🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | log inspection situation still under development | For MoarVM see #moarvm
Set by lizmat on 22 May 2021.
Geth rakudo: vrurg++ created pull request #4849:
Use nqp::rethrow instead of .rethrow		 01:22
lizmat Files=1352, Tests=117130, 290 wallclock secs (35.54 usr 9.41 sys + 4050.22 cusr 329.84 csys = 4425.01 CPU) 09:18
Geth DBIish: massa++ created pull request #223:
Look up mariadb lib		 10:29
Geth DBIish: massa++ created pull request #224:
Eliminate error when integer field was null		 10:37
Geth rakudo: 1b75e3fa0c | (Elizabeth Mattijsen)++ | 4 files
Streamline CompUnit::DependencySpecification logic

Specifically with regards to the xxx-matcher logic. Before, one had to explicitely specify True to indicate no interest in matching that matcher (which would essentially match against True, and thus always pass). And if something *was* specified for "version" or
  "api", then at several places a Version object was made to get the
... (10 more lines)		 12:08
Geth rakudo: 17573da271 | Coke++ | src/core.c/CompUnit/DependencySpecification.pm6
typo in comments
 12:27
nine lizmat: that True was by design: github.com/rakudo/rakudo/blob/mast...cification 12:28
lizmat .version-matcher returns True if not specified, or explicitely specified 12:29
so it's not that exact code, but functionally the same
Geth rakudo: d24dd0ab10 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/Repository/Installation.pm6
Streamline CURI.files

Mostly about not creating intermediate variables, and making sure that we only return meta info for existing files to begin with, instead of first returning the meta and *then* check if the file exists.
Also add a BIND-KEY method to the MetaAssigner role so that we can transparently bind to something that looks like a hash.		 14:04
DBIish/rbt.mariadb: ddfa8b5a9c | (Humberto Massa)++ | lib/DBDish/mysql.pm6
Look up mariadb lib

Makes it look for mariadb lib before it looks for mysql lib, rendering extra configuration unnecessary in recent versions of Debian/Ubuntu		 14:29
rakudo: 1b4977c67d | (Elizabeth Mattijsen)++ | src/core.c/Slip.pm6
Shortcut .map/.grep/.first on Empty

The empty Slip (aka Empty) is returned in many cases. This commit adds Slip candidates for .map / .grep / .first that will check whether the invocant is indeed the special Empty case, and then just return the appropriate thing (Empty in case of .map / .grep, Nil in case of
  .first). In all other cases, run the normal handling of .map / .grep
  / .first.
Makes .map on the empty Slip about 4x as fast.		 14:40
DBIish/main: ddfa8b5a9c | (Humberto Massa)++ | lib/DBDish/mysql.pm6
Look up mariadb lib

Makes it look for mariadb lib before it looks for mysql lib, rendering extra configuration unnecessary in recent versions of Debian/Ubuntu		 14:47
Geth rakudo: 8f5a07d036 | (Elizabeth Mattijsen)++ | src/core.c/Slip.pm6
Also short-cut Empty.head / .tail

These also can be handled very quickly if they're called on Empty
Also remove the names from the signature: the positionals are only needed for dispatch, not actually needed in that method itself.		 15:13
roast: fa6dd583ed | (Elizabeth Mattijsen)++ | S11-repository/cur-candidates.t
Re-arrange CUR testing a bit

Having only a single test with lots of subtest inside subtest, is not very nice when trying to debug issues.		 15:56
rakudo: 59c9d80b3e | (Elizabeth Mattijsen)++ | src/core.c/CompUnit.pm6
Fix CompUnit.WHICH

The CompUnit::Repository::Distribution class *does* exist, but not at that point when compiling the setting just yet. So it needs to be stubbed.		 16:29
rakudo: d9b24c9d71 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/Repository/Installation.pm6
Streamline CURI!matching-dist

  - use a low-level hash, so we can use nqp::ifnull
  - only use a single protect block
  - introduce a TWEAK method to set up seen hash
  - also use that to initialize the lock now that we have a TWEAK		 17:18
rakudo: 953d4fae54 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/Repository/Installation.pm6
Streamline CURI.resolve

  - no returns
  - hash access to a hash at compile time
  - only call next-repo once		 17:53
[Tux] Rakudo v2022.03-138-gd9b24c9d7 (v6.d) on MoarVM 2022.03-13-ga3476e286
csv-ip5xs0.837 - 0.940
csv-ip5xs-205.175 - 6.236
csv-parser3.730 - 3.742
csv-test-xs-200.402 - 0.423
test7.159 - 7.174
test-t1.477 - 1.540
test-t --race0.867 - 0.874
test-t-2021.088 - 21.498
test-t-20 --race7.402 - 7.960
 18:01
Geth rakudo: 94576f0916 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/Repository/Installation.pm6
Streamline CURI.need

  - use native has for caching
  - use binding instead of assigning
  - use more descriptive variable names
  - add more comments
  - focus on fast-path cached and already precompiled cases		 19:30
Geth rakudo: b578d15f83 | (Elizabeth Mattijsen)++ | src/core.c/CompUnit/Repository/Installation.pm6
Streamline CURI.resource

  - use native hash for caching, allowing for ifnull usage
  - make sure that caching distribution metas is thread-safe
  - remove check for repo versions < 2
Note that this removes the check for repo's of repository level 0 ... (7 more lines)		 21:26
lizmat and that concludes my hacking for today&

Powered by Cro and the Raku® Programming Language.

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

Thank you!