Geth rakudo: 8943f7c546 | (Zoffix Znet)++ | t/08-performance/01-iterators.t
Add ticket reference to test
 00:30
roast: 41f2ce6459 | (Zoffix Znet)++ | S32-io/dir.t
Remove outdated comments
 01:07
roast: b41e6a0e61 | (Zoffix Znet)++ | S32-io/dir.t
[v6.d REVIEW] Remove bogus fudge/adjust test

1) .dirname returns the dir portion of the path IO::Path was
   created with and dir("t/") would have that as simply "t".
   It's up to the user to absolutify the paths if they need to,
   so test's expectation for .dirname to contain $*CWD is incorrect
2) Fudge says Rakudo doesn't add '/' to $*CWD, which seems perfectly
   fine to me. Also, merely stringifying an IO::Path object and
   appending a string with a dir isn't the right way to obtain
   a child of a path. .child/.add are for that.		 01:13
JRaspass AlexDaniel: I've bumped code-golf.io to the latest git version of rakudo after the combinations slow down in 2017-12 so hopefully all your blog post examples will be nice 'n' fast now :-) 01:15
AlexDaniel oh that's very nice of you!
JRaspass It's the least I could do, ta for all the publicity! 01:16
AlexDaniel JRaspass: at least I can sleep well now looking at “12 Days of Christmas” task :D 01:21
JRaspass oh darn, did you beat me? :-P 01:22
AlexDaniel by 10 characters
JRaspass ooo, guess I shan't be sleeping then 01:23
Geth roast: e10e1fd400 | (Zoffix Znet)++ | S32-list/permutations.t
[v6.d REVIEW] Test return data directly

Instead of converting it to another type and stringifying.
Partial origs:
  github.com/perl6/roast/commit/d64d...944438a40f
  github.com/perl6/roast/commit/ff44...d982b9cbf2
AlexDaniel at least. Some of the solutions I can't be bothered unless somebody does better :) 01:24
to improve*
JRaspass: I never thought that you follow the progress so carefully. The advent post I can understand, but the combinations slow down? :) 01:26
JRaspass yeah im a lurker on this room, I just thought id finally install an IRC client and say hi 01:27
AlexDaniel 🙋 01:28
JRaspass who lacks emoji fonts it seems, what's that?
AlexDaniel u: 🙋
unicodable6 AlexDaniel, U+1F64B HAPPY PERSON RAISING ONE HAND [So] (🙋)
JRaspass hah, nice
apologies if im going off topic in this room, im hardly a perl6 dev 01:29
AlexDaniel it's quiet today, so no probs :) 01:30
JRaspass its the early hours of the morning over here, ill crash soon
AlexDaniel JRaspass: oh, that too. I should be doing that also, maybe. irclog.perlgeek.de/perl6-dev/2017-...i_15588284 01:32
JRaspass ah you're west of me, that explains that weird hour github issues i get from you ;-)
teatime lol, how do you read your graph?
AlexDaniel teatime: open the image at 100%, there are some explanations on top left 01:33
teatime you've been tracking it for a long time
AlexDaniel not even 10 years. We'll get there. The question is how I'd be making the graph then :( 01:34
even today it's a bit too wide, and each day is 1 pixel anyway
although… 4k monitors?
Geth roast: 6a595ec7af | (Zoffix Znet)++ | S07-hyperrace/hyper.t
[v6.d REVIEW] Simplify test code

