timotimo Zoffix: it can still use pick, but first srand 00:01
AlexDaniel .tell samcv if I'm getting “should eventually be unreachable” thingy in my code, what does it mean? 00:05
yoleaux2 AlexDaniel: I'll pass your message to samcv.
Zoffix timotimo: good plan 00:06
hm, actually... It's not lexical and'll affect all randomness, right? Rather not pop into there lest other tests somewhere else in file use randomness one day and will give non-random stuff. 00:08
IO::Handle has `.t` method that says whether the handle is to a TTY... how to get such a handle? 00:30
m: say $*OUT.t 00:31
camelia False
Zoffix IO Action Plan current status: 14 pages long. Hoping to finish and make it available for review in 24 hours. 01:03
Zoffix drops to bed
geekosaur on unix you open a /dev/tty*, /dev/pty* (requires opening the master first), or /dev/tty for direct connection to controlling terminal, or on some systems /dev/cua* for direct access serial port (ignoring carrier detect) 01:09
or, $*IN $*OUT $*ERR if running in a terminal and not redirected 01:10
however, libuv has a tty abstraction that moarvm doesn't currently bind
Zoffix Ah. Thanks. $*OUT.t and "/dev/tty".IO.open.t.say give me True. 01:11
[Coke] Zoffix: I JUST posted your feb report on the TPF site. Sorry for the delay, keep up the good work. :) 01:57
Zoffix [Coke]++ thanks. 01:58
[Coke] Also, the March 2017 Call for Proposals just went out. 01:59
Zoffix .tell samcv TPF posted call for proposals. Last day is April 2nd. Ensure TPF confirms your proposal went in this time :) news.perlfoundation.org/2017/03/cal....html#more 02:00
yoleaux2 Zoffix: I'll pass your message to samcv.
samcv thanks Zoffix :)
yoleaux2 02:00Z <Zoffix> samcv: TPF posted call for proposals. Last day is April 2nd. Ensure TPF confirms your proposal went in this time :) news.perlfoundation.org/2017/03/cal....html#more
Zoffix just rejected PR# 666 :) 02:01
Out with evil! 02:02
Geth nqp: 67a74d2491 | (Samantha McVey)++ | tools/build/MOAR_REVISION
Bump MoarVM to fix bug with case insensitive regex w/ synthetics

a8448142 Improve description for "should eventually be unreachable" error 498a21d8 Fix case insensitive string compare bug when synthetics in haystack c4de954b teach adjust_nursery about smallbigints		 04:47
¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...-ga8448142
rakudo/nom: e87179ddc0 | (Samantha McVey)++ | tools/build/NQP_REVISION
Bump Moar/NQP to fix bug with case insensitive regex w/ synthetics
 05:02
¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....-g67a74d24
samcv everything should be good now :) no more 05:03
m: say (88.chr ~ 875.chr ~ 8413.chr) ~~ /:i x /
camelia should eventually be unreachable
in block <unit> at <tmp> line 1
samcv plus that error is now actually not going to be reached and actually describe wtf it does
masak wow, that's a confusing error message :) 05:34
samcv yes. when i read it i thought something would have gone horrrrribly wrong
but turns out not quite so horrible. just a horrible message 05:35
m: say (88.chr ~ 875.chr ~ 8413.chr) ~~ /:i x /
camelia Nil
samcv yay and that's fixed now too
gotta add that to roast. more tests are needed for unicode for sure
wow we only have 36 tests :O github.com/perl6/roast/blob/master...norecase.t 05:37
for m:i// 05:38
interesting. if i run t/spec/S05-modifier/ignorecase.t enough times some of the 'blahst' ~~ m:i/st/ tests pass 05:43
interesting there would be any possible different between runs. but it is a tricky situation, though i'm not sure why that would happen
i mean see how it colud produce inconsisntent results based on what we currently have, but don't understand not getting consistently inconsisntent results (though 95% time will fail all 11 tests) 05:44
hmm having a problem with the test harness. everything is fine running the test with `perl6` but if i use perl 5 harness or perl 6 harness it doesn't notice the last two tests 06:08
and say you planned 38 tests but ran 36
here's my copy. gist.github.com/samcv/5a98a3febea6...9c0f1046cb goes in t/spec/S05-modifier/ignorecase.t 06:09
running it with `prove -v -e perl6 ignorecase.t` works fine as well, and shows it running all 38 tests 06:11
Woodi hmm, is something like $seq.grep(/\d/).ifelse(*>0;*+1;*-1). ... allowed ? I gues programmable GPUs have such possibility... or switch :) would be *realy* handy for parallel one-liners... 08:58
also how hyper/race handle side efects from map eg: my $counter = 0; ... . map( ...; $counter++ ). ... in parallele ? 08:59
timotimo Woodi: you put side effects in, your computer explodes, and its your fault 09:33
[Tux] This is Rakudo version 2017.03-29-ge87179ddc built on MoarVM version 2017.03-25-ga8448142 09:53
csv-ip5xs 3.012
test 13.577
test-t 5.064 - 5.139
csv-parser 13.595
Geth nqp/master: 4 commits pushed by (Pawel Murias)++ 10:02
Zoffix Got response for my bus ad thing: "For production purposes, 1 ad would cost you $ 275 + HST" 10:39
... maybe when I become rich and famous, I'll make it a point to run 1 bus ad, but for now I'll pass :) 10:40
dogbert17_ Zoffix: aren't you supposed to be asleep at this time? 10:58
Zoffix dogbert17_: 7am. I'm supposed to be getting dressed for work. 10:59
FWIW:
gfldex – org.remove_member
Removed themselves from the Perl 6 organization
Germany 2 days ago
Whoever has the skills to handle humans. May want to take a look into that and restore the membership. 11:00
s/handle/deal with/; 11:03
Whatever the right word supposed to be used.
