Camelia, the Perl 6 bug

IRC log for #parrot, 2012-05-02

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:09 whiteknight good evening, #parrot
00:18 benabik o/ whiteknight
00:18 whiteknight hello benabik
00:18 whiteknight I've officially started my annual "try to use vim" marathon
00:19 whiteknight so expect foul language
00:19 autark joined #parrot
00:20 * benabik speaks fluent Vim.
00:20 benabik (and foul language)
00:20 benabik dinner &
00:21 cotto quack
00:23 kid51 joined #parrot
00:35 whiteknight quack?
00:35 benabik honk
00:35 Coke fowl language.
00:39 whiteknight Coke: I've got a commit coming up for your branch that makes me very very happy
00:42 alester joined #parrot
00:49 whiteknight alester: you know what bugs me about ack?
00:51 whiteknight alester: the debian package installs it as ack-grep, and I always have to set up the symlink
00:55 benabik ack...-grep?
00:57 sorear whiteknight: alester has no control over that
00:57 whiteknight I know. I don't have any complaints about things alester has control over
00:57 sorear benabik: names in Debian are assigned on a first come first serve basis
00:58 sorear benabik: with very few exceptions (git, chromium), old programs are never renamed to make room for new ones
00:58 sorear and even git-scm and chromium-browser didn't get the short names until after a long and acrimonious debate
00:58 sorear git = GNU Interactive Tools, a distant relative of Norton Commander
01:07 Coke whiteknight: yessssssssssssssss?
01:08 whiteknight Coke: I'm still working on it. I need to setup a text editor and a few tools first
01:08 whiteknight I just reinstalled ubuntu, so I lost everything I need to finish this like flex, bison, etc
01:23 nbrown joined #parrot
01:36 alester joined #parrot
01:36 alester whiteknight: I know, it sucks.
01:37 whiteknight alester: you need to go put your foot down. You're practically a celebrity around here
01:37 alester Where's "here"?  In Debian packaging land?
01:37 alester I'm just glad they didn't call it "perl-grep"
01:37 alester or something with "perl" in the name
01:38 whiteknight alester: "here" is #parrot. You're a celebrity here
01:48 Coke whiteknight: is it likely that I will duplicate your work if I go hack on coke/rm_pasm?
01:48 whiteknight Coke: I'm editing compilers/imcc/* and src/pmc/imccompiler.pmc
01:48 whiteknight and a little bit on src/interp/api.c
01:48 Coke ah, the hard bits. ok.
01:49 whiteknight see? I'm good for something
01:49 * Coke wanders off to get the laptop
01:52 dalek parrot/coke/rm_pasm: 62008a2 | Whiteknight++ | / (11 files):
01:52 dalek parrot/coke/rm_pasm: Rip out the PASM compreg from IMCC. Rip out a few internal flags to differentiate between PIR and PASM at the interface. Rip PASM mode out of IMCCompiler. A few other small hack jobs
01:52 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/62008a2116
01:53 whiteknight blah, I hate diffs like this, because there's no way to tell github that certain files shouldn't be diff'd
01:54 whiteknight and then we see several hundred lines of bison output garbage
02:09 benabik You can use .gitattributes to set it -diff
02:09 benabik No idea if github pays attention to it.
02:16 Coke whiteknight++
02:17 cotto ~~
02:20 nbrown hello cotto
02:21 nbrown how's the thinking going?
02:28 JimmyZ joined #parrot
02:29 JimmyZ hello cotto
02:29 cotto nbrown, trying to decide between fixing the existing M0 arg tests and just going ahead with figuring out the memory layout
02:30 nbrown oh, I fixed them in my branch. I thought I told you last night
02:30 nbrown check out the last couple commits to https://github.com/nbrown/​parrot/tree/m0-c-args-fix
02:31 JimmyZ cotto: I'd like to change PC to +4, because the perl one is using a double dimensional array, while the C one is not.
02:32 JimmyZ cotto: I prefer the C one
02:32 nbrown JimmyZ: why does the internal representation of the ops data matter? isn't that just a implementation detail?
