Camelia, the Perl 6 bug

IRC log for #moarvm, 2013-07-16

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

All times shown according to UTC.

Time Nick Message
01:42 FROGGS_ joined #moarvm
01:56 benabik joined #moarvm
02:23 patspam joined #moarvm
02:39 benabik joined #moarvm
04:17 ggoebel2 joined #moarvm
04:46 crab2313 joined #moarvm
04:48 birdwindupbird joined #moarvm
05:38 dalek MoarVM: ee06c36 | jimmy++ | src/io/fileops.c:
05:38 dalek MoarVM: fixed Win32 build with MSVC
05:38 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/ee06c36db1
06:48 FROGGS_ joined #moarvm
06:52 FROGGS JimmyZ++
07:04 dalek MoarVM: 7cf3a8d | jimmy++ | src/io/fileops.c:
07:04 dalek MoarVM: small code cleanup
07:04 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/7cf3a8d26d
07:14 FROGGS JimmyZ: is the bigint stuff I did okay on windows?
07:14 * FROGGS feels like testing on windows too
07:16 JimmyZ FROGGS: It builds
07:16 JimmyZ I didn't run test though
07:17 JimmyZ the bigint test was mostly commented :P
07:19 benabik Makes it easy to pass!
07:19 * benabik should comment out all problem tests.  Would have saved me a few weeks work lately.
07:19 FROGGS JimmyZ: I only have two failing bigint tests using nqp's test file
07:20 FROGGS (due to nqp::box_i)
07:21 * JimmyZ tests
07:23 FROGGS JimmyZ: ahh, und you have to 'use npq-mo' instead of 'use nqpmo'
07:23 FROGGS first line in the test file
07:26 JimmyZ FROGGS++, timely help
07:27 * JimmyZ don't know why use nqpmo worked before
07:28 JimmyZ FROGGS: I get 3 failed test: 17, 18, 20
07:28 JimmyZ not ok 17 - Bit ops (RT 109740) , not ok 18 - can box to a complex type with a P6bigint target, not ok 20 - addition works on boxing type
07:29 FROGGS hmmm, 17 passes here
07:35 JimmyZ FROGGS: 17 test, I got -1 , expect 1
08:32 sorear joined #moarvm
09:12 dalek MoarVM: a300174 | jimmy++ | src/io/fileops.c:
09:12 dalek MoarVM: avoids some memory leak
09:12 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/a3001746e3
09:18 dalek MoarVM: 29ffbdd | jimmy++ | src/io/fileops.c:
09:18 dalek MoarVM: avoids a potential memory leak
09:18 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/29ffbdda37
09:28 JimmyZ FROGGS: looks like it failed because two_complement_* is not ported
09:29 FROGGS hmmm, then why does it work here? weird :o)
09:30 JimmyZ FROGGS: because of different platform and compiler?
09:31 JimmyZ :P
09:33 Util joined #moarvm
09:37 FROGGS hummm, there is no MSVC project file, is there?
09:37 JimmyZ you don't need it
09:37 FROGGS yeah, I see now
09:39 JimmyZ open 'VS2012 x64(x86) Cross Tools Command Prompt', and run perl Configure.pl there
09:42 FROGGS I executed vcvars32
09:43 FROGGS ohhh wow, that is slow
09:43 JimmyZ nmake doesn't support -j, FYI :(
09:44 FROGGS I didnt use -j on my linux for moarvm yet fwiw
09:45 FROGGS but compiling it only takes 13s on my linux box, but compiling a single apr source file seems to take about three seconds
09:46 FROGGS but to be fair: the windows is an xp 32 bit, and the linux is 64bit
09:47 * JimmyZ uses MSVC because apr doesn't support gcc on win32 :)
09:47 FROGGS another reason to get rid of apr :o)
09:48 JimmyZ :P
09:50 FROGGS selfhosting++ # time nqp nqp-moar-cc.nqp -e '1' -> real0m1.852s
09:52 FROGGS nqp@jvm has an identical startup time, and just doing "nqp -e '1'" takes 0.08s
09:52 * JimmyZ is waiting for selfhosting all the time :-)
09:52 FROGGS WHY WAITING???
