
07:58 rurban joined #perl11
09:11 anton joined #perl11
09:21 sivoais joined #perl11
12:48 rurban joined #perl11
14:20 rurban1 joined #perl11
16:59 rurban joined #perl11
20:48 willthechill joined #perl11
21:52 esaym153 rurban: ah, so I need to recompile with -DPERL_FAKE_SIGNATURE?
21:52 esaym153 seems like that was supposed to be default in 5.23, guess it didn't make it?
22:15 willthechill joined #perl11
23:08 bpmedley joined #perl11

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