#perl6-dev on 30 June 2018 - Raku Programming Language Log

Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_logs/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by AlexDaniel on 12 June 2018.

00:11 MasterDuke left 00:51 MasterDuke joined

Geth

nqp/new-runner-enable-execname: 2e65fa1f0d | 陈梓立++ (committed using GitHub Web editor) | t/nqp/114-pod-panic.t
Update 114-pod-panic.t

01:51

02:01 MasterDuke left

Geth

nqp/W4anD0eR96-patch-1: 1c59d2b68c | 陈梓立++ (committed using GitHub Web editor) | t/nqp/022-optional-args.t
Test invalid syntax on optional args

02:16

nqp: W4anD0eR96++ created pull request #480:
Test invalid syntax on optional args

02:17 MasterDuke joined

Geth

rakudo: jstuder-gh++ created pull request #1995:
Issue 1975: Fix WIN32 Proc::Async process termination issue

02:29

MasterDuke

nqp: say(nqp::tc("hello world"))

02:30

camelia

HELLO WORLD

MasterDuke

nqp: say(nqp::tclc("hello world"))

02:32

camelia

Hello world

MasterDuke

m: say "hello world".tc

camelia

Hello world

MasterDuke

m: say "hello world".tclc

camelia

Hello world

02:51 perlpilot left 02:54 perlpilot joined

AlexDaniel

6c: say Rat.new(10⁴⁰⁰, 9⁹⁹⁹)

03:19

committable6

AlexDaniel, gist.github.com/3f3c4f435f5d19d499...efdf1eefd1

AlexDaniel

↑ that's an interesting progression

03:34 AlexDaniel left, AlexDaniel joined

Geth

nqp: 1c59d2b68c | 陈梓立++ (committed using GitHub Web editor) | t/nqp/022-optional-args.t
Test invalid syntax on optional args

04:08

nqp: 0d797a89f3 | 陈梓立++ (committed using GitHub Web editor) | t/nqp/022-optional-args.t
Merge pull request #480 from perl6/W4anD0eR96-patch-1

Test invalid syntax on optional args

nqp/new-runner-enable-execname: 7fe69fc6cd | 陈梓立++ (committed using GitHub Web editor) | t/nqp/114-pod-panic.t
Update 114-pod-panic.t

04:10

nqp/W4anD0eR96-patch-2: 28908b2d5b | 陈梓立++ (committed using GitHub Web editor) | t/nqp/114-pod-panic.t
Update 114-pod-panic.t

04:12

rakudo: zoffixznet self-assigned Crazy Rats stringify as 0.00000…000Inf github.com/rakudo/rakudo/issues/1996
ddf496c72b | (Elizabeth Mattijsen)++ | 2 files

  - Set(|Hash) perlifies as Foo.new(...)
  - Bag(|Hash)|Mix(|Hash) perlifies as (...).Baggy (as coercer)
  - No longer specical case frequency/weight of 1
This seems inconsistent, but there is a reason for this. ... (22 more lines)

04:13

04:17 AlexDaniel left, AlexDaniel joined 04:58 perlpilot left

lizmat

Files=1242, Tests=76407, 320 wallclock secs (15.31 usr 5.24 sys + 2225.32 cusr 221.69 csys = 2467.56 CPU)

05:44

06:23 CIAvash[m] left, AlexDaniel`` left 06:24 ilmari[m] left 06:26 benchable6 left, greppable6 left, squashable6 left, bisectable6 left, releasable6 left, releasable6 joined, squashable6 joined, greppable6 joined, bisectable6 joined, benchable6 joined 06:31 lizmat left 06:54 lizmat joined 07:05 MasterDuke left

lizmat

after a night's sleep, I've come to the conclusion that:

07:06

1. we cannot have similar .perl for Baggy and Setty

07:07

2. that the problem is really in the .perl ification of QuantHashes, *not* in the itemization of Pairs

we want the following both to yield the same Bag:

