Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2017-02-03

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:28 pyrimidine joined #moarvm
02:10 pyrimidine joined #moarvm
02:12 KDr2 joined #moarvm
02:29 pyrimidine joined #moarvm
02:34 pyrimidine joined #moarvm
02:48 ilbot3 joined #moarvm
02:48 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
03:35 pyrimidine joined #moarvm
03:47 travis-ci joined #moarvm
03:47 travis-ci MoarVM build failed. Bart Wiegmans 'Merge pull request #517 from lucasbuchala/nfg-typo
03:47 travis-ci https://travis-ci.org/MoarVM/MoarVM/builds/197701952 https://github.com/MoarVM/MoarVM/compare/70d4bd53e62a...66961d88a803
03:47 travis-ci left #moarvm
05:35 pyrimidi_ joined #moarvm
06:35 domidumont joined #moarvm
06:42 domidumont joined #moarvm
06:58 samcv I seem to be getting an issue where assigning to a hash is applying the operation to _two_ different keys
06:58 samcv but only when the hash table is super populated
07:04 samcv oh wait i think i know what's causing it...
07:15 domidumont joined #moarvm
07:19 brrt joined #moarvm
07:20 domidumont joined #moarvm
07:24 domidumont joined #moarvm
07:25 brrt so, wtf
07:25 brrt one of the QAST subtest is broken
07:26 samcv when did that happen?
07:26 brrt not sure
07:26 brrt but travis was complaining about it this morning
07:27 brrt aha, that file has had a *lot* of edits
07:35 pyrimidine joined #moarvm
07:51 domidumont joined #moarvm
07:58 brrt joined #moarvm
08:26 brrt joined #moarvm
08:54 Geth ¦ MoarVM: samcv++ created pull request #520: Apply the radix speed improvement to MVM_bigint_radix as well
08:54 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/pull/520
08:55 samcv jnthn, you said you gave me commit access but i wasn't able to apply the change from this pr ^ it said i didn't have permission to push
09:00 arnsholt Check the URL of the repo? If it's HTTP pushing tends to not work
09:00 samcv no it's git
09:01 samcv upstream        git@github.com:MoarVM/MoarVM.git (push)
09:03 arnsholt In that case, sounds like you've not been added after all
09:08 arnsholt Doesn't look like I've the privs to add you either, I'm afraid
09:40 brrt it appears that the QAST test worked before pmurias changed it, and I'm not sure why the change was mad
09:40 brrt i'm going to contact him
09:42 brrt samcv: i'll have a look at your PR and see if I can get you a commitbit; I'm not sure if I can though
09:44 brrt samcv: I can't because I'm a regular member, not an owner; pmichaud and jnthn are; i can't see you in the list yet
09:44 samcv kk
10:58 agentzh joined #moarvm
11:04 Geth joined #moarvm
11:11 pyrimidine joined #moarvm
11:49 brrt well, the change appears good to me, so i'll merge it
11:55 brokenchicken Seems mine has all failures due to that qast test: https://github.com/MoarVM/MoarVM/pull/518
11:55 brokenchicken I did stresstest it on rakudo though and it was clean
12:10 jnthn samcv: Just checked on GitHub, it stays you have an invitation that is still pending
12:10 samcv hmm
12:10 jnthn "Invitations are sent via email and can be accepted at https://github.com/MoarVM"
12:10 samcv i get so many email
12:10 jnthn :P
12:11 jnthn Try at the URL?
12:11 samcv found it
12:11 jnthn :)
12:12 jnthn Was sure I'd given you access... :)
12:12 Geth ¦ MoarVM: f7ac91ccb2 | (Samantha McVey)++ | src/math/bigintops.c
12:12 Geth ¦ MoarVM: Apply the radix speed improvement to MVM_bigint_radix as well
12:12 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/f7ac91ccb2
12:12 jnthn \o/
12:18 samcv moarvm has no logo right?
12:19 jnthn No
12:20 jnthn Not as a matter of policy, just because everyone was too busy building a VM or something :P
12:20 samcv heh
12:20 jnthn I'm not sure what we'd have as a logo :)
12:20 samcv maybe just MVM?