lizmat toons.gotblah.com/archive/geekandp...3-2301.jpg :-) 11:30
jnthn :P 11:31
Today's fun discovery: when we install scripts on Windows, we make a .bat script to launch them. This works OK, until you try to kill such a process from, say, Java, at which point it kills the cmd.exe running the batch script and leaves the moar process intact :/ 11:34
Bizzarely, it also somehow keeps on getting the output, despite the process in the middle in theory being gone, so I've no idea how that works 11:35
geekosaur you'll continue to get output until all processes that inherited the pipe close it 11:39
jnthn Oh, right, it inherits, not acts as a pass-through 11:40
geekosaur same happens on unix if you use a shell to launch something and then kill just the shell. but unix has process groups so you can arrange to kill the entire process group
jnthn *nod*
Guess we may have the issue there also, then, thanks to the shell scripts we produce
geekosaur (I don't know if whatever libuv abstraction you're using creates process groups automatically on unix) 11:43
Zoffix What does IO::Handle.lock/.unlock supposed to do? 12:22
If a handle is in `:w` mode, it dies with failed to .lock; if it's in read mode, .lock don't seem to be doing anything 12:23
Hm, just calls to undocumented nqp::lockfh($!PIO, $flag)/nqp::unlockfh($!PIO); 12:24
Zoffix digs the sauce
it's fcntl stuff 12:45
jnthn lizmat: Did you have chance to look at my gist at all? :) 12:50
lizmat jnthn: not this morning: architect visit and some schlepping :-) 12:53
I left two more comments yesterday, which github seems to have eaten :-( 12:54
one was about spello's and too tired to look at it last night
the other one was suggesting Async as name of the role
aka AsyncHyper and AsyncRace
Zoffix: re nqp::lock: that's the engine behind the Lock class 12:56
ah, sorry, read wrong: you're talking about lockfh 12:57
Zoffix :)
lizmat please ignore me :-)
jnthn lizmat: Parallel would be more fitting than Async here
lizmat jnthn: maybe just Para ? 12:58
ParaHyper ParaRace
Conc ?
which would match ConcBlockingQueue 12:59
jnthn Well, but hyper and race already imply parallel...
lizmat but a specific type of parallelism
Zoffix keeps getting bitten by .foo: "something" ?? $x++ !! $y++ 14:20
And no way for computer to figure out what I actually mean :) 14:21
.foo("something") ?? …
DrForr .foo:? # Consume just the next token? 14:22
jnthn Just use parens? :P 14:23
Zoffix Yeah, parens are the fix :)
lizmat m: my $don't-do-this = 42; dd $don't-do-this # blogs.perl.org/users/morandimus/201...y-day.html 14:59
camelia Int $don't-do-this = 42
jnthn Yeah, saw that one :) 15:01
Zoffix heh 15:02
jnthn m: my $user = 'jnthn'; say "$user's beer"
camelia 5===SORRY!5=== Error while compiling <tmp>
Variable '$user's' is not declared. Did you mean '$user'?
at <tmp>:1
------> 3my $user = 'jnthn'; say "7⏏5$user's beer"
Zoffix Good error.