02:34 JimmyZ nbrown: PC+1 in current C one is wrong, but the perl interp is right
02:35 nbrown and I'm asking why is PC+1 wrong? To me +1 makes a lot of sense from a language design point of view: the second op is PC+1 past the first op
02:35 cotto JimmyZ, why do you say that?  It makes sense to disallow the possibility of executing a misaligned op.
02:35 nbrown the fact that it's location 4 bytes down the address space seems irrelevant
02:36 JimmyZ the perl interp PC is always point to a instruction
02:36 JimmyZ but the C interp is not
02:36 JimmyZ that's why you must always *4
02:37 nbrown you have to *4 because there's essentially a physical and logical PC
02:37 cotto yes
02:38 nbrown I don't see why the logical should be the same as the physical
02:38 JimmyZ nbrown:  the perl interp reads every  instruction and store in a double dimensional array, the C interp reads instruction and stores it in a bytecode flow
02:38 nbrown yes
02:39 cotto nbrown, I have yet to see that args test pass, even with your changes
02:39 cotto I don't like this test.
02:39 nbrown cotto: hmmmm, I wonder why it passes for me
02:40 nbrown cotto: I'm not strongly attached to it, but it's hard to debug when it doesn't fail
02:40 nbrown *for me
02:40 cotto nbrown, yeah
02:40 dalek parrot/coke/rm_pasm: eb1015f | coke++ | docs/ (13 files):
02:40 dalek parrot/coke/rm_pasm: remove some references to PASM in the docs
02:40 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/eb1015f178
02:40 JimmyZ the PC of  C interp is  point to a instruction, which is out of spec too.
02:41 JimmyZ *is not
02:42 nbrown JimmyZ: ok, so we should change the C implementation or make the concept of the logical PC clearer
02:42 cotto wfm
02:42 cotto once I get this stupid test working
02:43 JimmyZ cotto: I think bytecode flow is better than double dimensional array, so I prefer to change spec to make PC + 4
02:43 nbrown JimmyZ: I still prefer the logical PC concept because it allows flexibility in the implementation but makes manipulating the PC manually predictable
02:44 * JimmyZ likes faster load and execution :)
02:44 Coke whiteknight seems to have broken the build in coke/pasm
02:45 Coke msg whiteknight: on coke/rm_pasm: ./pbc_to_exe parrot-nqp.pbc -> init_pmc() not implemented in class 'IMCCompiler'
02:45 aloha OK. I'll deliver the message.
02:45 nbrown JimmyZ: is you concern about execution speed the fact that there's a bunch of extra multiplies?
02:46 JimmyZ nbrown: the machine code is not +1 too,
02:46 cotto afk for a bit
02:47 JimmyZ machine code usually should be a data flow, not  a array
02:48 JimmyZ any reason why PC must  to be +1?
02:48 nbrown JimmyZ: fair enough. I still prefer +1, but there's no issue with either way, it just needs to be documented clearly
02:49 nbrown JimmyZ: not really. I was trying to get the C and perl implementations to pass the same test suite, so one had to win and fixing the C implementation seemed simpler
02:49 JimmyZ well, I prefer data flow because it always faster than a double dimensional array
02:49 cotto I like +1 because there's less ambiguity and potential for breakage.  That's about all.
02:50 nbrown JimmyZ: we're still doing the data flow, we're just treating PC logically (+1) and physically (+4)
02:50 JimmyZ OK
02:51 nbrown JimmyZ: the logical exists for the programmer and higher level languages like m1, while the physical is something only m0 needs to know about
02:51 JimmyZ only if the C interp is not like the perl interp :)
02:51 JimmyZ I'm fine
02:51 nbrown JimmyZ: and it isn't
02:52 * JimmyZ hopes it won't be changed in the future too
02:52 nbrown JimmyZ: not if I have any say
02:53 nbrown the current C ops loop is pretty simple and I like it that way :)
02:53 JimmyZ yes
02:54 nbrown it makes debugging the interpreter not too bad
02:54 JimmyZ that's why prefer it
02:54 JimmyZ *why I
02:55 dalek parrot/coke/rm_pasm: a5f1118 | coke++ | docs/book/draft/ch10_opcode_reference.pod:
02:55 dalek parrot/coke/rm_pasm: Remove (stale anyway) PASM xref info
02:55 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/a5f1118136
03:05 Coke urrr. what is the difference between:
03:05 Coke git diff docs/book/pir/ch02_getting_started.pod  docs/book/draft/ch02_getting_started.pod
03:06 Coke ... I struck PASM from one of them and when I was doing the other i noticed it was verrrry similar.
