Camelia, the Perl 6 bug

IRC log for #moarvm, 2013-08-08

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

All times shown according to UTC.

Time Nick Message
00:31 FROGGS joined #moarvm
01:14 ggoebel joined #moarvm
01:38 benabik joined #moarvm
01:42 Alpha64_ joined #moarvm
01:45 Alpha64__ joined #moarvm
02:07 FROGGS joined #moarvm
02:21 Alpha64__ left #moarvm
02:34 dalek MoarVM/readlineintfh2: d3a19e8 | jimmy++ | 3rdparty/linenoise/linenoise.c:
02:34 dalek MoarVM/readlineintfh2: code cleanups
02:34 dalek MoarVM/readlineintfh2: review: https://github.com/MoarVM/MoarVM/commit/d3a19e8f00
04:41 birdwindupbird joined #moarvm
06:02 * diakopter is not surviving refactoring hell
06:11 JimmyZ aloha
06:14 FROGGS joined #moarvm
06:15 JimmyZ FROGGS: good morning
06:19 JimmyZ I take an look at BuildEnvironment.pm on Moarvm and uv.gyp, and think it's easy to create BuildEnvironment.pm file for libuv with support linux and windows
06:19 diakopter JimmyZ: actually
06:19 diakopter segomos is working on that
06:20 diakopter I forgot to mention
06:20 JimmyZ oh
06:20 JimmyZ great
06:20 diakopter hopefully he will submit a pull request soon
06:20 JimmyZ e
06:20 diakopter segomos: *pull request to a new branch
06:21 JimmyZ yeah, uv.gyp is about 500 lines
06:23 JimmyZ it' aslo create some .proj or .vcxproj files, but I really don't like it
06:23 JimmyZ I like Makefile :P
06:36 diakopter Method 'ws' not found for invocant of class 'QAST::MASTRegexCompiler'
06:39 diakopter could someone implement ws?
06:42 JimmyZ diakopter: you can, only one line
06:42 diakopter paste it? :)
06:42 JimmyZ # a :rxtype<ws> node is a normal subrule call
06:43 JimmyZ method ws($node) { self.subrule($node) }
06:44 diakopter ok
06:44 diakopter thx
06:44 JimmyZ np
06:46 diakopter Method 'dba' not found for invocant of class 'QAST::MASTRegexCompiler'
06:46 not_gerd joined #moarvm
06:46 not_gerd o/
06:46 diakopter not_gerd: howdy :)
06:46 diakopter not_gerd: did you notice I mentioned you in moarvm source code?
06:46 not_gerd diakopter: yes, I saw ;)
06:46 diakopter oh good :)
06:47 diakopter did I do it right?
06:47 not_gerd it was a pleasant surprise
06:47 not_gerd diakopter: all good
06:47 not_gerd JimmyZ: https://github.com/MoarVM/MoarVM/pull/43
06:47 diakopter use the utf8 parser I mean
06:47 JimmyZ I saw too, some string code
06:48 not_gerd diakopter: sure, the code was written to be used
06:48 not_gerd just don't sue me if there's a bug ;)
06:49 diakopter supposedly the artistic license protects from that
06:49 crab2313 joined #moarvm
06:49 diakopter JimmyZ: where did you find the ws method to copy
06:50 not_gerd JimmyZ: is that correct: https://github.com/MoarVM/MoarVM/blob/readli​neintfh2/3rdparty/linenoise/linenoise.c#L271
06:50 not_gerd assignment in if()s are.. iffy ;)
06:50 JimmyZ diakopter: nqp\src\vm\parrot\QAST\Compiler.nqp
06:52 JimmyZ not_gerd: it's right, in if() is not accepted :P
06:52 diakopter just needs another ()
06:52 JimmyZ s/in/but in/
06:52 diakopter (says clang)
06:54 diakopter .tell jnthn I have some corrections for your syllabus for the rakudo/nqp workshop.. you forgot to s/Day/Week/g
06:54 yoleaux diakopter: I'll pass your message to jnthn.
06:54 diakopter .tell lizmat I have some corrections for your syllabus for the rakudo/nqp workshop.. you forgot to s/Day/Week/g
06:54 yoleaux diakopter: I'll pass your message to lizmat.