09:52 FROGGS sorry :P
09:53 JimmyZ because I don't need to debug parrot :P
09:53 JimmyZ after selfhosting
09:54 FROGGS you do the selfhosting, jnthn++ does threads and I add some missing ops... done way before christmas!
09:56 JimmyZ selfhosting is hard to me, it needs some small fixes here and there, and some small fixes is not small actually :P
09:57 JimmyZ i.e: serialization writing
09:58 jnthn Serialization writing isn't so bad in so far as there's a test suite :)
09:58 JimmyZ jnhn: Will hash repr be typed?
09:58 jnthn JimmyZ: Some time in the future, perhaps.
09:59 jnthn JimmyZ: Not until Rakudo development works out how to make use of having that. :)
09:59 JimmyZ jnthn: there are some codes in Serialization writing, depends on typed hash, methinks
10:00 jnthn Yeah, but nothing that leaks outside of there. We don't have to use the hash repr in C code, we can use hashes directly.
10:01 JimmyZ I think I can port some Serialization writing to MoarVM, except typed hash code part
10:02 JimmyZ jnthn: or you will have some refactor in Serialization writing?
10:02 jnthn Not sure just yet...
10:02 * JimmyZ decommutes
10:02 jnthn Probably nothing overly crazy, though.
10:02 diakopter FROGGS: you forgot me in the before-Christmas plan
10:02 jnthn diakopter: p5interop! :P
10:02 FROGGS diakopter: awww
10:02 JimmyZ diakopter: yeah
10:03 diakopter jnthn: speaking of which, I have a hague grant app pending to tpf/karen for that
10:03 colomon joined #moarvm
10:03 FROGGS diakopter: yes, of course, P5@moarvm \o/
10:03 diakopter FROGGS: well, I wasn't thinking just that
10:03 FROGGS dyncall too, what else?
10:04 diakopter well, I was expecting to take on the bulk of replacing apr.. also, finishing nfg
10:04 * jnthn wouldn't mind doing the libuv bit also, fwiw.
10:04 tadzik MOAR the things!
10:04 jnthn NFG++
10:04 FROGGS ohh well, that is much todo till tehn
10:05 FROGGS then*
10:05 jnthn I want to get things bootstrapped on Moar before that, though. ;)
10:05 jnthn (Before I look at libuv things, that is.)
10:05 FROGGS yeah
10:06 diakopter jnthn: karen wants you to approve more of the p5interop design before allocating (essentially the rest of) the hague grant
10:06 jnthn The rest of? :)
10:06 diakopter jnthn: so can we discuss it sometime?
10:06 diakopter jnthn: yeah; there's not much left
10:06 jnthn diakopter: Really? Hm. :)
10:07 jnthn diakopter: Yes, sometime, but probably some Perl 5 person should weigh in too...
10:07 FROGGS lizmat++ might be able to?
10:07 diakopter well I ran the whole design by nwc10, who seemed to handwavey approve,ish, and TimToady has listened to the whole thing several times at least and given some feedback
10:08 diakopter lizmat was going to be the grant manager
10:08 jnthn nwc10++ is who I was pondering suggesting :)
10:08 diakopter but really it's quite straightforward
10:08 diakopter there are a couple of fiddly bits around types and metaobjects
10:09 diakopter but all the basic mechanics really aren't tricky
10:09 diakopter [after a great deal of study of p5 src code and thinking]
10:10 diakopter jnthn: anyway, can we discuss it sometime?
10:11 diakopter I've asked FROGGS too.. so hopefully he and tadzik can participate; sorear said he doesn't have time to help out with it
10:11 jnthn diakopter: Yes, we can
10:12 diakopter let's set a time sometime in the next week|2
10:12 FROGGS diakopter: I can read it if you send me the google doc linky, but I'm not sure if I can be of any help
10:12 tadzik I'm not sure I'll be of much use, other than being an additional pair of ears
10:12 tadzik I once pondered a thought of P5 interop while being drunk or 80% blackbery thing
10:12 tadzik it made perfect sense while being so simple in design
10:13 diakopter FROGGS: the various iterations of the design docs are to the non-slushiness point of it being faster to implement them than to keep them up-to-date and thorough
10:13 tadzik I don't remember a thing
10:13 diakopter tadzik: you're funny this week :D
10:13 tadzik it's because of this python thing
10:13 diakopter regex to filter a googolplex of books
10:14 diakopter jnthn: do you know of a 2-hour block we can reserve in the next week or two? I was hoping to use google hangout or skype
10:15 * diakopter suddenly remembers to install skype on the new computer
10:15 jnthn diakopter: We could do skype
10:15 jnthn Maybe towards end of this week?