03:06 Coke msg alvis check out: git diff docs/book/pir/ch02_getting_started.pod docs/book/draft/ch02_getting_started.pod
03:06 aloha OK. I'll deliver the message.
03:10 dalek parrot/coke/rm_pasm: 5ff60df | coke++ | docs/ (12 files):
03:10 dalek parrot/coke/rm_pasm: remove/update more PASM references
03:10 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/5ff60dfff9
04:22 cotto the thing finally passes
04:22 cotto well, the C implementation does
04:22 cotto stupid multiple implementations, exposing bugs in each other
04:58 dalek parrot/m0-c-args-fix: 827748e | nbrown++ | t/m0/integration/m0_args.m0:
04:58 dalek parrot/m0-c-args-fix: Fail m0_args if argv[1] doesn't have the right number of bytes
04:58 dalek parrot/m0-c-args-fix: review: https://github.com/parrot/parrot/commit/827748e9e5
04:58 dalek parrot/m0-c-args-fix: e6473c6 | nbrown++ | t/m0/integration/m0_args.m0:
04:58 cotto and now all tests pass on both implementations
04:58 dalek parrot/m0-c-args-fix: Treat argv[1] as an m0 string instead of a C string
04:58 dalek parrot/m0-c-args-fix: review: https://github.com/parrot/parrot/commit/e6473c6fb8
04:59 dalek parrot/m0-c-args-fix: 9d8bde2 | cotto++ | src/m0/perl5/m0_interp.pl:
04:59 dalek parrot/m0-c-args-fix: fix perl5-m0's encoding of cli arguments
04:59 dalek parrot/m0-c-args-fix: review: https://github.com/parrot/parrot/commit/9d8bde2b0a
04:59 dalek parrot/m0-c-args-fix: 07708ab | cotto++ | src/m0/c/m0_ops.c:
04:59 dalek parrot/m0-c-args-fix: add a debugging macro for c-m0
04:59 dalek parrot/m0-c-args-fix: review: https://github.com/parrot/parrot/commit/07708ab298
04:59 dalek parrot/m0-c-args-fix: 273be9f | cotto++ | t/m0/integration/m0_args.m0:
04:59 dalek parrot/m0-c-args-fix: fix the args test, add some disabled debugging code
04:59 dalek parrot/m0-c-args-fix: review: https://github.com/parrot/parrot/commit/273be9f1dd
04:59 cotto if only nbrown or I could figure out why they were passing before on his machine
05:00 cotto nbrown, please test that on your machine
06:34 fperrad joined #parrot
07:08 brrt joined #parrot
07:13 dalek parrot/m0: 2638f9f | cotto++ | / (3 files):
07:13 dalek parrot/m0: update cm0 to encode cli arguments as M0 strings
07:13 dalek parrot/m0:
07:13 dalek parrot/m0: The code emits a warning and needs a bit of cleanup, but the new
07:13 dalek parrot/m0: args_basic test successfully prints out its first arg.  Issues with the
07:13 dalek parrot/m0: m0_args.t are probably due a bug in that test.
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/2638f9f96f
07:13 dalek parrot/m0: 525be62 | cotto++ | src/m0/perl5/m0_interp.pl:
07:13 dalek parrot/m0: remove a stray newline from the p5 m0 interp
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/525be621c6
07:13 dalek parrot/m0: 6811dd3 | nbrown++ | src/m0/c/m0_interp.c:
07:13 dalek parrot/m0: Use 32 bit integers for the string length and encoding
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/6811dd3a62
07:13 dalek parrot/m0: c571a65 | nbrown++ | src/m0/c/m0_interp.c:
07:13 dalek parrot/m0: Make interp_argv be uint64_t* instead of char**
07:13 dalek parrot/m0: The pointers in interp_argv need to be uniformly 8 bytes each. This will
07:13 dalek parrot/m0: make m0_args_basic access the correct argument.
