github.com/moarvm/moarvm | IRC logs at irclog.perlgeek.de/moarvm/today
Set by moderator on 31 May 2013.
00:11 lee__ joined 02:22 FROGGS joined 04:16 japhb_ joined 06:10 `patch` joined 07:47 cognominal joined 09:00 tgt joined 09:24 Guest1337 joined
nwc10 for reasons I don't understand (so probably a 1 char error) the most recent commit which changes to dynamic libraries has the side effect of messing up the APR configuration such that apr.h isn't built 09:55
(more than one Linux distribution, and it's not because apr.h is installed in /usr/include) 09:56
10:00 cognominal joined 10:08 tomyan joined
nwc10 t/qast/qast_arguments.t fails tests 10 and 12 on 32 bit Linux (at least little endian) 10:35
given that test 10 fails because it gets 562949953421337, when it expected 25, and
$ perl -le 'printf "%x\\n", 562949953421337'
2000000000019
10:35 timotimo joined
nwc10 it's something to do with reading 64 bits, when only 32 were actually written 10:36
(and possibly the other 32 bits were part of some other value)
timotimo i'm having build trouble; apparently the location of apr.h is different in debian from others; mine is in /usr/include/apr-1.0/apr.h and the build script doesn't expect that. suggested fix? 10:39
nwc10 APR is bundled
on HEAD, something's a bit borked
timotimo is there a known good commit for me to try out?
nwc10 change into ./3rdparty/apr and run ./configure there
19a12faab4164b1412d65bdf7480af570476790d might be better 10:40
but everywhere I've built, APR isn't installed by the system
eternaleye nwc10: `apr-1-config --includes` will give the proper include path for apr
nwc10 so I don't know if the MoarVM build system can cope with a machine where APR is installed, particularly if it's a "wrong" version
eternaleye nwc10: apr doesn't use pkg-config yet, more's the pity 10:41
nwc10: There's also --version, so we can check if it's the right one.
nwc10: In general, if you want to find build deps, try pkg-config first, then specific ${foo}-config tools, then hardcoding various locations 10:43
Lessons from packaging and admin.
(If you're on CMake or autotools things change slightly, though)
timotimo i was able to make it after configuring apr 10:50
make: *** [NQPP5QRegex.pbc] Segmentation fault 11:11
whoops?
(nqp-cc/Configure.pl --gen-nqp)
11:35 BrowserUk joined
BrowserUk What are the "essential units" of P6 code? 11:36
timotimo bells and whistles 11:45
nwc10 I think everyone who can answer that question sensibly is in Austin Texas, and probably asleep 11:48
BrowserUk nwc10: I'll leave my machine up for a few hours and see what comes. 11:50
nwc10 also, I'm failing to think of my own flippant funny answer :-(
eternaleye BrowserUk: I'd say the essential *nature* is a combination of "Different things should look different, similar things should look similar", "There's more than one way to do it", and "Difficult things should be easy, impossible things should merely be difficult." - That is, exactly the same as Perl 5, but with a few more years of experience behind it. 11:53
timotimo you should deny that bells and whistles make up all of Perl6 and suggest something like nooks and crannies instead
nwc10 I was thinking about "line noise, but now in Unicode" 11:54
eternaleye timotimo: Not touching *that* one with a barge pole, to paraphrase an old ASR post :P
nwc10 but I believe that the context of the question is about what low level OPs a Perl 6 VM should be built from 11:55
timotimo i interpreted it like nwc10 did
but instead of being helpful i decided to say something silly 11:56
eternaleye Well, in that case I'd say either "Basic math, string handling, platform callout capability, and control flow" or "About the same as any other language VM"
moritz t/moar/threads.t fails test 6 here 12:08
is that known/expected? 12:09
nwc10 moritz: I found that it failed 2 tests when build "optimised" and not when not 12:11
12:11 colomon joined 12:27 tgt joined
BrowserUk How about: Blocks & Expressions. 12:32
OR evenL Blocks of expressions. 12:33
12:54 lizmat joined 13:34 tomyan joined 13:36 grondilu joined
BrowserUk Of course, expressions come down to operators and arguments -- which as P6 internally is OO -- translates to objects & methods ( and arguments; which or more objects.) 13:42
But blocks can contain other blocks. Conditional block; looping blocks; bare blocks; 13:44
But what if a block is just another object (say of class Code). 13:45
Then an "if statement" is just a call to the 'if' method of the controlled block; passing another block that is the condition argument. 13:46
grondilu In file included from src/moarvm.h:5:0, from src/core/args.c:1: 13:47
3rdparty/apr/include/apr_general.h:28:17: fatal error: apr.h: No such file or directory
compilation terminated.
