MasterDuke hmm, i'm getting a 'java.io.IOException: Class not found' exception, but i thought it would look inside the jar (CORE.setting.jar). if i extract the jar into the rakudo src director it finds the class 00:32
./perl6-j --ll-exception -e 'say 1/0' 05:38
Attempt to divide 1 by zero using div
in throw (SETTING::src/core/Exception.pm:64)
in throw (SETTING::src/core/Failure.pm:40)
psch that looks pretty good i'd say :) 05:39
MasterDuke it's not pretty
psch is the indent-inconsistency copied correctly? 05:40
MasterDuke no, that's just a copy/paste error
there a still some things to work through. it still requires CORE.setting.jar to be unzipped 05:41
psch oh
maybe you're looking in the wrong classloader?
MasterDuke no idea 05:43
it's time for me to sleep, but i'll gist the diff 05:45
psch cool, i'll have a look 05:46
MasterDuke gist.github.com/MasterDuke17/6a984...e191617fe1 05:51
heh, not only does it require the jar to be unzipped, but it only works if you run from the directory where you unzipped it 05:55
well, at least some progress was made today, but hopefully the rest of the problems (and bad Java code) can be fixed 05:56
psch .tell MasterDuke SourceFilenameAttribute didn't make it into the gist. but fwiw, you should get it work if instead of feeding the class name into the new ClassReader you get the byte[] (via 'mumble.getClassLoader().getResourceAsStream(className)') 06:19
yoleaux2 psch: I'll pass your message to MasterDuke.
psch .tell MasterDuke well, except getResourceAsStream returns an InputStream, but that usually can be converted to a byte[] rather easily :) 06:20
yoleaux2 psch: I'll pass your message to MasterDuke.
psch .tell orrr you use the InputStream directly, 'cause i just notice ClassReader has a constructor that takes an InputStream... 06:21
yoleaux2 psch: I'll pass your message to orrr.
psch err
.tell MasterDuke orrr you use the InputStream directly, 'cause i just notice ClassReader has a constructor that takes an InputStream...
yoleaux2 psch: I'll pass your message to MasterDuke.
[TuxCM] This is Rakudo version 2016.11-5-ge9487d6 built on MoarVM version 2016.11-8-gb582aeb 09:07
csv-ip5xs 3.118
test 13.771
test-t 6.541
csv-parser 14.087
FROGGS o/ 09:08
lizmat Files=1154, Tests=53760, 211 wallclock secs (12.78 usr 3.73 sys + 1287.16 cusr 123.43 csys = 1427.10 CPU) 11:07
FROGGS perl tools/build/gen-version.pl > gen/moar/main-version.nqp 11:16
fatal: No tags can describe '2432c469763dd01cf0f72b6b37fa0e4bf9c86b5f'.
FROGGS reads this as "It is sooo disgusting!" 11:17
dalek ast: 5ee33ec | niner++ | S05-capture/match-object.t:
Test equality and identity of Match objects
 11:53
kudo/nom: 7f26e8b | niner++ | src/core/Match.pm:
Fix Match pretending to be a value type when it is not

Comparing the identity of Match objects wrongly resulted in them being treated as equals. Happened because Match is a subclass of Capture which is a value type and therefor provides a WHICH method. Matches are most certainly no value types as they are mutable. So we circumvent Capture's WHICH method.		 11:55
lizmat nine: good catch! 11:59
nine vendethiel found the bug :) 12:01
vendethiel well, nine++ for fixing it so quickly 12:02
nine What scares me is that both the test code and the fix worked on the very first try with no typos or other errors. I'm not sure that has ever happened before ;)
Of course it's been a rather trivial bug :) 12:03
jnthn Match objects are mutable? 12:29
yoleaux2 19 Nov 2016 23:58Z <lizmat> jnthn: was there a reason why native shaped str arrays are not implemented? or was it an oversight that they aren't?
jnthn Ah, I guess .ast is... 12:30
lizmat: They just didn't seem especially useful, and time was short. :) 12:31
I've no problem with adding support for them.
lizmat jnthn: does that require work at the VM level ?
jnthn Shouldn't. 12:33
After all, nqp::list_s already exists :)
nine jnthn: yes .made is 12:36
MasterDuke would native shaped str arrays be useful for Text::CSV? 12:40
yoleaux2 06:19Z <psch> MasterDuke: SourceFilenameAttribute didn't make it into the gist. but fwiw, you should get it work if instead of feeding the class name into the new ClassReader you get the byte[] (via 'mumble.getClassLoader().getResourceAsStream(className)')
06:20Z <psch> MasterDuke: well, except getResourceAsStream returns an InputStream, but that usually can be converted to a byte[] rather easily :)
06:21Z <psch> MasterDuke: orrr you use the InputStream directly, 'cause i just notice ClassReader has a constructor that takes an InputStream...
lizmat jnthn: ok, will look at doing that 12:47
MasterDuke: it might, there are a number of Str @ in CSV.pm 12:50
dalek kudo/nom: 939d273 | lizmat++ | src/core/operators.pm:
Make indirect type lookup 3x as fast