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/c571a6548a
07:13 dalek parrot/m0: 6618c48 | cotto++ | t/m0/integration/m0_args_basic.m0:
07:13 dalek parrot/m0: remove bogus test.  m0_args tests this more effectively
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/6618c48e18
07:13 dalek parrot/m0: 5140899 | cotto++ | src/m0/c/m0_ops.c:
07:13 dalek parrot/m0: macroize op invocations to make debugging a bit less painful
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/5140899c41
07:13 dalek parrot/m0: 827748e | nbrown++ | t/m0/integration/m0_args.m0:
07:13 cotto so much for the heuristic
07:13 dalek parrot/m0: Fail m0_args if argv[1] doesn't have the right number of bytes
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/827748e9e5
07:13 dalek parrot/m0: e6473c6 | nbrown++ | t/m0/integration/m0_args.m0:
07:13 dalek parrot/m0: Treat argv[1] as an m0 string instead of a C string
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/e6473c6fb8
07:13 dalek parrot/m0: 9d8bde2 | cotto++ | src/m0/perl5/m0_interp.pl:
07:13 dalek parrot/m0: fix perl5-m0's encoding of cli arguments
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/9d8bde2b0a
07:13 dalek parrot/m0: 07708ab | cotto++ | src/m0/c/m0_ops.c:
07:13 dalek parrot/m0: add a debugging macro for c-m0
07:13 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/07708ab298
07:14 dalek parrot/m0: 273be9f | cotto++ | t/m0/integration/m0_args.m0:
07:14 dalek parrot/m0: fix the args test, add some disabled debugging code
07:14 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/273be9f1dd
07:14 cotto I guess I'll just have to deal with the karma.
07:15 tadzik wow, commits
07:15 tadzik commits everywhere!
07:18 cotto I'm big on recycling.  I even recycle commits.
07:18 cotto It's environmentally friendly *and* lazy.
07:20 tadzik renewable karma
07:33 dalek parrot/m0: df7bafd | cotto++ | docs/pdds/draft/pdd32_m0.pod:
07:33 dalek parrot/m0: remove reference to load_const, fix some formatting
07:33 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/df7bafd478
07:33 dalek parrot/m0: c1e846b | cotto++ | docs/pdds/draft/pdd32_m0.pod:
07:33 dalek parrot/m0: try to clarify how string constants work
07:33 dalek parrot/m0: review: https://github.com/parrot/parrot/commit/c1e846be9e
07:33 cotto nice to have that cleaned up a bit
08:25 lucian joined #parrot
09:44 nbrown cotto: both cm0 and perl m0 pass all tests for me. Looking at your changes I have no idea how it passed before
09:45 nbrown cotto++
09:56 dalek rakudo/nom: 7836d2f | moritz++ | / (3 files):
09:56 dalek rakudo/nom: implement $?LINE and $?FILE
09:56 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/7836d2f277
10:11 dalek rakudo/nom: 97f9756 | moritz++ | src/Perl6/Actions.pm:
10:11 dalek rakudo/nom: fix exception throwing from previous commit
10:11 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/97f9756d18
10:37 JimmyZ joined #parrot
10:38 JimmyZ good before evening, #parrot
10:45 JimmyZ Coke: ping
10:49 dalek parrot/coke/rm_pasm: ddd8294 | jimmy++ | examples/nci/QtHelloWorld.pasm:
10:49 dalek parrot/coke/rm_pasm: removed QtHelloWorld.pasm
10:49 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/ddd8294ad4
10:51 dalek parrot/coke/rm_pasm: 141099a | jimmy++ | examples/nci/xlibtest.nqp:
10:51 dalek parrot/coke/rm_pasm: removed xlibtest.nqp, the old nqp is not exist anymore
10:51 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/141099a9f6
11:07 JimmyZ nbrown: ping?
11:28 schm00ster joined #parrot
12:18 jashwanth joined #parrot
12:23 Coke JimmyZ: pong
12:23 JimmyZ Coke: I'd like to remove some examples, which is pasm
12:25 Coke my only concern with that is possiby translating the examples from pasm to something else - if they're truly showing off PASM, kill them. If they're showing how to, say, use a library that doesn't have another sample in another language, think about rewriting it.
12:26 JimmyZ Coke: some examples are 6 years old, and I don't think anyone will be interested with it
12:27 Coke use your discretion, that's fine.
