[TuxCM] This is Rakudo version 2016.10-285-gee8ae92 built on MoarVM version 2016.10-55-g20c8591 07:17
csv-ip5xs 3.081
test 13.988
test-t 6.545
csv-parser 14.669
dalek ast: bfd9438 | timotimo++ | integration/weird-errors.t:
don't segfault when null str goes into hash access

not satisfied with putting it into this file, though.		 10:42
viki m: say -100.base(16).parse-base(16) # Zoffix-- 11:30
camelia rakudo-moar ee8ae9: OUTPUT«-100␤»
viki oh
m: say (-100).base(16).parse-base(16) # Zoffix--
camelia rakudo-moar ee8ae9: OUTPUT«Invalid base-16 character: ⏏-64␤ in block <unit> at <tmp> line 1␤␤Actually thrown at:␤ in block <unit> at <tmp> line 1␤␤»
MasterDuke_ .ask jnthn do you have any thoughts on my line directive PR github.com/rakudo/rakudo/pull/919 and Zoffix's suggestion? 12:49
yoleaux2 MasterDuke_: I'll pass your message to jnthn.
MasterDuke_ .ask lizmat do you have any thoughts on my line directive PR github.com/rakudo/rakudo/pull/919 and Zoffix's suggestion?
yoleaux2 MasterDuke_: I'll pass your message to lizmat.
lizmat . 12:58
yoleaux2 12:49Z <MasterDuke_> lizmat: do you have any thoughts on my line directive PR github.com/rakudo/rakudo/pull/919 and Zoffix's suggestion?
lizmat MasterDuke_: left a comment 13:00
MasterDuke_ lizmat: thanks 13:19
dalek p: 0538cae | MasterDuke17++ | src/ (5 files):
Initial support for line directives

This will allow backtraces to report the original file and line number instead of m-CORE.setting and its line number.
This also works in EVALs, making it easier to know which one has died.		 13:40
nqp: 438b8e3 | (Zoffix Znet)++ | src/ (5 files):
nqp: Merge pull request #319 from MasterDuke17/source_file_line_numbers
nqp:
nqp: Initial support for line directives
kudo/nom: 2e9e89b | (Zoffix Znet)++ | tools/build/NQP_REVISION:
Bump NQP version

To bring in line directives PR github.com/perl6/nqp/pull/319		 13:42
kudo/nom: 5b0d81b | MasterDuke17++ | / (2 files):
Changes needed to support line directives

