Camelia, the Perl 6 bug

IRC log for #parrot, 2009-04-19

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:06 dalek rakudo: da03c6c | pmichaud++ | build/PARROT_REVISION:
00:06 dalek rakudo: Bump PARROT_REVISION to get latest ** <expr> changes.
00:06 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​a03c6ca9f0605d95d077553e948e4534fec9439
00:06 shorten dalek's url is at http://xrl.us/bephob
00:06 Limbic_Region joined #parrot
00:09 AndyA joined #parrot
00:09 DietCoke joined #parrot
00:10 Coke .
00:10 Limbic_Region ./
00:11 Theory joined #parrot
00:11 cotto ./.
00:27 bsdz_mob joined #parrot
00:29 dalek parrot: r38200 | pmichaud++ | trunk/compilers/pge/PGE/Perl6Regex.pir:
00:29 dalek parrot: [pge]:  Get ** <expr> quantifier to (mostly) handle :sigspace.
00:29 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38200/
00:30 dalek rakudo: d0310a3 | pmichaud++ | build/PARROT_REVISION:
00:30 dalek rakudo: Bump PARROT_REVISION once again.
00:30 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​0310a3b04a61195e3486edb393f6f6d1526a141
00:30 shorten dalek's url is at http://xrl.us/bephqd
00:36 dalek parrot: r38201 | pmichaud++ | trunk/t/compilers/pge/perl6regex/rx_quantifiers:
00:36 dalek parrot: [pge]:  Updates to rx_quantifiers for new ** <expr> quantifier.
00:36 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38201/
00:37 rdice joined #parrot
01:09 davidfetter joined #parrot
01:26 dalek parrot: r38202 | coke++ | trunk/t/codingstd/copyright.t:
01:26 dalek parrot: [t] for release, re-make this a TODO test.
01:26 dalek parrot: an occasionally passing TODO beats an occasionally failing test.
01:26 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38202/
01:26 dalek parrot: r38203 | coke++ | trunk/compilers/pge/PGE/Exp.pir:
01:26 dalek parrot: [cage] remove trailing whitespace
01:26 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38203/
01:30 cotto coke++
01:32 Coke with that, I think 'fulltest' is working for me again.
01:37 Coke going to add a top level nav item to www.parrot.org that points to Trac. someone suggest a name?
01:38 Whiteknight "Trac"
01:38 Coke sadly, that's all I came up with too. =-)
01:38 Whiteknight or, "TEH TRAKZ"
01:39 Whiteknight a bit of a node towards common internet hacker culture
01:39 Whiteknight s/node/nod/
01:43 Coke there. juggled the buttons while I was in there.
01:44 Coke need to reboot...
02:04 DietCoke joined #parrot
02:35 janus joined #parrot
04:07 Andy joined #parrot
04:57 Theory joined #parrot
05:07 dalek parrot: r38204 | fperrad++ | trunk/tools/dev/fetch_languages.pl:
05:07 dalek parrot: [languages] Cardinal is on Github
05:07 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38204/
05:10 dalek tracwiki: v89 | fperrad++ | Languages
05:10 dalek tracwiki: Cardinal is on Github
05:10 dalek tracwiki: https://trac.parrot.org/parrot/wiki/​Languages?version=89&amp;action=diff
05:10 shorten dalek's url is at http://xrl.us/bepibv
05:14 dalek tracwiki: v90 | fperrad++ | Languages
05:14 dalek tracwiki: Add XML
05:14 dalek tracwiki: https://trac.parrot.org/parrot/wiki/​Languages?version=90&amp;action=diff
05:14 shorten dalek's url is at http://xrl.us/bepib5
05:31 flh joined #parrot
05:37 TiMBuS joined #parrot
05:48 helozjisky joined #parrot
06:12 iblechbot joined #parrot
06:37 helozjisky joined #parrot
07:16 szabgab joined #parrot
08:20 barney joined #parrot
08:33 dalek rakudo: 0458077 | pmichaud++ | docs/spectest-progress.csv:
08:33 dalek rakudo: spectest-progress.csv update: 372 files, 10430 passing, 0 failing
08:33 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/0​458077065f2c42d39e3f12759c66025e2d764ce
08:33 shorten dalek's url is at http://xrl.us/bepiku
08:39 helozjisky joined #parrot
08:49 uniejo joined #parrot
08:58 dalek rakudo: 65b3bb5 | jnthn++ | src/parser/actions.pm:
08:58 dalek rakudo: Implement --> for specifying return type.