06:57 dalek MoarVM/readlineintfh2: 2df25bb | (Gerhard R)++ | 3rdparty/linenoise/linenoise.c:
06:57 dalek MoarVM/readlineintfh2: [3rdparty/linenoise] Get rid of some magic numbers
06:57 dalek MoarVM/readlineintfh2: review: https://github.com/MoarVM/MoarVM/commit/2df25bb938
06:57 dalek MoarVM/readlineintfh2: 88aa8d9 | (Gerhard R)++ | 3rdparty/linenoise/linenoise.c:
06:57 dalek MoarVM/readlineintfh2: [3rdparty/linenoise] Change WIN32 to the more portable _WIN32
06:57 dalek MoarVM/readlineintfh2: review: https://github.com/MoarVM/MoarVM/commit/88aa8d9b7d
06:57 dalek MoarVM/readlineintfh2: 939a8c0 | (Gerhard R)++ | 3rdparty/linenoise/linenoise.c:
06:57 dalek MoarVM/readlineintfh2: [3rdparty/linenoise] Pull in io.h on win32 for _isatty
06:57 dalek MoarVM/readlineintfh2: review: https://github.com/MoarVM/MoarVM/commit/939a8c05ce
06:57 dalek MoarVM/readlineintfh2: 008b2f8 | jimmy++ | 3rdparty/linenoise/linenoise.c:
06:57 dalek MoarVM/readlineintfh2: Merge pull request #43 from gerdr/gerdr/linenoise
06:57 dalek MoarVM/readlineintfh2:
06:57 dalek MoarVM/readlineintfh2: Cleanup and win32 fixes for linenoise
06:57 dalek MoarVM/readlineintfh2: review: https://github.com/MoarVM/MoarVM/commit/008b2f8387
06:58 diakopter ugh, I need a disassembler, not just a bytecode dumper
06:59 JimmyZ or moarvm debug
06:59 JimmyZ *debugger
06:59 diakopter yeah that too
07:00 JimmyZ not_gerd++
07:00 FROGGS joined #moarvm
07:14 FROGGS diakopter / JimmyZ / jnthn: do you mind if the cross compiler without the --output option would spit its result to stdout rather to temp.moarvm?
07:19 diakopter seems fine to me. just make sure to add the output option to the Test driver
07:19 diakopter for the moar tests
07:20 FROGGS k
07:21 JimmyZ to a.out ?
07:21 JimmyZ :P
07:21 FROGGS hehe >.<
07:22 diakopter to *GEN by default, let's say
07:23 * diakopter hopes no one gets that one
07:23 FROGGS O.o
07:24 diakopter as/400
07:30 FROGGS diakopter: are you talking in bytecode to me?
07:31 diakopter as/400 is an ibm systsem
07:33 FROGGS I had an IBM PCD-2M once
07:34 diakopter *cystem
09:13 crab2313 joined #moarvm
09:53 diakopter I might have an nqp ready to bootstrap soon
09:54 lizmat oh wow!
09:55 JimmyZ diakopter++
09:55 JimmyZ I'm looking forward to it
09:56 FROGGS diakopter++
09:56 FROGGS diakopter++
09:56 FROGGS diakopter++
09:57 diakopter At Frame 16, Instruction 380, op 'no_op' has invalid number (4) of operands; needs 0.
09:57 FROGGS you deserve more but that'd be catched as spam
09:57 diakopter hrm
09:57 diakopter no-op these four operands please
09:58 * not_gerd is working on basic cross-compilation and MinGW64 support
09:59 diakopter wat
10:00 not_gerd APR doesn't like MinGW64
10:00 not_gerd I'm hoping to fix that, as well as building MoarVM in case of buildsys != hostsys
10:01 not_gerd ie adding --build and --host to Configure.pl and passing this down to the 3rdparty configure scripts
10:01 JimmyZ not_gerd: we will do apr => libuv
10:02 not_gerd JimmyZ: sure, but what's the ETA?
10:02 JimmyZ for now, I'm using nmake and msvc
10:02 FROGGS not_gerd: I'd guess not more than two months...
10:03 FROGGS @others: tell me if my gut-feeling is wrong :o)
10:03 diakopter eh
10:03 diakopter I'd say more
10:03 FROGGS k
10:03 FROGGS diakopter: what is your expectation?