12:27 Coke just don't break the build like whiteknight. ;)
12:27 JimmyZ Coke: :P
12:28 Coke 謝謝
12:28 JimmyZ 不用客气
12:30 dalek nqp: 42329b4 | moritz++ | src/HLL/CommandLine.pm:
12:30 dalek nqp: remove some debugging output
12:30 dalek nqp: review: https://github.com/perl6/nqp/commit/42329b433c
12:30 dalek nqp: 88f00e1 | moritz++ | / (2 files):
12:30 dalek nqp: fix and test multiple uses of same command line option
12:30 dalek nqp: review: https://github.com/perl6/nqp/commit/88f00e1569
12:31 * Coke needs to step up his chinese studies as time permits.
12:31 Coke my hanzi skills consist of cutting and pasting from google translate, and I can say "I speak a little Mandarin." but not much else. ;)
12:35 whiteknight joined #parrot
12:36 whiteknight good morning, #parrot
12:36 moritz oh shark whiteknight
12:36 whiteknight ?
12:37 moritz we realized on #perl6 that 'hai' means 'shark' in several languages
12:37 moritz so 'oh hai' became 'oh shark'
12:38 whiteknight nice
12:39 Coke oh snap.
12:39 preflex_ joined #parrot
12:41 dalek rakudo/nom: 93a44d7 | moritz++ | / (4 files):
12:41 dalek rakudo/nom: Implement -M option
12:41 dalek rakudo/nom:
12:41 dalek rakudo/nom: Currently only works for proper modules, not pragmas like fatal or
12:41 dalek rakudo/nom: MONKEY_TYPING.
12:41 dalek rakudo/nom:
12:41 dalek rakudo/nom: Also bumps NQP revision to get command line parser fixes
12:41 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/93a44d7422
12:54 jashwanth joined #parrot
12:58 PacoAir joined #parrot
13:06 davidfetter joined #parrot
13:18 dalek parrot/coke/rm_pasm: e5d21d4 | jimmy++ | examples/benchmarks/addit (4 files):
13:18 dalek parrot/coke/rm_pasm: removed addit.pasm from examples/benchmarks
13:18 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/e5d21d4bdc
13:20 dalek parrot/coke/rm_pasm: a270008 | jimmy++ | examples/benchmarks/mops (3 files):
13:20 dalek parrot/coke/rm_pasm: removed mops* files from examples/benchmarks
13:20 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/a270008986
13:29 dalek parrot/coke/rm_pasm: 2540839 | jimmy++ | examples/benchmarks/oo4.p (2 files):
13:29 dalek parrot/coke/rm_pasm: changed oo4.pasm to oo4.pir
13:29 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/2540839d5c
13:30 jashwanth joined #parrot
13:31 dalek parrot/coke/rm_pasm: d8fb0d8 | jimmy++ | examples/benchmarks/oo4.pir:
13:31 dalek parrot/coke/rm_pasm: change pasm to pir in oo4.pir
13:31 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/d8fb0d8dda
13:34 dalek parrot/coke/rm_pasm: f548f94 | jimmy++ | examples/mops/ (10 files):
13:34 dalek parrot/coke/rm_pasm: removed mop dir from examples, there are no pirs and no pasms
13:34 dalek parrot/coke/rm_pasm: review: https://github.com/parrot/parrot/commit/f548f9430a
13:42 dalek rakudo/nom: 0059aa6 | duff++ | src/Perl6/Grammar.pm:
13:42 dalek rakudo/nom: Fix typo in subrule
13:42 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/0059aa6188
13:44 dngor joined #parrot
14:30 dngor_ joined #parrot
14:31 bluescreen joined #parrot
14:44 contingencyplan joined #parrot
14:50 jashwanth joined #parrot
15:14 lucian joined #parrot
15:16 jsut joined #parrot
15:19 benabik joined #parrot
15:19 davidfetter joined #parrot
15:19 dalek winxed: eeba2b5 | NotFound++ | winxedst2.winxed:
15:19 dalek winxed: fix multi assign modifiers
15:19 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/eeba2b5aff
15:19 dalek winxed: 38728bf | NotFound++ | winxedst2.winxed:
15:19 dalek winxed: allow multi assign in inline
15:19 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/38728bfbc1
15:20 benabik ~~
15:24 JimmyZ joined #parrot
15:25 dmalcolm joined #parrot
15:34 JimmyZ_ joined #parrot
15:38 whiteknight hello benabik
15:38 benabik 'lo whiteknight
15:39 benabik whiteknight: If you thought upgrading Ubuntu was fun, try updating a homebrew automated installer.  I love how they changed configuration and package names ever so slightly.