And don't create objects we never use. Orig: github.com/perl6/roast/commit/2212...0c908d2cc4
timotimo AlexDaniel: i think it should be "for you to have some fun" 01:39
i wonder if we've had swearwords like "fucking" in the calendar posts until now
oh, wow, it's in that song 01:40
and it's not likely that it was supposed to be "pumpkin pie"?
AlexDaniel yeah, the post is about fooling around so the language is kinda part of it 01:41
we did have a swearword at least once… I wonder who used it /o\ 01:42
timotimo: fwiw feel free to edit it 01:43
the post I mean
timotimo cool 01:46
i wasn't aware of most of the tricks in that post
m: for ^0x10FFFF { next without .unival; say .chr.codes / .unival.codes } 01:48
evalable6 (signal SIGHUP) 0.333333
0.333333
0.333333
0.333333
0.333333
0.333333
0.333333
0.333333
0.…
timotimo, Full output: gist.github.com/509c8f742ba157d0fa...2b5bb686c4
timotimo ah, yes. not terribly helpful :)
m: ^0x10FFFF .map({ next without .unival; .chr.codes / .unival.codes }).unique.say 01:49
evalable6 (signal SIGHUP) «timed out after 10 seconds»
timotimo m: ^0x10FFFF .hyper.map({ next without .unival; .chr.codes / .unival.codes }).unique.say
evalable6 (signal SIGHUP) «timed out after 10 seconds»
teatime never realized he could use a space like that instead of wrapping the range in () 01:50
AlexDaniel teatime: also, that may save one character I think!
timotimo with a proper :batch and :degree it runs in 18 seconds on my machine 01:51
(0.333333 1 0.25 0.666667 0.166667 0.2 0.5 0.142857 0.125 0.111111 0.076923 0.090909)
(兆 0.076923 1000000000000) (𖭠 0.090909 10000000000) (𖭡 0.076923 1000000000000) 01:53
with the littlest bit of knowledge about utf8 i could have used something much faster than .chr.codes 01:54
Geth roast: 6315201cc9 | (Zoffix Znet)++ | S07-hyperrace/hyper.t
[v6.d REVIEW] Simplify / test actual data

  - Get rid of useless variables
  - Break up overly long lines
  - Test actual return values instead of stringified versions
Plan decreased due to some tests wrapped in subtests; no tests were removed.		 01:58
Hotkeys "But what about 1000000000000 and 兆? In any case, code-golf.io is unicode-aware, so the length of any of these characters will be 1." 02:00
oooh that's direty
dirty*
AlexDaniel++ 02:01
Geth roast: c8442d826c | (Zoffix Znet)++ | MISC/bug-coverage.t
Add more .count-only/.bool-only tests

Test both method and sub versions of combinations/permutations as they [can be] are implemented using different chain of iterators.		 02:39
rakudo: 57d164b464 | (Zoffix Znet)++ | src/core/Rakudo/Iterator.pm
Add missing words in comment
 03:02
Zoffix It's not quite the plan any more. The metacpan fork was limping all along and last time I got the update about it it was "stalled but not yet dead", which is why CPAN support was added to mp6o. My crystal ball says mp6o's featureset will slightly develop further for ~2 years, then there'll be move to write proper Rakudo version of the site. Perl 5 version just isn't working out, the way I see it. 03:13
manchicken, CPAN is CPAN. It has support for both Pumpkin Perl 5 and Rakudo Perl 6 modules. It's one of our ecosystems, with the other being GitHub/GitLab based, called "p6c". The modules.perl6.org ("mp6o") is the front-end for both of those ecosystems 03:14
ugh wc
Geth rakudo: cda4a4d76a | (Zoffix Znet)++ | t/08-performance/01-iterators.t
Fix &combinations.bool-only test

Don't numify before boolifying; that uses wrong iterator method		 03:29
rakudo: 2c27eeaa34 | (Zoffix Znet)++ | t/08-performance/01-iterators.t
Test &permutation offers .count/bool-only opts
 03:37
roast: 7c8c01a98d | (Zoffix Znet)++ | MISC/bug-coverage.t
Fix incorrect .permutations use
 03:48
Zoffix Jesus. Sure made a billion mistakes this week :| need to sleep more often
Geth rakudo: d4f53ed5d9 | (Zoffix Znet)++ | 2 files
Implement .[count|bool]-only opts on List.permutations

Partially addresses github.com/rakudo/rakudo/issues/1331		 03:49
Zoffix ...and another one. Forgot the tests /o\ 03:50
Geth rakudo: 19604c2d21 | (Zoffix Znet)++ | t/08-performance/01-iterators.t
Test List.permutation got iterator opt methods