12:21 samcv idk would be cool to have just a basic something as a logo on github
12:21 samcv or even just the word MoarVM
12:22 jnthn Yeah :)
12:25 jnthn Maybe brokenchicken has design skillz and an idea? :)
12:26 KDr2 joined #moarvm
12:27 samcv well here is just MVM with like all the effects applied :P https://a.uguu.se/2I7sqln75Nk4_MVM.png
12:27 samcv hah
12:27 samcv just checked all the boxes. inner glow,shadow, pattern some other random things :P
12:27 samcv i'm sure a better one can be made
12:28 jnthn :D
12:29 jnthn It looks oddly like the letters were cut out of a doormat...
12:29 samcv haha
12:32 timotimo you know how there's a V inside the M?
12:36 samcv we should just get an actual doormat
12:36 samcv and go wild. and eventually by the end we will end up with a logo
12:42 timotimo after all, we are the kiddos who high-five every time a new emoji is coined!
12:43 jnthn So...once a year? :P
12:43 jnthn Or maybe a couple of times, 'cus we can see Unicode drafts too :P
12:44 timotimo well, the high-fiving itself is quite gratuitous
12:44 timotimo since it's our defining characteristic to people on the outside of our cabal
12:51 samcv why apple always gotta mess up emoji tho
12:51 samcv like replacing the pistol emoji with a water gun
12:52 samcv timotimo, also there's no high five emoji :\
12:53 pyrimidine joined #moarvm
12:55 timotimo "you are like a baby, watch this"
12:55 * timotimo o/⁵o samcv
12:56 timotimo just to be sure, i don't mean to step on anyones foot, just making light of a situation i'm not entirely grokking
12:58 jnthn Yeah, well, I can see from the post they had a sub-optimal experience
12:58 timotimo aye, they have :(
13:00 timotimo i'm not sure the person who suggested that "users should use msi instead" is even a perl6 regular?
13:00 jnthn No idea
13:01 timotimo also, on the one hand, it doesn't seem fair to be judged by how random people on twitter talk about us, on the other hand that just feeds the "perl6 people are an exclusive membership-only club"
13:01 timotimo also i didn't like the allegation of racism in that one post
13:01 jnthn I've no idea what that bit was all about.
13:02 timotimo though i'm very white, so it could just be i'm totally blind to racism
13:03 jnthn At the end of the day, though, I see volunteers freely giving their time to solve all kinds of real, and sometimes difficult, problems people run in to when using Perl 6. For sure we have some fun along the way too.
13:03 jnthn But yeah, if you're going to categorize a bunch of people who work on hard stuff as sitting around laughing at the latest cat face emoji all day...well, you're probably not going to endear yourself to them. :)
13:04 jnthn lunch; bbiab
13:05 * samcv looks away from the latest cat emoji to read this channel
13:06 samcv wow i'm up to 76% memory usage somehow. i have 16GB
13:07 stmuk joined #moarvm
13:07 pyrimidine joined #moarvm
13:08 samcv timotimo, where is that windows perl 6 discussion which the msi is mentioned?
13:08 brokenchicken jnthn: timotimo had a logo and I liked the idea.
13:08 timotimo https://twitter.com/joshuamy/status/826685435606806528 - it's not easy to navigate the fucking crap that is twitter threads, but here you go
13:09 timotimo well, it's certainly not as crappy on the twitter.com frontend compared to inside tweetdeck
13:09 pyrimidine joined #moarvm
13:10 timotimo brokenchicken: you're refering to the crab?
13:11 brokenchicken I think it was a butterfly growing out of MVM or something or other?
13:11 samcv @jnthnwrthngtn Not whining. Sharing my experiences. I am not going to interact with #perl6 for a while.
13:11 samcv :\
13:11 samcv pretty rude
13:12 brokenchicken huh
13:13 brokenchicken looks like there's more drama I missed.
13:13 samcv is it true if they just installed the rakudo star distro they wouldn't have had problems?
13:14 brokenchicken samcv: the original tweet from Jan 4th where I actually told him to use a user release instead of a dev commit is true.
