Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2015-06-23

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

All times shown according to UTC.

Time Nick Message
00:45 vendethiel joined #moarvm
01:15 vendethiel joined #moarvm
01:40 vendethiel joined #moarvm
02:11 TEttinger joined #moarvm
02:15 vendethiel joined #moarvm
02:25 TEttinger joined #moarvm
03:02 pyrimidi_ joined #moarvm
06:40 colomon joined #moarvm
06:40 FROGGS joined #moarvm
06:57 FROGGS nebuchadnezzar: is it possible that moarvm@kfreebsd lacks the dependency on libkvm?
06:59 zakharyas joined #moarvm
07:03 nebuchadnezzar FROGGS: maybe
07:03 nebuchadnezzar FROGGS: the dependency are really lite
07:05 nebuchadnezzar FROGGS: bingo https://paste.debian.net/255187/
07:06 FROGGS nebuchadnezzar: can you adjust that for the 2015-06 release?
07:06 nebuchadnezzar sure
07:06 FROGGS it is the only dependency where have there I believe
07:07 FROGGS we also need librt and libdl, but that's standard, right?
07:07 FROGGS I mean, it is a dependency on linux too, and it works on linux
07:07 nebuchadnezzar FROGGS: if not it would have failed on other arches
07:08 FROGGS yeah, I think so too
07:09 FROGGS Build-Depends should probably contain libkvm-dev then
07:11 FROGGS ahh, it is libkvm0
07:11 nebuchadnezzar FROGGS: not a problem, I do not have access to my SSH keys from work, so I can't do it now
07:12 FROGGS yes, sure
07:12 FROGGS but the packages are called: libkavm0 and libkvm-dev
07:13 FROGGS I always struggle to find the right packages
07:13 nebuchadnezzar FROGGS: Build-Depends on libkvm-dev I think https://packages.debian.org/unstable/libkvm-dev
07:13 FROGGS aye
07:24 nebuchadnezzar thanks a lot, I'm planning to setup a kfreebsd VM on my OpenNebula to test before pushing it
07:33 FROGGS nebuchadnezzar: is it easy to set up a debian package build bot?
07:59 brrt joined #moarvm
08:00 brrt so far, it works
08:01 nebuchadnezzar FROGGS: I'll just install a VM, clone and dpkg-buildpackage ;-)
08:12 brrt ok, bugfix, but many success
08:13 brrt https://github.com/MoarVM/dynasm/tree/dasm-fix <- can now use full GPR addressing on x64
08:14 brrt that tree is not suitable for merging with luajit, by the way
08:15 * brrt can stop dreaming of bits, again
08:16 brrt if we'd be doing a hackathon this would be a good moment to get drinks
08:19 FROGGS joined #moarvm
08:57 dalek MoarVM/even-moar-jit: 5a93fdf | brrt++ | 3rdparty/dynasm:
08:57 dalek MoarVM/even-moar-jit: Add new DynASM version as submodule
08:57 dalek MoarVM/even-moar-jit: review: https://github.com/MoarVM/MoarVM/commit/5a93fdfc48
09:01 jnthn morning o/
09:01 JimmyZ \o
09:01 jnthn brrt++ # dynasm hacking
09:01 brrt :-)
09:02 brrt good morning
09:02 brrt on the road already, or still in sweden?
09:02 JimmyZ it still is nice we choose DynASM :P
09:03 masak brrt++ # nananananananananananananananana dynasm!
09:03 brrt hey, there may still be bugs though :-)
09:04 brrt yes, actually, the alternative would've been knowing *every* bit of x86 instruction encoding, and now i only need to know 80% or so :-P
09:04 brrt that's an overstatement too
09:07 jnthn brrt: Still in Sweden, leaving on Friday.
09:08 brrt rather exciting, an international move
09:08 brrt :-)
09:09 jnthn :-)
09:10 jnthn Will be nice to populate the new place with some more of my stuffs :)
09:10 nwc10 yay
09:10 jnthn Not least my comfortable keyboard and nice dev box and big screen :)
09:10 nwc10 https://ln.wikipedia.org/wiki/Shekia
09:12 jnthn brrt: Quite reassured that the patches, though they look plenty of work, came out nice and small :)
09:13 nwc10 that's a good thing. Sort of related to "It costs a lot of money to look this cheap" and "I would have written a shorter letter, but I did not have the time."