08:58 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​5b3bb556df56c0098f1717e8b77af20759b8096
08:58 shorten dalek's url is at http://xrl.us/bepimg
08:58 dalek rakudo: 80f0d32 | jnthn++ | :
08:58 dalek rakudo: Merge branch 'master' of git@github.com:/rakudo/rakudo
08:58 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/8​0f0d32e4a0513c7a4acfae92568cadd9449a6fd
08:58 shorten dalek's url is at http://xrl.us/bepimi
09:02 nopaste "tene" at 166.70.38.237 pasted "interpolate hash as named args for jnthn++" (14 lines) at http://nopaste.snit.ch/16310
09:12 dalek rakudo: 0dfcf49 | (Stephen Weeks)++ | src/parser/actions.pm:
09:12 dalek rakudo: Implement |%args for interpolating a hash into named arguments to a sub call.
09:12 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/0​dfcf497357e5090a310a8ffd53cfac5b76e64dc
09:12 shorten dalek's url is at http://xrl.us/bepimt
09:18 uniejo joined #parrot
09:30 flh joined #parrot
09:37 dalek tracwiki: v1 | cotto++ | KnownIMCCBugs
09:37 dalek tracwiki: initial version, no bugs so far
09:37 dalek tracwiki: https://trac.parrot.org/parrot/wiki/Kn​ownIMCCBugs?version=1&amp;action=diff
09:37 shorten dalek's url is at http://xrl.us/bepim5
10:03 TimToady joined #parrot
10:08 dalek pipp: d42faf1 | (Bernhard Schmalhofer)++ | t/pmc/resource.t:
10:08 dalek pipp: add test for dereference
10:08 dalek pipp: review: http://github.com/bschmalhofer/pipp/commit​/d42faf18726c86ca6fef2a509904aa03e4c7555f
10:08 shorten dalek's url is at http://xrl.us/bepin6
10:10 dalek parrot: r38205 | fperrad++ | trunk/NEWS:
10:10 dalek parrot: [release] Updates to NEWS from commit log since last release.
10:10 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38205/
10:14 pmichaud tene:  your flattening name patch would only work for hashes directly, yes?
10:14 pmichaud (the one in 16310)
10:14 pmichaud i.e., it wouldn't work for    |(hash @array)
10:15 pmichaud (oh, I see conversation in other channel)
10:15 moritz good localtime()
10:17 jonathan hi moritz
10:29 TiMBuS joined #parrot
10:37 TiMBuS joined #parrot
10:39 nopaste "jonathan" at 87.238.45.55 pasted "masak - patch for you" (39 lines) at http://nopaste.snit.ch/16311
10:55 mikehh_ joined #parrot
10:59 TiMBuS joined #parrot
11:31 rblasch joined #parrot
12:06 uniejo joined #parrot
12:17 rdice joined #parrot
12:18 iblechbot joined #parrot
12:23 uniejo joined #parrot
12:28 PacoLinux_ joined #parrot
13:02 dalek rakudo: cc6b141 | jnthn++ | src/parser/actions.pm:
13:02 dalek rakudo: our T sub foo() { ... } form of setting the return value now akshually sets it.
13:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​c6b1411908564aa814c61babd780e21c74ae6b4
13:03 shorten dalek's url is at http://xrl.us/bepisi
13:03 dalek rakudo: 25f7728 | jnthn++ | :
13:03 dalek rakudo: Merge branch 'master' of git@github.com:/rakudo/rakudo
13:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/2​5f7728ad466eef6de0dcce78cc71c87b7fb66dd
13:03 shorten dalek's url is at http://xrl.us/bepisk
13:03 dalek rakudo: d803a10 | jnthn++ | src/classes/Role.pir:
13:03 dalek rakudo: Implement subtyping based on role parameters, so now Positional[Int] ~~ Positional[Num], for example.
13:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​803a1056a032c688d4030ab64539e0ea678bbcd
13:03 shorten dalek's url is at http://xrl.us/bepisp
13:03 dalek rakudo: df86625 | jnthn++ | t/spectest.data:
13:03 dalek rakudo: Mark a test that needs ICU as such, and add S12-role/parameter-subtyping.t.
13:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/d​f866252af41ed14fdbdfe3878f1d33129a6b1ff
13:03 shorten dalek's url is at http://xrl.us/bepisr
13:11 dalek parrot: r38206 | fperrad++ | trunk/config/gen/makefiles/docs.in:
13:11 dalek parrot: [docs] add dependency for target html
13:11 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38206/
13:24 dalek parrot: r38207 | jonathan++ | trunk/runtime/parrot/library/P6object.pir:
13:24 dalek parrot: [p6object] Refactor ACCEPTS to better handle non-P6 objects and maybe a little performance win too.
13:24 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38207/
13:30 uniejo joined #parrot
13:41 DietCoke joined #parrot
13:52 Whiteknight joined #parrot
13:54 mikehh how do you apply a patch henerated by svn diff?
13:54 mikehh s/hen/gen/
13:57 dalek parrot: r38208 | pmichaud++ | trunk (4 files):
13:57 dalek parrot: [pge]:  Add another dumper format to assist syntax highlighters.
13:57 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38208/
14:02 mikehh sorry my mind when blank for a moment - forgot the -p0
14:04 nopaste "he" at 158.38.152.119 pasted "Fix for NetBSD/OpenBSD powerpc has correct logic but is misplaced (my bad) -- hopefully last fix to trac 512" (58 lines) at http://nopaste.snit.ch/16312
14:10 Theory joined #parrot
14:11 rg he: are these jit fixes? is jit working on ppc?
14:14 Infinoid rg: it's a portability fix for netbsd/openbsd, apparently jit is working well enough to have broken on those platforms.  see TT #512
14:14 Infinoid he: Thanks for the patch.  Just so I can write a useful commit message, what does this fix?
14:16 Infinoid rg: ...and TT #502, and TT #454, and TT #421
14:16 Infinoid ...and RT #40959 :)
14:17 he Infinoid: The test for the various symbols needs to come after <elf.h> is included...
14:17 Infinoid ah, ok, thanks.
14:18 he rg: it's not jit fixes, but yes, jit appears to work on ppc, at least enough to complete the tests.
14:19 rg interesting. did you by chance do a make testj before we hab to disable the nci testsfor jit/i386 failure?
14:19 rg s/hab/had/
14:20 he rg: the testing i'm doing is with 1.0.0.  I don't know if that precedes the event you're referring to.
14:20 rg hmm that precedes *both* events i'd like to refer to ;)
14:20 bobke joined #parrot
14:20 he heh, ok.
14:21 dalek rakudo: 4a11dd5 | pmichaud++ |  (2 files):
14:21 dalek rakudo: Allow 'dumper' option to pick dump style.
14:21 dalek rakudo: Bump PARROT_REVISION.
14:21 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/4​a11dd5b9db83229ab98a733ca332ac7754c96fa
14:21 shorten dalek's url is at http://xrl.us/bepiyg
14:23 Whiteknight wow, I didn't realize that ppc was JIT-capable
14:25 rg a lot of the code is there is by grunblatt, paultcochrane and leo
14:26 rg unfortunately none of them seems to be active anymore :(
14:26 Whiteknight well, that's how most of the JIT platforms are. Those guys wrote a lot of early parrot
14:27 dalek parrot: r38209 | Infinoid++ | trunk/src/exec_save.c:
14:27 dalek parrot: [ppc] Move the TT #512 fix down a bit, it needs to be below the inclusion of <elf.h>.