10:16 jnthn afaik I'm teaching Mon-Wed next week
10:16 diakopter 8 a.m. PDT is good for me any day
10:16 diakopter er, 7
10:16 diakopter (it's 3:16 a.m. now)
10:16 diakopter -7, I think
10:17 diakopter pick a day and let me know
10:18 diakopter hopefully nwc10, TimToady, lizmat, & others could attend too
10:18 diakopter so that would almost certainly be google hangout
10:19 * diakopter [OT] uses jira-rest-java-client-core-2.0.0-m19.jar to add some custom features to JIRA this week
10:30 FROGGS -9 Berlin time
11:00 _ilbot joined #moarvm
11:00 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
11:37 colomon joined #moarvm
11:56 colomon_ joined #moarvm
11:57 birdwindupbird joined #moarvm
12:12 colomon joined #moarvm
13:06 colomon joined #moarvm
15:10 diakopter jnthn: ping
15:14 diakopter .tell jnthn see http://www.aqute.biz/Bnd/Bnd for ease of automating jar bundling/creation (it's apache licensed). Also clsd (class dumper) from the same toolkit.
15:14 yoleaux diakopter: I'll pass your message to jnthn.
16:13 FROGGS joined #moarvm
17:10 flussence joined #moarvm
17:10 diakopter joined #moarvm
17:20 benabik joined #moarvm
17:30 diakopter jnthn: ping
17:34 FROGGS PARROT VM: Could not load bytecode
17:34 FROGGS Could not load oplib `nqp_ops'
17:34 FROGGS >.<
17:35 diakopter happens
17:35 jnthn diakopter: o/
17:35 jnthn cooking, but it's easy so aboutish :)
17:36 diakopter jnthn: now I forgot what I was going to ask
17:36 diakopter ergh.
17:36 * diakopter tries to remember
17:37 FROGGS uhh, works now
17:37 FROGGS yeah, "nmake install" is a step one should not skip
17:38 FROGGS seriously, I can't image developing on windows...
17:38 FROGGS imagine*
17:40 colomon joined #moarvm
17:49 diakopter FROGGS: it's AWESOME
17:50 FROGGS no
17:50 jnthn Yeah, nobody who develops in Windows can be productive. :)
17:50 FROGGS yre you guys using bash or something similar? the usual command line just sucks
17:50 diakopter <- no exception
17:50 diakopter FROGGS: well, there's Powershell, which has been around for many years now
17:51 FROGGS jnthn: I was a windows dev till 2010 or so
17:51 diakopter has many Perl features
17:51 diakopter and list comprehension thingies
17:51 * FROGGS is stuck on windows xp
17:51 diakopter and pipes
17:51 diakopter there's a version of powershell that installs on XP I'm pretty sure
17:52 benabik http://www.johndcook.com/blog/2009/11/​02/powershell-2-0-for-windows-xp-etc/  (dunno if there's a 3.0 or whatever)
17:53 FROGGS I will try that
17:53 benabik Oh hey, guess there is.  http://blogs.msdn.com/b/powershell/archive/​2008/12/30/download-windows-powershell.aspx
17:53 FROGGS thanks guys
17:53 diakopter FROGGS: actually I wouldn't recommend it
17:53 diakopter it's almost entirely useful mostly for sysadmin things
17:53 diakopter not really for data-y things
17:53 FROGGS hmm
17:53 benabik PowerShell is way better than cmd even just for running things.
17:53 diakopter heh.
17:54 benabik dunno about scripting, but I'd rather have a psh window than cmd window to type in.
17:55 diakopter benabik: sounds like you use it far more than I have... ;)
17:55 benabik diakopter: It's almost like a real shell.  Kept me from having to install all of cygwin or something stupid when I was trying to test Parrot on Windows.
17:57 diakopter sigils and all?
17:58 benabik Like I said, have tried exactly no scripting with it.  But it has all the nice little completions and editing and such I needed to stay sane.