09:14 brrt hah, yes, i'm happy about that too
09:14 jnthn aye
09:15 brrt otoh, it's not quite ready for merging with luajit, and it's also not complete (as in, not all cases of addressing are done)
09:15 brrt but the most important one for our use case is
09:15 nwc10 https://tpi.wikipedia.org/wiki/Sek_Ripablik
09:15 nwc10 https://kaa.wikipedia.org/wiki/Shexiya
09:16 brrt what languages are these
09:16 nwc10 er, I'm not sure
09:16 nwc10 https://ss.wikipedia.org/wiki/Shekhi
09:17 nwc10 but it seems, if you try hard enough, almost any country can start with an S
09:17 jnthn What on earth language is ss...
09:18 jnthn ooh, Swazi
09:18 jnthn Yet another one I know nothing about
09:19 jnthn "iRiphablikhi yeShekhi"
09:19 jnthn That's some curious capitlaization...
09:20 jnthn ooh yay, boxman will show up in 30 mins.
09:21 jnthn This is especially good 'cus I failed to have the foresight to go and buy something last night for today's lunch...
09:22 nwc10 boxman collects boxes?
09:23 nwc10 or boxman delivers more boxes?
09:23 brrt http://www.strchr.com/x86_machine_code_statistics <- is interesting and recognisable
09:23 jnthn Collects :)
09:24 nwc10 and then your all alone with the things that belong to the flat?
09:24 nwc10 and what you're travelling with
09:28 nwc10 brrt: it reminds me of what I noticed about ARM object files - hand written assembler looks very different from compiler generated assembler. The compiler seems to spend a lot of its time doing loads and stores
09:28 nwc10 the human has figured out how to get all the working data into the available registers, and codes accordingly
09:28 * brrt nods
09:28 nwc10 (or, I guess, didn't write it in assembler)
09:29 brrt well, in case of the moar-jit, a lot of time is also spent on maintainig VM invariants
09:29 brrt like, 'this structure always points to this frame'
09:30 brrt and of course that's a lot of what i'm trying to achieve :-)
09:49 jnthn phew, done
09:51 jnthn Looking at more examples of .ss, it seems the lowercase letters at the start of words are some kind of grammatical "marker".
09:56 masak jnthn: I was trying to find out, and tell you. :) but that sounds reasonable.
09:58 jnthn "Live weYurophu." is another example
09:58 jnthn Yurophu is clearly "Europe"
10:02 masak maybe they just add on whatever sounds nice. :)
10:11 brrt lunch &
10:18 Ven joined #moarvm
10:25 nwc10 jnthn: gosh, important question I'd not realised - with everything boxed up and in transit, how are you making coffee for the next few days?
10:35 jnthn nwc10: That is a rather disturbing problem...
10:36 Ven joined #moarvm
11:08 Ven joined #moarvm
11:21 jnthn Phew, enough packing for now. I'll have lunch, and then spend the afternoon hacking on Moar/Perl 6 things. :)
11:26 FROGGS joined #moarvm
11:37 FROGGS brrt++
11:41 FROGGS jnthn: weYurophu is equivalent how you'd create such words in klingon
11:42 FROGGS though, things like 'I do' or 'you are' are syllables that can be in front or after a noun, but attached to it
11:46 FROGGS like:
11:46 FROGGS I don't understand  -  jIyajbe'
11:46 FROGGS I don't know  -  jISovbe'
11:46 FROGGS I ...mumble... not
11:47 FROGGS (I understand  -  jIyaj)
12:03 * masak .oO( kiu elspezas tiom da tempo lerni tute artefaritan lingvon? ) :P
12:05 TEttinger jnthn: the aeropress makes good coffee and doesn't take any effort to pack
12:05 TEttinger I don't know if they sell them in local stores, I bought two on amazon for about $20 each
12:06 TEttinger also, thanks for your help, everyone, I now can compile and run the very basics of a lisp :D
12:07 jnthn \o/
12:07 jnthn .oO( What is the Czech Republic called in Klingon? :) )
12:08 TEttinger ... estYurophu
12:09 TEttinger MIGHTY KLINGON WARRIORS CARE NOT FOR LOCAL BORDER DISTINCTIONS
12:14 tadzik I didn't know that I know more Klingon speakers than Esperanto speakers
12:14 tadzik but having thought about it for a moment, I probably shouldn't be surprised
12:14 FROGGS tadzik: you are just visiting the wrong channels :o)
12:16 tadzik you spelled "right" wrong :P
12:16 * jnthn gets himself latest builds of all the things
12:26 masak tadzik: I think I know a handful of Klingon speakers, and somewhere between 500 and 1000 Esperanto speakers.