Fix: github.com/rakudo/rakudo/commit/d4f53ed5d9 Partially addresses github.com/rakudo/rakudo/issues/1331		 03:51
Zoffix .ask lizmat does this commit look sane? github.com/rakudo/rakudo/commit/d4f53ed5d9 It's part of fixing R#1331 and if it looks fine, I plan to add something similar to List.combinations (thought not yet clear what it'll look like) 03:52
yoleaux Zoffix: I'll pass your message to lizmat.
synopsebot R#1331 [open]: github.com/rakudo/rakudo/issues/1331 [perf] .count-only/.bool-only exist on combinations/permutations subs, but not methods
Geth roast: 78c3c865eb | (Zoffix Znet)++ | 3 files
[v6.d REVIEW] Merge race and hyper tests

  - 90% of the tests are identical. Put them all in a loop and shove
   method name to use into a variable. Add routine that sorts
   the results when we're testing `race`.
  - Toss JVM fudges; the entire test files were `#moar`-marked in Rakudo's
   spectest.data file
  - No tests were removed		 05:51
rakudo: f10e8b1346 | (Zoffix Znet)++ | t/spectest.data
Reflect race - hyper test files merge

  github.com/perl6/roast/commit/78c3c865eb
roast: a034c77f15 | (Zoffix Znet)++ | S02-types/instants-and-durations.t
[v6.d REVIEW] Formatting fixes

  - Toss Perl-5-ism and useless variables
  - Improve test descriptions
Orig: github.com/perl6/roast/commit/8650...cafa615259		 06:02
rakudo: b7f93bee16 | (Zoffix Znet)++ | t/05-messages/02-errors.t
Test (:::[]) does not spill guts

Moving error-message-textual-content-test from roast. Orig: github.com/perl6/roast/commit/5dcd...1b7b6abc9f		 06:10
roast: fa6247ddee | (Zoffix Znet)++ | integration/weird-errors.t
[v6.d REVIEW] DSETEM with `(:::[])`

DSETEM = "Do Not Spec Exact Text of Error Messages". These proved of little value and were interfering with readability changes in 6.c. Keep these in Rakudo's tests for now. If other impls need it, we can always do an Appendix set of tests to the spec that impls can optionally follow.
Test moved to rakudo's suite:
  github.com/rakudo/rakudo/commit/b7f93bee16
Orig: github.com/perl6/roast/commit/5dcd...1b7b6abc9f		 06:15
roast: 1a4548d3b3 | (Zoffix Znet)++ | S32-list/seq.t
Avoid double-heredoc

  - Easier to understand
  - Doesn't blow up highlights for the entire file in atom		 06:19
roast: cb96abffca | (Zoffix Znet)++ | S32-list/seq.t
[v6.d REVIEW] Simplify consumed Seq .perl.EVAL

  - Use fewer things
  - Use more precise tests		 06:30
roast: 17e1dda1fb | (Zoffix Znet)++ | S11-modules/require.t
[v6.d REVIEW] Move test closer to thing being tested

Orig: github.com/perl6/roast/commit/25c4...4bc34c449f		 06:34
travis-ci Rakudo build passed. Zoffix Znet 'Reflect race - hyper test files merge 06:50
travis-ci.org/rakudo/rakudo/builds/320559820 github.com/rakudo/rakudo/compare/1...0e8b134696
Geth roast: 6fa5f6481f | (Zoffix Znet)++ | S11-modules/require.t
[v6.d REVIEW] Use preciser test routines

So they give better output when failing. Orig: github.com/perl6/roast/commit/91f9...39d974379b		 07:11
roast: 85265395f5 | (Zoffix Znet)++ | S32-str/split.t
[v6.d REVIEW] Test actual data instead of stringified

Orig: github.com/perl6/roast/commit/10b2...96fde63574		 07:20
roast: 3518bd06fc | (Zoffix Znet)++ | S03-operators/buf.t
Remove trailing whitespace
 07:29
roast: c44a3015fa | (Zoffix Znet)++ | S03-operators/buf.t
[v6.d REVIEW] Improve tests for Buf.append and ilk