make: *** [src/core/args.o] Error 1
I guess I should install the apache portable runtime, right? 13:48
BrowserUk Ditto for a While Loop(1 argument block); and a for loop (3 argument blocks); and a foreach loop ( 1 argument block)
nwc10 grondilu: try cd 3rdparty/apr && ./configure 13:49
something is messing up the configuration
BrowserUk So, expressions are just (lists of) objects+methods+arguments; and blocks are the same.
So the VM opcodes could be just those required to create classes; add methods and invoke them. 13:50
grondilu is running cd 3rdparty/apr && ./configure 13:51
./configure ended with no error 13:52
nwc10 make should work
grondilu runs make again (in the top directory) 13:53
moritz I have a grammar fix. CAN HAZ COMMIT BIT PLZ?
(or do you folks prefer pull requests?)
eternaleye nwc10: One thing I'd *really* like would be --with-apr={system,bundled}, because if anyone ends up packaging this for my distro it'll be me. And bundled libs are an abomination. 13:54
grondilu ld failed now
/usr/bin/ld: cannot find -lapr-1
collect2: error: ld returned 1 exit status
make: *** [moarvm] Error 1
I don't have APR installed (I don't think so anyway). Shouldn't the configure script have dectected that? 13:55
grondilu checks apt-cache policy libapr1
libapr1: Installed: (none) Candidate: 1.4.6-4
nwc10 eternaleye: I don't have commit access. I'm just the one awake. I guess jnthn and a couple of others would need to be "persuaded", but I suspect that they'll agree 13:56
for all I know someone will do it at the hackathon this week
eternaleye Mm 13:58
pmichaud good moarning 14:07
moritz good morning, pmichmoar
pmichaud (apr) I'll be pushing/trying really hard to come up with a system where apr isn't bundled.
I'm thinking I'll do a similar thing in the nqp repo... bundling other packages in source repos is really not a good idea. We can do better. 14:08
grondilu no sure is related but the configure script in 3rdparty/apr does not parse my os version correctly
3.9-2.towo-siduction-686
moritz can't parse that either 14:09
grondilu ^ uname -r
grondilu is running siduction (siduction.org)
(sid-derived OS) 14:10
pmichaud moritz: github is down at the moment 14:11
moritz pmichaud: no hurry 14:12
pmichaud angry unicorns all over the place.
14:17 colomon joined 14:18 rurban joined 14:21 icovnik joined 14:22 birdwindupbird joined
grondilu line 18844, I replaced the sed command with s/\\(.\\)\\.\\(.\\)[\\.-]\\(.).*/\\1\\2\\3/ 14:23
instead of s/\\(.\\)\\.\\(.\\)\\.\\(.).*/\\1\\2\\3/ 14:24
moritz are there any moarvm LHFs? 14:28
colomon a question many of us at the hackathon want to know. :) 14:30
14:35 FROGGS joined 14:50 hoelzro joined 15:00 japhb_ joined
jnthn morning o/ 15:09
whoa, lots to backog here 15:10
FROGGS moarning *g* 15:28
15:28 lizmat joined
moritz github is back up agian... CAN HAZ COMMIT BIT PULEZ? kthx! 15:33
masak JimmyZ++ # github.com/MoarVM/MoarVM/pull/1 15:35
pmichaud moritz: you have commitbit 15:40
FROGGS jnthn: btw, you dont have to care about build problems, I have experience with cross platform stuff due to my p5 SDL module
jnthn FROGGS: You're offering to improve the build things? :) \\o/
Nice :)
pmichaud does this mean I don't have to do it? ;-)
FROGGS well, I usually have 10 or more VMs for different OSes to do sometesting 15:41
jnthn +1 if somebody wants to implement --with-apr={system,bundled} 15:47
(still backlogging...)
