timotimo not sure if there's a better way; i would like shorter syntax as well 00:01
discord6 <theangryepicbanana> it would be cool if making a sub in a class automatically made the sub a static method 00:02
TreyHarris Well, an 'is static' trait's been banging around as a perennial proposal; hopefully problem-solving will be able to keep institutional memory, because I can't remember why 'is static' was a bad idea :-)
discord6 <theangryepicbanana> since that kinda already makes sense
<theangryepicbanana> well I mean having an attribute is cool and all, but it'd be ideal to have something as short as sub or method
TreyHarris m: class B { method foo (B:U: --> Str) { return "Bee" }; our sub foo(--> Str) { return "no bee" } }; say B.foo; say B::foo(); 00:07
camelia Bee
no bee
TreyHarris theangryepicbanana: you see the ambiguity ↑ 00:08
timotimo having a scoping qualifier seems like the wrong abstraction to me
the method is looked up the same way, it just checks the invocant type differently 00:09
TreyHarris well, I just did that because it's a pain to export and import on a single line...
I mean, it's probably not best practice for a class to export routines, but it's probably also essential to some edge cases 00:10
and multis are of course frequently "exported" one way or another
timotimo i think i hallucinated someone suggesting "our" for "static methods" 00:11
TreyHarris like, just now here?
I'm sure someone proposed it at one point--in fact, I think I may have written a reply if it was on p6-l showing why our subs are useful 00:12
discord6 <theangryepicbanana> perhaps a Ruby-like syntax could be used like method self.something(...) {...}?
TreyHarris Oh good gravy, about to search for that mail I was able to see T&M's latest--he's still hammering on .^name "lying" and why can't you ask a native exactly what signedness and length it is? phew. 00:14
discord6 <theangryepicbanana> that would however mean that pretty much every class with a custom new method wouldn't work anymore
timotimo m: my uint32 $foo = 9; say $foo.VAR.^name
camelia IntLexRef
timotimo m: my uint32 $foo = 9; say $foo.VAR.^methods 00:15
camelia No such method 'methods' for invocant of type 'Perl6::Metamodel::NativeRefHOW'
in block <unit> at <tmp> line 1
timotimo m: my uint32 $foo = 9; use nqp; say nqp::objprimbits($foo.VAR)
camelia 0
timotimo m: my uint32 $foo = 9; use nqp; say nqp::objprimbits($foo)
camelia 0
timotimo m: my uint32 $foo = 9; use nqp; say nqp::objprimbits($foo.VAR<>) 00:16
camelia 0
timotimo m: my uint32 $foo = 9; use nqp; say nqp::objprimbits($foo<>)
camelia 0
timotimo ehh.
TreyHarris docs say you can't
you can do over/underflow heuristics to figure it out
if -1 assignment produces a positive value that gives you your answer completely in one step; for signed values you have to suss it out 00:19
well, for ints. for floats it would be a serious PITA
timotimo m: my uint32 $foo = 9; use nqp; say $foo.VAR.of 00:21
camelia No such method 'of' for invocant of type 'Int'
in block <unit> at <tmp> line 1
TreyHarris rofl... never mind; searching my back mail, I see that I have explained 'our sub' six times since 2016, but each time it was to T&M. 00:28
I don't see "is static" ever mentioned on p6-l or p6-u. weird. coulda sworn that was a perennial idea 00:31
TreyHarris Do it Python-style, `method foo() is classmethod`? (I can't decide whether the Python @staticmethod is entirely obviated already or just mostly. I think "mostly", but the problem space is more than filled up by multis.) 00:36
nine theangryepicbanana: no, a sub is a sub and a method is a method. When I declare a sub inside a class it's because I want a sub that's visible in that lexical scope. I do that precicely because I don't want the class' users to have access to it. 07:26
I wonder if class methods really are so common that they would warrant special (short) syntax. It's not like we're Java where there's only methods. We already have our subs and exporting to make functions available. That doesn't seem to leave many use cases for class methods. Mainly those where you want subclasses to be able to override them. 07:29
Now a shorter syntax for object-only methods (::$?CLASS:D) would be very welcome. In hindsight, I'd even wish for that to be the default, but that ship has probably sailed... 07:30
Or not, we could still change it with a new language version 07:31
AlexDaniel lol what, now we're getting spam in the old-issue-tracker repo? :D 09:19
is RT cursed or something
this is probably like the second time I see spam on github 09:21
lizmat Files=1302, Tests=109837, 210 wallclock secs (28.20 usr 8.03 sys + 2942.75 cusr 262.98 csys = 3241.96 CPU) 09:52
AlexDaniel oh 10:21
so now GitHub reports Raku files as Raku
awesome
Geth rakudo: 6df78ca3ed | (Elizabeth Mattijsen)++ | src/core.c/SetHash.pm6
Add SetHash.set/unset methods