14:27 dalek parrot: Patch courtesy of heidnes++.
14:27 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38209/
14:31 Infinoid he: We plan to release 1.1 in a couple of days.  Is there a chance I could convince you to add your various targets to the PLATFORMS file?
14:33 rg ah but the code we're currently having trouble with (build_call_func) is i386 only
14:38 Infinoid rg: What's the problem?  Is there a ticket?
14:39 rg infinoid: tt#551. we were discussing that last night
14:42 Infinoid Thanks.  I'll see if I can reproduce it.
14:43 rg infinoid: the question is how to fix. the general feeling was that jit is unmaintainable and should probably be replaced (or ripped out if there is no replacement)
14:46 Infinoid That sounds like more of a longterm fix
14:47 Infinoid odd, I'm on a jitcapable linux/x86 box but I can't reproduce that.
14:47 rg the tests are currently todo-ed
14:47 Infinoid oh.
14:49 davidfetter joined #parrot
14:49 he Infinoid: sure.
14:50 he Infinoid: I just wished I was able to squish the "fails on NetBSD/macppc 5.0 and NetBSD/i386 5.0" (the latter is newly discovered)
14:51 Infinoid 1.1 should still work on more netbsds than 1.0 did, I think :)
14:52 he I've taken a look at the difference between "working" and "non-working", and it looks like Parrot_find_global_n() falls into the PMC_IS_NULL(ns) case.
14:53 he I've looked, and the namespace is full of these "Null" functions, such as add = 0x82a9ed0 <Parrot_Null_add>,
14:53 Infinoid cotto's #551 patch works for me, trying fulltest now
14:54 he If I understand correctly, that's supposed to be the global name space.
14:54 Infinoid you're referring to vtable functions?
14:54 he However, that's about as far as I've come.
14:54 he Yep.
14:55 Infinoid those come from the Null PMC
14:55 Infinoid I'd love to get a look at some gdb output, if you have some
14:55 he In the working case, they're different, e.g. add = 0x5775f0 <Parrot_default_add>,
14:56 he BTW, the PLATFORM file does not contain os-version, it appears.
14:56 Infinoid which probably means the PMC you're looking at is of a different type
14:56 Infinoid sure it does.  look at the entries for darwin, cygwin, aix
14:57 he I can "nopaste" my non-working gdb session.
14:57 he one moment.
14:57 Infinoid thanks
14:57 Infinoid (though interestingly cygwin is listed under both "supported platforms" and "extra platforms")
14:57 Infinoid is cygwin a parrot core platform?
14:57 rg infinoid: the point is not that the patch works, but that it's not a good solution. for one get_pointer/set_pointer are not complementary anymore with the patch.
14:58 Infinoid One possible short term fix, then, is to subclass or copy the Pointer PMC to something specific which has the right semantics, and documents that.
14:58 Infinoid I dunno the right answer, it's just a thought.
14:59 rg i've suggested a short term fix that removes the handling of V from the jitted call frames
14:59 rg it's the most simple workaround
14:59 rg i also have a solution that instead introduces a new method to pointer, but it would return void ** and that needs 3 infrastructure changes
15:00 Infinoid Having recently gotten into a (nearly) never-ending discussion about bolting more weird functionality onto already-complicated-and-overused base types (in my case, ManagedStruct), I'm a bit wary of changing the semantics of Pointer
15:01 rg whiteknight mentioned that allison or chromatic won't be happy about the simple workaround (removing functionality)
15:01 Infinoid rg: Do any of those infrastructure changes require deprecation cycles?
15:01 nopaste "he" at 158.38.152.119 pasted "GDB session of parrot on NetBSD/i386 5.0 where parrot build fails, for Infinoid" (963 lines) at http://nopaste.snit.ch/16313
15:01 rg i believe no
15:02 Infinoid If it's C stuff, the existence or nonexistence of PARROT_API is a pretty good indicator of whether something needs a depcycle.
