🦋 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: ... | Logs available at irclogs.raku.org/raku-dev/live.html | For MoarVM see #moarvm
Set by lizmat on 8 June 2022.
Geth rakudo: 683ddf4767 | (Vadim Belman)++ | tools/templates/NQP_REVISION
Bump NQP

Prepare for #4993.
00:10
Geth rakudo: 71de22321d | (Vadim Belman)++ | 25 files
Implement nominalizable transparency for definites

The following will now produce correct true output:
   subset Foo of Int();
   say Foo:D.^archetypes.coercive;
... (25 more lines)
01:12
rakudo: 9b13a55a12 | (Vadim Belman)++ (committed using GitHub Web editor) | 25 files
Merge pull request #4993 from vrurg/make-generics-better

Implement nominalizable transparency for definites
nine vrurg: which archetypes method actually uses the $obj parameter? I still cannot find any 07:04
lizmat @vrurg all Inline::Perl5 tests in roast fail for me after pull / rebuild 07:05
lizmat @vrurg all Inline::Perl5 tests in roast fail for me after pull / rebuild 07:06
nine That's the backwards incompatibility we're talking about. 07:13
lizmat meh 07:42
so that implies any installation running with Inline::Perl5 will break if they upgrade to Rakudo 2022.08 without upgrading Inline::Perl5 as well ? 07:44
nine yes 07:54
lizmat meh again :-(
nine And I do not even understand why that change is needed. Why pass a parameter to a method when it doesn't get used?
lizmat so that you can customize it with that parameter ? 07:55
nine But none of the archetypes methods does that.
lizmat well, we still have a few weeks to revert... 07:56
or change
Kaiepi i double checked the hs_log_* produced by the CORE.d.setting failure on the jvm 08:26
usually i kinda gloss over module loader errors because they can happen for entirely unrelated reasons at times, but i noticed this was happening a bit later than is normal for one of those 08:27
it happens in the nqp::loadbytecode op, but not until attempting to get a constructor for a loaded class 08:28
or rather, at a different time 08:29
this line in particular github.com/Raku/nqp/blob/9858768b7...r.java#L71 08:31
thing i don't get is this is happening rather late in the loading of bytecode... why such a huge chunk of memory? 08:32
the files are already loaded, and decompressed
Geth rakudo: 49e04c6193 | (Tom Browder)++ (committed using GitHub Web editor) | CREDITS
Update CREDITS
12:07
vrurg lizmat, nine: archetypes has to have a parameters otherwise there is no solution for examples in the PR (github.com/rakudo/rakudo/pull/4993) 13:07
Unfortunately, in many locations where generics are instantiated, either additional checks for definites are to be introduced (performance), or a single "unused" attribute of archetypes makes the cost much smaller. 13:10
The problem is that definites implemented as parameterizations. Therefore, in order to determine archetypes of underlying type of a definite, one has to have the type itself to resolve its parameters. Since definites are ubiquitous, parameterization spares us memory and performances on creating new metamodel object for each new definite. 13:13
lizmat, nine: Ah, and one more. This is another reason why I see Inline::Perl5 as a part of Rakudo. As long as spectests depend on it to be complete. 13:22
nine vrurg: I do see the effect of your change, i.e. the difference between passing that argument and not. But I can't for the life of mine find the place in the code that actually uses that parameter. 14:10
Ah, god damn it, there it is: github.com/rakudo/rakudo/pull/4993...92c1278R11 14:11
vrurg I see... Right, that's the place. 14:12
BTW, speaking of backward compat, archetypes are not specced and therefore remain implementation detail, as most of the metamodel is. 14:13
Geth rakudo: pamplemoussecache++ created pull request #5015:
Update all ranged copyrights from 2021 to 2022
16:57
Geth rakudo: 36b0a635ff | (Madeleine Goebel)++ (committed using GitHub Web editor) | 8 files
Update all ranged copyrights from 2021 to 2022 (#5015)

Co-authored-by: Madeleine Goebel [email@hidden.address] Co-authored-by: Madeleine Goebel <mgoebel@biocircuit.com>
17:39