As proposed about 8 months ago in:
   github.com/rakudo/rakudo/pull/2942
Since nobody had a big objection against it, and I think it will make working with SetHashes much easier, I decided to just put it in
  (as the Pull Request had bitrotted).
Please revert if disagreeing.		 11:05
rakudo/SetHash-set-unset: b9d9943c34 | (Elizabeth Mattijsen)++ | src/core/SetHash.pm6
Give SetHash a set/unset method

Inspired by seeing things like:
   %sethash{$_} = True for @to-be-added;
in the wild. Instead, one can now do: ... (5 more lines)		 11:12
lizmat huh, what was that ? 11:13
anyway, branch now deleted
sena_kun lizmat: I did it automagically and then thought that something important may be still there and restored. :S 11:17
lizmat sena_kun: nothing important in there, so it's gone now 11:24
travis-ci Rakudo build errored. Elizabeth Mattijsen 'Add SetHash.set/unset methods 12:00
travis-ci.org/rakudo/rakudo/builds/645896933 github.com/rakudo/rakudo/compare/3...f78ca3ed72
Geth rakudo: ce5b5dea55 | (Ben Davies)++ | src/core.c/IO/Socket/INET.pm6
Don't attempt to split ports from paths for for UNIX sockets

It's possible for a valid path containing a ':' to wind up as something else if the platform allows colons in filenames like this. Since UNIX sockets don't use ports in the first place, don't treat them as if they do.		 13:59
timotimo gist.github.com/timo/6e8a259891326...4ae0d5d1fa - a good chunk of frames in the core.c setting are short: 7 instructions is the msot common, followed by 4 instructions, then 8. 14:13
there's even 137 frames that have only 2 instructions 14:14
the first few i found are called !LEXICAL_FIXUP_CSCOPE, i don't think i've seen that before 14:15
many have an empty name, though 14:16
oh, the numbers start at 0, so all the instruction counts need to be read as "off by one"
very many of the ones that end at 3 are !LEXICAL_FIXUP, many of the ones ending at 4 are the "execution of block eliminated by optimizer" blocks 14:21
if it's about 1.35k of the "internal error" reporting blocks we can save about 4 bytes each by not putting a return_o directly after the die :D 14:24
or are nqp::die resumable?
ok, they apparently are 14:25
Geth rakudo: 62cb756d91 | (Ben Davies)++ | src/core.c/IO/Socket/INET.pm6
Don't call nqp::getport when listening on UNIX sockets
 14:54
travis-ci Rakudo build passed. Ben Davies 'Don't attempt to split ports from paths for for UNIX sockets 14:56
travis-ci.org/rakudo/rakudo/builds/645958354 github.com/rakudo/rakudo/compare/6...5b5dea5501
travis-ci Rakudo build passed. Ben Davies 'Don't call nqp::getport when listening on UNIX sockets' 15:49
travis-ci.org/rakudo/rakudo/builds/645982295 github.com/rakudo/rakudo/compare/c...cb756d9124
Geth rakudo: c1b8ff60f8 | (Elizabeth Mattijsen)++ | 2 files
Add BagHash.add/remove

Similar to SetHash.set/unset . Please revert if you disagree.		 18:38
TreyHarris Can anyone here kick camelia? I'm guessing since AlexDaniel` disappeared at the same time it may be more than an issue of kicking something, though 19:53
er, "restart", didn't mean "kick" in the IRC sense
AlexDaniel` I don't maintain camelia 19:54
and that was my laptop shutting down, has nothing to do with the bot :)
TreyHarris AlexDaniel`: I didn't think so, but the times were so coincident I was guessing a big part of the Internet got partitioned. That's good to know at least 19:55
(I didn't think you maintained camelia, I mean)
moritz restarting it now 20:06
TreyHarris I recently switched over from the last rakudo-star to a self-build. Assuming my PATH was `...${rakudo-star}/latest/install/bin:${rakudo-star}/latest/install/share/perl6/site/bin` and it's now `...:${rakudo}/bin:${rakudo}/install/share/perl6/site/bin`, and I have zef built anew in the latter, how can I best get my old modules rebuilt? Is there anything else from rakudo-star I need to add? 20:57
jnthn I just pulled/built latest Rakudo and I get one spectest failure that looks like this: 22:20
1..3
Error in mp_exptmod: Value out of range
in block at t/spec/S32-num/expmod.rakudo.moar line 31
in block <unit> at t/spec/S32-num/expmod.rakudo.moar line 20
The last test, pretty much
MasterDuke jnthn: yep, i forgot to change the test after the moar ltm v1.2.0 upgrade 22:44

Powered by Cro and the Raku® Programming Language.

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

Thank you!