An extra blank line after a directive throws off the numbering.
The m-CORE.setting file is no longer seen in a backtrace, so we need another way of determining if we are in the settings. This isn't a great solution, but should work for now.
kudo/nom: b068e3a | MasterDuke17++ | tools/build/gen-cat.nqp:
Mark the line directives for Rakudo core src files
viki s/This isn't a great solution//; # that has been amended 13:44
cognominal m: class A { method Str { 1 } }; say A.new.Str # should not it be some kind of compile time type enforcement of returned value when defining casting methods,
camelia rakudo-moar ee8ae9: OUTPUT«1␤»
cognominal obviously there is none
viki Well, it's really a casting method. It's just a method. 13:45
m: class is-prime {}; sub (is-prime(Int) $x) { say $x }(31337) 13:46
camelia rakudo-moar ee8ae9: OUTPUT«True␤»
viki ^ any method can be a "casting method" really :) 13:47
s/it's really/it's not really/;
timotimo the fact that methods called Int cast something to Int is really just a convention
cognominal timotimo: yes that´s my point. Should it be when the method name is a type ? 13:48
viki My opinion: definitely not
jnthn It's just a convention. 13:49
Ah, timotimo already said that :)
But anyway, I agree. I'd not expect an automatic return type enforcement here
cognominal ok, just asking. I don´t yet have an opinion on the subject 13:51
viki cognominal: adding such an enforcement basically means we can never add or change a core class. Because someone may have a method named like that and our enforcement would break that method. Also, it's just an assumption that a method name that matches a core type name necessarily means you want to cast it. I could have a .DateTime method on an object that stamps it with current date or tells me whether it 13:52
has been stampted.
(granted, I'd name that method .date-time, but there are these strange people who use CamelCase :))) 13:53
timotimo date'time :P 13:54
like 0'00"00
viki :)
lizmat afk& 14:00
viki ZOFVM: Files=1203, Tests=130161, 134 wallclock secs (20.46 usr 2.79 sys + 2354.09 cusr 195.34 csys = 2572.68 CPU) 14:01
cognominal updated S13 to hopefully clarify the matter 14:04
viki Well, specs are speculations—historical documents. A lot of the stuff in 'em is outdated. 14:08
m: say "{.file}:{.line}" with &say 14:53
camelia rakudo-moar f03ef5: OUTPUT«SETTING::src/core/io_operators.pm:20␤»
viki hehe. neat. MasterDuke_++
s: &say
SourceBaby viki, Sauce is at github.com/rakudo/rakudo/blob/f03e...gue.pm#L16
viki heh. time to update CoreHackers::Sourcery, I guess
pmurias viki: does a lot of stuff in the specs need updating? 14:55
[Coke] you mean the syn? 14:56
viki pmurias: well, almost a year ago we decided keeping them up-to-date was wasted effort. So I'd guess a lot of them are out of date, but *none* need updating.
[Coke] It's probably not getting updated at this point. We're considering them historical docs. Inspirational, not definitive.
pmurias [Coke]: yes 14:58
viki: so the current docs live at docs.perl6.org?
[Coke] as current as we have, yes. 14:59
viki pmurias: yes, roast is The Official Perl 6 Specification and docs.perl6.org is the human-readable version that depends on the former
(that is, if the two disagree, the roast is the authoratitive source)
dalek p: 798eeb3 | jnthn++ | tools/build/MOAR_REVISION:
Bump MOAR_REVISION for latest MoarVM.
 15:05
viki \o/ 15:07
dalek kudo/nom: 395f369 | jnthn++ | tools/build/NQP_REVISION:
Bump for latest MoarVM.

This includes fixes to a SEGV in multi-threaded programs where strings internally represented as strands were used in various ways (typically as hash keys). As a bonus, hashes now use 4 (32-bit) or 8 (64-bit) bytes less memory per item stored, so a hash of 1000 keys will be up to 8KB smaller.		 15:08
jnthn Hope that description helps the changelog :) 15:09
viki It does. jnthn++ 15:10
dalek ast: 6a40128 | jnthn++ | S17-promise/start.t:
Test covering RT #129781.
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=129781
dalek p: 6dadbe3 | (Pawel Murias)++ | src/vm/js/nqp-runtime/core.js:
[js] settypecheckmode with nqp::const::TYPE_CHECK_NEEDS_ACCEPTS.
 15:15
p: 687bdf4 | (Pawel Murias)++ | t/nqp/103-typecache.t:
Test the type check cache.
p: 4965e17 | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (4 files):
[js] Improve the VMArray repr as we work on using it instead of the NQPArray class.
p: d8f1a24 | (Pawel Murias)++ | t/serialization/01-basic.t:
Test serialization of a VMArray.
 15:16
p: e61fa7c | (Pawel Murias)++ | src/vm/js/ (2 files):
[js] Remove old code for doing a splice on raw js arrays, call the $$splice method directly.
 15:31
p: 1f7721e | (Pawel Murias)++ | src/vm/js/ (2 files):
[js] Use $obj.$$iterator instead of nqp.op.iterator($obj) everywhere.
 15:44
kudo/nom: 4ccb2f3 | jnthn++ | src/core/Grammar.pm:
Make `parse` try a bit harder to get a match.