  - Use more appropriate test routines
  - Use more random item values (e.g. in case there's bug that just
   pushes .elems instead of the value itself)
Orig: github.com/perl6/roast/commit/a41c...9dc0581555		 07:31
roast: e1fe3edf32 | (Zoffix Znet)++ | S32-temporal/Date.t
[v.6 REVIEW] Improve DateTime subclass tests

  - Use proper test routines
  - Test type of created object
  - Lexicalize the test class
Orig: github.com/perl6/roast/commit/e144...f36ae36ed5		 07:36
roast: 4f924634f5 | (Zoffix Znet)++ | S32-temporal/Date.t
[v6.d REVIEW] Lexicalize test types

To avoid potential conflicts in other places of the test file
roast: 8846ba7407 | (Zoffix Znet)++ | S32-temporal/DateTime.t
[v6.d REVIEW] Improve DateTime subclass tests

  (earlier commit was actually improvements for Date sublclass test, not DateTime)
  - Use proper test routines
  - Test type of created object
  - Lexicalize the test classes
Orig: github.com/perl6/roast/commit/edaa...c5ed95a624		 07:40
roast: 53aa3810b8 | (Zoffix Znet)++ | S01-perl-5-integration/context.t
[v6.d REVIEW] Use more appropriate test routine
 07:42
roast: bbf4314142 | (Zoffix Znet)++ | S32-str/split.t
[v6.d REVIEW] Move vim thingie to end of file
 07:44
roast: 84d90464aa | (Zoffix Znet)++ | integration/error-reporting.t
[v6.d REVIEW] Don't spec ::AdHoc compilation error
 08:07
roast: 991061b0d9 | (Zoffix Znet)++ | S12-attributes/native.t
[v6.d REVIEW] Use more appropriate test routines

Orig: github.com/perl6/roast/commit/eb2b...0ac690e193		 08:10
roast: f17e8a6503 | (Zoffix Znet)++ | S02-names-vars/signature.t
[v6.d REVIEW] Fix up capture binding + postfix:<++> tests

  - Add parens to avoid a warning
  - We now have proper exception type; use that		 08:15
roast: fd0b683207 | (Zoffix Znet)++ | S02-names-vars/signature.t
[v6.d REVIEW] Use more appropriate test routines

Orig: github.com/perl6/roast/commit/2701...0647f60ca9		 08:42
roast: 46bd9a77b2 | (Zoffix Znet)++ | S03-operators/buf.t
[v6.d REVIEW] Fix code typo

Made in github.com/perl6/roast/commit/c44a...b821da5409		 09:15
roast: b429572460 | (Zoffix Znet)++ | S32-temporal/DateTime.t
[v6.d REVIEW] Fix wrong variable name

Forgot to run the test before commiting previous commit		 09:17
rakudo: a04c616471 | (Zoffix Znet)++ | 2 files
Improve cmp-ok Str metaop error message

Suggest to the user they need to use the &[...] Callable form. Fixes github.com/rakudo/rakudo/issues/1334
roast: 61c882c452 | (Zoffix Znet)++ | S03-operators/buf.t
Temporarily revert previous commits