jnthn :)
Maybe a little unforunate we use single quotes around a variable name when the variable can contain '
But that nit aside, yes, it's clear it's interpreted the whole thing as a variable name 15:03
Zoffix Curious, is there any point to role IO {}? I will propose to nix its `umask` method and it'll then be entirely empty... 15:31
jnthn If it's just empty, probably not
Zoffix OK. 15:32
jnthn I didn't even know it as a role :) 15:33
Zoffix heh
jnthn *was 15:34
[Coke] SmokeMachine: it was your CLA that bounced, yes? 16:03
Zoffix Yes. 16:04
[Coke] I can confirm I've received it. Pending them fixing the mail alias, it's entered in the list of CLAs, feel free to give them a commit bit. 16:07
SmokeMachine [Coke]: yes 16:14
[Coke] welcome aboard. 16:18
Zoffix \o/ 16:20
SmokeMachine \o/
:)
Zoffix .tell jnthn SmokeMachine's CLA has been received by TPF, and per [Coke] ( irclog.perlgeek.de/perl6-dev/2017-...i_14318800 ) a commit bit can be given. The github username's "FCO" github.com/FCO 16:32
yoleaux2 Zoffix: I'll pass your message to jnthn.
SmokeMachine :) 16:37
Zoffix [Coke]: is one person allowed to do more than 1 TPF grant at a time? I have no current plans to do so, but just curious of the rules. 17:05
[Coke] rules are available on the site. pretty sure there's no explicit rule, but I imagine you'd have to be convincing, (and funding would have to be available)
www.perlfoundation.org/rules_of_operation\ 17:06
er, minus the \
perlpilot
.oO( "I can't believe TPF gave Zoffix a grant in the first place, now he's applying for another?!?!?" ;-) )
Zoffix perlpilot: I explicitly said that's not what I plan on doing. 17:10
Yeah, rules don't state anything on the limit of grants. And I recall $10,000 being the upper limit to the requested amount, though I don't see it now on the rules_of_operation page...
perlpilot Zoffix: yeah, I'm teasing 17:11
Zoffix Why do people do that? It's weird. 17:12
mst it was ... a silly joke, as identified by the double interrobang and the ;-) 17:13
Zoffix I hope this page is outdated :) Last entries are from 5 years ago: www.perlfoundation.org/sponsors 17:14
perlpilot I seem to remember that there were some donations last year. 17:16
booking.com maybe? 17:17
Zoffix Ah, the 10K limit is mentioned in www.perlfoundation.org/ian_hague_pe...ent_grants 17:21
"The Grants Committee has a constitutional limit of USD$10,000 on any given grant, and we envision Perl 6 development grants larger than this."
[Coke] those limits are for that particular fund which is done. 17:22
OH.
misread, yes.
"quick respond before you've read the whole thing" -- my brain, probably. 17:23
Zoffix And more explicitly stated on www.perlfoundation.org/grants_committee_charter 17:24
[Coke] see, you're better at this than I am. 17:28
perlpilot [Coke]: OOC, how does one become a grants committee member? 17:33
[Coke] www.perlfoundation.org/grants_committee_charter - section 3, paragraph 3. 17:37
perlpilot So, if someone wanted to become a member, they'd have to petition one of the existing members to nominate them? 17:39
[Coke] Yes. We just elected a new member, actually. 17:40
news.perlfoundation.org/2017/03/new...r-sec.html
and before: news.perlfoundation.org/2017/01/gra...new-m.html 17:41
perlpilot So, I guess www.perlfoundation.org/who_s_who needs an update?
Hmm. from the committee charter you linked, it says the grants committee shall have at most 11 members but the Jan 25 post says there are 13 members 17:44
[Coke] I think the implication is "voting" memebers. 17:57
*members
perlpilot I thought charters were for being explicit rather than implicit :-) 20:00
But "voting members" makes sense.