While a Grammar built out of regexes isn't a particularly wise idea, if we do have a `regex TOP` (or the `rule` we specify to `parse` is a
  `regex`) then we should at leave give it a chance to do the required
backtracking to produce a result that eats the whole string. It costs nothing for non-regex grammars to add this.		 16:33
ast: 7b838ea | jnthn++ | S05-grammar/parse_and_parsefile.t:
Test for RT #130081.
 16:34
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130081
jnthn m: use Test; { my $p = Promise.new; $p.break("OH NOES"); try await $p; is $!.message, "OH NOES", "Promise broken with string form of .break conveys correct message"; } 17:13
camelia rakudo-moar 4ccb2f: OUTPUT«not ok 1 - Promise broken with string form of .break conveys correct message␤␤# Failed test 'Promise broken with string form of .break conveys correct message'␤# at <tmp> line 1␤# expected: 'OH NOES'␤# got: 'bindexpayload needs a VMException…»
[Coke] hey, that looks familiar! :) 17:15
jnthn plan 2999999999; 17:16
...seems I need to keep working on these vim skills :P
timotimo hehe
jnthn [Coke]: Yeah, it's RT #130064; spectesting the fix now
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130064
timotimo ctrl-a will increase the next number the cursor can find
jnthn And wanted to make sure that my test really covered it
timotimo (on the same line)
jnthn o.O 17:17
timotimo ctrl-x is for decreasing
jnthn omg
That's awesome
timotimo++
timotimo it's neat, yeah :)
^A timotimo ;)
also, did you know that when you search with /foobar/e, the cursor will end up on the last matched character instead of the first 17:18
so you can start appending immediately
jnthn Didn't know about /e, no :) 17:20
Also cool
timotimo very helpful, indeed indeed
viki
.oO( pfft... vim users and their crazy shortcuts! )
timotimo who will go ahead and write a perl6-compatible regex engine in vim? :P 17:22
*cough* vimscript *cough*
jnthn is seeing some backtrace/line number or similar fails in spectst
And can't track them back to anything I've done; appeared after a fetch 17:23
One each in S32-basics/warn.t and integration/error-reporting.t
dalek kudo/nom: 58a4826 | jnthn++ | src/core/Exception.pm:
Allow `.rethrow` to work on unthrown exceptions.