17:59 FROGGS well, it start when you want to copy the output of a script, in a bash you dont have implicit linebreaks within your string when it is longer than the window buffer
18:01 FROGGS and that you cant maximize the cmd window, you have to calculate the number of chars on your own
18:02 FROGGS is it still like that on windows 7/8 ?
18:02 diakopter you can change the cols/rows, via the gui
18:04 diakopter jnthn: oh yeah, I remembered
18:04 FROGGS yeah, but you have to calculate those rows/cols on your own first, that is the odd thing
18:04 FROGGS E:\MoarVM\nqp-cc>nqp\nqp nqp-moar-cc.nqp -e 'say(1)'
18:04 FROGGS Malformed UTF-8 string
18:04 FROGGS current instr.: 'slurp' pc 657 (src\stage2\gen\NQPCORE.setting.pir:302) (src\stage2\NQPCORE.setting:647)
18:04 FROGGS :/
18:05 FROGGS has somebody else a 32bit os?
18:06 diakopter FROGGS: erm cmd.exe doesn't use '
18:06 diakopter it uses "
18:07 FROGGS E:\MoarVM\nqp-cc>nqp\nqp nqp-moar-cc.nqp -e "say(1)"
18:07 FROGGS Malformed UTF-8 string
18:07 FROGGS current instr.: 'slurp' pc 657 (src\stage2\gen\NQPCORE.setting.pir:302) (src\stage2\NQPCORE.setting:647)
18:07 FROGGS ó.ò
18:07 diakopter you can't run nqp that way
18:07 diakopter you have to add that dir to the path
18:08 FROGGS E:\MoarVM\nqp-cc>nqp\nqp -e "say(1)"
18:08 FROGGS 1
18:08 diakopter oh.
18:08 FROGGS well, I can change tha path to include nqp's dir
18:10 FROGGS okay, path for comfort, but didnt help my problem
18:10 benabik psh also acts as a modern terminal emulator, IIRC.  Resizing windows and highlighting and everything.
18:10 FROGGS I'd guess that the byte ops for utf8 might be broken on 32bit
18:11 FROGGS that would be cool!
18:16 diakopter FROGGS: I think you'd have to add the install\bin to the path, not the root of the nqp checkout
18:17 FROGGS k, will test in a bit
18:20 diakopter FROGGS: which C compiler are you using
18:22 diakopter jnthn: anyway, I was going to ask whether you'd seen the other stuff in the dyncall branch - the realclean target I mean
18:24 FROGGS diakopter: cl
18:25 diakopter FROGGS: I haven't tested nqp/parrot on 32-bit cl in years
18:25 diakopter (and I doubt anyone else has either)
18:31 FROGGS well, I had a working rakudo from 2012.10 or so
18:32 FROGGS and nqp is working, but not the cross compiler, I'll trak it down when my kids are sleeping
18:54 jnthn diakopter: not yet
19:02 dalek MoarVM/dyncall: 3a1c4cf | diakopter++ | / (3 files):
19:02 dalek MoarVM/dyncall: make dyncall configure respect verbose flag
19:02 dalek MoarVM/dyncall: review: https://github.com/MoarVM/MoarVM/commit/3a1c4cf653
19:16 FROGGS HAHA
19:16 FROGGS I know now why nqp's slurp fails
19:17 benabik No lips?
19:17 FROGGS it tries to slurp a file called temp.output, which contains the output of "del /?", with german umlauts
19:17 jnthn No straw?
19:17 jnthn hah
19:17 FROGGS hehe
19:17 FROGGS how comes that?
19:18 FROGGS there is a temp.moarvm too
19:18 jnthn dunno, sounds odd
19:19 FROGGS yeah
19:23 FROGGS jnthn: MoarVM_/nqp-cc/nqp-moar-cc.nqp:68:            pir::spawnw__Is("del /? >temp.output 2>&1");
19:23 diakopter what about it
19:24 diakopter heh.
19:24 FROGGS it generates a file non utf8 conform file
19:24 diakopter need to make it find the Command Extensions string in german too :)
19:25 diakopter FROGGS: run this at the command prompt:
19:25 diakopter chcp 65001
19:25 FROGGS diakopter: https://gist.github.com/FR​OGGS/2e014a3aa692d980a4b4
19:26 diakopter FROGGS: translate?