12:28 TEttinger I know worf
12:28 brrt joined #moarvm
12:28 lizmat jnthn: re latest stuff, did we consider upgrading unicode to 8.0 yet ?
12:29 brrt Q'apla!
12:29 brrt anyway, jnthn, i'm probably an utter heretic
12:30 brrt but i quite enjoy an occasional cup of powdered espresso coffee
12:32 nwc10 lizmat: yes, I blieve jnthn has been thinking about it
12:33 TEttinger brrt: I don't even drink coffee, I take no-doz as pills (less acidic in my stomach)
12:33 lizmat if it would be something "mechanical" like importing some data files and running a few scripts, maybe someone else can do that  :-)
12:33 nwc10 It never seems to be quite that simple
12:33 brrt no-doz?
12:33 lizmat if it is described somewhere, ther is...
12:33 lizmat *that
12:34 TEttinger brrt: caffeine in pill form
12:34 brrt oh, that's horrible
12:34 TEttinger I like coffee, it just messes up my stomach
12:34 brrt no offense meant, but caffeine is a stronger drug than many people realize
12:35 TEttinger I take less than most people get from energy drinks
12:35 TEttinger monster energy messes me up
12:35 brrt monster energy messes *everyone* up :-)
12:36 tadzik brrt: doesn't that make coffee just as bad?
12:36 brrt but yeah, i'm not encouraging energy drinks either
12:36 tadzik (I'm not saying that it isn't :))
12:36 brrt well, yes, it's one of these moderation things; and also the pleasure/effects tradeoff i guess? for me at least
12:37 brrt coffee actually tastes nice, i mean to say
12:37 tadzik I see
12:37 brrt but using caffeine as an all-purpose booster is not a very good idea, and not very effective either
12:37 jnthn lizmat: It's not quite mechanical in that somebody has to read the "what chnaged" summary of the spec and know if it's anything more than updating the database.
12:37 jnthn lizmat: I did that part a few days ago.
12:37 tadzik I don't really like the taste of coffee, but I *love* how it smells after grinding
12:38 jnthn And it probably is just updating the DB
12:38 tadzik maybe I should grind it, smell it, throw it away and take a pill instead
12:38 jnthn I guess I may as well do it now while it's on my mind :)
12:38 TEttinger brrt: oh I agree.  that's why I take vyvanse (prescription slow-acting amphetamine)
12:38 lizmat jnthn++  :-)
12:38 brrt lol TEttinger
12:39 jnthn grmbl, I'm clogged up today
12:39 jnthn I need meaner anti-histamine or something...
12:39 TEttinger vyvanse doesn't help energy, it does make my focus basically superhuman
12:39 brrt the company of computer programmers is the company of drug users :-D
12:40 TEttinger well this is interesting: https://en.wikipedia.org/wiki/Amphetamine#Enhancing_performance
12:42 TEttinger at low doses, things like vyvanse (which are not the same as crystal meth, but similar if you compare 100x different doses) cause unambiguous cognitive performance improvements.  as in, no nervousness or panicky feelings like caffeine also causes at too high doses
12:43 psch interesting that the study is only from this year
12:43 psch oh, it's the review and meta-analysis that's from this year, i misread
12:43 TEttinger yeah. it's difficult to study currently because it amphetamines are prescription medicines now, but more highly regulated
12:44 TEttinger not as hard as studies on marijuana (it's fascinating how some epileptic people benefit from it even if the strain has no recreational effects)
12:45 jnthn $ perl6-m -e "say ord qq[\c[SIGN OF THE HORNS]]"
12:45 jnthn 129304
12:45 TEttinger hahaha
12:45 jnthn Well, thta Unicode 8 update wasn't so bad
12:45 TEttinger is that the metal fingers?
