»ö« Welcome to Perl 6! | perl6.org/ | evalbot usage: 'perl6: say 3;' or rakudo:, niecza:, std:, or /msg p6eval perl6: ... | irclog: irc.perl6.org/ | UTF-8 is our friend!
Set by sorear on 4 February 2011.
00:00 raiph joined 00:01 PZt left, snearch left, drbean joined 00:07 icwiener joined 00:12 drbean left
ggoebel \o 00:13
Happy New Year's
00:16 packetknife joined 00:17 drbean joined
ggoebel sorear: regarding niecza and publicity... never underestimate the seemingly little things. A simple yet clever visually appealing logo. A pithy quotable value statement, website, etc. 00:22
00:22 icwiener left
ggoebel a simple memorable searchable name 00:23
00:23 icwiener joined
ggoebel my greatest fear for niecza is you getting hit by a bus or losing interest 00:23
tadzik I don't see how niecza is not simple or memorable :)
but again, I'm Polish and I have no problem with Brzęczyszczykiewicz either :)
ggoebel that might be it... I always hear Nietsche when I type it... and usually flip the vowels and consonants 00:24
tadzik it's almost like Nietsche, but the 'e' is audible, like the 'e' in Perl 00:27
cognominal nom: my %h; say "Brzęczyszczykiewicz".split('').map: { %h{$_}++ }; say %h 00:49
p6eval nom a7fd89: OUTPUT«0 0 0 0 0 1 0 0 2 1 3 1 0 0 0 0 1 2 4␤("B" => 1, "r" => 1, "z" => 5, "ę" => 1, "c" => 3, "y" => 2, "s" => 1, "k" => 1, "i" => 2, "e" => 1, "w" => 1).hash␤»
cognominal nom: my %h; "Brzęczyszczykiewicz".split('').map: { %h{$_}++ }; say %h
p6eval nom a7fd89: OUTPUT«().hash␤»
cognominal seems that some active code is optimized out :(
or I need sleep 00:50
Timbus peers in
Timbus slaps cognominal's hand for doing that
cognominal ?? 00:51
Timbus nom: my %h; %h{$_}++ for "Brzęczyszczykiewicz".split(''); say %h
p6eval nom a7fd89: OUTPUT«("B" => 1, "r" => 1, "z" => 5, "ę" => 1, "c" => 3, "y" => 2, "s" => 1, "k" => 1, "i" => 2, "e" => 1, "w" => 1).hash␤»
cognominal I want to use map for the mere side effects. 00:52
at least to get a warming. 00:53
Timbus maybe a /for method on lists is not as crazy as i thought
.for
tadzik split('') => comb 00:54
00:56 PZt joined
cognominal "Civilization advances by extending the number of important operations which we can perform without thinking of them". But programmers, when using a new language, advance by rethinking how to write important operations instead of using muscle memory. 00:56
sorear some people think that all list generating functions should be eager when the result is ignored 00:57
I do not like this at all
raiph sorear: what are your reasons for wanting niecza publicity? 00:58
cognominal what would be the equivalent code in haskell? 00:59
sorear maybe use mapM_ 01:00
raiph: a lot of people outside the echo chamber think perl6 means parrot
cognominal I want Perl6 on V8 :) 01:01
sorear ehehehe... I doubt that will work well
raiph sorear: outside the echo chamber they think perl6 is a blue norwegian ;) 01:02
sorear Perl6 can't be efficiently simulated on foreign VMs
cognominal sorear: but that would mean Perl 6 on chrome (polychrome?) and world domination 01:04
ggoebel sorear: if the motivation behind niecza is "make it fast"... in layman's terms what are the hurdles which need to be overcome in the short, mid, and long terms? 01:05
dalek ast: cb88c13 | (Solomon Foster)++ | S02-types/anon_block.t:
Fudge for niecza.
 01:10
Timbus how much effort would it take to lift nqp from parrot and put it in llvm 01:11
dalek ecza: 260335a | (Solomon Foster)++ | t/spectest.data:
Turn on S02-types/anon_block.t.
Timbus gc and utility libraries aside
i mean just getting it to compile itself into llvm for now 01:12
you can always plug in a hans boehm i guess.. 01:13
cognominal Timbus: indeed I think .map as a degenerated form of the .grep.map idiom I use a lot. I use them both for side effecta. The former being your .for method. 01:14
Timbus im not even a functional programmer, but side effects like that dont sit right for me
01:16 drbean left
cognominal bad habits are hard to die 01:16
that's the perl 4 programmer in me\ 01:17
Timbus fortunately i missed that stage
colomon ah, perl 4! 01:20
01:22 drbean joined
colomon niecza: say Order::Increase 01:23
p6eval niecza v13-43-gd7bbde0: OUTPUT«Unhandled exception: Unable to resolve method postcircumfix:<( )> in class Any␤ at /tmp/0WFpfPLDi4 line 1 (mainline @ 2) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2613 (ANON @ 3) ␤ at /home/p6eval/niecza/lib/CORE.setting line 2614 (module-CORE @ 61…
cognominal my first immortalized perl post date from 1991 groups.google.com/forum/?hl=fr#!se...7fT_e8iOIJ 01:24
*dates
colomon woah, I didn't learn perl until late 1994
cognominal I note I did not know all the magic switches to gobble stdin. 01:26
colomon I still don't. ;) 01:27
01:28 raiph left
tadzik you mean the -plne stuff? 01:29
cognominal I have lost the mails dating from the time where I was [email@hidden.address] [email@hidden.address] was quite a mouthful.
tadzik: yup 01:30
tadzik heh, I ususally put them there at random until it starts to work
01:35 bluescreen10 left
[Coke] cognominal: you at Oracle now? 01:39
cognominal nope, I have quitted Sun a long time ago... 01:43
geekosaur wonders when mono-sgen will be fixed in macports (long delay then bus error; mono itself works) 01:47
01:51 cspencer left 01:53 cspencer joined
dalek ast: 9ee6b08 | (Solomon Foster)++ | S12-class/attributes.t:
Fudge for niecza. Please note that IMO the last set of tests is at best misguided...
 01:56
ecza: cc39754 | (Solomon Foster)++ | t/spectest.data:
Turn on S12-class/attributes.t.
 01:57
sorear Timbus: look at the "6model" repo... nqp already has ports to luajit, java, and c# 02:03
Timbus yeah but th- wait luajit 02:04
02:04 Chillance left
Timbus .. yeah but they're not quite as low level as llvm 02:04
sorear is younger than Perl itself
02:05 Chillance joined
Timbus also do these ports compile and run rakudo? 02:05
tadzik I've not heard of any attempts 02:06
[Coke] t/spec/S16-filehandles/filetest.t has failures in niecza. 02:07
I just added that file to spectest.data on 12/24 02:08
colomon [Coke]: works locally for me.
sorear Timbus: no.
Timbus wouldn't have thought so 02:09
colomon [Coke]: do you have the very latest niecza? 02:11
TimToady perl6: say +(1 .. '10') 02:24
p6eval niecza v13-43-gd7bbde0: OUTPUT«0␤»
..pugs b927740, rakudo a7fd89: OUTPUT«10␤»
geekosaur *blink* 02:27
Timbus i.. okay 02:29
colomon TimToady: was there ever an official ruling on what to do with that case? 02:30
TimToady S04:3414 covers that 02:32
(in the context of numeric arg on the left)
colomon S03:3414, you mean 02:33
TimToady er, yeah, sorry
colomon no worries, TimToady++ 02:34
Is the effect intended to be, if the left-hand-side is Numeric and the right-hand-side is not, you cast the right-hand-side to Numeric? 02:35
TimToady yes
colomon so 0 .. '1.5' is 0 .. 1.5 and not 0 .. 1
TimToady it should be like prefix:<+> 02:36
colomon I ask because the text says "the right argument is coerced to the type of the left argument"
TimToady hmm
that's probably a specbug 02:37
colomon so should it be something like: 02:38
1) both numeric, take as is
2) first numeric, prefix:<+> the second
3) cast the right argument to the type of the left argument
(that's a sequence of whens, not a list of instructions. :) 02:39
dalek ecs: 8fcd8d6 | larry++ | S03-operators.pod:
spec 0 .. '1.5' better, colomon++
 02:40
colomon TimToady++ 02:41
TimToady 3 is not correct, it's "cast narrower to wider"
cover by "as above" in the spec
colomon gotcha 02:42
colomon is now hacking niecza 02:43
er, colomon is now helping his wife unload an ungodly quantity of groceries 02:49
02:50 tokuhirom joined 02:51 cspencer left 02:57 icwiener_ joined
colomon ..... not that I need a definition today, but how do you define "wider"? 02:59
03:01 icwiener left
colomon for instance, assume you've got a Rat and a Num. Which is wider? 03:02
niecza> say +(1 .. '10')
10
sorear num is "wider" 03:04
03:04 drbean left
sorear (1/2) + 1.5e0 ~~ Num 03:04
colomon ........... ! 03:05
\
guess that makes a sort of sense
afk moment'
back 03:08
03:10 drbean joined 03:13 am0c left 03:14 lutok left
dalek ecza: ab2db52 | sorear++ | lib/ (2 files):
Implement prefix:<sleep>
 03:24
Timbus i.imgur.com/SlD8L.png 03:36
:I thanks visual studio
03:37 Patterner left 03:38 c-lesh joined
c-lesh hello 03:38
03:38 Psyche^ joined, Psyche^ is now known as Patterner
colomon c-lesh: o/ 03:38
c-lesh I was wondering if there is a way to compile Java into Parrot bytecode.
Timbus yes: by hand 03:39
:D
c-lesh No, I mean is there a translator for Parrot?
Timbus none I am aware of, 03:40
c-lesh Is it true that Parrot can translate some of its bytecode into native code? If so then would Perl 6 be good for writing games? 03:41
Timbus you'd be better off using niecza imo 03:42
c-lesh niecza compiles into native code? or Parrot bytecode?
Timbus to clr code 03:43
it's for mono/.net
which has a jit
c-lesh ah, so clr does translate some of its code to native
or all of it? 03:44
Timbus most of it
c-lesh wow!
Timbus but the other advantage is you can use .net libraries
c-lesh that's nice
Timbus whereas rakudo needs a bit of work in that area 03:45
c-lesh my understanding is that Parrot does not yet compile its bytecode to native, right?
Timbus correct, it used to have a jit
but it was removed
c-lesh why?
Timbus maintenance nightmare
c-lesh oh, I see 03:46
I'm just wondering how one would translate bytecode to native, say in c
Timbus it makes more sense for parrot to target something else that jit's better (llvm was the idea for a while)
well, you know what a jit is?
c-lesh yes 03:47
Timbus well.. that's how :P
c-lesh I'm not sure how to translate bytecode to native at runtime
there is an __emit__ macro in C?
jits are written in C, right? 03:48
Timbus they can be written in anything, really 03:49
c-lesh I'll have to look at some source code.