10:04 diakopter just that the api will be difficult to get right
10:04 FROGGS k
10:07 diakopter C:\Users\mwilson\src\MoarVM>perl6 tools\update_ops.p6
10:07 diakopter Parsed 8 op banks with 532 total ops
10:07 diakopter Wrote ops.h, ops.c, and Ops.nqp
10:10 jnthn joined #moarvm
10:10 diakopter jnthn: wb!
10:15 woosley left #moarvm
11:14 crab2313 joined #moarvm
11:22 JimmyZ good evening
11:26 JimmyZ FROGGS: ping
11:30 * JimmyZ ponders to fork libuv and patch it
11:30 FROGGS JimmyZ: ola
11:30 FROGGS what do you wanna patch libuv for?
11:31 crab2313 joined #moarvm
11:32 JimmyZ FROGGS: https://gist.github.com/zhuomingliang/6077134, 'no' parot
11:32 JimmyZ *part
11:33 FROGGS ahh, I see
11:42 JimmyZ windows support will be always annoying :(
11:43 FROGGS yeah, but needed
11:44 JimmyZ want to join in ?
11:44 JimmyZ :P
11:45 FROGGS maybe... I'll decide after seeing your first commits :P
11:48 JimmyZ I maybe do it this weekend
11:49 FROGGS cool, by benchmarks should be done by then :o)
11:50 JimmyZ good
12:19 benabik joined #moarvm
12:23 arnsholt joined #moarvm
12:41 not_gerd mostly works: perl Configure.pl --build i686-pc-cygwin --host x86_64-w64-mingw32 --mingw && make
12:54 JimmyZ not_gerd: that's great
12:58 not_gerd strawberry perl doesn't work yet because we rely on $^O instead of %config in places where we shouldn't
12:59 * not_gerd also needs to properly fix APR
13:01 * JimmyZ doesn't know who use mingw32 to patch MoarVM
13:31 not_gerd thoughts on configuration refactor: https://gist.github.com/gerdr/6184541
13:36 JimmyZ not_gerd: I think we can steal some idea from uv.gyp in libuv
13:36 JimmyZ a json file
13:38 JimmyZ some add_defines(), add_link_tags(), add_file_to_compile_source() function would be fine, methinks
13:39 JimmyZ or add_link_setting
13:39 JimmyZ and add_link_library()
13:40 JimmyZ and add_include_dir[s]()
13:47 not_gerd anyway, I'm off...
13:47 not_gerd o/
13:47 not_gerd left #moarvm
14:06 crab2313 joined #moarvm
14:34 colomon joined #moarvm
15:37 FROGGS joined #moarvm
17:45 diakopter JimmyZ: but...
17:45 diakopter JimmyZ: segomos is working on the libuv thing
17:47 segomos i was going to play around with it, i can work on other stuff too
17:47 segomos if needed
17:48 FROGGS well, hacking in missing features and binding it to moarvm are two different task (which can happen totally in parallel)
17:48 FROGGS tasks*
18:13 diakopter this is true
18:25 segomos cool, have libuv building with moarvm - i'll work on wtf i'm actually doing..i need to dig into this and figure out why apr wants a memory pool
18:26 diakopter well we might actually have to keep apr in addition to libuv
18:26 diakopter if jimmyz's list is correct
18:26 segomos for?
18:26 diakopter subprocess interactions, maybe
18:26 segomos process stuff?
18:26 diakopter guid/uid yeah
18:26 segomos okay
18:26 diakopter not sure
18:26 diakopter need jnthn to chime in
18:26 jnthn I'd rather not keep both.
18:27 segomos i didn't remove APR and i don't have libuv stuff building OS specific, i just included darwin and a couple others
18:27 diakopter darwin, linux, windows, all 64 is enough for now
18:27 diakopter 32 later if we want
18:27 segomos yea win won't build, i don't have the # statements in there for OS yet
18:27 diakopter segomos++
18:27 segomos ./lz
18:28 segomos lunch time though
18:37 diakopter jnthn: I made a bunch of progress on the refactor. it builds cleanly now, at least, and passed all the moar tests
18:37 diakopter except two thread ones that it's now failing (at least it no longer hangs!)