13:15 brokenchicken samcv: but it's apparently now broken in 2017.01 and the prose now tries to paint me as an evil "leading Perl 6 developer" who told the user to use a release but release is broken and I'm oh so refusing to take any input.
13:15 brokenchicken timotimo: do you know what I'm talking about? I think it was a draft or something. I could make it proper.
13:16 samcv also https://twitter.com/sinan_unur/status/826221362859634688
13:16 samcv really :\
13:17 samcv so they just download it and run roast just so they can say "look at all these things it fails"?
13:17 samcv makes me angry
13:17 samcv also do they know that roast tracks git branch?
13:18 brokenchicken Yes, it's not 6.c tests, but still-to-be-vetted master branch. And the failures are known: https://github.com/perl6/roast/issues/197 many of them are poor tests and not an indication of problem with Rakudo on Windows
13:18 samcv yeah exactly
13:19 samcv and then complains about a module that doesn't work with windows doesn't work
13:19 nine samcv: there are 7.4 billion people on earth. I could care less about this one individual being happy with Perl 6. He seems to try so hard to be unhappy, so let him have his desires.
13:19 samcv he does try pretty hard
13:19 nine s/could care/couldn't care/
13:23 timotimo brokenchicken: i just plugged in my wacom tablet
13:25 brokenchicken Heh. That reminds me I still have mine somwhere!
13:26 brokenchicken Last used 5 years ago :o https://scontent-lga3-1.xx.fbcdn.net/v/t1.0-9/75019_4849133318784_1632425265_n.jpg?oh=44446d57ddaaaf3e8356931c614224d0&oe=590C4F3B
13:28 timotimo http://imgur.com/NAXuBRE
13:28 timotimo .o( richard dawkins? )
13:29 samcv m: $_ = ['this' => 'that', 'other' => 'thing']; say .».key.all ~~ Str
13:29 camelia rakudo-moar f24968: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Malformed postfix call (only alphabetic methods may be detached)␤at <tmp>:1␤------> s' => 'that', 'other' => 'thing']; say .⏏».key.all ~~ Str␤»
13:29 brokenchicken timotimo: yeah
13:29 samcv is there a reason it doesn't let me do this?
13:29 samcv other than no fun allowed
13:29 brokenchicken timotimo: is that the same one you showed last time?
13:29 brokenchicken This one does look like a crab, but I thought it was a butterfly heh
13:29 timotimo the one i showed back then was drawn on my whiteboard
13:29 samcv nice crab
13:29 timotimo http://i.imgur.com/gMyYJjN.jpg
13:30 timotimo attack its weak point for massive damage
13:30 brokenchicken heh
13:30 timotimo tut-tut, no fun allowed
13:31 timotimo (until you file that bug on rakudobug@)
13:34 jnthn Reflecting on things a bit over lunch: I suspect that for various people - certainly me included - in the Perl 6 community, there's a bit of a siege mentality still. Thanks to years and years of hearing "it's vapourware", "it's fucked", giving a Perl 6 talk and having the conf orga say "well, nice if it ever happens", some folks trying to undermine it all by proposing a Perl 7, and so many more things.
13:36 jnthn Hopefully that will lessen with time. But generally - for *any* team of people I've seen building software, not just us - if the environment they're in is hostile, the result is defensiveness.
13:36 timotimo jnthn: would you be okay with an MVM crab? :)
13:37 timotimo not saying mvm is crap!
13:37 nine You can generalize that to humans in...general. You get attacked, you defend. It takes hard training to not get defensive.
13:37 brokenchicken Aren't crabs slow and move the wrong way? :)
13:37 timotimo no, everything else moves the wrong way
13:37 timotimo and crabs can be pretty fast
13:37 brokenchicken :D
13:38 jnthn nine: Yes, it's something I have to conciously work at.
13:38 timotimo ifyou have a crazy amount of self-confidence (or just confidence) you can probably just "tank it with your face" whenever attacks come in
13:39 samcv how do i negate this?
13:39 samcv <:alpha + [»]>
13:39 nine timotimo: in general, you don't tank it, you step aside.
13:39 timotimo negate all of it?