Timbus the difference between bytecode and native code is.. well, one is native bytecode
c-lesh I'm writing an interpreter myself. It compiles to bytecode but I would like to compile this to native.
Timbus oh i see
well, let me warn you ahead of time that it's a lot of effort 03:50
please consider an existing jit library. llvm, GNU lightning
c-lesh Thanks, I'll check that out! 03:51
Did Parrot use any of the jit libraries you mentioned or did it implement its own? 03:53
Timbus it was it's own 03:54
.. its
c-lesh So is there a chance that the Parrot community will use some of these libraries? 03:55
Timbus eventually, yes i would hope so.
it really depends on what happens to parrot. 03:56
03:56 icwiener_ left 03:57 icwiener_ joined
c-lesh What is Larry Wall doing these days? Does he still work on Perl? 03:57
colomon On Perl 6, sure. 03:58
c-lesh That's cool!
colomon He was on here about 70 minutes ago
c-lesh Wow! Too bad I missed him!
Ok, so I'm curious about the history of Perl. Does Perl 1.0 have a garbage collector? 03:59
geekosaur actually still here now, just being quiet and unobvious
as is his wont 04:00
geekosaur unfamiliar with perl < 3.0 but believes there's a reasonable history in the perl5 pods 04:01
04:01 kaleem joined 04:03 hundskatt left
Timbus huh c-lesh www.gnu.org/software/dotgnu/libjit-...ibjit.html check out this here. I hadn't seen it before but the tutorials are a helpful start 04:07
blog.fallingsnow.net/2008/05/23/sim...with-llvm/ and this. took me a while to dig that one up :/ 04:15
c-lesh Thanks!
sorear c-lesh: "unmaintainable" was only half of the storyy 04:17
c-lesh What was the other problem?
sorear c-lesh: it generated terrible code
c-lesh Not efficient and fast? 04:18
sorear it's extremely easy to write a naive native code compiler
04:18 kaleem left
sorear but naive native code is actually not much faster than bytecode 04:18
and is MUCH larger
c-lesh Really!!!?
How can that be? 04:19
sorear trade a lot of ram for a small amount of speed... not much win
c-lesh: performance of native code is ... complicated
c-lesh: say you have one bytecode instruction, which gets turned into 10 machine instructions by your naive compiler
c-lesh So, for example, I write a game in, say, Python (I imagine this language is bytecode interpreted.) It would probably run fast. 04:20
sorear a good C compiler could turn the switch-clause into 15 instructions, but because of better scheduling that runs faster
Timbus jits come into their own when they unbox and such. the actual loop construct of an interpreter isn't much overhead no
c-lesh I see. 04:21
So why are jits even used?
sorear c-lesh: there is no "fast", only "fast enough"
c-lesh: because a jit starts to win once you have about 10 man-years into it
Timbus well jits do add speed
compare javascript performance to 4 years ago 04:22
sorear c-lesh: you're oversimplifying, I feel
c-lesh So a game written in python that takes advantage of graphics hardware acceleration can be faster than a game written in C that does not take advantage of graphics hardware acceleration.
Timbus c-lesh, easily
sorear c-lesh: I want you to stop thinking about "this will be fast"
c-lesh I have seen a lot of javascript games!
Wow! This really cleared things up! Thanks! 04:23
sorear c-lesh: unless you're trying to make the next DF, calculation speed is not normally a big issue for games 04:24
c-lesh I understand branch prediction is good these days as an interpreter would have to compare its op codes.
especially for 2D games like Mario 04:25
sorear even for 3D games
c-lesh These are the games I'm writing for my interpreter
3D games?
Are 3D games possible in python?
sorear do you understand Amdahl's Law? 04:26
c-lesh not really
sorear if a program spends 90% of its time in module A and 10% of its time in module B, no amount of optimizing module B can ever give you more than a 10% speedup
this is Amdahl's Law
c-lesh ah 04:27
sorear I consider it common sense, but that's probably just because I've used it for so long
c-lesh I'm new to the world of virtual machines, so to speak.
Actually, I'm writing one to make games.
sorear so if you're writing a "typical" 3D game, 90% of the work is being done by the GPU, inside OpenGL
so optimizing your code won't help much 04:28
you can only make it faster by reducing the number of glXYZ calls
c-lesh So taking advantage of hardware acceleration is key - not code optimization.
sorear (why are you writing a VM?)
c-lesh I'm writing it for fun and because I want to learn about them. 04:29
sorear c-lesh: yes, but I have to stress that every case is different
when you have a good understanding of what's expensive and what's cheap, you'll naturally write faster code
experience matters, a lot
c-lesh My VM takes advantage of hardware acceleration and I have gotten my first game to run quite fast on my netbook. 04:30
I have been writing code since 1998.
C++, Java, C, Perl, Python, Java, etc.
sorear I'm sure you can access OpenGL from Python
c-lesh Sure.
Timbus panda3d and pygame come to mind 04:31
c-lesh I have used pygame before.
04:31 icwiener_ left
sorear probably longer than me. 04:32
c-lesh Python uses what kind of garbage collection? stop-the-world, etc.
sorear python uses "fake" kind of garbage collection
04:32 icwiener_ joined
c-lesh I started using python on 2008 but didn't like the indentation stuff. 04:32
fake?
sorear reference counting
fastish, but wastes memory and leaks like a sieve if you mess up cycle handling 04:33
c-lesh So that means that memory is only marked as available?
sorear perl 5 does the same thing
c-lesh Parrot?
sorear Parrot has a real GC
Timbus parrot uses a generational gc
c-lesh does this run whenever, like for example in a game loop? 04:34
Timbus yep
c-lesh is there a System.gc() like in Java?
Timbus couldn't answer that since I havent used parrot in awhile, but probably 04:35
sorear I would like to gently dissuade you from basing anything mission-critical on Parrot
c-lesh My VM has no garbage collection and the programmer accesses the memory directly like in assembly.
kshannon Yes.
c-lesh sorry
sorear e.g. last time I tried this, a simple sub call is 2 µs on Parrot and <50ns on Mono
s/mission/performance/
Timbus mono is pretty boss.
still mad android didnt use it 04:36
c-lesh It's like of like a cross between assembly and C. I think it's cool.
sorear what do you think of c--?
c-lesh I write stuff for Android.
c--?
Timbus www.cminusminus.org/ 04:37
sorear it's like a cross between assembly and C. I think it's cool.
although it's lost a lot of mindshare to LLVM IR 04:38
Timbus i just liked the name
c-lesh c-- looks cool but my VM is not assembly like. It has no registers and uses DMA (direct memory access). It encodes all expressions and structures in bytes and executes these. 04:39
sorear I do not think DMA means what you think it means
c-lesh what does it mean then?
sorear DMA refers to a situation where hardware *other than the CPU* can access memory 04:40
like say you have a network card that writes packets directly into RAM
geekosaur dma means devices can access memory directly insted of having the cpu poll and copy data
c-lesh I see. Like what the SNES does.
Does that mean that one does not have to transfer from memory to register to memory again? 04:41
Timbus in a roundabout way. yes?
sorear yes, but you'd be stupid not to 04:42
c-lesh why?
sorear because transfer from memory to register to memory again is a million times faster than using DMA
c-lesh ?
sorear and I might not be exagerating, it's at least tens of thousands of times faster
04:42 kaleem joined
c-lesh but the snes uses it because according to the manual it is faster to block data transfers 04:43
sorear that was 20 years ago
Timbus but my pc.. it is not a snes
that would be awesome
c-lesh so dma is outdated 04:44
sorear *MEGA HEAD DESK*
c-lesh i have looked at snes assembly code and architecture
sorear I give up. geekosaur, you're up
geekosaur not sure he can do any better
c-lesh these people hack Super Mario World all the time: www.smwcentral.net
04:45 kaleem left
geekosaur it depends on the system and bus design, plus even relatively slow DMA can be a net win if the system has enough other stuff to do while the DMA is occurring 04:46
c-lesh So in a typical VM why are registers even used for register based VMs. For an addition why not just use arbitrary memory locations? 1 instruction with 3 params is better than 2 instructions, right?
I guess snes used this sinces its hardware was memory mapped instead of port mapped. 04:47
geekosaur instruction decoding takes longer, memory fetch cycles may slow things down
in modern system, performance isn't so much a function of individual design choices as the performance of the entire system on the intended load 04:48
c-lesh i see
geekosaur and individually "slower" designs may net a faster overall system if they are used appropriately 04:49
Timbus the closer you get to the wire, the less sense it makes too. -_-
c-lesh interesting
Timbus simple jz loops often outperform repne even though it is a "single" instruction. why? pppfft 04:50
idk
c-lesh pppft?
Timbus thats my answer
geekosaur I think that's just onomatopoeia 04:51
c-lesh i'm not familiar with the instruction repne
Timbus repeat until not equal(?) its been so long
geekosaur it's complex. early on, certainly, jz loops often outperformed rep loops; now its more dpeendent on what you're doing
c-lesh I'm learning quite a bit. Thanks! 04:52
geekosaur x86 architecture has a number of "repeat" opcodes that act like automatic loops
c-lesh This would be the CISC architecture.
RISC is better.
geekosaur but, as they're designed to be general, they can be higher overhead than a well thought out explicit loop
c-lesh So VM try to mirror the RISC architecture. 04:53
sorear "RISC is better" ... this statement is so so very wrong
c-lesh ?
sorear it's complicated, like everything else 04:54
geekosaur and, for example, a rep loop was almost always slower than the corresponding explicit loop on 8086/8088 and 80186; they got better in 80286 and again in 80386
c-lesh interesting
geekosaur indeed. RISC is better if you code carefully to the strengths of the architecture, and hilariously bad if you code naively
sorear RISC *was* better when it first came out. but the supposed advantages of RISC don't stand up to processor evolution
geekosaur also required more complex code generators 04:55
sorear the original MIPS design was extremely efficient, but it exposed a lot of the internal structure of the original MIPS processor in the ISA
geekosaur but CISC architectures evolved faster than code generators did, for the most part
sorear once you start making small changes to the chip, like adding a pipeline stage, the code generation has to completely change
geekosaur yep 04:56
sorear a modern processor needs a recoding unit to isolate the user from changes in the processor
c-lesh that's were cisc comes in, right?
sorear right 04:57
geekosaur there is less of a difference in that case, but the real problem is the red queen's race between RISC architecture changes and code generators
c-lesh so modern processors like dual cores uses cisc
Timbus looks at dual core phone
i.. hmm..
c-lesh VMs also tend to emulate cisc, right?