15:03 dalek rakudo: 6161852 | (Carl Masak)++ | src/setting/IO/Socket (2 files):
15:03 dalek rakudo: added server bits to IO::Socket::INET
15:03 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​1618521b40cb4853c69f5bb73fb6a23326ffd0b
15:03 shorten dalek's url is at http://xrl.us/bepi3d
15:03 rg no, it's all about parsing a void ** return and adding a new vtable method
15:04 Whiteknight that's just my anticipation, I don't actually know what either of them will think about it
15:05 nopaste "he" at 158.38.152.119 pasted "GDB session of parrot on NetBSD/amd64 5.0 where parrot build succeeds, for Infinoid" (447 lines) at http://nopaste.snit.ch/16314
15:05 he "Included for good measure"
15:08 Infinoid Oh, is this with parrot 1.0.0?  I've been trying to follow along with svn HEAD, but the line numbers look different
15:13 he Infinoid: yep, this is all with 1.0.0.
15:14 he Do you think it would make a difference with svn HEAD?
15:14 he I could relatively easily test that...
15:15 Infinoid I don't know, HEAD is just what I have easily available.  so far things have matched up pretty well, but reading gdb output is kinda slow :)
15:16 Infinoid wait, what?  this looks like your CONTEXT(interp)->current_namespace is PMCNULL
15:17 he Yep.
15:17 he My reaction as well.  "How did that happen?"
15:17 Infinoid Yeah, we should work backward from there
15:19 Infinoid jeez... Parrot_NameSpace_init has been called at least 100 times so far for this simple test script
15:20 nopaste "he" at 158.38.152.119 pasted "Some suggested NetBSD entries for PLATFORMS (I hope I got them right)" (18 lines) at http://nopaste.snit.ch/16315
15:21 Infinoid thanks, looks good.
15:24 dalek rakudo: f03edcc | (Carl Masak)++ | src/builtins/control.pir:
15:24 dalek rakudo: [src/builtins/control.pir] switched to op form of join
15:24 dalek rakudo: The method form is in the setting, and not accessible at this early stage.
15:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/f​03edcc0b0ec48819a1d576fcc5d575c19f3660c
15:24 shorten dalek's url is at http://xrl.us/bepi3w
15:24 dalek rakudo: c271e16 | (Carl Masak)++ | build/gen_setting_pm.pl:
15:24 dalek rakudo: [build/gen_setting_pm.pl] added support for multi-jointed type names
15:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/c​271e16392faecf04f9d20e55563433eb44cda42
15:24 shorten dalek's url is at http://xrl.us/bepi3y
15:26 Infinoid Do your netbsd5.0-i386-gcc-4.1.3 and netbsd5.0-ppc-gcc-4.1.3 platforms not compile?
15:26 Infinoid (due to this null namespace issue?)
15:31 rg infinoid: i've now commented on the ticked with my view of things (and patches).
15:38 nopaste "Infinoid" at 75.5.243.250 pasted "he: backtrace of where the relevant NameSpace PMC is allocated from (on known working linux/x86-64 platform)" (18 lines) at http://nopaste.snit.ch/16316
15:40 Infinoid can you put a breakpoint in src/hll.c:Parrot_register_HLL and see if the call to Parrot_make_namespace_keyed_str returns PMCNULL?
15:41 Infinoid on my box, a breakpoint on src/hll.c:195 is called twice, and it looks like the first time is the relevant one.
15:48 Infinoid oh, they're both for the "parrot" hll namespace, but the second one is for a second interpreter.  The first one is what matters.
15:50 nopaste "he" at 158.38.152.119 pasted "More gdb session output, stepping through Parrot_register_HLL()" (360 lines) at http://nopaste.snit.ch/16317
15:50 he Oddly, it appears it did not return a PMCNULL.
15:52 Infinoid incidentally, for STRING values like hll_name, hll_name->strstart is what you want to print
15:53 Infinoid and to figure out the type of a PMC, I usually just print pmc->vtable->init
15:56 Infinoid what's the type of interp->HLL_namespace?
15:57 Infinoid hmm, nevermind, if that were wrong, parrot would have crashed earlier.