12:45 jnthn Let's see how the test suite likes it
12:45 jnthn TEttinger: yes!
12:45 psch awesome
12:45 jnthn Unicode is now officially METAL
12:45 TEttinger jnthn++
12:45 jnthn I guess I should check we can actually build NQP and Rakudo as well as run the test suite. :)
12:46 TEttinger naaaah
12:46 TEttinger metal is justification enough
12:46 brrt epilepsy is weird
12:46 TEttinger brrt, yeah one of my best friends in high school had rough tics
12:46 brrt one of my rats has it
12:47 brrt one would not necessarly expect amarijuana to have a beneficial effect though
12:47 TEttinger I like rats.  rats apparently like my birdfeeder.  cats like anything on he birdfeeder.
12:48 TEttinger last time I touched a rat, it was inside the birdfeeder, getting every last morsel, and I gave it a scritch scratch on its incredibly soft back
12:48 TEttinger then washed my hands :)
12:50 TEttinger it's interesting how opossums are much uglier creatures than rats, pretty much without doubt, because the opossum uses that as a defense mechanism (it needs to look and smell dead)
12:50 TEttinger but they both have the hairless gripping tail
12:51 TEttinger drawing this back to perl, it makes you wonder about what makes a language visually appealing vs. actually effective
12:52 TEttinger (opossums are the only surviving marsupial in north america.  there used to be tons in prehistoric south america)
12:53 brrt aren't there anymore?
12:53 TEttinger there are a small amount of relatives of opossums in south america, I think. most are in australia
12:54 TEttinger there are possums that somehow got to new zealand, but they're non-native
12:54 brrt anyway, perl5 (and perl6) give you tons of clues on what something is and what you're doing it in a very short amount of space
12:54 TEttinger yep!
12:54 brrt imho that's hugely valuable and something which really contributes (for me) to the feeling that i'm 'speaking' the language
12:54 brrt rather than composing programs in it
12:55 jnthn Well, all the S15 stuff passes with Unicode 8, at least
12:57 brrt \o/ :-)
12:58 TEttinger in biological terms, perl is similar to a rat in terms of adaptability; flexible, able to adapt to any environment, uses relatively few resources -- where a language like Java, with a 1600-page standard library documentation printout, is more similar to an elephant; able to do a few things a rat would find difficult (jumping is not one of them), but occupying a tiny fraction of their pre-human territory
13:00 jnthn Darn it, my spectest repo was dirty so I did the test against outdated tests...
13:00 TEttinger test the tests
13:00 masak TEttinger: so you're saying that the arrival of humans made Java endangered? :P
13:00 * jnthn sets it off again...
13:00 brrt if only
13:01 jnthn Maybe the arrival of Java endangered humans...
13:01 TEttinger the closest equivalent to early hominids, who found elephants really good sources of spareribs and other tasty meat, in a programming field might be scripting languages in general
13:02 TEttinger monkeys and primates had been around for a while
13:02 brrt the biggest weakness of rats are their short lifes, offset by their high reproductive rate
13:04 TEttinger but early hominids developed a way to compensate for a relatively weak body, with tool use.  you could compare it to how JS had been around but unsuitable for writing lots in, then it became an everywhere platform that suddenly got a good JIT too...
13:05 TEttinger a less verbose language with a good JIT, versus a more verbose language with a good JIT, it's no wonder java's getting less usage
13:06 TEttinger (even on android, the android V8 JS engine outperforms their Dalvik Java runtime)
13:06 TEttinger (sometimes)
13:06 brrt that sometimes is kind of important :-)
13:06 TEttinger also, Dalvik is pretty awful
13:07 TEttinger they "creatively interpreted" what the static keyword does...
13:07 brrt 'optimised for a different scenario' is how we prefer to call it :-P
13:08 TEttinger static variables last longer than your application does, which is ok if you expect to shut down and start up in a perfectly optimal scenario.  but not if those static variables have native associated state, like an OpenGL context, in which case those variables suddenly refer to junk VRAM
13:08 TEttinger this has no equivalent on desktop java
13:09 TEttinger it was made up entirely by google and it's a pain.