sorear c-lesh: VMs do whatever they want 04:58
CPU design is fundamentally different from fast VM design
geekosaur and having to make sure you pick the right code generator for your current arch, or having to pick a general one that runs ok everywhere instead of very well natively and horribly if you copy the binary to an ostensibly similar system with a different cpu stepping
c-lesh true, my VM doesn't even have registers or simple instructions
sorear because transistors are *always on* while in a VM, *one instruction* is run at a time
c-lesh some instruction is my VM can be up to 30 bytes! 04:59
sorear think about this: your processor has, say, 10^8 transistors clocked 10^9 times per second. Think about how many logical operations happen per second, and how much useful computation occurs.
c-lesh they include expressions as their parameters
sorear CPUs do a *lot* of redundant computation in order to minimize *logical depth* 05:00
c-lesh i see
sorear also, dual core systems have been around since forever. "Modern cpus like dual cores" is ... not really meaningful
c-lesh since when?
sorear I don't know. I have a copy of the Motorola 68020 system programmer's manual in my room and it talks about concurrent processing using CAS 05:01
c-lesh P4 was not dual core
05:02 cooper left
sorear so at least 1984 05:02
but I'm sure this goes back much further, perhaps to the 60s
c-lesh 1984? That long? I though they came out in the early 2000s.
geekosaur if you only care about particular x86 implementations, sure 05:03
05:03 cooper joined
geekosaur dual cores on the same chunk of silicon is recent 05:03
sorear it wasn't that long ago when a single core had to be split across many chunks of silicon 05:04
c-lesh i see there were coprocessors before
sorear especially back in the 50s and 60s when cores used discrete logic
c-lesh cpu, gpu, spu
geekosaur although even there it was more the "commonly available and easy for implementors to inegrate" that was the enhancement
sorear geekosaur: were there any single-die multi-{ARMv1,6502,8080} systems back in the day? 05:05
geekosaur I don't think so
sorear geekosaur: for embedded systems wanting the ultimate fault isolation
geekosaur single-die could be done as a tour de force, but not in any kind of quantity, IIRC 05:06
but there were multiple CPU mainframes in the 60s and multiple CPU microcomputers in the 70s 05:07
including hybrid logic machines such as the OSI C8D and the Commodore SuperPET 05:08
c-lesh not familar with those
geekosaur (6502/6800/8080 in the latter, IIRC)
sorear I kind of want to see what the maximum clock speed of a 35nm 6502 would be, and how many of them / combined TFLOPs you can get onto a P4-sized chip 05:09
(ignoring the issue that small local memories are a huge pain for programming) 05:10
c-lesh 64K
that sucks
sorear I'm talking more like 4K
*small*
geekosaur what sucked even more about the 6500 series was that the stack was hardcoded to page 2
c-lesh 4K - I wrote pong in 256 bytes :)
geekosaur that is, 256 bytes at $0200 05:11
c-lesh there are clones of pong written in 256 bytes
256 bytes was for the OS?
$0200
so 64K - 256 was for the program or programs 05:12
sorear c-lesh: $0200 is how we used to write 0x0200
means that the stack occupies locations 512 to 767
c-lesh I know but that was the starting address of where the program has to be loaded in 64K architectures. 05:13
sorear also, 64k *addresses* does not mean 64k *bytes of memory*
c-lesh there was an architecture that had only 64K of memory and 16 bit addressing without segmentation 05:14
i think that was the 8080
sorear segmentation was *rare*, c-lesh
geekosaur on 6500 your programs usually loded above $0300 or $0400; page 0 was used for indexing (!) and page 2 for stack, page 1 could hold a bootstrap but rarely an entire OS
sorear *most* of the 16-bit systems didn't have anything even remotely like the 8086's segment registers 05:15
c-lesh how big was a page
sorear page means 256 bytes
on the 6502/8080/6800 and related systems
c-lesh where was the os loaded to
geekosaur depended on the OS. CP/M on 8080 and successors loaded the program at $0100 and kept the OS in high memory, letting the program overwrite much of the OS while running and then reloading that part from disk when the program exited 05:16
c-lesh virtual memory 05:17
so not much of the os had to be loaded
not even the shell, right?
sorear Virtual memory for microcomputers did not exist at this point in time.
geekosaur doing virtual memory well required CPU support that didn't exit back then. this is more like explicitl overlays where a program or the OS had to explicitly load stuff into the overlay area and swap it around explicitly if needed 05:18
c-lesh but overwriting the os is dangerous
sorear CPU support for virtual memory on personal computers didn't really exist until the 286
geekosaur indeed. but, well, that was what was available at the time
sorear well after this period
05:19 tokuhirom left
c-lesh how could the os be overwritten and maintain the stability of the system? 05:21
sorear these weren't multiprogrammed systems.
c-lesh so if the os was overwritten and the program exited what happened, was the shell still there? 05:22
sorear The program crashes, the user reboots the system 05:23
Timbus did.. did they reboot it with a crank handle 05:24
:v
geekosaur the shell was one of the things CP/M explicitly allowed to be overwritten during program execution; proper program exit allowed the OS to reload the shell
but if the program crashed, well, no memory protection or etc., so you got to hit the reset button or flip the power switch
sorear I imagine the reloader only took a few dozen instructions
geekosaur it was just the way things worked
sorear It was a very different culture than we see today in the Land of UNIX Clones 05:25
(imo NT counts as a UNIX clone, even if it's really an OS/2 clone)
geekosaur weird hybrid of OS/2 and VMS and Unix, really 05:26
but with a horrid lot of backward compatibility stuff, some of which microsoft is *still* trying to get rid of 05:27
sorear Timbus: Usually the way things worked back then was that you had a normally-open push button switch that connected the \RESET line on the CPU to ground
If you were lucky you had another switch that did that for \NMI
Timbus geekosaur, and Wine is *still* trying to add
heh 05:28
sorear (the closest thing to a "real" 70s-80s micro I've used was a 68040 Macintosh running System 7.1.)
c-lesh the program would call a sub to reload the shell
right? 05:29
sorear the Macintosh was a 32-bit system with 24 address lines; the bulk of the OS was stored in ROM, and the NMI vector pointed to a debugger
c-lesh ROM as in BIOS 05:30
sorear ROM as in read-only memory
BIOS is a term specific to PC clones
c-lesh but doesn't the BIOS chip have some interrupts that allow files to be opened, etc. you know, os specific tasks 05:31
sorear no, no, no, no
the BIOS chip in a PC is just memory
it stores 32-128 kilobytes of data
c-lesh no programs? 05:32
sorear the address decode logic on the motherboard causes some addresses to be routed to the chip
in particular, the PC starts executing instructions at $FFFF0
(there are some real horrors with initial CS setting that I'm not going into here)
so usually the BIOS memory range includes, at least, $F0000 to $FFFFF 05:33
c-lesh memory mapped, right?
sorear the BIOS doesn't know what files are
it does include a disk driver
c-lesh that's true, just sectors
disk driver to read sectors?
sorear the bios loads sector (512byte) 0 of the floppy disk at address $7A00 (?), checks the last 2 bytes for the magic signature $55AA, and if so jumps to address $7A00 05:34
might be $7C00 or $7E00
been a while since I wrote a boot sector
c-lesh you wrote a boot sector, for what? 05:35
sorear the BIOS also contains subroutines that the OS can use to access hardware
I wasted a few months of my life writing a toy OS... please don't judge
c-lesh so the bios contains basic hardware drivers
sorear yes
c-lesh that sounds like fun
how did you debug the boot sector?
sorear I used a PC emulator 05:36
c-lesh that's cool 05:37
sorear the BIOS *chip* does not interact with interrupts in any way
c-lesh i see
sorear when the x86 chip sees an instruction like $CD $13, it acts as though it had received an interrupt 13h from the bus 05:38
so it loads a long address from $0004C and calls it after saving the old CS:IP and flags
it's called an interrupt but it's really more like a subroutine call
c-lesh i see 05:39
that's interesting
sorear anyway the standard PC BIOS sets up interrupts 10h to 1Fh to different services
c-lesh so the instruction initiates the interrupt
sorear "interrupt" 10h is a disk driver, interrupt 13h controls the display
the idea was that you would write an OS, and the OS would use the BIOS-provided services so that IBM could change the PC hardware 05:41
sort of like how CISC makes it possible for processors to evolve
many problems can be solved by adding a layer of indirection
c-lesh interesting
sorear ... it didn't quite work, modern OSes generally don't use the BIOS for a variety of reasons
geekosaur non-reentrant, non-interruptible, often quite slow compared to direct access, often uses scratchpad memory that interferes with OS memory 05:43
c-lesh i see
geekosaur although, I would say the BIOS is in many ways back; ACPI in particular
sorear also, writing code to work around bugs in 10 PC clone BIOSes is just as hard as writing code to work with 10 diffierent pieces of PC clone hardware 05:44
geekosaur another issue is, with disk drivers the generic BIOS entry point doesn't generally allow for intelligent scheduling of reads/writes which can be done by an OS's disk driver
yep
c-lesh so the bios is not really used 05:46
sorear another issue is that the BIOS uses 8086 instructions
this was especially a problem for the 286, because once a 286 has been switched to native mode, the only way to reenter 8086 compatibility mode is to reboot (!)
the 386 and later processors can switch back and forth freely, but it's still a performance and complexity cost 05:47
"new" subsystems like ACPI and VBE often provide an alternate entry point that uses 386 instructions 05:48
c-lesh real mode 05:53
sorear rakudo: say "foo".IO ~~ :s 06:03
p6eval rakudo a7fd89: OUTPUT«Bool::False␤»
sorear rakudo: say "README".IO ~~ :s
p6eval rakudo a7fd89: OUTPUT«Bool::False␤»
sorear rakudo: say "/".IO ~~ :s
p6eval rakudo a7fd89: OUTPUT«Bool::True␤»
sorear rakudo: say "/".IO.s
p6eval rakudo a7fd89: OUTPUT«Bool::True␤»
sorear rakudo: say "NQP_VERSION".IO.s 06:04
p6eval rakudo a7fd89: OUTPUT«Bool::False␤»
sorear rakudo: say "/etc/debian_version".IO.s
p6eval rakudo a7fd89: OUTPUT«Bool::True␤»
sorear rakudo: say "/etc/debian_version".IO.^methods(:local)
p6eval rakudo a7fd89: OUTPUT«open close eof get getc lines read seek tell write opened print say slurp d e f l r s t w x z copy chmod ins chomp path␤»
c-lesh cool 06:09
sorear yes, I mean real mode 06:10
it seemed like the unneccessary obfuscation though, so I didn't say it 06:11
c-lesh: disclaimer: I'm 21 years old. Everything I say about computer architectures in the 70s, don't assume I have first-hand knowledge 06:12
06:13 tokuhirom joined 06:15 fridim_ left
c-lesh well, i'm going to sleep thanks for all the knowledge 06:22
06:23 c-lesh left 06:31 icwiener_ left 06:32 icwiener_ joined 06:43 icwiener_ left 06:52 icwiener_ joined 06:57 lateau_ joined 06:58 wtw joined 06:59 fridim_ joined 07:09 packetknife left 07:13 hundskatt joined 07:28 sayu joined 07:33 xjiujiu joined 07:36 kaleem joined 07:39 baest joined 07:41 fridim_ left
frettled Good moaning! 07:46
sorear hi
07:52 Mowah joined 08:04 xjiujiu left 08:08 tokuhirom left
Woodi hi ppls today :) 08:14
just backlogged :)
sorear, geekosaur: pls, PLS, write blogs ! 08:15
colomon: me think part of fractal work goes to performance testing and this was marketed. and would be nice to spread knowledge about more application-way using of Perl6 GUI 08:21
sorear Woodi: about what? 08:24
I have no idea which of the things I've done are worth blogging about.