07:08

(a => 42, b => 666).Bag

[a => 42, b => 666].Bag

if however we would make the interpretation of Pairs dependent on the itemization, then the latter would change in semantics

07:09

as the array iterator delivers containers, rather than values

so: perlification of Settys will use 'Set.new()', as that will take all values given at "face value" (aka no special handling for Pairs)

07:11

perlification of Baggys will use ().Bag syntax *and* not special case when the value is one

07:12

if a Baggy then contains a Pair as a key with value 1, it will be rendered as ":foo => 1" rather than as ":foo"

07:13

removing the ambiguity between a => 42 => 1 and a => 42

07:35 stmuk_ joined, ilmari[m] joined

[TuxCM]

Rakudo version 2018.06-44-g3cbf25e0f - MoarVM version 2018.06-34-g411f73fdc

csv-ip5xs	0.896	-	0.936
csv-ip5xs-20	7.406	-	7.886
csv-parser	24.259	-	24.894
csv-test-xs-20	0.427	-	0.464
test	8.650	-	8.799
test-t	2.300	-	2.384
test-t --race	0.966	-	0.987
test-t-20	42.268	-	42.331
test-t-20 --race	14.309	-	15.343

07:46

08:03 AlexDaniel`` joined, CIAvash[m] joined

10:14 pmurias joined 10:53 MasterDuke joined

MasterDuke

pmurias: around?

10:59

dogbert17

stupid question of the day. Is Hash.kv known to be slower than using Hash.keys and then retrieving the value using the key?

11:01

m: my %colors = ("00" => ["1","2"], "11" => ["0","2"]); for ^1000000 { for %colors.kv -> $k, $v {my $s = "test" ~ $v[1]; }}; say time - INIT time

11:03

camelia

dogbert17

m: my %colors = ("00" => ["1","2"], "11" => ["0","2"]); for ^1000000 { for %colors.keys -> $k {my $s = "test" ~ %colors{$k}[1]; }}; say time - INIT time

camelia

11:04

lizmat

dogbert17: I am not surprised: the iterator for .kv is much more complex, as it needs to keep state between alternating between key and value

11:05

iterating over a hash could maybe be faster using .pairs (or nothing, as .pairs is implicit)

11:06

and then get the .key / .value

dogbert17

lizmat: thx, so if you're after speed, kv should be avoided in preference of keys or pairs then

lizmat

well, benchmark to be sure, I would say

11:07

dogbert17

lizmat

but .kv is simply more complicated, so slower

dogbert17

m: my %colors = ("00" => ["1","2"], "11" => ["0","2"]); for ^1000000 { for %colors.pairs -> $p {my $s = "test" ~ $p.value[1]; }}; say time - INIT time # like this

11:08

camelia

dogbert17

if I did that right it runs quite a bit faster than kv

11:09

lizmat

dogbert17: s/time/now

dogbert17

oops

m: my %colors = ("00" => ["1","2"], "11" => ["0","2"]); for ^1000000 { for %colors.pairs -> $p {my $s = "test" ~ $p.value[1]; }}; say now - INIT now # like this

camelia

5.233219

lizmat

yes, otherwise you only get second granularity

11:10

dogbert17

m: my %colors = ("00" => ["1","2"], "11" => ["0","2"]); for ^1000000 { for %colors.kv -> $k, $v {my $s = "test" ~ $v[1]; }}; say now - INIT now

camelia

9.0061334

dogbert17

cool, I'm looking at onw of rindolf's benchmarks where p5 crushes p6. He uses kv a lot there

11:11

MasterDuke

pmurias: is the argument to the @Deserializer annotation needed/used for anything?

11:13

pmurias

MasterDuke: it's used when building the truffle ast

11:17

when passing the array structure to nqp::runtruffle the argument is used for the ast name (that's in the first array element)

11:18

MasterDuke: if the argument is not present it's generated from the class name

MasterDuke

pmurias: not all the annotations had an argument, but that could be ok then?

11:19

pmurias

MasterDuke: yes, it's generated by default

MasterDuke: so for NQPConcatNode it ends up being "concat"