15:40 whiteknight I've been bitten several times by weird little renaming decisions, especially by ubuntu
15:40 whiteknight or moving install paths for common libraries, or changing the names of .h files, etc
15:41 dalek winxed: 05af985 | NotFound++ | winxedst2.winxed:
15:41 dalek winxed: add a class for function parameter list
15:41 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/05af9852a3
15:50 benabik The fun one for me was that they changed the installer `netcfg/disable_dhcp` to `netcfg/disable_autoconfig`.  So the installer happily went and tried to use DHCP instead of all the static information my config script gave it
15:54 benabik Of course, this doesn't cause errors at the networking step, but much later.  And doesn't occur on my VM setup which does use DHCP.  Blah.
15:55 dukeleto ~~
15:57 whiteknight hello dukeleto
15:57 dukeleto whiteknight: hola!
15:57 whiteknight dukeleto: FYI, I've been seeing some PLA failures, which I think are related to the Ubuntu 12.04 update. I'm going to track those down tonight or this weekend.
15:57 whiteknight dukeleto: and jashwanth was seeing some segfaults in PLA, but I suspect it's from an improperly installed parrot and a libparrot mismatch
15:58 dukeleto whiteknight: sounds fun
15:58 dukeleto whiteknight: both parties signed docs yesterday and I am waiting for it to record + get keys today
15:59 dukeleto whiteknight: i avoid ubuntu updates like the plague. I am still on 10.04 and 10.10
15:59 benabik dukeleto: Yay!  \o//
16:00 dukeleto benabik: thanks! It has been a crazy few weeks.
16:00 * benabik was so excited he grew an extra arm, apparently.
16:00 dukeleto \\\\\o////// # cthulhu wave
16:01 dukeleto whiteknight: i will pull down a fresh PLA+parrot and see if I can get some smoke for you
16:02 whiteknight dukeleto: congrats! I sincerely hope it works out for you this time
16:03 benabik Hm. New record load: 1071
16:03 whiteknight unfortunately, it's not over till those keys are in your hand
16:03 whiteknight benabik: saw an article that you might find interesting about static analysis: http://lwn.net/SubscriberLi​nk/493599/7621ec4c8ab14f15/
16:03 whiteknight actually, it's about how different internal representations (GCC's GIMPLE vs LLVM's AST) are suited for it
16:06 benabik I wonder what the overall structure of GIMPLE is.
16:06 whiteknight honestly, I have no idea
16:06 benabik I would personally dispute the idea that the AST is the best form for static analysis.  It's good for some things, but not others.
16:07 whiteknight dukeleto: regarding your aversion to Ubuntu upgrades, I think I need to get into that mindset myself
16:07 whiteknight dukeleto: The upgrades are always too painful. When the upgrade works I end up with weird software changes, and when the upgrade doesn't work I end up with emergency restore CDs and a few missing hours
16:10 dukeleto whiteknight: yep. That happened to me one too many times.
16:10 PerlJam fwiw, my laptop upgrade to ubuntu 12.04 went super smooth.
16:11 whiteknight PerlJam: I know it's smooth for most people, but it always is a disaster for me
16:11 whiteknight and I haven't been able to figure out why there's a discrepancy
16:11 PerlJam well, this is the first time it's been smooth for me.
16:28 cotto ~~
16:46 Coke ubuntu is on 12 already? eesh
16:47 Coke my upgrade to 10.x was horrible (broke sleep), and I've pretty much never recovered.
16:47 * Coke will try it, mebbe it'll make me happy
16:47 dukeleto Coke: i doubt it
16:48 dukeleto Coke: all the GUI stuff got changed to look like Windows or some junk
16:48 Coke dukeleto: given that it was changed before that to look like a cell phone, I think I'll be ok.