Woodi sorear, geekosaur you can easily blog about 'technology', just find problem or 'possible improvement' you want to comment, add historical view, add comparision of existing implementations... 08:25
sorear: me think, for the start, would be nice some jurnalist ask some questions to you :)
08:26 mj41 joined
Timbus it was an alright read 08:27
gotta admit
Woodi sorear: I have question for example: how stable or rock solid (in room temperature) are basic niecza/Perl6 features/functionalities. eg. are 'if', 'say', misc loop production ready ? 08:28
so scripting with that could be done 08:29
what other features are stable ?
how fast is niecza on linux with mono ?
Timbus about thiiiis fast *gestures with arms* 08:30
Woodi sorear: how fast would be web server or web backend using niecza ?
which databases you can access today ?
sorear: pls write some library or app (crud) ppls can copy 08:31
sorear Woodi: you're not going to be able to break if/say/misc loops. Those are super-super-solid 08:32
niecza on linux with mono - varies a bit, comparable to Perl 5 but a bit slower
08:33 x3nU left
sorear web server - no clue 08:34
Woodi do niecza GUI include business grid/table ?
Timbus a what
Woodi relational table/data from db presenting widget :) 08:36
how stable is GUI ? small, stable Excel can be coded ? 08:37
how stable text manipulation, regexes, grammars are ? 08:38
08:39 x3nU joined
moritz I'll be traveling most of the day, hope to be online in ~6h again 08:39
ciao
Woodi sorear: problem with niecza is that it use worse then alien technology :) but one or two small but usefull apps will make it will be used
08:39 x3nU is now known as Guest13145
Woodi tee 08:43
08:46 risou_awy is now known as risou 08:47 Mowah left
Woodi anyway, ppl, do not forget moust of us have winter now so low on sunlight which make better feeling. bears can at least sleep over that season :) 08:57
but hot cup of tee make me usually better :) 08:58
09:03 sayu left
sorear the drink is spelled "tea" 09:07
Woodi sorear: just found on Rosettacode that Niecza was 'using Perl5' initially. and in Niecza announcement that it is 'project focusing on optimization research'. is that goal done or to what point ? why architecture and how changed ?
sorear smiles for the first time today
Woodi damn
sorear Woodi: I'm always asking how niecza can be made faster. 09:08
And use less memory
Woodi I realy need to re learn basics...
sorear: but what other Perl6 implementations can learn from niecza ? 09:09
sorear how to implement features
having two implementations helps to keep the spec on top 09:10
but if you don't mind I'd like to sleep now
Woodi sorear: no problem, I can still put questions, etc here and hope it helps somehow someday :) 09:11
sorear -> sleep
09:14 Mowah joined 09:26 kranius joined
masak morning, #perl6! 09:26
Woodi hallo 09:34
masak: how you think, niecza could use niecza.perl6.org for now ? 09:36
09:37 sftp left, molaf joined
Woodi I could submit it to freshmeat.net/freecode too 09:38
masak er. 09:48
09:48 ruoso left
masak Woodi: (1) how do you mean, "use niecza.perl6.org"? (2) what is it you think you could submit to freshmeat, and why? 09:48
stand down, soldier. 09:49
don't make decisions that are really up to sorear.
Woodi ad 1. just thinked about simple web page for now. 09:50
ad 2. you must see it yourself :) it portal with announcements of opensource project, works seence '90 09:51
masak I know what freshmeat is.
Woodi url was freashmeat.net and stil works
masak my question was more, why do you think you should submit something Niecza-related to freshmeat?
09:52 icwiener_ left
Woodi why you use Reddit ? :) another place :) 09:52
09:52 icwiener_ joined 09:54 kaleem left
masak I know you're eager to publish sorear, but I really think it's better if he starts the process himself. 09:54
otherwise we'll have one or more unmaintained pages claiming to contain up-to-date information on Niecza.
Woodi Ithink sorear would like community will do simple and boring task for him... 09:55
masak yeah, don't. 09:56
Woodi k, I will ask sorear first
later 09:59
10:05 molaf left, arlinius left 10:07 kaleem joined 10:17 arlinius joined 10:24 sayu joined
kshannon good * #perl6 11:07
11:08 Mowah left
tadzik hello #perl6 11:24
11:25 ruoso joined, lateau_ left 11:27 kaleem left
tadzik why is sleep() a sub:prefix rather than just a sub? 11:28
Juerd I'll guess: precedence 11:30
sleep 5, say "hi"
Like Perl 5's ($)-prototyped subs 11:31
</guess>
11:34 sftp joined 11:50 LlamaRider joined 11:54 Trashlord left 11:58 kaleem joined 12:01 sayu_ joined 12:02 sayu left, Trashlord joined
masak Juerd: yeah. 12:09
Juerd: also, &time. 12:10
I don't remember how we did with &rand.
masak checks the symbols in STD
S03 gives 'sleep abs sin temp let' as examples. 12:12
tadzik hrm
masak ah;
There is no unary C<rand> prefix in Perl 6, though there is a C<.rand>
method call and an argumentless C<rand> term.
perl6: say rand 5
p6eval niecza v13-46-gab2db52: OUTPUT«===SORRY!===␤␤Unsupported use of rand(N); in Perl 6 please use N.rand or (1..N).pick at /tmp/UdmCAm4STk line 1:␤------> say rand⏏ 5␤␤Parse failed␤␤»
..pugs b927740: OUTPUT«3.425864913733676␤»
..rakudo a7fd89: OUTPUT«===SORRY!===␤Unsupported use of rand(N); in Perl 6 please use N.rand or (1..N).pick at line 1, near " 5"␤»
masak \o/
12:14 kaleem left
Timbus nom: say 5.rand 12:24
p6eval nom a7fd89: OUTPUT«4.18286033244387␤»
Timbus so... 'why'
masak maybe because there's the realization that unary-prefix subs are exceptions to the general rule of listop subs, and we want to cut down on them as much as possible? 12:26
12:27 snearch joined
masak without, for some reason, going all the way down to 0. :P 12:27
maybe going all the way down to 0 would count as having an ideological axe to grind...
Timbus rand is a listop? 12:28
12:28 snearch left
Timbus i think something flew over my head there 12:29
it is far too hot for thinking
12:30 kaleem joined
LlamaRider aren't listops just shorthand for using associative operators? 12:31
masak Timbus: &rand as a sub is deprecated in Perl 6. use the method instead. 12:34
LlamaRider: I did not understand that question.
masak wants to reject rt.perl.org/rt3/Ticket/Display.html?id=107292
also, rt.perl.org/rt3/Ticket/Display.html?id=107330 is a valid ticket but belongs in github.com/perl6/nqp/issues 12:35
dalek ast: af353d6 | (Solomon Foster)++ | S0 (3 files):
Refudge a bit to account for new, harder fails in Niecza.
12:36 tokuhirom joined
LlamaRider masak: scratch that question, i was thinking only of infix ops that end up as list ops 12:38
masak ah, yes. 12:42
well, even something like 'say' is a listop, long as you don't use parens. 12:43
nom: enum LionelRichie <me naturally together you>; sub it(*@a) { @a }; say you; say me; say it together, naturally 12:46
p6eval nom a7fd89: OUTPUT«LionelRichie::you␤LionelRichie::me␤together naturally␤»
12:48 icwiener_ left
Timbus <masak> Timbus: &rand as a sub is deprecated in Perl 6. use the method instead. 12:50
:I copout
12:52 icwiener_ joined
dalek ecza: 3171346 | (Solomon Foster)++ | lib/CORE.setting:
Do numeric casting as needed in Range constructor.
 12:53
geekosaur thinks sleep is special-cased because you know it's never going to do anything sensible with a list, and 5.sleep is just weird 12:55
masak Timbus: that was in response to your question "rand is a listop?", which doesn't make much sense since the sub's deprecated.
12:58 icwiener_ left
masak I guess it wasn't a listop before it got deprecated... 12:59
pugs: say rand 4, 5
p6eval pugs b927740: OUTPUT«3.76167444512248045␤»
masak pugs: say rand 4, ":::", 5 13:00
p6eval pugs b927740: OUTPUT«1.895408839918673:::5␤»
masak right.
13:00 snearch joined
masak geekosaur: I think it would be more consistent to have sleep parse as a listop, but to complain (at parse time) whenever more than one argument was passed to it. as it is 'say 2, sleep 3, 4, 5' is hard to read and understand. 13:02
13:02 icwiener_ joined
masak though I guess it's no more tricky than scalar assignment :) 13:08
it all depends on what expectations one has. 13:09
13:20 snearch left 13:38 spine joined 13:40 kaleem left
dalek ast: 35ebf63 | (Solomon Foster)++ | S03-operators/range.t:
Add tests for string and array RHS of numeric ranges.