13:39 samcv <!:alpha> is like what i want, but i'm not sure how to have it negate all of them
13:39 samcv ye
13:39 brokenchicken <-:alpha - [»]> ?
13:40 timotimo i'd expect what brokenchicken just wrote
13:40 nine timotimo: by not getting defensive, the attack runs into a void. And then you come back from the side and get your arguments placed. Getting defensive or blocking is just a form of counter-attack which provokes defensive behavior.
13:40 brokenchicken m: say "a.,432»«as" ~~ m:g/<-:alpha - [»]>/
13:40 camelia rakudo-moar f24968: OUTPUT«(「.」 「,」 「4」 「3」 「2」 「«」)␤»
13:41 samcv yeah that works perfectly :)
13:41 timotimo mhm
13:42 nine Classic turn around phrasing is "Yes, I can understand you very well. Can you understand how I feel about ...?". It's hard to attack or get defensive when you are taken seriously and being faced with a question that evokes an emphatic response.
13:43 samcv also guys <:ws> is 2x faster than <ws>
13:43 samcv we can make cursor 2x faster
13:43 samcv in places
13:43 timotimo yeah, <ws> can do a bunch more, though
13:43 samcv oh?
13:43 samcv yeah i see what you mean
13:43 samcv what about the other ones?
13:43 timotimo it's rather complicated
13:43 samcv yeah whitespace is complicated, but like <alpha> is fine to change to <:alpha>?
13:43 timotimo could be
13:44 samcv i know you can redefine whitespace
13:44 * nine is boarding the plane now
13:44 timotimo safe flight!
13:44 * [Coke] notes that brokenchicken is actually a leading Perl 6 developer, whether he wants to be or not. :)
13:44 timotimo gotta make sure that <:alpha> also plays well with the NFA generator, too
13:45 samcv why would it not?
13:45 timotimo dunno :)
13:47 samcv hmm changing <!alpha> to <-:alpha - [»]> gets me more problems
13:47 samcv Preceding context expects a term, but found infix . instead
13:47 samcv when trying to do .»
13:47 timotimo grammar can really be hard
13:48 jnthn Is it <:!alpha> maybe?
13:48 samcv jvm: say 'a' ~~ /<:alpha>/
13:48 samcv jnthn, it works fine otherwise
13:48 samcv i was just trying to fix .».
13:48 samcv so everything is fine that worked before
13:49 jnthn phone meeting; bbiab
13:49 samcv j: say 'a' ~~ /<:alpha>/
13:49 camelia rakudo-jvm fb4f16: OUTPUT«「a」␤»
13:49 samcv cool
13:49 samcv j: for ^1000 {'a' ~~ /<:alpha>/}; say now - INIT now;
13:49 camelia rakudo-jvm fb4f16: OUTPUT«0.185␤»
13:49 samcv j: for ^1000 {'a' ~~ /<alpha>/}; say now - INIT now;
13:49 camelia rakudo-jvm fb4f16: OUTPUT«0.402␤»
13:50 samcv j: for ^2000 {'a' ~~ /<alpha>/}; say now - INIT now;
13:50 camelia rakudo-jvm fb4f16: OUTPUT«0.932␤»
13:50 samcv j: for ^2000 {'a' ~~ /<:alpha>/}; say now - INIT now;
13:50 camelia rakudo-jvm fb4f16: OUTPUT«0.366␤»
13:52 samcv m: for ^2000 {'a' ~~ /<:alpha>/}; say now - INIT now;
13:52 camelia rakudo-moar f24968: OUTPUT«0.04570131␤»
13:52 samcv m: for ^10000 {'a' ~~ /<:alpha>/}; say now - INIT now;
13:52 camelia rakudo-moar f24968: OUTPUT«0.1672124␤»
13:53 samcv m: for ^10000 {'a' ~~ /<alpha>/}; say now - INIT now;
13:53 camelia rakudo-moar f24968: OUTPUT«0.2700231␤»
13:53 samcv yeah about 2x faster on MVM and like 220% faster on jvm
14:14 samcv i gotta go to bed, night all
14:14 brokenchicken night
14:43 brokenchicken Can someone merge https://github.com/MoarVM/MoarVM/pull/518 please? :)
14:43 brrt i can look at it
14:44 brokenchicken brrt++
14:44 brrt what. how was that in there for so long
14:44 brokenchicken The Travis failures are due to qast test stuff we talked about this morning.