Should have a positive effect on bare startup time, since we do at least 5 of these every time we start.		 13:24
masak I'm still flummoxed as to why people like the methos name ".made" at all... :P 13:41
method*
MasterDuke doesn't, fwiw 13:42
masak maybe it's just one of those unfortunate word-person combinations, where the chemistry totally doesn't work ;)
you can't have a Match object and go "ok, get me the made" with a straight face
lizmat well, it's the counterpart of "make" conceptually, no? 13:43
masak oh, I understand that bit
lizmat still prefer it to .ast
masak which is fine 13:44
we can live in a world where we prefer different things -- I'm ok with that :)
lizmat indeed :-) 13:45
afk&
masak maybe I would be more comfortable with ".made" if "make" also wasn't largely grabbed out of thin air
it's a short word meaning, roughly, "associate a piece of data with this Match object"
MasterDuke anybody have any idea why this line: "new ClassReader(info.compUnit.getClass().getClassLoader().getResourceAsStream(kls)).accept(" would throw "java.lang.ArrayIndexOutOfBoundsException: 60673"? 13:46
masak and if you don't know that bit, ".made" is doubly confusing in the code
MasterDuke finds it amusing that working on a Project Euler problem in Perl 6 led to working on Perl 6 itself, in Perl 6, Perl 5, NQP, Java, C++, C, HTML, Markup, Bash, and JavaScript 13:51
psch MasterDuke: ArrayIndexOutOfBounds in code dealing with objectweb.asm usually seems to mean something missing (or too much) on the stack in the generated classfile 13:52
MasterDuke and /me still hasn't solved the original PE problem!
psch: i haven't been able to get it working with getClassLoader(), etc. yet 13:53
but wouldn't i have had the same error the way it was doing it before? 13:54
psch i'm not sure 13:56
i mean, when you get the resource from the classloader you're walking a loaded class in memory
maybe that messes things up somewhere, somehow..?
MasterDuke psch: fyi, just added SourceFilenameAttribute.java to the gist. i wanted to just use an anonymous subclass of Attribute, but couldn't get that working
psch objectweb.asm error reporting is kinda lacking i find
MasterDuke i'm away for a bit, but if you're curious, i pushed my current (not quite working) code here: github.com/MasterDuke17/nqp/tree/s...ers_on_jvm 14:00
ugjka whos that usuckirc guy 16:02
getting k-lined
viki Hm, they joined my #zofbot channel too.... 16:09
nine MasterDuke: took me 7 months of working on module management from "I'd like to create rakudo rpm packages for openSUSE for installation on our servers" to finally being able to do just that. 17:18
MasterDuke: and I've yet to get around to creating packages for Perl 6 modules ;) 17:19
dalek kudo/nom: 2706221 | lizmat++ | tools/build/makeNATIVE_SHAPED_ARRAY.pl6:
Introducing makeNATIVE_SHAPED_ARRAY.pl6

Support script for generating the shaped(int|num|str)array roles. So far, only copied the current state of the code in native_array.pm		 20:11
kudo/nom: 486155e | lizmat++ | src/core/native_array.pm:
shaped(int|num|str)array roles are now generated

This also introduces the shapedstrarray role, but this is not used yet.		 20:13
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Introducing makeNATIVE_SHAPED_ARRAY.pl6 21:01
travis-ci.org/rakudo/rakudo/builds/177491853 github.com/rakudo/rakudo/compare/9...06221ac9c0
buggable [travis build above] ☠ Did not recognize some failures. Check results manually.
dalek kudo/nom: cf451f1 | lizmat++ | src/core/native_array.pm:
Use the same sig everywhere in shaped role
 21:06
kudo/nom: f25ab3a | lizmat++ | src/core/ShapedArray.pm:
Some more signature streamlining
kudo/nom: 0cb765f | lizmat++ | / (4 files):
Move shared AT-POS code to ShapedArrayCommon role
 22:09
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Some more signature streamlining' 22:11
travis-ci.org/rakudo/rakudo/builds/177500978 github.com/rakudo/rakudo/compare/4...5ab3a28e38
dalek kudo/nom: d07ab04 | lizmat++ | / (4 files):
Move shared ASSIGN-POS code to ShapedArrayCommon role
 22:34
kudo/nom: 0fc5580 | lizmat++ | src/core/ (3 files):
Move shared EXISTS-POS code to ShapedArrayCommon role
 22:46
lizmat .tell moritz in blogpost: "and nothing as fixed" s/as/was/ ? 22:51
yoleaux2 lizmat: I'll pass your message to moritz.
dalek kudo/nom: 014d4cf | lizmat++ | src/core/native_array.pm:
Activate native shaped str array

Alas, it's not working yet:
   $ 6 'use nqp; say nqp::multidimref_s(my str @a[5],nqp::list_i(0))'
   No str positional reference type registered for current HLL
So we need some more hookup work somewhere		 23:19
lizmat .tell jnthn something needs fixing in the bowels: github.com/rakudo/rakudo/commit/014d4cff2f 23:21
yoleaux2 lizmat: I'll pass your message to jnthn.
lizmat good night, #perl6-dev!

Powered by Cro and the Raku® Programming Language.

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

Thank you!