16:48 Coke I basially just want it for chrome, gcc, and ssh.
16:51 PacoAir joined #parrot
17:16 Coke aloha: msg jimmyz: some of the tests you converted in t/pmc/sub.t are failing - please fix them or back out the initial change to pir.
17:16 aloha Coke: OK. I'll deliver the message.
17:17 Coke whiteknight: Did you see the message that you broke nqp in branch?
17:18 whiteknight Coke: no I didn't see that message. I'll fix it tonight
17:19 Coke Danke.
17:24 whiteknight Oh, I suspect I didn't update pbc_to_exe after my api change
17:34 Coke did we /need/ an API change?
17:34 Coke seems like if we can avoid an api change, we'll be better off. (aside from something like "compile_block_of_PASM", of course.
17:35 cotto ~~
17:35 whiteknight Coke: I think we did, yes.
17:35 whiteknight Coke: we certainly wanted to get rid of that compreg, if we're not supporting it anymore
17:36 Coke I'd have to see what the change was.
17:37 Coke instead of removing init_int, why not just restrict the valid values?
17:39 Coke (I mean, obviously we'd only have PIR at that point, but it would avoid a rewrite, and leave the door open for something other than PIR.)
17:46 whiteknight we can keep init_int
17:48 dukeleto i am a fan of init_int
17:54 whiteknight it's unneeded in this case, but we can keep it for some backwards compatibility. I don't think that's the change that broke NQP anyway
17:55 dukeleto whiteknight: lots of HLLs, libraries, etc use init_int (including the elderly PL/Parrot)
17:59 Coke whiteknight: that's the change it complained about, yes.
18:02 dukeleto whiteknight: i am testing pla on the latest parrot+rosella master now
18:06 whiteknight dukeleto: I'm not removing init_int, I'm just removing it from IMCCompiler PMC
18:23 PacoAir joined #parrot
18:30 PacoAir joined #parrot
18:34 Coke whiteknight: except not . ;)
18:35 Coke that is, except, you're adding it back, neh?
18:49 whiteknight Coke: Yeah. I may back that entire commit out and save it for a later branch
18:51 whiteknight it's stuff we want done, but it might be too aggressive a step for this branch
18:51 alester joined #parrot
18:58 dalek winxed: 98bc7a9 | NotFound++ | winxedst2.winxed:
18:58 dalek winxed: allow local functions in inline
18:58 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/98bc7a9423
18:58 dalek winxed: df000e3 | NotFound++ | winxedst2.winxed:
18:58 dalek winxed: allow parameters and lexicals in local functions inside inline
18:58 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/df000e3d85
18:58 dalek winxed: 1f58aaf | NotFound++ | winxedst2.winxed:
18:58 dalek winxed: forgotten part of parameters in inlined local functions
18:58 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/1f58aaf8c9
18:58 dalek winxed: b165c17 | NotFound++ | winxedst2.winxed:
18:58 dalek winxed: allow return in inlined local functions
18:59 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/b165c17b89
19:14 benabik ack-grep --thpppt just doesn't seem as right.
19:22 bluescreen joined #parrot
19:32 dalek winxed: 5c6a1eb | NotFound++ | t/advanced/20experimental.t:
19:32 dalek winxed: tests for local inline new features
19:32 dalek winxed: review: https://github.com/NotFoun​d/winxed/commit/5c6a1ebfda
19:33 cotto $ alias|ack ack
19:33 cotto alias ack='ack-grep'
19:33 cotto problem solved
19:46 whiteknight blah: you know those days where you have straight-forward problems to work on, and you get them done and you feel like you've accomplished something? Today is not one of those days for me.
19:49 dukeleto whiteknight: i feel ya
19:49 dukeleto whiteknight: i am running the pla test suite on the latest parrot+rosella right now
19:51 dukeleto whiteknight: PASS on ubuntu 10.10 64bit
19:52 whiteknight nice. Let me know what you find. jashwanth was seeing a few errors that look like they might be problems in Parrot. I saw a few issues on 12.04 that look related to BLAS binding or symbol export, which I'm sure is something that got monkeyed with in the upgrade
19:52 whiteknight I know it worked when I was on 11.10, so getting it working in 12.04 too shouldn't be too hard. But jashwanth isn't on 12.04 anyway so it's low priority
19:53 whiteknight Making it work on whatever system jashwanth is using is my priority
19:55 whiteknight That means I'm probably going to have to set up a 32-bit ubuntu 11.10 VM tonight for testing
19:56 whiteknight ...which means I probably need to reinstall virtualbox
20:06 dukeleto whiteknight: oy vey
20:47 dukeleto I found a reproducible bug in "tail" today which causes it to core dump. Achievement unlocked.