14:46 brrt expected as much
14:47 brokenchicken And I stresstested it yesterday and also ran in loop to see there are no issues in jittings: https://irclog.perlgeek.de/moarvm/2017-02-02#i_14032981
14:49 travis-ci joined #moarvm
14:49 travis-ci MoarVM build errored. Samantha McVey 'Apply the radix speed improvement to MVM_bigint_radix as well'
14:49 travis-ci https://travis-ci.org/MoarVM/MoarVM/builds/197980028 https://github.com/MoarVM/MoarVM/compare/66961d88a803...f7ac91ccb25e
14:49 travis-ci left #moarvm
14:50 Geth ¦ MoarVM: 1c1746e525 | (Zoffix Znet)++ | src/core/interp.c
14:50 Geth ¦ MoarVM: Fix overflow in div_i op
14:50 Geth ¦ MoarVM:
14:50 Geth ¦ MoarVM: The op uses `int` which overflows on large numbers, making the op
14:50 Geth ¦ MoarVM: produce wrong results.
14:50 Geth ¦ MoarVM:
14:50 Geth ¦ MoarVM: Fix by using MVMint64 instead.
14:50 Geth ¦ MoarVM:
14:50 Geth ¦ MoarVM: Fixes RT#130686: https://rt.perl.org/Ticket/Display.html?id=130686
14:50 synopsebot6 Link:  https://rt.perl.org/rt3//Public/Bug/Display.html?id=130686
14:50 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/1c1746e525
14:50 Geth ¦ MoarVM: 9258acaffd | (Jonathan Worthington)++ | src/core/interp.c
14:50 Geth ¦ MoarVM: Merge pull request #518 from zoffixznet/fix_div_i
14:50 Geth ¦ MoarVM:
14:50 Geth ¦ MoarVM: Fix overflow in div_i op
14:50 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/9258acaffd
14:50 brokenchicken \o/
14:51 * jnthn wonders what's up with that qast.t
14:51 brokenchicken It was already fixed this morning.
14:51 jnthn ah, cool
14:54 jnthn Only just got around to backlogging #perl6-dev :)
15:00 pyrimidine joined #moarvm
15:00 agentzh joined #moarvm
15:25 dogbert17_ o/ has anyone seen ever this message? moar: 3rdparty/libuv/src/unix/stream.c:1499: uv_read_start: Assertion `((stream)->io_watcher.fd) >= 0' failed.
15:25 dogbert17_ s/seen ever/ever seen/
15:26 dogbert17_ when running 'make spectest HARNESS_TYPE=6'
15:29 domidumont joined #moarvm
15:32 jnthn No
15:33 jnthn Well, somebody may have, I ain't
15:42 dogbert17_ I did have moar built with --asan btw
15:55 geekosaur joined #moarvm
16:08 brrt joined #moarvm
16:47 domidumont joined #moarvm
16:59 brrt joined #moarvm
17:56 konobi joined #moarvm
17:56 [Coke] joined #moarvm
17:58 harrow joined #moarvm
17:59 Ven joined #moarvm
18:01 arnsholt joined #moarvm
18:23 pyrimidine joined #moarvm
18:40 agentzh joined #moarvm
18:57 zakharyas joined #moarvm
19:26 zakharyas1 joined #moarvm
20:19 FROGGS joined #moarvm
20:23 travis-ci joined #moarvm
20:23 travis-ci MoarVM build passed. Jonathan Worthington 'Merge pull request #518 from zoffixznet/fix_div_i
20:23 travis-ci https://travis-ci.org/MoarVM/MoarVM/builds/198026127 https://github.com/MoarVM/MoarVM/compare/f7ac91ccb25e...9258acaffd43
20:23 travis-ci left #moarvm
20:41 Ven joined #moarvm
21:08 pyrimidine joined #moarvm
21:11 lizmat joined #moarvm
22:20 timotimo so, how exactly is expmod with a negative exponent going to work?