 13:42
13:48 Mowah joined 13:58 daxim joined 14:31 whiteknight joined, whiteknight is now known as Guest25455 14:39 owlEyes joined 14:40 bao joined 14:41 orafu left 14:43 orafu joined 14:46 PerlJam joined
PerlJam greetings! 14:46
masak earthlings!
spine hi there, is there some way in rakudo to keep the list generated by the cross operator from flattening? 14:49
S03 says to uses lol context, but that doesn't seem to be implemented.
rakudo: say .perl for (1,2)X,(3,4) 14:50
p6eval rakudo a7fd89: OUTPUT«1␤3␤1␤4␤2␤3␤2␤4␤»
bao ? 14:51
masak spine: correct -- I don't think it's implemented. 14:52
bao: ¿
14:53 snearch joined
spine masak: so I'd be best off by implementing a version of <infix:X> myself, which makes items from the lists? 14:53
masak bao: 你来自中国吗？
spine: not sure what you'd be best off doing :) if you know the size of the items, you can simply iterate that size. 14:54
nom: for 1..10 X 'a'..'j' -> $a, $b { say "$a$b" } 14:55
p6eval nom a7fd89: OUTPUT«1a␤1b␤1c␤1d␤1e␤1f␤1g␤1h␤1i␤1j␤2a␤2b␤2c␤2d␤2e␤2f␤2g␤2h␤2i␤2j␤3a␤3b␤3c␤3d␤3e␤3f␤3g␤3h␤3i␤3j␤4a␤4b␤4c␤4d␤4e␤4f␤4g␤4h␤4i␤4j␤5a␤5b␤5c␤5d␤5e␤5f␤5g␤5h␤5i␤5j␤6a␤6b␤6c␤6d␤6e␤6f␤6g␤6h␤6i␤6j␤7a␤7b␤7c␤7d␤7e␤7f␤7g␤7h␤7i␤7j␤8a␤8b␤8c␤8d␤8e␤8f␤8g␤8h␤8i␤8j␤9a␤9b􏿽xE2
spine masak: OK, thanks. I'll see if that is enough. trying to write down a nice solution to the tree problem from your coding contest ;) 14:58
masak nice! 14:59
cognominal "Foundations of Language : Brain, Meaning, Grammar, Evolution" www.amazon.fr/dp/B001ODEPQ2 15:02
may be of interest to people here
better link : www.amazon.com/dp/B001ODEPQ2 15:03
15:04 bao left 15:10 kaare_ joined 15:13 wtw left 15:14 kaare_ left 15:15 kaare_ joined 15:16 kaare_ left, kaare_ joined 15:18 simcop2387 left 15:20 simcop2387 joined 15:23 sayu_ left 15:24 sayu joined 15:30 teeter joined
fsergot nom: for 1..10 X 'a'..'j' -> $a { say $a } 15:30
p6eval nom a7fd89: OUTPUT«1␤a␤1␤b␤1␤c␤1␤d␤1␤e␤1␤f␤1␤g␤1␤h␤1␤i␤1␤j␤2␤a␤2␤b␤2␤c␤2␤d␤2␤e␤2␤f␤2␤g␤2␤h␤2␤i␤2␤j␤3␤a␤3␤b␤3␤c␤3␤d␤3␤e␤3␤f␤3␤g␤3␤h␤3␤i␤3␤j␤4␤a␤4␤b␤4␤c␤4␤d␤4␤e␤4␤f␤4␤g␤4␤h␤4␤i␤4␤j␤5␤a␤5␤b␤5␤c␤5␤d␤5␤e␤5␤f␤5␤g␤5␤h␤5␤i␤5␤j␤6␤a␤
15:30 teeter left
fsergot nom: my @a := 1..10 X 'a'..'j'; say @a.WHAT; 15:30
p6eval nom a7fd89: OUTPUT«List()␤» 15:31
fsergot nom: my @a := 1..10 X 'a'..'j'; say @a.perl;
p6eval nom a7fd89: OUTPUT«((ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (ListIter.new(),), (…
fsergot nom: my @a := 1..10 X 'a'..'j'; say @a;
p6eval nom a7fd89: OUTPUT«1 a 1 b 1 c 1 d 1 e 1 f 1 g 1 h 1 i 1 j 2 a 2 b 2 c 2 d 2 e 2 f 2 g 2 h 2 i 2 j 3 a 3 b 3 c 3 d 3 e 3 f 3 g 3 h 3 i 3 j 4 a 4 b 4 c 4 d 4 e 4 f 4 g 4 h 4 i 4 j 5 a 5 b 5 c 5 d 5 e 5 f 5 g 5 h 5 i 5 j 6 a 6 b 6 c 6 d 6 e 6 f 6 g 6 h 6 i 6 j 7 a 7 b 7 c 7 d 7 e 7 f 7 g …
fsergot nom: my @a = 1..10 X 'a'..'j'; say @a; 15:33
p6eval nom a7fd89: OUTPUT«1 a 1 b 1 c 1 d 1 e 1 f 1 g 1 h 1 i 1 j 2 a 2 b 2 c 2 d 2 e 2 f 2 g 2 h 2 i 2 j 3 a 3 b 3 c 3 d 3 e 3 f 3 g 3 h 3 i 3 j 4 a 4 b 4 c 4 d 4 e 4 f 4 g 4 h 4 i 4 j 5 a 5 b 5 c 5 d 5 e 5 f 5 g 5 h 5 i 5 j 6 a 6 b 6 c 6 d 6 e 6 f 6 g 6 h 6 i 6 j 7 a 7 b 7 c 7 d 7 e 7 f 7 g …
fsergot Whats differences between := and =? 15:34
masak I'd just like to say I like fluent interfaces, and I think they could be used more than they are today: martinfowler.com/bliki/FluentInterface.html
fsergot: := binds a value to a variable, = assigns a value to the container associated with a variable.
15:35 Mowah left
masak fsergot: after the binding, you have variable->value. after the assignment, you have variable->container->value. 15:35
15:35 baest left, havinglargeballs joined
fsergot nom: my @a = 1..3 X 'a'..'c'; say @a.perl; 15:35
p6eval nom a7fd89: OUTPUT«Array.new(1, "a", 1, "b", 1, "c", 2, "a", 2, "b", 2, "c", 3, "a", 3, "b", 3, "c")␤»
masak fsergot: the container allows you to do more assignments later.
fsergot nom: my $a := 'a'; $a = 'b'; 15:36
p6eval nom a7fd89: OUTPUT«Cannot assign to a non-container␤ in block <anon> at /tmp/7UHz7Z9JDf:1␤ in <anon> at /tmp/7UHz7Z9JDf:1␤»
fsergot masak++ clear, thanks.
:)
the deeper in perl i am, the more interesting p6 is :) 15:38
masak best explanation I've given so far :) 15:39
fsergot: you might find strangelyconsistent.org/blog/plain-...ky-binding interesting, too.
fsergot masak: where, in life examples, i can use := ? :)
masak: thanks :)
masak hm, that's slightly badly formatted. sorry about that.
fsergot: that's too general a question. 15:40
15:40 Mowah joined
masak fsergot: but often binding involves less "copying" than assignment. I've seen people use if for performance reasons. 15:40
dalek ast: f6f7478 | (Solomon Foster)++ | S03-sequence/nonnumeric.t:
Fudge for niecza.
 15:41
fsergot thanks :) 15:42
dalek ecza: 1f4230f | (Solomon Foster)++ | t/spectest.data:
Turn on S03-sequence/nonnumeric.t.
 15:43
masak fsergot: for example strangelyconsistent.org/p6cc2010/p5-moritz/ uses binding for this reason. 15:45
binds the array rather than clone it and copy all of the strings in it. 15:46
spine fsergot; I've come to some problems with = when using lazy lists, e.g. my @a = map {$_}, 1..Inf; leads to an infinite loop, whereas my @a := map {$_}, 1..Inf; retains it's lazyness. 15:48
masak oh yes. that's a good reason, too.
array assignments are eager -- array bindings are lazy. 15:49
it's even possible to bind a lazy list to an array constant :)
TimToady niecza: constant @primes = 2, 3, { ($_ + 2, $_ + 4 ... *).first: -> $n { $n %% none @primes ... * > sqrt($n); } } ... *; say @primes[^20]; 15:50
p6eval niecza v13-47-g3171346: OUTPUT«2 3 5 7 11 13 17 19 23 29 31 37 41 43 47 53 59 61 67 71␤»
masak \o/
colomon niecza: constant @song := $*IN.lines; say @song[1]
p6eval niecza v13-47-g3171346: OUTPUT«===SORRY!===␤␤Malformed constant at /tmp/2QeX4cP_Fd line 1:␤------> constant @song ⏏:= $*IN.lines; say @song[1]␤␤Parse failed␤␤»
colomon niecza: constant @song = $*IN.lines; say @song[1]
p6eval niecza v13-47-g3171346: OUTPUT«Unhandled exception: Unable to resolve method lines in class Any␤ at /tmp/h2aTGpinkC line 1 (@song init @ 2) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/src/NieczaBackendDotnet.pm6 line 0 (downcall @ 0) ␤ at /home/p6eval/niecza/src…
TimToady the pseudo assignment works more like a binding here
masak but surely binding should be syntactially allowed, too? 15:51
[Coke] starts skipping through backscroll.
TimToady niecza: constant @song = lines($*IN); say @song[1]
p6eval niecza v13-47-g3171346: OUTPUT«Unhandled exception: Unable to resolve method lines in class Any␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /tmp/IRlIdmlY4Y line 1 (@song init @ 2) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/src/NieczaBackendDotnet.pm6 line 0 (do…
TimToady oh, should've expected that 15:52
colomon niecza: constant @song = $*IN.slurp; say @song[1]
p6eval niecza v13-47-g3171346: OUTPUT«Unhandled exception: Unable to resolve method slurp in class Any␤ at /tmp/R6diXdVUi1 line 1 (@song init @ 2) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/src/NieczaBackendDotnet.pm6 line 0 (downcall @ 0) ␤ at /home/p6eval/niecza/src…
colomon niecza: constant @song = lines(); say @song[1] 15:53
p6eval niecza v13-47-g3171346: OUTPUT«Unhandled exception: Unable to resolve method lines in class Any␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /tmp/qR9nBecSu4 line 1 (@song init @ 2) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/src/NieczaBackendDotnet.pm6 line 0 (do…
TimToady the problem is $*IN is not there at compile time
colomon ah
15:53 havinglargeballs left
colomon TimToady++ 15:53
15:54 daxim left
masak d'oh! 15:54
[Coke] masak: re: RT#107292 - is Perl6::Compiler in the specs anywhere? 15:55
masak nope.
it's a Rakudo thing. 15:56
15:56 bluescreen10 joined
masak I don't think the expectation that it'll be loadable at the user level is justified. 15:56
15:57 LlamaRider left
PerlJam Why not? Seems like it falls under making "hard things possible" 15:58
[Coke] PerlJam: because it's not in the spec, and is therefor one more thing we'd have to support if we refactored. 15:59
15:59 snearch left, snearch joined
masak nqp is a different language than Perl 6, so even if the user find the file, he'd have to load it through with some :lang<nqp> modifier. 16:00
s/find/found/
jnthn afternoon, #perl6 16:01
phenny jnthn: 21 Dec 20:01Z <moritz> tell jnthn at your convenience, please take a look at S12-attributes/mutators.t. It seems to die in $!varies +=2 with "No applicable candidates found to dispatch to for 'Numeric'", which I find quite weird, since $!varies is typed Int
jnthn: 22 Dec 12:18Z <moritz> ask jnthn if he wants to get some commits in before the release
jnthn: 27 Dec 02:12Z <sorear> tell jnthn irclog.perlgeek.de/perl6/2011-12-27#i_4883396 I'd like a second opinion
PerlJam [Coke]: I agree that those things are true, but that doesn't seem like a sufficient argument for not making it part of the spec.
[Coke] +1 from me on rejecting it, anyway.