20:47 moritz oh wow
20:49 tadzik nice
20:51 cotto dukeleto: wow
20:51 cotto what's the bug?
20:51 cotto and I'm pretty sure that covers at least two achievements
21:01 dukeleto cotto: while in screen, i tail -f a file and then press "CTRL-Shift-Insert"
21:01 dukeleto cotto: i have core files to prove it :)
21:02 dukeleto Incidentally, these core files were owned by the wrong user and then caused deployments to fail. SO MUCH WIN.
21:03 cotto my version is fine, sadly
21:03 cotto 8.5
21:03 PerlJam dukeleto: is that a tail problem or a screen problem?  Sounds like the latter.
21:07 bluescreen_ joined #parrot
21:08 alester joined #parrot
21:25 dukeleto PerlJam: not sure, actually
21:25 dukeleto PerlJam: probably screen. One less achievement, I guess.
21:28 dalek nqp/toqast: aba8043 | jnthn++ | src/QAST/Node.nqp:
21:28 dalek nqp/toqast: Slots for slurpy and named.
21:28 dalek nqp/toqast: review: https://github.com/perl6/nqp/commit/aba8043d6a
21:28 dalek nqp/toqast: 9191efb | jnthn++ | src/QAST/Compiler.nqp:
21:29 dalek nqp/toqast: Start to get support for local parameters in place, including natively typed ones.
21:29 dalek nqp/toqast: review: https://github.com/perl6/nqp/commit/9191efba26
21:29 dalek nqp/toqast: 85b021d | jnthn++ | t/qast/qast.t:
21:29 dalek nqp/toqast: A couple of tests for local parameters.
21:29 dalek nqp/toqast: review: https://github.com/perl6/nqp/commit/85b021d2d5
21:36 rurban joined #parrot
21:38 rurban packacking for cygwin: I have a  config/gen/opengl.pm patch (stricter X11 vs non-X11)
21:39 brrt joined #parrot
21:42 dukeleto rurban: send it our way!
21:43 rurban github issue?
21:43 rurban #761
21:43 rurban do I have to clone and patch it also, or is the src patch enough?
21:44 aloha (parrot/parrot) Issues opened : 761 (cygwin needs to strictly decide between X11 opengl and non-X11 opengl) by rurban : https://github.com/parrot/parrot/issues/761
21:44 dukeleto rurban++
21:45 rurban mingw might be influenced. They might to add the w32api header path extra.
21:46 rurban cygwin folks persuaded me to favor X11 over native opengl now.
22:04 bluescreen__ joined #parrot
22:54 whiteknight joined #parrot
22:54 whiteknight good evening, #parrot
22:59 cotto hi whiteknight
22:59 whiteknight hello cotto
23:18 whiteknight bleh, I don't even have autoconf installed. I thought that was a staple
23:19 cotto did you install build-essentials?
23:19 cotto istr it including that kind of thing
23:19 whiteknight I'm getting it now
23:21 whiteknight It's just weird that something so important wouldnt be included by default
23:21 cotto that's one of the downsides of a consumer-oriented distribution
23:22 cotto easy fix, of course
23:22 whiteknight Ubuntu is the closet thing to a "perfect" distro that I've found, so I can't complain too much
23:34 rurban Ubuntu with its GNOME3 mess? Debian maybe
23:35 whiteknight different things for different people, of course. I actually like unity
23:35 whiteknight I know that's not a very popular opinion
23:36 whiteknight rurban++ # Still working on cygwin
23:37 rurban got some hanging tests with cygwin. alarm, task. src/checkdepend.t also fails
23:49 aloha joined #parrot
23:58 aloha joined #parrot

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

Parrot | source cross referenced