The original tests have a precedence bug and the fixed precedence appears to uncover a bug in Buf.push/.unshift, but I'm not yet sure why the rest of the tests was passing and too sleepy to figure out today.		 09:34
travis-ci Rakudo build failed. Zoffix Znet 'Improve cmp-ok Str metaop error message 10:02
travis-ci.org/rakudo/rakudo/builds/320595112 github.com/rakudo/rakudo/compare/b...4c61647152
buggable [travis build above] ☠ All failures are due to: failed make test (6 failures). Across all jobs, only t/05-messages/02-errors.t test file failed.
lizmat good *, #perl6-dev 11:29
yoleaux 03:52Z <Zoffix> lizmat: does this commit look sane? github.com/rakudo/rakudo/commit/d4f53ed5d9 It's part of fixing R#1331 and if it looks fine, I plan to add something similar to List.combinations (thought not yet clear what it'll look like)
synopsebot R#1331 [open]: github.com/rakudo/rakudo/issues/1331 [perf] .count-only/.bool-only exist on combinations/permutations subs, but not methods
lizmat .tell Zoffix I'm slightly worried about the overhead this will incur for the normal case: if it is noticeable, then I would be in favour of removing count-only 11:31
yoleaux lizmat: I'll pass your message to Zoffix.
timotimo we may want to reword or reorder or resectionize the output of perl6 --help to make a few of the flags more obviously about the compiler rather than the user program. like the --output=name flag, which one might assume works like a redirect for program output, but in reality it does nothing at all unless you also have a --target=stage set 11:55
lizmat +1
timotimo also, at some point, sooner rather than later i'd expect, we'll want to have support for -i.foo and -i 11:56
lizmat timotimo: ?
timotimo working in-place with IO::CatHandle 11:57
with an extension passed it'll backup the original file, without it'll clobber
AlexDaniel +1+1+1 :) 11:58
I needed that several times
well, +i itself
m: say 42
camelia 42
AlexDaniel \o/ \o/ 11:59
timotimo $ s='a+b' perl -ne '$pos = length() - length($ENV{s}) - 1;
print if index($_, $ENV{s}) == $pos' eqns.txt
ugh, look at this ugly one-liner
isn't perl6 -ne '.say if .ends-with(%*ENV<s>)' much nicer 12:00
hm. i wonder if starts-with and ends-with and maybe other string-comparison-y operators will want to learn :ignoremark and :ignorecase
operators/operations 12:01
d'oh, just going through the output of seq 14323 14563435 with -ne '.say if $++ == 234' takes ~22 seconds in perl6 but in perl5 it takes like 2 12:09
Geth rakudo: 6cca27669f | (Elizabeth Mattijsen)++ | 3 files
Make List a value type if it only has value types

  - add a $!WHICH attribute to List (and therefor also on Array)
  - on first call to .WHICH, will freeze the '$!WHICH' value
   - if any of the elements is a container, it's not a value type
   - if any of the elements.WHICH is not an ValueObjAt, it's not a value type
   - if all value types, create a sha of the .WHICH strings as its .WHICH
... (18 more lines)		 13:02
lizmat the code for mixin in $!WHICH attribute on demand: gist.github.com/lizmat/4f9366ea226...e2a1d34b45 13:12
Geth roast: 5670acf676 | (Elizabeth Mattijsen)++ | 2 files
Follow semantic change of github.com/rakudo/rakudo/commit/6cca27669f

Lists consisting of only value types, are now also a value type. Please revert should 6cca27669f get reverted.		 13:29
timotimo lizmat: can we be sure that attrinited will work here? also, since autovivification of attributes and such is going away in the future, it'll have to be done differently then 13:54
Geth rakudo: Kaiepi++ created pull request #1335:
Ensure Configure.pl uses gmake for builds on BSDs		 14:24
lizmat timotimo: it worked fine for Set, Bag, Mix so far 14:34
timotimo m: say "which version" 14:40
camelia which version
timotimo m: my $a = (1, 2, 3, 4); use nqp; nqp::getattr(nqp::decont($a), List, '$!WHICH'); say $a.WHICH 14:41
camelia (Mu)
timotimo m: my $a = (1, 2, 3, 4); say $a.WHICH
camelia List|0B079F685A3B305504908F130B8523E6D103194F
timotimo the question is whether it's possible to cause that with user code
lizmat well, if that would be an issue. we would have seen it earlier with QuantHashes, as they use the same mechanism 14:44
timotimo that's fair
lizmat still I wonder whether ($a,$b) should decont $a and $b automatically or not 14:47
I guess that would be too much magic
timotimo it probably would, yeah
and how would you get The Other Meaning 14:48
lizmat yeah, true
timotimo does an array decont and then recont?
lizmat so ($a<>,$b<>) is then
yes
timotimo thought so
yeah, i'd imagined the <> operator
yoleaux Zoffix: take care of tossing deprecation in github.com/rakudo/rakudo/commit/9d9c7f9c3b and github.com/rakudo/rakudo/commit/cc6c05586b , bruh 14:49
lizmat blogs.perl.org/users/leon_timmerman...-5277.html # how Perl 5 is about to lose one of its most praise properties: stability 14:57
*praised
Geth rakudo: 729caa8f17 | (Elizabeth Mattijsen)++ | src/core/List.pm
Make sure <<$a b c>> is also a value type