13:10 masak TEttinger: that sounds weird and interesting. do you have a URL explaining that in more detail?
13:11 jnthn After the Unicode upgrade I get 3 new failures in S05-mass/properties-derived.t
13:11 TEttinger I'm not sure, I only know it from solving the recurring issues that people get from it in #libgdx here on freenode.  that's a java game lib that works on android and desktop (and others that aren't as often)
13:12 brrt jnthn: that could potentially be database disagreements?
13:13 jnthn I think it may be a bad test.
13:13 jnthn .u 0x9FCD
13:13 jnthn m: say uniname 0x9FCD
13:13 camelia rakudo-moar 7fbd54: OUTPUT«<CJK Ideograph>␤»
13:13 jnthn m: say uniprop 0x9FCD, 'Ideographic'
13:13 camelia rakudo-moar 7fbd54: OUTPUT«0␤»
13:13 masak wut
13:14 jnthn It's 1 in Unicode 8
13:15 masak aha.
13:15 TEttinger http://stackoverflow.com/a/1944564
13:15 TEttinger masak: ^
13:16 TEttinger m: '\u9FCD'
13:16 camelia rakudo-moar 7fbd54: ( no output )
13:16 TEttinger m: say '\u9FCD'
13:16 camelia rakudo-moar 7fbd54: OUTPUT«\u9FCD␤»
13:16 TEttinger what's the proper unicode escape?
13:16 jnthn In Unicode 8:
13:16 jnthn 4E00;<CJK Ideograph, First>;Lo;0;L;;;;;N;;;;;
13:16 jnthn 9FD5;<CJK Ideograph, Last>;Lo;0;L;;;;;N;;;;;
13:16 psch m: say "\u9FCD"
13:16 camelia rakudo-moar 7fbd54: OUTPUT«===SORRY!=== Error while compiling /tmp/laWQXSuY0V␤Unrecognized backslash sequence: '\u'␤at /tmp/laWQXSuY0V:1␤------> say "\⏏u9FCD"␤    expecting any of:␤        argument list␤        double quotes␤        term␤»
13:16 jnthn \x
13:17 psch m: say "\x9FCD"
13:17 camelia rakudo-moar 7fbd54: OUTPUT«鿍␤»
13:17 TEttinger m: say '\x9FCD'
13:17 psch yes...
13:17 camelia rakudo-moar 7fbd54: OUTPUT«\x9FCD␤»
13:17 TEttinger ah
13:17 TEttinger so it isn't in unifont
13:17 jnthn In Unicode 7:
13:17 jnthn 4E00;<CJK Ideograph, First>;Lo;0;L;;;;;N;;;;;
13:17 jnthn 9FCC;<CJK Ideograph, Last>;Lo;0;L;;;;;N;;;;;
13:18 jnthn So the test took one past the end of the valid range of ideographs
13:18 jnthn In Unicode *7*
13:18 TEttinger the numbers are different
13:18 jnthn But 8 adds more
13:18 TEttinger ok
13:18 jnthn So yeah, fragile test.
13:18 jnthn I suspect it should pick something that (a) exists, and (b) isn't ideographic :)
13:18 jnthn m: say uniprop 'A', 'Ideographic'
13:18 camelia rakudo-moar 7fbd54: OUTPUT«0␤»
13:18 jnthn Like that :)
13:19 jnthn If I just bump the number we'll probably have to play this game again for Unicode 9
13:19 TEttinger say uniprop "\x4F00", 'Ideographic'
13:19 TEttinger m: say uniprop "\x4F00", 'Ideographic'
13:19 camelia rakudo-moar 7fbd54: OUTPUT«1␤»
13:19 masak jnthn: +1
13:20 masak "don't write tests that pass only because something isn't the case yet" :)
13:20 jnthn Maybe the test author was thinking of the right bicep... :P
13:24 dalek MoarVM: 7575ca4 | jnthn++ | src/strings/unicode_ (2 files):
13:24 dalek MoarVM: Update to Unicode 8.
13:24 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/7575ca4bdf
13:33 AndChat|228864 joined #moarvm
14:21 njmurphy joined #moarvm
14:39 vendethiel joined #moarvm
15:04 jnthn Fixing RTs is a very good way to procrastinate on the VMArray refactor...