For example, those created with `.new`. Previously, we spat out an internal error when trying to do this. Fixes a regression in some cases of error reporting involving promises.
ast: 63f7e36 | jnthn++ | S17-promise/basic.t:
Test to cover RT #130064.
synopsebot6 Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130064
viki We did make a backtrace/line-number change today, but stresstest was clean after I merged it 17:25
jnthn Hmm 17:26
dogbert17 wonders if that might fix the oldie hexdump bug we discussed yesterday
viki pulls and fires up another run
m: use v6; warn("OH NOEZ"); say "alive" 17:27
camelia rakudo-moar 4ccb2f: OUTPUT«OH NOEZ␤ in block <unit> at <tmp> line 1␤alive␤»
timotimo the repl test hangs even without my string encoding stuff
viki 0.o
jnthn OK, I need to a break from the keybarod.
And then should cook us some dinner here, I guess... :-) 17:28
timotimo that sounds nice :)
jnthn bbl
viki Sweet. What are we having?
timotimo oh wait, you don't mean us as in all of the people in irc here ...
jnthn No, I meant wife and I :-)
viki :)
dogbert17 indian food? 17:29
jnthn Will make kofta (meatballs) and some spicy-ish sauce to go on them. :)
Yeah, the sauce qualifies it as Indian food I guess :) 17:30
viki kofta? Lol :) translate.google.ca/#ru/en/kofta
jnthn Though lots of places seem to have things they call kofta
timotimo allowing strings that are decoded from latin1 to become 8bit-per-char stored gets us about 3/4 of a megabyte less for perl6 -e '' 17:31
jnthn viki: Yes, my wife has pointed that out more than once :P
viki :)
jnthn timotimo: Nice saving!
timotimo oh 17:32
jnthn OK, really off for a bit... :)
timotimo actually closer to 1/2 or 1/3rd
viki All tests successful. 17:33
Files=1203, Tests=130164, 164 wallclock secs (20.62 usr 3.01 sys + 3092.72 cusr 283.71 csys = 3400.06 CPU)
Result: PASS
dalek p: 1c36493 | (Pawel Murias)++ | src/vm/js/Operations.nqp:
[js] Use a :method_call adverb to add_simple_op to avoid reptitive code.
 18:37
p: 876ccd5 | (Pawel Murias)++ | src/vm/js/Operations.nqp:
[js] Make :ctx work together with :method_call on add_simple_op.
p: 1178c84 | (Pawel Murias)++ | src/vm/js/Operations.nqp:
[js] Remove unneeded argument.
p: b0e6c9c | (Pawel Murias)++ | src/vm/js/Operations.nqp:
[js] Rename side-effects -> side_effects for consistency.
viki s: &say 19:34
SourceBaby viki, Sauce is at github.com/rakudo/rakudo/blob/58a4...ors.pm#L20
MasterDuke_ Undercover: help 19:39
Undercover MasterDuke_, Use s: trigger with args to give to sourcery sub. e.g. s: Int, 'base'. See modules.perl6.org/dist/CoreHackers::Sourcery
MasterDuke_ Undercover: s: Int, 'base' 19:40
viki: ^^^ how do you use Undercover? 19:41
c: Int, 'base'
Undercover MasterDuke_, The code is hit during stresstest [WARNING: this line is a proto! Check individual multies] See perl6.WTF/src_core_Int.pm.coverage.html#L64 for details
viki c: Int, 'base', \()
Undercover viki, Something's wrong: ␤ERR: Could not find candidate that can do \()␤ in sub sourcery at /home/zoffix/services/lib/CoreHackers-Sourcery/lib/CoreHackers/Sourcery.pm6 (CoreHackers::Sourcery) line 29␤ in block <unit> at -e line 6␤␤
viki c: Int, 'base', \(4) 19:42
Undercover viki, Something's wrong: ␤ERR: Could not find candidate that can do \(4)␤ in sub sourcery at /home/zoffix/services/lib/CoreHackers-Sourcery/lib/CoreHackers/Sourcery.pm6 (CoreHackers::Sourcery) line 29␤ in block <unit> at -e line 6␤␤
viki c: 4, 'base', \(4)
Undercover viki, The code is hit during stresstest See perl6.WTF/src_core_Int.pm.coverage.html#L65 for details
MasterDuke_ how up-to-date is the coverage report?
viki It's generated on the top of the home page: "Generated on 2016-10-07T06:04:53.940737-04:00"
You can also generate a fresher copy yourself: perl6.party/post/Perl-6-Core-Hackin...ityourself 19:47
MasterDuke_ yeah, i had done it a while ago, but couldn't remember if you had automated it 19:48
viki Well, to some extent. It's a single command to run on my home box :) 19:49
But I don't have a good box to run this automatically on.
... I guess I can set it to run in cron at home around 4AM
MasterDuke_ don't you have a beefy cloud vm? 19:50
viki Yeah, but I only run it when I need it. 19:51
Too expensive to run it 24/7
It would cost $486.05/month... my current usage bill is ~$17/month 19:52
MasterDuke_ yeah, that'd be a noticeable difference 19:54
dalek p: 8d18c6c | (Pawel Murias)++ | src/vm/js/nqp-runtime/null_s.js:
[js] make js-lint happy
 20:01
p: b802319 | (Pawel Murias)++ | src/vm/js/nqp-runtime/code-ref.js:
[js] Hack that allows code generated by rakudo.js to load the setting.
p: 17a5ff6 | (Pawel Murias)++ | src/vm/js/nqp-runtime/core.js:
[js] Emit a warning on stderr instead of stdout so it doesn't get mixed with normal output when running tests.
mst viki: note that you can get a significantly powerful (though not as much as your insane VM) box for rental for reasonable money at hetzner 20:40
viki: this may or may not be worthwile to you, of course, but I thought I'd mention
viki Thanks, but I'm good :) I don't think many people need coverage for perl6 for it to be critical that we generate it automatically and I will look into setting up it in cron on my home box when I don't use it 20:42
nine viki: would be happy to give you resources on the server camelia's running on. Maybe just using camelia's VM would be a reasonable way, too. 20:53
mst yeah, I was going to say, if we need somewhere to stick 2am cron jobs I can use a dedicated user account on a shadowcat server if that's helpful 20:54
viki Thanks. I'll keep that in mind if the sticking it as a 4am job on my home computer doesn't work out :)
mst yeah, if that works that's obviously simpler 21:03
but shadowcat has all sorts of shit stashed in the corners of our production boxes that make one or other bit of the perl community's lives easier

Powered by Cro and the Raku® Programming Language.

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

Thank you!