18:37 jnthn diakopter++
18:38 diakopter but I can't get it to run the nqp tests...
18:38 diakopter oh wait, I'll just disable those additional .nqp in the Makefile as they're not necessary
18:38 jnthn aww
18:39 diakopter well it's just b/c it wasn't finishing building NQP.pm
18:39 diakopter dies with (too many operands (4) to noop)
18:39 diakopter darn.
18:40 diakopter now every nqp test is dying with  This representation (MVMCode) does not support elems
18:40 diakopter musta done something wrong-o
18:40 diakopter time to take a look at the multi-thousand line diff
18:43 lizmat small steps, grasshopper, small steps!
18:43 diakopter can't really
18:43 diakopter huge refactor
18:44 diakopter can't have StaticFrames and CompUnits be both repr/GC-managed and not, simultaneously
18:49 diakopter jnthn: I found a bunch of places MVM_ASSIGN_REF needed to be... at least I *think* so.
18:49 diakopter we'll need to review this commit thoroughly, perhaps together
18:49 diakopter once it actually works...
18:52 diakopter I found a C trick I didn't know about
18:52 diakopter you can typedef a named struct to another name before actually declaring it
18:52 diakopter highly convenient
18:55 diakopter @#%&)@(*# this will take hours to track down
18:55 diakopter (using solely the code-reading method)
19:53 segomos what c trick?
19:54 diakopter (the next line in the irclog)
19:54 diakopter (seemed neat to me) :)
20:58 colomon joined #moarvm
21:10 segomos i like the biers at the bottom of some of these pl filez
21:18 diakopter jnthn: someone noticed your .pm results
21:21 jnthn ;)
21:22 jnthn Feels fair. Hack on configure, get beer tips.
21:34 segomos i feel weird about this shit compiling
21:34 segomos i also don't know how to test
21:34 diakopter ? :)
21:34 diakopter jnthn: did you look at not_gerd's gist above?
21:34 segomos build uvlib, hack around in the build stuff, changed the open/close methods in io/fileops.c  to use uvlib and it compiles fine
21:35 diakopter oh heh.
21:35 diakopter well what objects does it place in the OSHandle
21:35 diakopter after opening a filehandle
21:35 segomos in MVMOSHandle ?
21:36 segomos i added a uv_fs_t* to the struct called file_req and it assigns that
21:36 diakopter ok
21:36 diakopter is that object threadsafe?
21:36 diakopter if not, need to add a mutex on the object also
21:36 segomos i'll check it out
21:36 diakopter (was going to need to do that with apr also)
21:39 diakopter jnthn: okay, I tracked it down (the problem I was having)
21:39 jnthn diakopter: No, didn't look yet
21:39 diakopter jnthn: it has to do with the void :want
21:40 diakopter well, how I "fixed
21:40 diakopter "fixed" it
21:41 FROGGS ohh
21:41 FROGGS I'm eager to see your commit then :o)
21:52 segomos do you have an fileops tests?
21:57 FROGGS I've got something locally...
21:57 FROGGS ohh, no, *I had* something
21:58 FROGGS I did a git clean -xdf a week ago >.<
21:58 FROGGS ohh, hold on
22:03 dalek MoarVM: cfc1eaa | (Tobias Leich)++ | nqp-cc/t/nqp/77- (2 files):
22:03 dalek MoarVM: added file-io tests
22:03 dalek MoarVM:
22:03 dalek MoarVM: These need to be fudged to work on nqp@jvm, since it has no chdir yet.
22:03 dalek MoarVM: review: https://github.com/MoarVM/MoarVM/commit/cfc1eaa971
22:03 FROGGS segomos: ^^
22:29 diakopter FROGGS: no, I was saying I *didn't* fix it
22:30 FROGGS I read it as: the problem is how you "fixed" it... so when you push your stuff, it will be properly fixed...
22:30 diakopter the problem is that how I fixed it was wrong
22:31 diakopter or in some way related to some other problem that's causing the breakage
22:32 FROGGS right, I got it already
22:32 diakopter ah, oh :)
22:32 segomos FROGGS: thanks, i'll take a look
22:45 segomos diakopter: is there a reason that MVM_file_info tries to open the file first?
22:45 benabik joined #moarvm
23:52 FROGGS joined #moarvm

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