#raku-dev on 21 January 2020 - Raku Programming Language Log

jnthn

patrickb: In principle, yes, I can mentor...will have to ponder project ideas a bit.

00:12

patrickb

jnthn, samcv, MasterDuke: Thank you for taking the time to think about mentoring and ideas!

00:30

I'm off to bed. Night everyone!

00:44

00:44 patrickb left 07:15 domidumont joined

nine

How much time does mentoring need? How much latency is still OK? I'd be interested, just not 100 % sure I have the time.

07:44

As to project ideas, that's much harder. I usually act on my ideas :) The projects I haven't done yet are more on the edge of or even beyond my current abilities, so probably not good for a newbie.

07:46

09:18 b2gills left 09:25 sena_kun joined 09:33 b2gills joined

sena_kun

releasable6, status

10:20

releasable6

sena_kun, Next release will happen when it's ready. 1 blocker. 0 out of 508 commits logged

sena_kun, Details: gist.github.com/48a922a744fb4a0105...1d2cbaa861

nine

I don't think a bug in Documentable is a real blocker for the release

10:21

sena_kun

of course. :)

the current state of things is: 1)we are very ready for moarvm point release; 2)right now I am updating the release branch to be up to date; 3)after moarvm release I am cutting a release ASAP

10:22

10:34 Altai-man_ joined 10:36 sena_kun left

lizmat

Files=1301, Tests=109805, 210 wallclock secs (28.50 usr 8.04 sys + 2943.37 cusr 241.31 csys = 3221.22 CPU)

10:46

Altai-man_ ++

10:47

Geth

rakudo: e3cbd32c6b | (Elizabeth Mattijsen)++ | src/core.c/Str.pm6
Add Version coercer to Str

This not only allows you to just do `"foo".Version` to get a Version object, this also allows you to use `Version(Str)` in a signature.

11:03

11:16 squashable6 left 11:19 squashable6 joined

Geth

rakudo: 11a8d57f6d | (Elizabeth Mattijsen)++ | src/core.c/Perl.pm6
Save 2 Scalar allocations at each startup

And make the version cache a little more optimised

11:40

nqp/compressed_lex_loc_types_serialization: f7631eb747 | (Timo Paulssen)++ | src/vm/moar/QAST/QASTCompilerMAST.nqp
"Compress" types of locals and lexicals in serialization

"obj" registers/lexicals are overwhelmingly common,
so leaving their data out is worthwhile.
The result is a 1.85% size reduction of CORE.c.setting

11:54

Altai-man_

phew

12:04

changelog is ready, now $dayjob

lizmat

Altai-man_++

12:08

12:28 lucasb joined 12:33 MasterDuke left 12:34 sena_kun joined 12:36 Altai-man_ left

lizmat

Looks like "is built(:bind)" needs some extra work:

13:14

m: role A { has $!a is built(:bind); method TWEAK() { $!a := 42 } }; class B does A { }; dd B.new # yeah, great!

camelia

B.new(a => 42)

lizmat

m: role A { has $!a is built(:bind) }; class B does A { method TWEAK() { $!a := 42 } }; dd B.new # not so great

camelia

Type check failed in binding; expected NQPMu but got Int (42)
in method TWEAK at <tmp> line 1
in block <unit> at <tmp> line 1

lizmat

I wonder if it step one causing this issue:

13:15

m: role A { has $!a is built(:bind); method TWEAK() { $!a := 42 } }; class B does A { }; use BUILDPLAN; note BUILDPLAN(B)

camelia

class B BUILDPLAN:
0: nqp::bindattr(obj,B,'$!a',:$a) if possible
1: vivify nqp::getattr(obj,B,'$!a') if part of a mixin
2: call obj.TWEAK

lizmat

golfed it down to:

13:26

m: role A { has Int $!a }; class B does A { method TWEAK() { $!a := 42 } }; dd B.new

camelia

Type check failed in binding; expected NQPMu but got Int (42)
in method TWEAK at <tmp> line 1
in block <unit> at <tmp> line 1

lizmat

aka nothing to do with "is built(:bind)"

13:27

I guess if we don't install a container for the attribute, then there's nothing to type check against, and NQPMu is what we get to match against

jnthn ^^^ is that assertion correct ?

m: role A { has Int $!a }; class B does A { method TWEAK() { use nqp; nqp::bind($!a,42) } }; dd B.new # naughty workaround

13:30

camelia

B.new

jnthn

lizmat: No, that looks wrong (leaking an NQPMu is certainly wrong)