[Coke] perlpilot: added it as a GC issue. 20:02
masak heh -- in Perl 5, why is it `while (/.../g)` and not `for (/.../g)`? :) 20:41
that one just tripped me up and I readed to resort to perldoc
perlpilot You can do either depending on what you're trying to achieve :) 20:42
masak well, I did `for`, and my $1 etc ended up containing only info about the last match ;) 20:43
(not what I was trying to achieve)
geekosaur because it doesn't return a list; it's an iterator that needs its crank turned to produce the next match 20:45
(otherwise it does everything up front and if there are 100000 matches in a long string slurped from a file your machine will become very unhappy) 20:46
also some questions about how you handle $1 and friends 20:47
lizmat Zoffix: there have been other donations, specifically to the Perl 5 and to the Perl 6 core development funds 20:49
Zoffix sweet
Geth rakudo/nom: 1bacc61031 | (Elizabeth Mattijsen)++ | 4 files
Replace all Foo.CREATE by nqp::create(Foo)

It is *at least* 14x faster. This is caused by Mu.CREATE not being jitted. By using the direct nqp::create equivalent, we completely remove the need to call this (currently) unoptimizable method. So fewer frames, and thus (I assume) lower memory pressure as well.		 21:15
Zoffix :o 21:34
lots of boost 21:35
jnthn Hm, I wonder if we shouldn't just be compiling .CREATE into nqp::create :) 21:47
Probably we should
Zoffix Is there good reasoning for IO::Handle.encoding attribute to be mutable? Feels like it shouldn't be like the other public ones, but is there a usecase where switching encoding or bin/non-bin is useful? IO::Handle's $!PIO is mutable, so there's no real multi-thread safety attained from making it immutable. 21:48
Zoffix writes down proposal to make it immutable; some things like .Supply aren't affected for bin-non-bin change, so it feels problematic to allow mutation 21:54
jnthn Zoffix: I can't really think of one 21:56
lizmat Zoffix: well, in the socket world, the header is usually ISO-8859-1, but the body could be UTF-8
so I've known people wanting to change encoding on the handle 21:57
jnthn Well, there's that, yeah :)
fwiw, my expectation is that we'll gradually move towards a model where the VM only does binary I/O, and IO::Handle coordinates the decoding
lizmat and this could well extend to files
jnthn (Which will still be VM-backed for the common encodings)
But we'll be able to support user-space ones also then. 21:58
But since we'll be pulling that stuff into Perl 6 space, we will have quite a lot more control over it
I'd probably make .encoding immutable and have an explicit change-encoding method or so, perhaps.
samcv Zoffix, have you worked on the TAP for perl 6 at all? 22:00
i'm getting a problem when using `prove` or using harness6 will not see the last few tests i run
but if i run the file manually it works perfectly
err maybe prove is fine actually. in verbose mode 22:01
but if i do `make t/spec/S05-modifier/ignorecase.t` it says bad plan and it doesn't see the output of the last two tests 22:02
Zoffix lizmat: thanks. Will leave it as is then. 22:10
samcv: I fixed a few bugs, but there are no docs for it and I found the code incomprehensible. 22:11
samcv :(
the only difference this test has when i run it using `perl6` is that it says `# Looks like you failed 10 tests of 38` at the end 22:12
but the other test doesn't
Zoffix samcv: can you paste the ignorecase.t version you're running? I'm gonna build HEAD and give it a whirl
(after a dinner break)
samcv here you go Zoffix irclog.perlgeek.de/perl6-dev/2017-...i_14316624
Zoffix thanks; will take a look 22:13
samcv thank you
oh Zoffix it looks like the todo on line 87 on that file. if i remove it the problem goes away 22:15
so it must be related to the fudging or something
somehow causing it not to run the tests after them at lal
yeah Zoffix if i move them above the todo then it runs them fine. 22:20
Zoffix I think it's the fudger that's problematic in this case, the .rakudo.moar file don't got those tests in it 22:21
samcv: yeah, this version works: gist.github.com/zoffixznet/bdd5468...93eedd1fd3 22:26
I changed the todo count to just 1, not 10
I think it sees just 1 test but is looking for 10, so something makes it choke 'cause it can't find them, so it loses the last 2
samcv yeah maybe 22:27
i'm just gonna re-order it
to put the other tests above it
Zoffix samcv: just change the todo count to 1
'cause if you simply move other tests, in the future when we add more tests to the end, they'll be disappearing just like these 2
samcv ah 22:28
good plan Zoffix
*go to google* search with this korean text. google gives me Error 400 Your client has issued a malformed or illegal request. :( 22:45
idk maybe because it has newlines in it? or something weird
Geth roast: 7b14317673 | (Samantha McVey)++ | 2 files
Rename test-gen.p6 to more descriptive mass-roundtrip-test-gen.p6
 23:34
roast: 1524fa25fa | (Samantha McVey)++ | S05-modifier/ignorecase.t
Add tests for case insensitive regex for synthetic graphemes

These tests were prompted by this commit:
  github.com/MoarVM/MoarVM/commit/49...c6285eafa2
  AlexDaniel++ for finding this bug
AlexDaniel m: say (88.chr ~ 875.chr ~ 8413.chr) ~~ /:ignoremark :i x / 23:38
camelia ｢Xͫ⃝｣
Zoffix suggests 88.chr to be "X", to make it clearer to the reader an 'x' is part of the string and the regex should match 23:42
Well, I lied. 23:49
01:03 <Zoffix> IO Action Plan current status: 14 pages long. Hoping to finish and make it available for review in 24 hours. 23:50
23:49 <Zoffix> Well, I lied.
Still need to sort out a couple of routines and then polish off the assembled plan, but tired for tonight. 23:51
But definitely in the next 24 hours :)
Zoffix &

Powered by Cro and the Raku® Programming Language.

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

Thank you!