15:06 AndChat|228864 for NSA?😊
15:09 jnthn jnthn: In part, yeah
15:10 jnthn Though most immediately to deal with the fact that the current factoring of MVMArray isn't memory safe in the face of multi-threaded abuse.
15:10 jnthn It doesn't have to give even slightly sensible answer under such conditions, but it should never SEGV.
15:11 * JimmyZ_ nods
15:19 Ven_ joined #moarvm
15:34 timotimo jnthn: do you want me to make the code-gen work for localref/local sound more interesting?
15:35 timotimo you'll probably have to dive deep into the negative space between operations and find out what part of the huge machine composed of many cogs of different sizes to figure out where a "decont" materializes out of nowhere
15:35 timotimo ... that sentence didn't grammar
15:36 * timotimo goes for takes his grammar pill
15:36 jnthn timotimo: Heh, that task does get more exciting as it'll combine nicely with brrt++'s JIT work :)
15:42 TEttinger perl 6: now 98% made without the use of thaumaturgy!
15:43 TEttinger * blood sacrifices to aztec gods don't count
15:44 TEttinger interesting thing to note: the aztecs believed that literally all of their gods were dead, which explained perfectly reasonably why those gods could be in charge of the afterlife
15:48 TEttinger and with multiple preceding worlds, the gods of the mortal world had themselves worshipped dead gods in a preceding version of reality, and I don't know what the view was on whether a world and an afterlife ago was a thing.  it makes you wonder if semantic versioning is much older than we thought :P
15:52 TEttinger CHANGELOG FOR WORLD 5.0.0: Breaking Changes are many, first you need to refer to Tlaloc and any of his children in the cosmos::afterlife package, and the cosmos::material package is... hang on, someone committed all the antimatter out of cosmos::material. DO NOT USE 5.0.0, use 5.1.0.
16:05 timotimo about the x86 machine code statistics article: i would have liked to see the instructions multiplied by their length and then graphed as a pie
16:18 TimToady joined #moarvm
16:22 Ven joined #moarvm
16:48 dalek MoarVM: a982f1b | jnthn++ | src/core/interp.c:
16:48 dalek MoarVM: Fix SEGVs that can occur with code object ops.
16:48 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/a982f1b269
16:52 JimmyZ_ jnthn: IS_CONCRETE should be first since REPR(o) may be NULL, iirc
16:52 jnthn IS_CONCRETE isn't a nullness check
16:53 timotimo right, we have MVM_is_null for that
16:53 jnthn Even a VMNull has a REPR
16:53 timotimo yup
16:53 timotimo but is_null also checks for a null pointer
16:53 timotimo BBIAB
16:53 JimmyZ_ someone said it should be first two years ago.😊
16:54 JimmyZ_ before vmnull though
17:00 * JimmyZ_ sleeps
17:09 dalek MoarVM: ef9d045 | jnthn++ | src/core/frame.c:
17:09 dalek MoarVM: Fix missing concreteness checks in invocation.
17:09 dalek MoarVM:
17:09 dalek MoarVM: These were to blame for Sub(0) in Rakudo leading to a SEGV.
17:09 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/ef9d045538
17:17 nebuchadnezzar héhé, Debian kfreebsd installed
17:19 jnthn .tell brrt
17:19 jnthn oops
17:19 jnthn .tell brrt in the blog "To make manners more difficult" ==> matters
17:20 jnthn brrt++ # blogging
17:47 Ven joined #moarvm
18:43 FROGGS joined #moarvm
21:03 dalek MoarVM: 6dc0f3f | (Steve Mynott)++ | docs/moar.pod:
21:03 dalek MoarVM: correct typo
21:03 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/6dc0f3fc90
21:03 dalek MoarVM: 2681ac2 | jnthn++ | docs/moar.pod:
21:03 dalek MoarVM: Merge pull request #225 from stmuk/stmuk
21:03 dalek MoarVM:
21:03 dalek MoarVM: correct typo
21:03 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/2681ac2481
21:16 FROGGS_ joined #moarvm
23:17 TEttinger joined #moarvm

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