FROGGS k 15:52
jnthn: what should be the default?
pmichaud I don't like that formulation 15:55
by default it should use the system apr
but there may also want to be a --gen-apr, which means to generate one locally 15:56
similar to how we do --gen-nqp and --gen-parrot
iirc, --with-xyz is supposed to provide the location of xyz
checking
oh, the standard also allows keywords to be there 15:58
16:11 dalek joined
moritz now if somebody adds feather3.perl6.nl:5000/dalek?t=freenode,moarvm to a github post-commit hook, dalek should report commits in here 16:12
jnthn I think I maybve already did 16:15
checking
masak .oO( dalek me maybe ) 16:16
jnthn ah, I'll hit the tst button
dalek arVM: 036ce9c | jimmy++ | src/6model/reprs/P6num.c:
fixed copy-pasto
arVM: b91d79f | (Carl Mäsak)++ | src/6model/reprs/P6num.c:
Merge pull request #1 from zhuomingliang/patch-1

fixed copy-pasto
arVM: 4bafb0a | moritz++ | docs/arg-passing.markdown:
[docs] grammar fix
jnthn yay
masak \\o/
16:23 vmspb joined
pmichaud I'd be okay with --with-apr=gen then, where "gen" means download an appropriate version of apr 16:42
jnthn pmichaud: You really want it out of the repo? :) 16:43
pmichaud out of the main repo, yes. 16:44
jnthn ok
pmichaud I'm thinking we should handle 3rdparty libs similar to how we currently handle parrot/nqp in nqp/rakudo
i.e., if you have it, use it, otherwise fetch from a known location and build real quick
16:50 Guest1337 joined
pmichaud but somehow I feel like --gen-apr is more consistent with --gen-parrot and --gen-nqp 16:56
FROGGS true 17:02
jnthn Works for me. 17:03
17:30 tomyan joined
pmichaud also, --gen-apr might even allow for version specifiers, similar to what we do with the other --gen-*'s 17:31
FROGGS yeah, good point 17:44
I'll look into it these days
jnthn ++FROGGS 17:45
FROGGS I made an issue: github.com/MoarVM/MoarVM/issues/2 17:48
colomon github down again? I got a unicorn with that link, and now ... nope, third try worked to pull my source. 17:50
still get a unicorn for FROGGS issue... 17:52
FROGGS O.o 17:54
everything fine for me
masak github keeps coming and going, it seems. 17:55
17:59 BrowserUk left 18:08 colomon joined 18:30 FROGGS joined 20:08 BrowserUk joined 20:10 BrowserUk left
timotimo i'm still getting a segfault when trying to --gen-nqp 20:11
jnthn timotimo: Do you have an old install hanging around?
timotimo might be, yes 20:12
well, very probably
can i use rak-jvm-support for regular parrot rakudo?
jnthn timotimo: You can build a Parrot Rakudo provided you use the jvm-support branch of Rakudo 20:13
I'm gonna try and get these merged soon
timotimo neat. 20:14
20:57 tomyan joined 20:58 eiro joined
eiro hello everyone 20:59
colomon o/
20:59 arnsholt joined
jnthn o/ 21:09
dalek arVM: 1b4bd2d | (Marc Chantreux)++ | .gitignore:
git ignore tags file
 21:23
arVM: 74e2c80 | (Marc Chantreux)++ | build/Config/APR.pm:
build the apr lib even in linux
arVM: eca8ac5 | jonathan++ | / (2 files):
Merge pull request #3 from eiro/master

now compile under linux
21:30 Guest1337 joined 22:09 BrowserUk joined 22:18 Guest1337 joined 22:29 labster joined, tomyan joined 23:02 BrowserUk left
timotimo it seems i had to get a newer parrot to make the nqp cross compiler compile 23:05
er, make nqp compile
23:18 colomon joined 23:34 FROGGS joined 23:42 japhb_ joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!