PerlJam: then you're forcing every implementation to use our (currently internal) API. 16:02
jnthn Just dropping by to say I'm still alive. ;-)
[Coke] jnthn: \o/
jnthn Will be back home in a few days, and then back to my normal level of hacktivity. :)
PerlJam [Coke]: no, I'm saying make the API public. (at least I'm entertaining the idea a little more than the rest of you :)
masak jnthn! \o/ 16:03
[Coke] so, don't make it spec, but make it supported? Still -1 from me. ;) 16:04
16:05 owlEyes left
jnthn Loading NQP code will need a :lang<nqp>. 16:05
PerlJam I'm perfectly happy rejecting that ticket for the reasons cited, I just think the idea has some merit that should be considered. 16:07
jnthn The Perl 6 grammar almost certainly should be accessible from Perl 6 space, with an appropriate use statement. The exact interface to doing so is something we probably want to try and standardize over implementations. 16:08
Though not a priority at the moment, for me anyway.
16:09 phenny joined
jnthn huh, how come Pretence got changed to Pretense in a7fd89ea? :) 16:12
masak goodness knows!
masak pretends to be innocent
jnthn www.thefreedictionary.com/pretence is correct, y'know :P
oh, pretense is the American spelling.
masak aye. 16:13
the commit comment says as much!
jnthn But my MOP is proudly British! :P
PerlJam jnthn: so ... do you "kick to the curb" or "kick to the kerb"? 16:14
(or would you were you so inclined to do such a thing )
:)
jnthn Well, it's spelt kerb of course :P 16:15
But "kick to the kerb" sounds a bit of a gangsta thing to be doing to me :P
jnthn hasn't actually heard that expression before :)
PerlJam it's probably american in origin, so it should be "kick to the curb" :) 16:16
masak jnthn: it is? I can see no other instances of British spelling in the MOP. that's why I thought "Pretence" was an honest mistake.
jnthn masak: I was mostly kidding about.
masak: Well, it's not so much mistake as "jnthn simply doesn't think about these things" 16:17
PerlJam was just surprised at the "americanization" of the MOP ;)
ruoso happy new year... (for all of those in the gregorian or julian calendars :) )
masak PerlJam: it wasn't so much "americanization" as... a silly commit.
PerlJam: now they have to credit me in the January release. mwhahaha!
PerlJam heh!
jnthn masak: Bah, you were meant to do that by getting us macros. :P 16:18
masak ok, I'll do some of that too, then.
jnthn yay :)
jnthn well get back to his NCI work
*will
Friday or weekend at the latest.
Doing family/friend/visiting stuff for last few days here :) 16:19
masak jnthn: make sure you relax really well, and come back with batteries charged to the brim! 16:20
ruoso jnthn: hey... I am getting some weird behaviors and segfaults with my junction-using sudoku solver
colomon jnthn, \o/ 16:21
PerlJam wished his xmas would have been of the battery-charging kind rather than the bettery-draining kind
s/wished/wishes/
jnthn ruoso: OK. An RT ticket with code and, if you can get it, a backtrace would be helpful.
ruoso: I'll look at it once I'm back do things...probably the weekend.
16:23 bbkr joined
ruoso submitted 16:27
jnthn Thanks! 16:29
Looking forward to getting back to Perl 6 stuff in a few days time :)
And with much more energy than I had by the end of December :)
Tuit supply is looking reasonable for Jan/Feb too :)
masak sounds wonderful. 16:30
jnthn: looking forward to you getting back!
jnthn :)
16:32 kaleem joined
PerlJam btw, anyone seen pmichaud? 16:33
Guest13145 hi
16:33 Guest13145 is now known as x3nU
masak hi, Guest25455^W x3nU 16:33
[Coke] PerlJam: if anyone had actually seen him, I'd expect it to be you! ;) 16:34
x3nU jnthn: someone told me to ask you
i can't compile rakudo with msvc 2010
pastebin.com/vnJy7JjP
there's no error message
jnthn 32 or 64?
x3nU 32
just "nqp.exe has stopped working" or something
jnthn Hm
x3nU like that
PerlJam [Coke]: I was actually thinking something similar when I asked the question :) 16:35
[Coke] I have a 32-bit windows system; I can try on mine.
jnthn How are you building?
I wonder if it's the "need to copy libparrot.dll in some cases" issue
x3nU perl Configure.pl --prefix=C:/Rakudo --gen-parrot --gen-nqp
it fails on configure
jnthn Try (in the NQP directory) copy ..\install\bin\libparrot.dll .
Then give it another go. 16:36
x3nU ok, i will check it :)
masak jnthn++
jnthn We have an RT for this issue (if this is the issue) 16:37
Just didn't get to it yet.
OK, back to relaxation mode here :) 16:40
Probably back about on Thursday evening, once I get home :)
o/
masak \o 16:41
x3nU bye :)
16:42 kaleem left
tadzik oh, I've just missed jnthn 16:42
16:44 tokuhirom3 joined, tokuhirom left 16:49 sayu left
masak tadzik: yes, but he promised to be back later in the week :) 16:50
x3nU jnthn's solution worked :D 17:03
17:03 fsergot left
x3nU jnthn++ 17:03
tadzik yeah, I can see :)
17:09 lateau_ joined 17:10 estrabd left 17:18 MayDaniel joined 17:24 fsergot joined
colomon std mavens: I'm looking at sort.t, and it defines a new multi for sub cmp (???) and then tests that sort uses it. There's at least two major errors with that idea, right? 17:33
17:34 c-lesh joined
c-lesh hello again 17:34
colomon \o
c-lesh i was wondering if anyone has looked at the source for Parrot 17:35
colomon I avoid it like the plague. ;) 17:36
c-lesh is it hard to read
is it object oriented
i was just wondering which module interprets the bytecode 17:37
colomon c-lesh: there's probably someone around here who can answer that. but there's an entire #parrot channel full of experts out there. 17:38
colomon doesn't recall whether it's on freenode or the perl IRC network.
tadzik perl 17:39
c-lesh does parrot development take place there 17:40
colomon I believe so.
c-lesh cool 17:41
colomon as far as I know, it's the equivalent of this channel, but centered on parrot
x3nU irc.perl.org #parrot
ruoso jnthn: is rakudo autothreading using Tasks in the parrot level?
c-lesh cool, i'll check it out! 17:42
I've been doing some research so that I can finish writing my VM.
17:43 drbean left
colomon ruoso: almost certainly not yet 17:43
c-lesh Whoa! A camel made out of code on irc.perl.org! What is it supposed to do?
x3nU just logo, when i said "irc.perl.org" i meant irc network to which you should connect 17:44
ach, you're using webirc
wait
i will give you proper link
c-lesh CTCP?
x3nU c-lesh: wbe02.mibbit.com/?settings=1c154d53...nnect=true 17:45
here you can reach parrot folks
c-lesh: that CTCP something meant that i was checking what irc client you were using
you are*
17:46 lateau_ left
c-lesh Thanks! 17:46
17:47 mj41 left
c-lesh they all seem to be out to lunch 17:47
colomon it is virtual new year's day... 17:48
ruoso colomon: it is interesting that the errors I'm getting are somewhat consistent with threading...
c-lesh haha :O
colomon is watching the rose bowl parade, and looking forward to football
ruoso: in what way? 17:49
17:50 drbean joined 17:51 thou joined
x3nU c-lesh: on IRC it is normal to wait for reply for pretty long time 17:51
just ask questions and wait...
c-lesh did that
i guess many are at work
i work from home, though
ruoso colomon: in the way that it fails in different points at different runs, and that sometimes it even succeeds
but it may be just memory corruption
colomon yes, memory corruption / GC issues are definitely a current rakudo problem. :( 17:52
c-lesh personally i detest garbage collection
why not implement something similar to C++ memory management and then have an "memory leaks" cleaned up at the end 17:53
colomon C++ memory management isn't very practical with a lot of the idioms we'd like to use in p6 17:54
tadzik 1) no one wants C++-like memory management 2) there'll be a hell load of those "memory leaks"
c-lesh instead of having a stack one could just make sure each subroutine has its own memory
colomon but I have to admit I'm frequently distressed at the side effects of NOT having C++-like memory management
c-lesh but the interpreter could check for those at the end of its execution and alert the programmer
this was the programmer could fix his/her mistakes 17:55
a garbage collector running in a time critical loop is not cool
17:56 Mowah left
ruoso altough it definitely consistently segfaults in the same point 17:59
#0 0xf6901c4c in mp_cmp_d () from /home/ruoso/devel/rakudo/install/lib/parrot/3.11.0-devel/dynext/nqp_bigint_ops.so
c-lesh .so is an assembly file?
ruoso my bet is rakudo is not built with debug by default 18:00
18:00 Mowah joined
colomon .so is a library file 18:00
thou $ nom if.pl 18:01
> take the white beer and the big yellow cheese from the kitchen and put it under the couch
(["take", ("beer" => ("white",).list.item, "cheese" => ("big", "yellow").list.item), {"prep" => "from", "noun" => ("kitchen" => ().list.item,)}], ["put", ("it" => ().list.item,), {"prep" => "under", "noun" => ("couch" => ().list.item,)}])
colomon object file library
tadzik thou: wow, what's that?
colomon thou++ 18:02
c-lesh raduko is running on parrot?
thou gist.github.com/1536817
tadzik: it's a hack :)
colomon thou: what will you do when "it" has two apparent antecedents? ;)
thou i thought it'd be fun to rewrite masak++ 's crypt.pl
tadzik c-lesh: yes
thou so this works OK, i could write some Noun class or whatever to make it usable 18:03
18:04 MayDaniel left 18:06 zby_home joined, achromic left
thou i was thinking about having a 2-pass method: first pass would map each word to a list of possible uses (e.g., "fire the rocket" -> (('fire' => ('transitive verb', 'singular noun')), ('the' => 'definite article'), ('rocket' => ('intransitive verb', 'singular noun'))), and then a 2nd pass to create a tree out of that. 18:07
but i don't know how to use a grammar for that. can i use the grammar on a stream of objects instead of characters? 18:09
i kind of think not
18:09 tokuhirom joined
colomon sort tests are making me see red 18:18
ruoso thou: as long as you can implement your stream of objects in terms of the Str API 18:26
thou hmmm 18:27
18:27 risou is now known as risou_awy
masak thou: ooh, rewrite crypt.pl? 18:30
in what?
thou masak, in perl6
just in a different way
ruoso thou: on the parsing natural languages topic... you do know that besides some very limited cases, this is a dead end, right?
masak thou: honestly, I've been thinking of rewriting it as well -- I was stupid not to develop it with TDD. 18:31
thou ruoso -- yeah, but i'm thinking of slightly more friendly than most Interactive Fiction
ruoso i.e.: "fruit flies like bananas"
thou sue
sure
time flies like an arrow
masak see strangelyconsistent.org/blog/what-i...e-blogging
18:31 snearch left, snearch joined
benabik thou: There's an IF system for the Z-machine (info com) called Inform that has a pretty darn good parser. 18:31
ruoso current trend on natural language processing is neural networks and lots and lots of data 18:32
thou benabik: yeah, i have used TADS and some others, a little bit, that are really good
ruoso all formally-described-grammar initiative fail too soon on real world usage
fsergot masak++
thou anyways, this was mainly just a thing to learn more about perl6 grammars
ruoso In that case, yes... it's lots of fun... I did that myself for portuguese once :) 18:33
thou and to have fun with a small but non-trivial project :-)
benabik thou: Inform was put together by someone who did a thesis in natural language processing and his documentation is fairly clear... 18:34
thou really, if i were wanting to make real IF, i should just use one of the existing systems. but i want to play with p6
:-) 18:35
benabik Well, he had a good chapter on how he did the parsing...
thou cool
benabik inform7.com/learn/papers/
thou thanks!
benabik Although none of those are the ones I was thinking of, I think. Pity.
Ah! This is what I was thinking of: www.inform-fiction.org/manual/html/s34.html 18:37
18:37 awoodland joined 18:38 awoodland left
colomon niecza: say (1.1,2,NaN,-3.05,0.1,Inf,42,-1e-07,-Inf).sort.perl 18:47
p6eval niecza v13-47-g3171346: OUTPUT«(Inf, -61/20, -Inf, -1E-07, 1/10, 2, 11/10, NaN, 42).list␤»
colomon :\ 18:48
niecza: say (1.1,2,NaN,-3.05,0.1,Inf,42,-1e-07,-Inf).sort({ $^a <=> $^b }).perl
p6eval niecza v13-47-g3171346: OUTPUT«(-Inf, -61/20, -1E-07, 1/10, 11/10, 2, 42, NaN, Inf).list␤»
thou benabik: the inform7 stuff is really interesting :-) 18:49
colomon niecza: say (1.1,2,-3.05,0.1,Inf,42,-1e-07,-Inf).sort.perl 18:50
p6eval niecza v13-47-g3171346: OUTPUT«(-61/20, Inf, -Inf, -1E-07, 1/10, 2, 11/10, 42).list␤»
masak colomon: I can see why you're not entirely pleased, yes. 18:51
colomon masak: with niecza or with the tests?
masak the test looks OK to me. 18:52
colomon is cheesed off at both at the moment
You can't sort NaN
masak ;)
no, you can't.
colomon yet there it is in sort.t 18:53
and defining a sub cmp should have never have an effect on the behavior of sort 18:54
benabik thou: I was less than impressed with it because it's trying to do natural language programming, which I find somewhat irritating. :-D 18:56
masak colomon: not unless &cmp called dynamically, which it isn't.
thou yeah, i guess if it makes sense anywhere, it's in a language for writing IF 18:57
masak or SHRDLU. 18:59
19:14 tokuhirom left, cognominal left 19:16 cognominal joined
colomon masak: what do you mean by dynamically? 19:17
(sorry to drop out there, I've got a three-year-old tackling me atm.)
masak :) 19:18
colomon now he's back to playing with legos 19:19
masak colomon: just like you can declare dynamically scoped variables, so you can define dynamically scoped subs. 19:20
which can then be overridden, and sort would find the overriding variant. 19:21
except &cmp isn't dynamical that way.
colomon ah, okay. 19:22
more importantly, &cmp has nothing whatsoever to do with sort
masak right :) 19:23
19:23 arlinius left
colomon huh. next sort question: why should sort(@a) work? 19:33
kshannon Ah, rakudo is depending on find_method dying with an appropriate error message rather than return NULL for quite a few things, but that's not the correct interface for VTABLE_find_method which should return NULL instead. There are quite a few places in the rakudo source which use VTABLE_find_method instead of STABLE(obj)->find_method; I'm currently patching VTABLE_find_method to catch exceptions and return NULL to play nice with non-6model parrot stuff (in
19:34 c-lesh left
kshannon Jeez, I didn't realize the line had got that long :( 19:34
colomon according to S32 containers, sort's first argument should be an ordering
oversight in S32?
kshannon colomon: or the slurpy block. 19:35
whoops, that's the method... 19:36
colomon nom is doing 19:40
multi sub sort(*@values) {
@values.at_pos(0).^does(Callable)
?? do { my $cmp := @values.shift; @values.sort($cmp) }
!! @values.sort;
guess I can make niecza do that too, and maybe TimToady can fix the spec. (Or tell us to stop it!) 19:42
flussence
.oO( wouldn't it make more sense to use 2 multis instead of a ternary op there? )
 19:44
19:45 Vlavv` joined 19:47 drbean left 19:49 Vlavv left
colomon niecza's sort really leaves a lot to be desired 19:52
since it doesn't appear to actually sort. :( 19:53
19:53 drbean joined
masak yeah, that's usually a minimal requirement. 19:54
you could consider it part of the "contract" of the &sort routine. 19:55
colomon this is really frustrating -- sort doesn't work right, and the tests in sort.t are insane
is ~(42,).sort: { 0 }, "42"
{ 0 }
?
masak 0 is a valid output from a comparator routine.
19:55 Moukeddar joined
sorear good * #perl6 19:55
masak it means, essentially, "objects $^a and $^b are equivalent" 19:56
colomon masak: it's valid output, but no current compiler can handle a 0-arg comparator
tadzik good star sorear
colomon masak: and with good reason: wtf does a 0-arg comparator mean?
masak good asterisk, sorear
colomon good multiplying, sorear
masak colomon: aye, that seems slightly insane.
tadzik now we see how #perl6 users pronounce '*' 19:57
masak "here's the obj..." -- "ZERO!"
geekosaur reminded of haskell-cafe thread where someone sked about pathological comparators and he suggested _ _ -> GT
colomon masak, geekosaur: the really crazy thing here is that the comparator has nothing to do with the point of the test, according to the comments. 19:58
geekosaur (amd all good p6ers should know that * is prononced Whatever...)
colomon geekosaur++
perl6: say (42,).WHAT
19:58 mj41 joined
p6eval rakudo a7fd89, niecza v13-47-g3171346: OUTPUT«Parcel()␤» 19:58
..pugs b927740: OUTPUT«Array␤»
colomon nom: say Parcel ~~ Positional 19:59
p6eval nom a7fd89: OUTPUT«Bool::True␤»
colomon I think Henry is trying to figure out how to piledriver me. 20:00
now he's lying on me, trying to make me hold a lego, and chanting "Rock Island" from The Music Man 20:02
PerlJam btw, is there a way to "unpack" a parameter and name it? sub foo ([$head, *@tail]) {} will give me the parts, but how do I refer to the aggregate? sub foo (@blah($head, *@tail)) {} is suggested by the tree node example 20:03
er, perhaps @blah[$head, *$tail] 20:04
masak both should work, I think.
dalek ast: 801c909 | (Solomon Foster)++ | S32-list/sort.t:
Trying to make less pathological, plus light fudging for niecza.
20:10 Moukeddar left 20:11 drbean left
dalek ecza: e224a62 | (Solomon Foster)++ | lib/CORE.setting:
Make sub sort recognize a callable first argument.
 20:14
20:17 wolfman2000 joined, drbean joined
masak wolfman2000! \o/ 20:17
wolfman2000 Afternoon
masak evenin'
colomon sorear: if you grab my latest patches, you can try sort.t and see where it is giving incorrect answers. :( 20:18
masak 27 people are now signed up for p6cc. 20:21
wolfman2000 p6cc? Alright, what am I behind on this time? 20:22
tadzik :)
masak wolfman2000: strangelyconsistent.org/blog/the-20...ng-contest
tadzik . o O ( 28 people are now signed up for p6cc )
masak :P
as heard on twitter, "solve great problems, win great books". 20:23
felher 28, wow. :) 20:24
masak well, 27 outside of tadzik's thought bubble :) 20:25
wolfman2000 I'll pass. Sad to say, but I don't know how to solve all of these.
tadzik I don't know either
felher Oh, ah, okay, just 27, thats _much_ less impressive ;P
thou anyone know a blog, etc. interpretation of perlcabal.org/syn/S12.html#Construc...ialization : when to use new, CREATE, BUILD, how they fit together....
20:25 alvis joined
tadzik what fun is it to solve things you know how to solve? :) 20:25
masak moritz and I didn't choose problems that we knew immediately how to solve :) 20:26
tadzik that's something to put in the new perldoc perloo
wolfman2000 fair enough
Still...first thing I need to do is re-get perl 6 on here if I am to even consider it
20:28 molaf joined
masak do that regardless. :) 20:28
20:29 PacoAir left
colomon just got charged $266 for a hammer made out of legos. 20:29
x3nU sounds fun
masak .oO( and every problem looks like a little LEGO nail... ) 20:30
20:30 snearch left
tadzik my first thought was: colomon has built a lego hammer which is illegal, and he got a ticket for that 20:30
I've built a Lego Pirate Ship, I guess they're already after me
good that we didn't build Lego Pirate Bay too 20:31
colomon my little guy appears to be a shrewd businessman. but not much of a hammer-maker, yet.
20:32 packetknife joined
thou e.g., i'm thinking of something like: class Car is Word does Scenery { has $.word = 'car'; ... }; if Word.^find("car") -> $word { say "I recognize $word"; } # Do I implement Word.BUILD to add 'car' to a hash of known words, so I can implement Word.^find()? 20:34
masak tadzik: LEGO Pirate Bay, enabling users to share material using LEGO-bittorrent :)
dang, that pun only works in Swedish... :) 20:35
phenny: sv en "bit"?
phenny masak: "piece" (en to en, translate.google.com)
thou hmmm, now that i spent the time to write that question, i'm thinking i might just want my Thing $car does Scenery .= new(:word('car')); 20:36
tadzik hehe
20:38 pothos left 20:39 pothos_ joined, y3llow_ joined, y3llow left 20:41 y3llow_ left, drbean left, y3llow joined 20:42 pothos_ left, pothos joined
thou still, i'm not sure even in class Thing where to implement storing the word; do I write Thing method new(|$args) { callsame; # Now get :$word out of $args and stuff it into %vocabulary } ? 20:43
i have a feeling i should look at some example code :-)
20:43 y3llow left 20:44 y3llow joined, pothos left 20:45 pothos joined
thou i see masak avoided this question with something like: grass => Grass.new(:name<grass>), 20:46
20:46 PacoAir joined, sudokode left 20:48 drbean joined 20:52 proller joined
masak thou: could also be solved using MOP and reflection, but I didn't want to do pull in that kind of heavy machinery. 20:52
thou i think what i want is simple. i'm going to write a quick example of it in P5. if i can remember how to do that. :-) 20:53
masak :) 20:56
colomon niecza: (2, 45, 6, 1, 3).sort.say 21:00
p6eval niecza v13-47-g3171346: OUTPUT«1 2 3 6 45␤»
colomon niecza: (2, 45, 6, 1, 3).sort(-> $a, $b { $b <=> $a }).say
p6eval niecza v13-47-g3171346: OUTPUT«45 6 3 2 1␤»
21:02 drbean left
colomon niecza: { say $^b; say $^a; }("hello", "world") 21:04
p6eval niecza v13-47-g3171346: OUTPUT«world␤hello␤»
masak niecza: { &^c(&^a, &^b) }("hello", "world", &say) 21:05
p6eval niecza v13-47-g3171346: OUTPUT«helloworld␤»
masak \o/
tadzik nice
21:07 spine left 21:09 drbean joined 21:18 alvis left
sorear good * #perl6 ... for real this time 21:20
masak no-one expects the sorear inquisition! 21:21
sorear masak: I think sleep as a prefix is a horrid fossil
masak sorear: yeah; I don't really see what the fuss is about there.