22:21 timotimo like, res = a ^(-1) mod c ...
22:21 timotimo is it even supposed to ever return anything non-integer?
22:22 timotimo because ^-1 is always just 1/a, and since it already coerces a and c to int ...
22:22 timotimo anyway, i think we just basically pass the values 1:1 into the libtommath routine
22:24 timotimo oh, okay, seems like the tommath manual suggests that negative values for the exponent are fine
22:25 geekosaur joined #moarvm
22:26 timotimo For negative values of X the operation is performed as
22:26 timotimo Y ≡ (G−1 mod P)
22:26 timotimo |X|
22:26 timotimo (mod P) provided that gcd(G, P) = 1
22:26 timotimo the |X| was supposed to be an exponentiation of the (G-1 mod P)
22:27 brokenchicken isn't mod an integer operation?
22:28 brokenchicken m: .5 mod 3
22:28 camelia rakudo-moar c98b3a: OUTPUT«WARNINGS for <tmp>:␤Useless use of "mod" in expression ".5 mod 3" in sink context (line 1)␤Cannot resolve caller infix:<div>(Rat, Int); none of these signatures match:␤    (Int:D \a, Int:D \b)␤    (int $a, int $b --> int)␤  in block <unit> at <tm…»
22:30 brokenchicken timotimo: so you're looking at the expmod ticket?
22:32 timotimo just a very cursory glance
22:32 timotimo gotta be AFK for a little bit now
22:32 timotimo but the idea would be: update libtommath and see if that fixes the problem
22:35 brokenchicken Ah. Probably too deep down for my skills :)
22:38 pyrimidine joined #moarvm
22:54 timotimo doesn't seem like there's been any change to that since 2015
22:58 timotimo it doesn't help that we don't seem to have a trivial gdb-callable thing that just outputs the value of an mp_int struct
22:58 timotimo that might be a good thing to have so we can figure out What The Hell Is Going On
22:59 samcv i've almost got this nailed down, UCD-gen.p6 if/else statements to save space for the cp to bitfield index
22:59 pyrimidine joined #moarvm
23:00 timotimo brokenchicken: i'd be interested to see where you'd take the "MVM crab" logo idea
23:01 brokenchicken ok :)
23:03 brokenchicken But despite my job title, don't expect incredible things :P
23:04 timotimo as long as it doesn't come out looking like richard dawkins, that should be fine
23:04 brokenchicken hehe :)
23:14 timotimo anyway
23:14 timotimo we found an issue in libtommath that also hits us there
23:14 timotimo and there's already a fix suggested there that prevents our infinite loop
23:26 samcv sometimes i get really amazed when my code works
23:26 samcv if this is right, the UCD database which has most things is 104K which is pretty good
23:27 samcv have a huge tower of if statements though :P
23:28 samcv yeah 107K is really tiny
23:29 samcv look at the minimap :P https://i.imgur.com/CE1yJwF.png
23:29 samcv will have to tweak it. atm it adds an if/else for every range of cp that has >10 cp's with the same values adjacent
23:29 timotimo amazing
23:29 samcv but 107K!!
23:29 timotimo reminds me of the highway that a langton's ant would build
23:30 brokenchicken :o
23:30 samcv from line 20,000 to 70,000
23:30 samcv though a lot of those are brackets and stuff
23:31 samcv it's sorta ridiculous. but it only takes up 107KB :P
23:32 timotimo let's hope that clang doesn't take 10 minutes to compile it
23:32 samcv naw
23:32 samcv it compiles really fast
23:32 samcv like 2 seconds
23:33 timotimo k
23:33 samcv hmm for some reason i bulit for 100 cp and it was off by one, so i subtracted one. but now i do a full build and i get negative index's in the table. hopefully if i remove the adding one and do another full build it won't be off
23:34 samcv err may be another problem since it did print out the numeric value of numbers 0 throug h9 properly by unicode property
23:43 samcv i noticed before i'd get weird things happening if i only generated up to like 100cp or something though

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary