github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today
Set by moderator on 13 July 2013.
diakopter jnthn: hi 00:21
jnthn diakopter: hi 00:22
03:35 benabik joined 08:02 FROGGS joined
FROGGS o/ 08:03
08:23 Util joined 08:24 tadzik joined, dalek joined 08:25 PerlJam joined 08:38 JimmyZ_ joined
JimmyZ_ Good moarvming, FROGGS 08:38
FROGGS hi JimmyZ 08:39
dalek arVM: a8f3943 | (Tobias Leich)++ | nqp-cc/.gitignore:
fix .gitignore, dir nqp-cc was somehow included
arVM: b5e4486 | (Tobias Leich)++ | nqp-cc/t/nqp/ (8 files):
pull in nqp-test updates that pass already
JimmyZ_ hello
FROGGS JimmyZ_: btw, I'm not sure what I should change for s/setencoding/set_encoding/..., there is getstdin, which has no underscore 08:40
JimmyZ_ FROGGS: I meant in fileops.[h|c], not the opcode 08:41
FROGGS only the C function? okay
JimmyZ_ FROGGS: yeah :P 08:42
the opcode and C function has different consistency ... 08:44
oh, nice, 05-comments.t passed magically, last I checked it's not and I don't know why :D 08:46
08:49 patspam joined
dalek arVM: f55a54c | (Tobias Leich)++ | src/ (3 files):
consistent naming, JimmyZ++
09:00
FROGGS ohh okay, I need several math ops in order to include sprintf... 09:01
JimmyZ_ :D 09:03
jnthn moarning o/ 10:01
JimmyZ_ moarvming, jnthn :P 10:03
FROGGS hi jnthn 10:05
10:35 colomon joined
FROGGS jnthn: for nqp::open($fn, $mode), it looks like I cant do the $mode translation at compile time, because the value of a variable wouldnt be known 10:52
jnthn: so I do the translation in the C function, okay?
jnthn ok 11:09
12:36 birdwindupbird joined
dalek arVM: e12cbc7 | (Tobias Leich)++ | / (12 files):
open_fh takes 2args like nqp::open()

Before it tried to translate the mode (str->int) at compile time, which doesnt work well when passing a variable.
12:43
arVM: c6ff08b | (Tobias Leich)++ | nqp-cc/nqp-src/NQPCORE.setting:
pull in everything from src/core/IO.pm
arVM: 767ae54 | (Tobias Leich)++ | nqp-cc/t/nqp/63-slurp.t:
we have lexical &slurp now, so run its test
12:55
jnthn \\o/ 12:56
FROGGS MoarVM is fun somehow 13:07
jnthn :)
jnthn will give it some more tuits once he's back from this vacation :) 13:08
FROGGS I'll care about the missing math ops that are needed for sprintf 13:10
dalek arVM: 8cefd22 | (Tobias Leich)++ | / (2 files):
added abs_n
13:16
diakopter I had a dream I saw a book on MoarVM at the bookstore (just published), but I opened it and it was about another VM that chose the same name 13:49
FROGGS :/ 13:55
JimmyZ_ :P 14:02
dalek arVM: e107233 | (Tobias Leich)++ | / (6 files):
fixed abs_n and added ceil_n and floor_n
15:40
16:04 ggoebel joined
diakopter FROGGS: no likey // 16:04
FROGGS: weird, e107233 isn't showing up on the web 16:05
er, or, my web browser broke. 16:06
timotimo yesterday i had to wait ~5 minutes for github to show my new branch 16:16
FROGGS yeah, few hours ago it had a hickup too 16:24
dalek arVM: 26e2670 | (Tobias Leich)++ | nqp-cc/src/QASTOperationsMAST.nqp:
mapped nqp::not_i
16:38
arVM: bcd772a | (Tobias Leich)++ | nqp-cc/src/QASTOperationsMAST.nqp:
correctly map nqp::neg_n
17:17
diakopter FROGGS: would you remove the // comments 17:58
FROGGS ohh, yes, of course
diakopter: sorry, these shouldnt be there 17:59
diakopter yah; that took a few reminders from jnthn before I stopped that :D 18:00
dalek arVM: 366fade | (Tobias Leich)++ | src/core/interp.c:
removed pointless comments, diakopter++
18:02
diakopter jnthn: someday when the optimizer gets good at statically resolving/JITing things, do you agree we can convert a *ton* of the special reprs to p6opaque (and their hundreds of nqp:: ops to methods)? 18:21
CFunction methods perhaps, but still. 18:22
it's just that classes with methods are so much easier to mentally hold than a large flat namespace of ops 18:23
[even if those classes are built up in C code binding directly to those CFunctions] 18:24
dalek arVM/dyncall: 7114f57 | diakopter++ | / (3 files):
initial attempt. fail.
18:25
diakopter oh you know what. 18:26
benabik .WHAT #?
diakopter :) trying to figure out why I'm failing so badly at Makefile 18:28
18:50 cognominal joined
arnsholt Because the syntax is bananas and the functionality a bit weird at times? =) 18:54
dalek arVM: d321e68 | (Tobias Leich)++ | / (2 files):
fix split() so that it gathers empty strings

This handles delimiters at start and end of string too.
18:56
arVM: 4577ad3 | (Tobias Leich)++ | nqp-cc/t/nqp/59-nqpop.t:
added 59-nqpop.t, only nqp::existspos is NYI
18:58
19:29 benabik joined 19:30 lizmat joined
jnthn diakopter: Well, the point of the ops is that they're REPR-specific, not type specific... 19:43
diakopter I was suggesting making "sealed" types for them 19:45
but nm; :) 19:46
jnthn
.oO( why do people think methods are a magically better way of calling a piece of VM code than functions... )
diakopter I tried to answer that immediately after my question ;) 19:47
dalek arVM/dyncall: a174449 | diakopter++ | / (5 files):
add verbose option to configure; decrease default output; make dyncall build
21:11
diakopter jnthn: yay, got dyncall building/linking
oh, also added a realclean target 21:12
dalek arVM: 3ed32f9 | (Tobias Leich)++ | / (7 files):
added pow_I
21:15
diakopter FROGGS: could you or someone test the dyncall branch on linux/mac
(just to see if dyncall configures/builds) 21:16
FROGGS will do it now
diakopter ooo cool :)
benabik OS X configure works, building now 21:17
diakopter o_O
FROGGS is that expected? Configuring dyncall ...................................* configure package dyncall 21:18
diakopter also I added a realclean target and --verbose flag to configure; maybe test those too? :)
FROGGS: prob not
benabik Yeah, also got that here.
FROGGS there are fours lines after that, like guessing os, then the last line OK
so maybe it is just a bit verbose 21:19
benabik Hm. I think the compile completed, but I don't think dyncall compiled? Trying after realclean
diakopter benabik: , FROGGS: thanks for testing :)
benabik cd 3rdparty/libtommath && @clang, @clang not found. 21:20
diakopter oh hm.
er
benabik (Trying again in master to see if that's broken too.)
FROGGS looks like dyncall got configured but wasnt made
diakopter it won't be
benabik: it won''t be broken there 21:21
benabik diakopter: Sure, probably not, but it's worth checking. Been a while since I've built.
FROGGS cd 3rdparty/libtommath && @gcc -c -o bncore.o -I. -D_REENTRANT -D_LARGEFILE64_SOURCE -Wparentheses -Wreturn-type -g bncore.c 21:22
/bin/sh: 1: @gcc: not found
dalek arVM/dyncall: 10da99f | diakopter++ | build/Config/BuildEnvironment.pm:
oops.
FROGGS ohh, yeah
benabik Also getting a pile of warnings. Maybe I should clean those up someday. Tends to make real errors harder to see.
diakopter FROGGS: try now
benabik: try now
FROGGS that is better
benabik Yes, compile completed. 21:23
FROGGS it made tommath and apr, but not dyncall
benabik Yeah, looks that way here too. 21:24
diakopter hrm
benabik Did you remove echoing everything from the Makefile? I see warnings, but no indications of what command generated them.
FROGGS gist.github.com/FROGGS/bab23a1536e983c610fb
diakopter benabik: herm
benabik: hrm
didn't mean to suppress any output on *n*x
benabik Maybe I should try pulling in Git's quieter makefile stuff. Although I don't know how that would work out in Windows. :-/ 21:25
benabik prefer lines of "CC <file>" and "LINK <file" instead of piles of "clang -piles -of -options filename -options other file"
diakopter well I tried with the --verbose option
arnsholt benabik: That's true, but occasionally it's useful to see the options being used as well 21:26
benabik arnsholt: Well, it's generally easier to enable them when needed than dig through them when it's not. Git allows "make V=1" to go back to the original way. 21:27
arnsholt True dat 21:28
benabik But it's all preference. I tried to put it in Parrot's Makefile and it got rejected.
FROGGS bed time for me.... gnight all 21:29
arnsholt Well, it might be worth discussing with jnthn
As long as we have V=1 or something similar I think I'd back it 21:30
benabik Eh. It's not a lot of work, so I'd probably just write it and then ask jnthn or someone else to try it on Windows. 21:31
diakopter oops :)
benabik To quote my mother: "What do you mean, oops?"
dalek arVM/dyncall: dd8cfd9 | diakopter++ | build/Makefile.in:
another oops
21:32
diakopter erm
hang on
benabik "make: *** No rule to make target `3rdparty/dyncall/dyncall/libdyncall_s.lib', needed by `moarvm'. Stop." 21:33
diakopter I said hang on :P
dalek arVM/dyncall: 8c608f3 | diakopter++ | build/Makefile.in:
another oops
diakopter ergh 21:34
oh heh. 21:35
benabik :-D "make: *** No rule to make target `3rdparty/dyncall/dyncallback/libdyncallback_s.lib', needed by `moarvm'. Stop."
dalek arVM/dyncall: f37dc6a | diakopter++ | build/ (2 files):
lib exts
21:38
benabik Link line for moarvm that includes libdyncall* stuff. :-) 21:39
diakopter it's not there?
dalek arVM/dyncall: b452f56 | diakopter++ | build/ (2 files):
more libext
21:41
diakopter (it is for me on linux)
benabik It is there. 21:45
dalek arVM/dyncall: 01460a5 | diakopter++ | .gitignore:
fix realclean
21:46
benabik Erm. Did you make realclean use `git clean`? 21:50
diakopter well, now it works on linux, but not on windows
benabik: why?
benabik diakopter: Your commit says "fix realclean" and modifies .gitignore.
diakopter oh heh 21:51
meant the other thing
benabik Ah, 'kay.
diakopter hrm. 21:58
there. 22:26
23:22 lizmat joined