Originally, anything with a $!todo would be considered a non-value type. This made <<$a b c>> not be a value type, because for some reason this is codegenned with an iterator. Now, we try to vivify the whole list when a .WHICH is requested: if that fails (usually because the List is lazy), the List is considered a non-value type.		 15:03
rakudo: 18557da950 | (Elizabeth Mattijsen)++ | src/core/List.pm
Make sure subclasses of List create consistent .WHICH

Well, at least consistent with the other .WHICH values so far, as in
  "Typename|stuff" rather than "Typenamestuff"		 15:36
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Make sure <<$a b c>> is also a value type 15:58
travis-ci.org/rakudo/rakudo/builds/320673070 github.com/rakudo/rakudo/compare/6...9caa8f17f4
buggable [travis build above] ☠ All failures are due to: failed make test (6 failures). Across all jobs, only t/05-messages/02-errors.t test file failed.
tbrowder .tell AlexDaniel i think i have all the pod config hash stuff ready for a PR (after a bit of tidying). i hope all the work i see on value arrays and typing doesn't interfere with it--we'll see--i don't think it will. i have tested all the stuff as it shows currently in the docs (including lists and hashes). 16:14
yoleaux tbrowder: I'll pass your message to AlexDaniel.
Geth rakudo: ec0021bbd6 | (Elizabeth Mattijsen)++ | t/05-messages/02-errors.t
Make sure we also expect STDOUT to give something
 16:18
lizmat I wonder if we could use a prefix op that would make valuetype an array so they could be easily used as keys in an object hash 16:35
I was thinking of a prefix:</> op 16:36
m: sub prefix:</>(\a) { a.List.item }; my @a = 1,2,3; my %h{Any}; %h{/@a} = 42; dd %h
camelia Hash[Any,Any] %h = (my Any %{Any} = ((1, 2, 3)) => 42)
lizmat wonders what TimToady jnthn timotimo masak moritz would think about that 16:37
m: sub prefix:</>(\a) { a.List.item }; my @a = 1,2,3; my %h{Any}; %h{/@a} = 42; dd %h{/(1,2,3)} 16:38
camelia Int %h = 42
travis-ci Rakudo build failed. Elizabeth Mattijsen 'Make sure subclasses of List create consistent .WHICH 16:44
travis-ci.org/rakudo/rakudo/builds/320683335 github.com/rakudo/rakudo/compare/7...557da95050
buggable [travis build above] ☠ All failures are due to: failed make test (6 failures). Across all jobs, only t/05-messages/02-errors.t test file failed.
moritz lizmat: prefix / collides with /.../ regexes 16:46
lizmat sigh, indeed... 16:53
apart from bikeshedding the form of the op, does the op make sense ?
Geth rakudo: 548dcf4656 | (Elizabeth Mattijsen)++ | src/core/Exception.pm
Reword X::Seq::Consumed message

