[Tux] Rakudo v2020.12-21-g94d685d95 (v6.d) on MoarVM 2020.12-14-g78f4f1d9e
csv-ip5xs0.805 - 0.819
csv-ip5xs-208.152 - 8.305
csv-parser26.333 - 27.142
csv-test-xs-200.385 - 0.388
test7.838 - 7.935
test-t1.890 - 2.112
test-t --race0.892 - 0.907
test-t-2031.851 - 32.972
test-t-20 --race9.165 - 9.361
 13:42
dogbert17 lizmat: is stuff like this supposed to work in your new slice impl? 14:52
m: dd ("a".."z")[(3, (4, (5,)))]:p
camelia (3 => "d",)
dogbert17 i.e. other cases from github.com/Raku/old-issue-tracker/issues/6233
lizmat dogbert17: yes, that should work 14:54
except that that piece of code runs into another improper use of nqp::slice 14:55
dogbert17 yeah, as soon as an adverb is added things seem to go wrong 14:59
lizmat oddly enough, this is *not* picked up by roast :-( 15:00
dogbert17 I guess we could borrow the examples from the bug report
although they also exposes errors so we would need to figure out what the correct results should be 15:01
lizmat $ r 'dd ("a".."z")[(3, (4, (5,)))]:p' 15:06
(3 => "d", (4 => "e", (5 => "f",)))
dogbert17 ^^
Geth rakudo/faster-slice-access: 563be8dc7a | (Elizabeth Mattijsen)++ | 2 files
Fix other improper uses of nqp::slice
 15:07
dogbert17 since I got a comment on my roast PR yesterday perhaps I should add your case above (and possibly the other array related ones as well) ?
Geth rakudo/faster-slice-access: c3c565e216 | (Elizabeth Mattijsen)++ | 2 files
How did that e get there? grrr
 15:09
lizmat dogbert17: yes please
dogbert17 I guess an amend is in order then 15:10
dogbert17 tries 15:11
Geth roast: dogbert17++ created pull request #712:
Add tests for nested array slices		 16:00
dogbert17 lizmat: I made an entirely new PR, git was against me :( 16:01
lizmat that's ok :-)
dogbert17 now all adverbs are tested 16:02
lizmat cool! 16:03
dogbert17 lizmat: FYI, running a spectest uncovers errors in t/spec/S32-array/exists-adverb.t and t/spec/S32-array/adverbs.t 16:15
haha, so much for all adverbs :-) 16:17
m: dd ("a".."z".list)[3, 30, 5]:exists:v 16:38
camelia Failure.new(exception => X::Adverb.new(what => "slice", source => "Range", unexpected => [], nogo => ["exists", "v"]), backtrace => Backtrace.new)
dogbert17 should stuff like this work?
lizmat nope: :exists and :v cannot work together as :exists implies no filtering, and :v does 16:46
:exists:!v should be legal though, that's one of the errors
Geth rakudo/faster-slice-access: aea57b1840 | (Elizabeth Mattijsen)++ | src/core.c/array_slice.pm6
Rince/repeat on the @a[*-1]:adverb candidates

Makes the simple @a[*-1]:adverb about 3.5x as fast, and the compound
  @a[*-1]:foo:bar case about 2x as fast.		 17:34
rakudo/faster-slice-access: ccf4f921df | (Elizabeth Mattijsen)++ | 9 files
Remove SLICE_ONE_LIST and SLICE_MORE_LIST subs

These implemenation details are no longer necessary.		 18:05
lizmat afk for a few hours&
Geth rakudo: 754b285053 | (Elizabeth Mattijsen)++ | 2 files
Add "is-implementatio-detail" delegation

As suggested by vrurg++		 22:03
lizmat argh,... wrong branch 22:04
MasterDuke lizmat: btw, 'rince' -> 'rinse' 22:05
lizmat oops
:-)
Geth rakudo: eb1eabf6a5 | (Elizabeth Mattijsen)++ | 2 files
Revert "Add "is-implementatio-detail" delegation"

This reverts commit 754b285053aed8bc29076ea2622638a9ef70c237.
Shouldn't have happened on the master branch.		 22:06
rakudo/class-implementation-detail: aa469bd08e | (Elizabeth Mattijsen)++ | 2 files
Add "is-implementatio-detail" delegation

As suggested by vrurg++		 22:07

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!