tadzik it just suprised me, reading the commit 21:22
sorear TimToady: should $*IN exist at compile time? 21:23
jnthn!!!
colomon niecza: (1.1,2,-3.05,0.1,Inf,42,-1e-07,-Inf).sort.say 21:24
p6eval niecza v13-49-ge224a62: OUTPUT«-3.05 Inf -Inf -1E-07 0.1 2 1.1 42␤»
masak sorear: jnthn's gone again... but he'll be back soon. 21:25
colomon niecza: (1.1,2,-3.05,0.1,Inf,42,-1e-07,-Inf).sort.say
p6eval niecza v13-49-ge224a62: OUTPUT«-3.05 Inf -Inf -1E-07 0.1 2 1.1 42␤»
sorear colomon: last time I looked at sorting in niecza, the main issue that came up was that niecza doesn't use a "stable sort" 21:27
colomon sorear: look again at those numbers. that's not a sort at all 21:28
sorear What. 21:31
niecza: say ('a'..'z').pick(*) 21:32
p6eval niecza v13-49-ge224a62: OUTPUT«a v c g n s e f o z d h k l b r t q i y w m p x u j␤»
sorear niecza: say ('a'..'z').pick(*).sort
p6eval niecza v13-49-ge224a62: OUTPUT«a b c d e f g h i j k l m n o p q r s t u v w x y z␤»
colomon -3.05 Inf -Inf -1E-07 is not sensible, either with numeric or string sort
sorear I wonder why Str sorting seems to work 21:34
colomon it seems like half the sorts work
errrr, 2/3rds by the sort.t count 21:35
sorear niecza: say -3.05 cmp Inf
p6eval niecza v13-49-ge224a62: OUTPUT«-Inf␤»
sorear niecza: say -3.05 cmp -Inf
p6eval niecza v13-49-ge224a62: OUTPUT«Inf␤»
sorear niecza: say (-3.05) cmp Inf
p6eval niecza v13-49-ge224a62: OUTPUT«-Inf␤»
colomon that's one thing I was wondering about: is your sort routine okay with values other than -1, 0, and 1 internally?
niecza: say (1..100).pick(5).sort 21:36
p6eval niecza v13-49-ge224a62: OUTPUT«4 24 52 73 98␤»
colomon niecza: say (1..100).pick(5).sort({ $^b <=> $^a }) 21:38
p6eval niecza v13-49-ge224a62: OUTPUT«99 59 52 48 35␤»
colomon niecza: say (1..100).pick(5).sort({ $^b cmp $^a })
p6eval niecza v13-49-ge224a62: OUTPUT«86 62 21 20 13␤»
colomon that's working fine.
sorear looks like it's using (int) internally
so if <cmp> returns 0.2, that gets counted as 0
colomon that would explain a lot
sorear also, C# thinks that (int) double.PositiveInfinity == int.MinValue 21:40
21:40 kaare_ left
colomon that's a bad combination, indeed. 21:40
I just checked the spec, and Comparator says "negative/zero/positive", not -1 0 1, so infix:<cmp> seems okay 21:41
btw, as I read it, zero-argument "comparators" are not allowed. 21:43
the given types listed for Ordering are Comparator (arity 2), KeyExtractor (arity 1), OrderingPair (one of each of the two prior, as a Pair), a Signature, or a Whatever 21:46
sorear niecza: for 0, 0.5, 0.5, :x, 0, :x -> $x, $y { say $x cmp $y } # look, non-transitive 21:47
p6eval niecza v13-49-ge224a62: OUTPUT«-0.5␤Unhandled exception: Attempted to access slot key of type object for Pair␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/lib/CORE.setting line 1381 (Pair.Str @ 3) ␤ at <unknown> line 0 (ExitRunloop @ 0) ␤ at /home/p6eval/niecza/li…
dalek ecza: e6a9ddc | sorear++ | lib/Kernel.cs:
Fix sorting of numbers differing by non-integral amounts
 21:54
ecza: d921607 | sorear++ | lib/CORE.setting:
Tweak cmp to be transitive with mixed number/non-number data
colomon sorear++ 21:57
kshannon BLOODY SEGFAULTS!! AAARRGGHH!!!!!! 22:00
colomon niecza: say { $^a cmp $^b } ~~ Callable 22:01
p6eval niecza v13-49-ge224a62: OUTPUT«Bool::True␤»
22:02 snearch joined
colomon niecza: say { $^b <=> $^a }(1, 10) 22:02
p6eval niecza v13-49-ge224a62: OUTPUT«1␤»
colomon niecza: say { $^a <=> $^b }(1, 10)
masak hugs kshannon
p6eval niecza v13-49-ge224a62: OUTPUT«-1␤»
masak kshannon: if they're reproducible, please report them. 22:03
kshannon It's to do with my exceptions catching patch for VTABLE_find_method for nqp github.com/perl6/nqp/pull/25 22:06
and I think I just found the reason. I need to somehow peel the runloops back to where they were on entry... 22:07
masak kshannon++ 22:09
'night, #perl6
22:09 snearch left 22:10 sayu joined
tadzik g'night masak 22:11
22:12 icwiener_ left
wolfman2000 Hmm...any reason that it looks like the last Rakudo release was around July of last year? I guess I was expecting something more...recent. 22:13
sorear sounds like you might be looking at Rakudo Star
wolfman2000 I am.
sorear Rakudo releases are monthly
Star was something totally different 22:14
wolfman2000 Which do you recommend?
22:15 munchor|afk joined, munchor|afk left
sorear dunno, I've never used star 22:16
tadzik wolfman2000: at the moment I'll recommend the monthly release 22:18
the new Star will probably happen sometime this month
wolfman2000 anyone want to try to recommend I just go straight to source?
colomon going straight to the source works quite well, IMO 22:19
wolfman2000 at least I'm used to working with source 22:20
22:23 proller left 22:24 pothos left, ruoso left, tokuhirom3 left, fhelmberger left, athomason left, TimToady left, ranguard left 22:25 pothos joined, tokuhirom3 joined, ruoso joined, fhelmberger joined, athomason joined, TimToady joined, ranguard joined, pothos left, molaf left
kshannon WooHoo! free_runloop_jump_point FTW! 22:25
22:25 pothos joined 22:29 orafu left 22:30 orafu joined, drbean left 22:35 drbean joined 22:36 packetknife left
wolfman2000 t/01-sanity/99-test-basic.t ........ Dubious, test returned 1 (wstat 256, 0x100) <-- felt I should report it 22:36
kshannon What happens if you run that test by itself? 22:40
wolfman2000 kshannon: I forgot how to run individual tests by themselves. 22:41
kshannon make t/01-sanity/99-test-basic.t
22:41 ruoso left
wolfman2000 pastie.org/3115020 22:42
22:42 orafu left
kshannon Ooops, that's from tadzik-- 5b58e03 Cleanup Test.pm 22:45
wolfman2000 I have plenty to relearn now
22:47 ruoso joined, sayu left
kshannon Fixed. github.com/rakudo/rakudo/pull/51 22:52
wolfman2000 Do I just git pull and then remake? Or do I have to configure again? 22:53
kshannon Well I don't have access to the rakudo repository yet, so you'll have to "git pull git://github.com/KrisShannon/rakudo.git add-back-reasonless-skip" 22:55
tadzik oh, gosh
merging that now
done 22:56
wolfman2000 Perhaps some of this should be explained on the how-to-get-rakudo page
dalek kudo/nom: 076bca2 | (Kris Shannon)++ | lib/Test.pm:
Add back argumentless skip to lib/Test.pm

t/01-sanity/99-test-basic.t breaks without it
kudo/nom: ea03112 | tadzik++ | lib/Test.pm:
Merge pull request #51 from KrisShannon/add-back-reasonless-skip

Add back argumentless skip to lib/Test.pm
wolfman2000 either way, glad I could help
kshannon Heh. OK, now you can just git pull.
no remake needed as it just touches lib/Test.pm 22:57
tadzik which gets compiled to lib/Test.pir
at the end of 'make', so that should be fast 22:58
kshannon make test will automatically rebuild the lib/Test.pir for you.
wolfman2000 all tests successful 23:01
kshannon \o/ 23:02
wolfman2000 I'll relearn parts of the language proper after some evening activities
dalek p: 390af1f | (Kris Shannon)++ | src/pmc/sixmodelobject.pmc:
Catch exceptions in VTABLE find_method

The sixmodel VTABLE find_method calls the find_method in the containers STABLE which will throw instead of returning NULL if it can't find the method.
Anything which is calling the VTABLE find_method is not going to be expecting an exception so catch any exception and return NULL.		 23:13
p: b9c26af | (Kris Shannon)++ | src/pmc/sixmodelobject.pmc:
May also need to peel runloops after catching an exception
23:14 PacoAir_ joined, PacoAir left, PacoAir_ is now known as PacoAir
japhb finally catches up with backlogging 23:14
kshannon That probably should have read "Anything **outside of 6model** which is calling the VTABLE find_method is not going to be expecting an exception so catch any exception and return NULL"
sorear o/ japhb 23:15
kshannon The 6model stuff in nqp and the rakudo nom code all seem to expect VTABLE_find_method to possibly throw. 23:16
The rakudo nom Failure type depends on it for how it works :(
23:17 PacoAir left
japhb sorear, apropos to the discussions re: VM performance, do you feel that you have already found most of the big (> 100%) performance boosts available for any Perl 6-on-mono implementation, or are there still significant areas that you have not researched/implemented? If you've got all the really big ones, what about medium-level (>10%) boosts? Is that still an area ripe with targets? 23:18
sorear japhb: there are definitely specific cases left where there is considerable room for improvement. 23:19
japhb sorear, Ah, fair enough. 23:20
23:20 zby_home left 23:25 molaf joined
colomon ack. I'm an idiot 23:28
23:34 mj41 left 23:38 Mowah left 23:40 Trashlord left 23:47 orafu joined 23:57 cognominal left
thou here's an example of using Moose BUILD() to take action during object creation; how to do this in rakudo nom? 23:58
gist.github.com/1552694
benabik submethod BUILD, IIRC
23:59 cognominal joined
thou i guess i'm not sure how to access the args, and it seems the object isn't created yet when submethod BUILD is called? 23:59
gist.github.com/1552700

Powered by Cro and the Raku® Programming Language.

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

Thank you!