Geth rakudo: vrurg++ created pull request #3834:
Partial fix of mixing in a role with attributes into a NQP class
00:35
Geth nqp/roast_657_attr_fix: 497c2e550e | (Vadim Belman)++ | src/how/NQPClassHOW.nqp
Support for typechecking against Raku roles

If a `NQPClassHOW` class get a Raku role mixin then typechecking fails for complex cases like parameterized roles. By incorporating
  `role_typecheck_list` provided by role's HOW we fix the problem.
NQP, Rakudo, and roast tests are passing.
01:54
nqp: vrurg++ created pull request #650:
Support for typechecking against Raku roles
01:55
Geth roast/roast_657: 82369b4a6c | (Vadim Belman)++ | S14-traits/package.t
Adapt the test to the syntax and semantics of Raku v6.c

Requires Raku/nqp#650 and rakudo/rakudo#3834 to pass.
I have removed trait `described` because the final implementation of traits doesn't need covering two different trait routine signatures.
... (6 more lines)
02:45
linkable6 NQP#650 [open]: github.com/Raku/nqp/pull/650 Support for typechecking against Raku roles
linkable6 RAKUDO#3834 [open]: github.com/rakudo/rakudo/pull/3834 Partial fix of mixing in a role with attributes into a NQP class
Geth roast/roast_657: abca43f97a | (Vadim Belman)++ | 2 files
Adapt test for Raku v6.c syntax and semantics

  - change trait signatures to match v6.c requirements
  - rename role `doc` into `Doc` for `is` to be able to distinguish
   between the two
  - because the role is applied to the container, invoke the `doc` method
   on `.VAR`
roast/roast_657: deb54cc64f | (Vadim Belman)++ | 2 files
Fix S11-modules/gh2979.t test

Export semantics requires a `Map` to be returned by `sub EXPORT` in order to preserve containers of the exported symbols.
[Tux] Rakudo version 2020.07-46-ga3af5833d - MoarVM version 2020.07-10-g7a2881a85
csv-ip5xs0.814 - 0.816
csv-ip5xs-207.987 - 8.398
csv-parser26.566 - 27.485
csv-test-xs-200.388 - 0.391
test7.493 - 7.712
test-t1.891 - 1.897
test-t --race0.828 - 0.872
test-t-2031.984 - 34.315
test-t-20 --race8.883 - 10.047
13:43
Geth nqp: 497c2e550e | (Vadim Belman)++ | src/how/NQPClassHOW.nqp
Support for typechecking against Raku roles

If a `NQPClassHOW` class get a Raku role mixin then typechecking fails for complex cases like parameterized roles. By incorporating
  `role_typecheck_list` provided by role's HOW we fix the problem.
NQP, Rakudo, and roast tests are passing.
15:09
nqp: 717b0442db | (Vadim Belman)++ (committed using GitHub Web editor) | src/how/NQPClassHOW.nqp
Merge pull request #650 from Raku/roast_657_attr_fix

Support for typechecking against Raku roles
rakudo: ea0c98cf16 | (Vadim Belman)++ | 3 files
Rework tools/update-passing-test-data.pl to support backends

Also improve output formatting, use File::Spec for pathname manipulations, and few other improvements.
rakudo-limited.pl now receives the Rakudo binary to use as its firs command line argument.
Make autounfudge.pl work with updated rakudo-limited.pl
15:13
rakudo: 17f7f156a8 | (Vadim Belman)++ (committed using GitHub Web editor) | 3 files
Merge pull request #3829 from vrurg/roast_657

Rework tools/update-passing-test-data.pl to support backends
JJMerelo You've probably seen this, but TPF is looking for communication people news.perlfoundation.org/post/commu...n_outreach 15:36
There's also this grant proposal by tony-o++ on a new ecosystem. news.perlfoundation.org/post/grant..._ecosystem
[Coke] I just added a ton of questions to tony-o's proposal. 15:43
JJMerelo [Coke]++ 15:45
[Coke] also, may I interest you in writing something for the 20th aniversary calendar? github.com/Raku/advent/tree/master/20th 15:46
[Coke] Probably not, but thank you. 15:49
ugexe > If this is intended to be the primary ecosystem, why does this need a plugin to zef? -- [Coke]: everything in zef is a plugin/handler basically, so essentially it would be a plugin in that respect. however, what should probably be clarified is the difference between the consumer plugin (what would probably ship with zef by default, although turning it on by default does mean additionally latency, so its 16:13
not free), but also like a more authoring specific plugin to do things like push new releases to that ecosystem using appropriate auth.
Geth nqp: 755fa3f97c | (Christian Bartolomäus)++ | 3 files
[JVM] Implement op 'execname'

This is a bit hacky, but it should enable us to use nqp::execname in Rakudo's code base uniformly on all backends.
Currently we use nqp::jvmgetproperties(){nqp.execname} for NQP and nqp::jvmgetproperties(){perl6.execname} for Rakudo. ... (5 more lines)
18:51
gfldex raku: my @a = 'a', Empty, 'b'; say @a.keys; for @a { .say }; say @a».item; say @a.map({$_}); 22:51
evalable6 (0 1)
a
b
[a b]
(a b)
gfldex raku: my @a is default(Empty); @a[0]='a'; @a[2]='b'; say @a.keys; for @a { .say }; say @a».item; say @a.map({$_}); 22:53
evalable6 (0 1 2)
a
()
b
[a b]
(a b)
gfldex So Empty as a default is handled differently then an Empty in a container in an Array. 22:54
That is at least an ENODOC.
In fact, the docs are not well filled with explanations of Empty. 22:55
raku: my @a := 'a', Empty, 'b'; say @a.keys; for @a { .say }; say @a».item; say @a.map({$_});
evalable6 (0 1)
a
b
(a b)
(a b)
gfldex raku: my @a := ('a', Empty, 'b'); say @a.keys; for @a { .say }; say @a».item; say @a.map({$_}); 22:56
evalable6 (0 1)
a
b
(a b)
(a b)