16:00 Infinoid So the value was allocated there and put into an RPA, at HLL_namespace[0] (idx == 0).  Apparently it's fished out of the RPA at some point and ends up assigned to CONTEXT(interp)->current_namespace, which must be the part that breaks
16:06 Theory joined #parrot
16:13 he interp->HLL_namespace->vtable->init is (init_method_t) 0x83c2820 <Parrot_ResizablePMCArray_init>
16:13 Infinoid yeah, it's an RPA
16:16 Infinoid the namespace PMC is stored in there for a while, and there are an awful lot of accesses.  The one we care about is where it's copied to the sub PMC, which happens in Parrot_store_sub_in_namespace (if you breakpoint on that, it will be the second hit, presumably the first one is for "main")
16:16 Infinoid tracking this stuff is like hunting down international jewel thieves, but infinitely less exciting. :)
16:18 Infinoid It's this line:  get_namespace_pmc(interp, sub_pmc);
16:19 he Hm, the first call to get_namespace_pmc() appears to return a null pmc.
16:20 he (type_method_t) 0x82a6410 <Parrot_Null_type>
16:20 he That's ns->vtable->type
16:22 he As does the second invocation.
16:22 Infinoid that must be the bug
16:23 Infinoid it just returns nsroot, here
16:23 Infinoid but that isn't pmcnull for me.
16:25 pmichaud rakudo:  say "\c[INFINITY NEGATED WITH VERTICAL BAR]"
16:26 rg pmichaud: that's still down
16:26 nopaste "he" at 158.38.152.119 pasted "More gdb session output, stepping through get_namespace_pmc()" (42 lines) at http://nopaste.snit.ch/16318
16:27 he nsroot is pmcnull for me.
16:28 Infinoid time for some watchpointing, I think.  one moment
16:32 he Hmm, how do I do PMC_int_val() in the debugger?
16:32 he hll_id appears to be 0.
16:32 he ...in Parrot_get_HLL_namespace(), and it can return PMCNULL.
16:36 Infinoid hll_id should be 0, it's the root namespace (named "parrot")
16:36 he Oh, interp->ctx.state->current_namespace->cache->_i is  {_int_val = -1164939200, _int_val2 = -1163229224}, i.e. apparently un-initialized.
16:37 Infinoid Parrot_get_HLL_namespace() with hll_id=0 returns the correct namespace for me
16:37 he So of course if (PMC_int_val(interp->HLL_namespace) < hll_id) always succeeds.
16:38 he In Parrot_get_HLL_namespace()
16:38 he Where's that supposed to be initialized?
16:38 he Oh, wait.
16:39 Infinoid that was from Parrot_register_HLL, which you verified was valid
16:39 he Still un-initialized, interp->HLL_namespace->cache->_i is {_int_val = -1163211336, _int_val2 = 0}
16:40 Infinoid HLL_namespace isn't an integer, so no problem there
16:40 nopaste "Infinoid" at 75.5.243.250 pasted "Here's a gdb script which will hopefully set a watchpoint to see when the root namespace pointer was changed. (Run it in a new gdb session)" (10 lines) at http://nopaste.snit.ch/16319
16:40 Infinoid those values are a union.  it might be uninitialized, or you might be looking at a pointer in disguise
16:41 * particle1 needs cotto's email address
16:44 DietCoke cotto?
16:44 purl cotto is probably sleeping.
16:44 Infinoid no, cotto is Christoph Otto <christoph@mksig.org>
16:44 purl okay, Infinoid.
16:46 he Hm, at the end of Parrot_register_HLL(), interp->HLL_namespace->cache._i is {_int_val = -1163211336, _int_val2 = 0}
16:46 particle Infinoid++
16:46 he ...which is consistent with what I see above.
16:47 Infinoid he: All of those values are deprecated and probably unused in this case.
16:47 Infinoid If it were used, you'd probably want _ptrs, not _i
16:48 he Uh?  Parrot_get_HLL_namespace() uses PMC_int_val() to decide whether it should return PMCNULL.