Having seen some discussions recently, e.g:
   github.com/perl6/roast/commit/760dbe8465d7a5		 17:15
pmurias .seen jntn 17:16
yoleaux 21 Dec 2017 22:52Z <bartolin> pmurias: I fiddled with the test for nqp::cleardispatcher you added yesterday. If I'm not mistaken the workaround I added does not change the purpose of the test -- so I hope my change is okay for you.
I haven't seen jntn around.
pmurias .seen jnthn
yoleaux I saw jnthn 22 Dec 2017 13:34Z in #perl6-dev: <jnthn> Though looking up the Perl 6 Int type, for example, and then using nqp::box_i, would do it
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Make sure we also expect STDOUT to give something' 17:33
travis-ci.org/rakudo/rakudo/builds/320694266 github.com/rakudo/rakudo/compare/1...0021bbd6b1
lizmat afk& 17:46
travis-ci Rakudo build passed. Elizabeth Mattijsen 'Reword X::Seq::Consumed message 18:14
travis-ci.org/rakudo/rakudo/builds/320709820 github.com/rakudo/rakudo/compare/e...8dcf46561e
Zoffix I think .push and .unshift on Bufs should be removed. 19:41
yoleaux 11:31Z <lizmat> Zoffix: I'm slightly worried about the overhead this will incur for the normal case: if it is noticeable, then I would be in favour of removing count-only
Zoffix They're exactly like .append/.prepend, except they also come with a footgun that if you're not in full control of the argument you're passing, occasionally, the methods will throw
m: my $b := Buf.new; $b.push: 1 19:42
camelia ( no output )
Zoffix m: my $b := Buf.new; $b.push: 1 x 1
camelia Type check failed in push to Buf; expected uint8 but got Str ("1")
in block <unit> at <tmp> line 1
Zoffix m: my $b := Buf.new; $b.push: 1 xx 1
camelia Type check failed in push to Buf; expected uint8 but got Seq ($((1,).Seq))
in block <unit> at <tmp> line 1
Zoffix So, is that how we'd document them? (they aren't right now) ".push is like .append, but to be on a safe side always use .append"? 19:43
Going the other way, we could make them DWIM, but then that'd raise the question why it doesn't DWIM with, say, `(my Str @a).push` and there it makes more sense not to DWIM, so to me it seems like removing .push/.unshift from Bufs is more consistent than that. Yes, users will on occasion try to use .push and fail, but that's an easier failure to stumble upon than accidentally passing an iterable 19:45
m: -> :$stuff { say Buf.new.push: $stuff }(:stuff<1>) 19:46
camelia Buf:0x<01>
Zoffix m: -> :$stuff { say Buf.new.push: $stuff }(:stuff<1 2>)
camelia Type check failed in push to Buf; expected uint8 but got List ($(IntStr.new(1, "1"),...)
in block <unit> at <tmp> line 1
Zoffix ^ like in that code for example.
And we actually have 2 bugs in our own roast due to this behaviour: github.com/perl6/roast/commit/a41c...9dc0581555 19:56
Ah, nm. Not "due to this behaviour"
heh, chat logs' "plain text" doesn't work on Windows. Browser downloads the file (and it contains wrong new lines) 19:58
jnthn Zoffix: I think .push/.unshift on Buf is fine for single items (integers) at least 19:59
Well, with multiple args it's fine too if each is an integer 20:00
Perhaps should honor Slip
Zoffix Alright. 20:01
jnthn But since one can't put anything other than an integer into a Buf then $buf.push(@foo) is doomed to always fail
Or should be
That's what append is for
m: my Int @a; my @y = 1,2; @a.push(@y)
camelia Type check failed in assignment to @a; expected Int but got Array ($[1, 2])
in block <unit> at <tmp> line 1
jnthn Should be consistent with that, I'd think 20:02
m: my Int @a; my @y = 1,2; @a.append(@y)
camelia ( no output )
jnthn With that being fine
m: my @a; my @y = 1,2; @a.push(@y); dd @a
camelia Array @a = [[1, 2],]
jnthn And note that makes a nested array
The reason we have append is 'cus @a.push(@b.pop) should always transfer 1 item 20:03
Geth roast: 1f38f4acdb | (Zoffix Znet)++ | S03-operators/buf.t
Revert "Temporarily revert previous commits"

This reverts commit 61c882c452590b573b2cb5df3f60202a09d94442.
The rest of the tests was passing, because the precedence errors in the original tests always caused .append/.prepend to be called.
roast: 44719c8113 | (Zoffix Znet)++ | S03-operators/buf.t
[v6.d REVIEW] Fix erroneous | causing explosions