19:27 FROGGS command ".." not found
19:27 diakopter I don't understand the output there
19:27 diakopter why is it saying >>temp.output
19:28 diakopter on the second line
19:28 FROGGS diakopter: debug output I added
19:28 FROGGS btw, the temp.output still has the german text which does not contain the word "Extensions"
19:29 diakopter that should be easy to fix :P
19:29 FROGGS diakopter: what exactly should it do?
19:29 diakopter it's just checking to see if on windows
19:30 diakopter there wasn't a better way when that was written
19:30 diakopter feel free to make it use the right way if there's one now
19:30 FROGGS ohh
19:30 FROGGS that is a weird way to do that
19:30 FROGGS parrot should be able to answer that question, no?
19:30 diakopter I didn't know how at the time
19:31 FROGGS hmmm, okay, will investigate
19:32 diakopter FROGGS: ... but at least that way is backend-agnostic
19:34 FROGGS it does a pir::spawnw__Is
19:34 diakopter ohlol.
19:34 diakopter nm.
19:35 FROGGS I'll try to provide something parrot-ish and JVM-ish
19:35 diakopter search for Extensions in the repo; it's a couple other places too
19:38 FROGGS k
19:38 FROGGS thanks, will do
19:43 FROGGS E:\MoarVM\nqp-cc>nqp -e "my %conf := pir::getinterp__P()[pir::co​nst::IGLOBALS_CONFIG_HASH]; say(%conf<platform>);"
19:43 FROGGS win32
19:43 FROGGS diakopter: do you get the same?
19:44 diakopter FROGGS: I .... actually don't have nqp built on this windows machine yet
19:44 FROGGS k, np
19:48 diakopter chipdude_: ping
19:50 FROGGS yay, it works
20:09 diakopter jnthn: are rakudo-jvm's collections threadsafe?
20:12 diakopter jnthn: (if I had two threads competing to push/pop 10,000 times each (starting with 10,000 elements), would it always come out as size 10,000 in the end?
20:12 diakopter )
20:12 FROGGS I cant add fudge lines to nqp-mo-cc.nqp
20:12 FROGGS :/
20:14 jnthn diakopter: no
20:15 jnthn diakopter: Things are at a *very* early point :)
20:15 FROGGS jnthn: the fudging happens when concatenating the source files?
20:16 FROGGS yeah, looks like
20:16 * diakopter would bet $500 that the speed of that test would not be detectably affected by adding a java fast lock to those reprs
20:16 diakopter ... up to the number of cores, ish
20:24 diakopter jnthn: I mean putting this inside each of the repr ops: synchronized(this) { everything else }
20:25 diakopter (obviously doesn't scale at all)
20:26 prammer joined #moarvm
20:29 jnthn diakopter: aye, though I suspect it may be a bit more involved given that Perl 6 arrays are more than just a VMArray :)
20:29 jnthn oh, that reminds me of a design question for TimToady++ that I need to write up coherently...
20:29 diakopter hrm, I guess you'd need an explicit lock
20:31 FROGGS gah, I just add the parrot code as it is and put the jvm version commented underneath
21:02 dalek MoarVM: 6b55554 | (Tobias Leich)++ | nqp-cc/ (2 files):
21:03 dalek MoarVM: query vm for osname
21:03 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/6b55554fc3
22:47 FROGGS joined #moarvm
22:47 dalek joined #moarvm
22:47 sorear joined #moarvm
22:47 ggoebel2 joined #moarvm
22:47 patspam joined #moarvm
22:47 lizmat joined #moarvm
22:47 PerlJam joined #moarvm
22:47 tadzik joined #moarvm
22:47 jlaire joined #moarvm
22:47 itz joined #moarvm
22:47 TimToady joined #moarvm
22:47 ChanServ joined #moarvm
22:47 chipdude_ joined #moarvm
22:47 eternaleye joined #moarvm
22:47 rblackwe joined #moarvm
22:47 hoelzro joined #moarvm
22:47 jnthn joined #moarvm
22:47 Tene joined #moarvm
22:47 yoleaux joined #moarvm
22:47 JimmyZ joined #moarvm
22:47 timotimo joined #moarvm
22:47 arnsholt joined #moarvm
22:47 harrow` joined #moarvm
22:47 felher joined #moarvm
22:47 BinGOs joined #moarvm
22:47 __sri joined #moarvm
23:37 benabik joined #moarvm

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