Orig: github.com/perl6/roast/commit/a41c...9dc0581555 Explosions found after fixing precedence errors in
  github.com/perl6/roast/commit/1f38f4acdb		 20:05
rakudo: 4f86567168 | (Ben Davies)++ | Configure.pl
Ensure Configure.pl uses gmake for builds on BSDs
 20:15
rakudo: 78166e1ef8 | (Zoffix Znet)++ (committed using GitHub Web editor) | Configure.pl
Merge pull request #1335 from Kaiepi/configure

Ensure Configure.pl uses gmake for builds on BSDs
synopsebot RAKUDO#1335 [closed]: github.com/rakudo/rakudo/pull/1335 Ensure Configure.pl uses gmake for builds on BSDs
pmurias jnthn: hi 20:18
jnthn: when we are making a thunk for orelse (and similiar), it seems we are migrating stuff into the thunk being created: github.com/rakudo/rakudo/blob/mast....nqp#L9067 20:19
jnthn: does it make sense to do that, as p6capturelex doesn't work correctly in that case
jnthn Hm, why doesn't it work correctly? 20:22
pmurias jnthn: when we have nested orelse: Foo orelse $x orelse $x, the second orelse block is nested inside the first one 20:25
Zoffix Don't know if related, but there's this bug RT#132337
synopsebot RT#132337 [new]: rt.perl.org/Ticket/Display.html?id=132337 [BUG] Code block inside `andthen`/`orelse` doesn't close over lexical variables correctly
pmurias jnthn: evil moar autoclosing hackery "fixes" it 20:26
jnthn Given there's a bug report that Zoffix just linked to, it probably doesn't really fix it :)
Yeah, that looks like some mis-nesting is happening or some such 20:28
It's OK if the second block ends up nested inside the first provided it gets closure cloned appropriately inside of there
pmurias jnthn: are we really migrating the blocks in the right direction? 20:30
jnthn No idea 20:31
Also, are they being migrated in a sufficiently fine-grained way, 'cus I think the migration mechanism is statement level.
pmurias as we are migrating stuff from $*W.cur_lexpad() into the $block (the thunk we are creating)
jnthn: if it's non-obvious I'll try to get some proper sleep and I'll check tomorrow what spectests break if we reverse the migration direction ;) 20:34
jnthn :)
Yeah, it's not clear to me what it's doing or should be :) 20:35
Geth rakudo: e0af68a08a | (Zoffix Znet)++ | 2 files
Remove deprecation fudge for native type instantiations

Deprecation period has expired, so now all forms are fatal		 20:43
roast: dbc1929dc7 | (Zoffix Znet)++ | README.md
Mention errata branches
 21:09
rakudo: 9606aea22a | (Zoffix Znet)++ | tools/build/Makefile-common-rules.in
Fix issues with roast checkout

Unsure what the original purpose of setting the push_url stuff was about, but on any clean checkout I always have to change origin URLs, if I want to push anything, as it gives errors about push access.
Fix by not messing with anything and just checking out over HTTPs.		 21:10
[Coke] wonders if there's a way to get the acme advent highlighter to treat a section as a <pre> 21:58
[Coke] realizes he can just make it a literal pre. 22:00
[Coke] puts perl6advent.wordpress.com/?p=7045&...eview=true up 22:11
i *think* it's scheduled...
[Coke] pushes github.com/coke/p6-cube-three 22:40
timotimo [Coke]: reading your post 23:38
do you think gather for { take foo } is cleaner than do for { foo }?
[Coke]: i wonder if "[eq] %!Sides{$side}" is nicer than "%!Sides{$side}.all eq %!Sides{$side}[4]" 23:40
.o( also, .all eq .[4] given %!Sides{$side} ) 23:41
timotimo wonders about double-width letters for the side names and ansi background colors for the squares 23:47
nice post, [Coke]++

Powered by Cro and the Raku® Programming Language.

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

Thank you!