16:49 he And my header file says #define PMC_int_val(pmc)      (pmc)->cache._i._int_val
16:49 Infinoid I see it using VTABLE_get_pmc_keyed_int, not PMC_int_val
16:50 Infinoid guess this changed between 1.0.0 and HEAD.  Let me look at 1.0.0
16:51 he Yep, it's changed between 1.0.0 and HEAD.
16:51 he Maybe it's time for me to try to build parrot HEAD on one of the failing variants.
16:51 Infinoid Please test HEAD.  That looks like a difference which may be significant for your test case
16:51 Infinoid (the PMCNULL clause is removed entirely)
16:53 he OK, will do.  (Building subversion will take a bit of time, it pulls in all kind of other stuff in my case...)
16:53 Infinoid No problem, thanks
17:04 Andy joined #parrot
17:06 helozjisky joined #parrot
17:10 bsdz joined #parrot
17:19 flh joined #parrot
17:30 rg damn, forgot to cc parrot-dev again. can someone close RT#42312?
17:31 moritz sure
17:31 moritz done
17:31 rg thanks
17:33 tetragon joined #parrot
17:38 LylePerl joined #parrot
17:39 dalek parrot: r38210 | fperrad++ | trunk/compilers/pct/src/PCT/Dumper.pir:
17:39 dalek parrot: [codingstd] remove trailing space
17:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38210/
17:40 he Infinoid: btw, did that ppc NetBSD/OpenBSD fix go in? (I don't think I saw it)
17:42 rg he: was it the one from r38209?
17:42 Infinoid Yeah, that was r38209
17:43 he The good news is that parrot from HEAD built OK on NetBSD/i386 5.0_RC3.
17:43 he 1.0.0 had problems on same (ref. debugging session)
17:43 Infinoid Great, how's it test?
17:44 LylePerl Is string substitution a case of using index, substr and a loop? Or is there something easier I haven't found yet?
17:44 he Chances are that this will fix it for OpenBSD as well, as originally reported in Trac 464.
17:44 he Running test now.
17:46 he I've googled for the "Null PMC access in get_pmc_keyed_str()", and apparently some FreeBSD users talked about a similar problem.
17:47 he Test result: All tests successful, 14 tests and 571 subtests skipped.
17:47 bsdz LylePerl: just a guess but does regexs in pcre.pir help at all?
17:47 rg he: who did? i don't remember anything about a problem with get_pmc_keyed_str
17:48 LylePerl bsdz: I'll check, thanks
17:48 he rg: wasn't reported here, but e.g. http://groups.google.com/group/perl.perl6.co​mpiler/browse_thread/thread/851eb3cba79f5caf
17:48 shorten he's url is at http://xrl.us/bepjfu
17:48 Infinoid he: Thanks, I'll update netbsd5.0-i386-gcc-4.1.3.  I'm holding off on netbsd5.0-powerpc-gcc-4.1.3 for now, until I get a report on whether it works there
17:49 he Infinoid: checking it out now.
17:49 Infinoid also, it's kinda strange that netbsd5.0-x86_64-gcc-4.1.3 worked when the other two processor types failed
17:50 bsdz LylePerl: t/library/pcre.t might(?) also be useful
17:50 contingencyplan joined #parrot
17:50 he True.  One difference is that i386 and ppc does jit, while x86_64 doesn't.
17:50 he Although I don't understand how that can make a difference.
17:50 Infinoid true.  But I think that shouldn't matter for this code
17:51 Infinoid Sounds like blind luck. :)
17:51 Infinoid something to do with struct member layout or alignment, perhaps
17:51 he That could very well be.
17:52 rg i guess i'll need to try rakudo more ;)
18:01 LylePerl bsdz: pcre doesn't do substitution
18:03 bsdz LylePerl: ah okay. sorry about that
18:09 dalek parrot: r38211 | Infinoid++ | trunk/PLATFORMS:
18:09 dalek parrot: Update PLATFORMS with the boxes I have.
18:09 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38211/
18:09 dalek parrot: r38212 | Infinoid++ | trunk/PLATFORMS:
18:09 dalek parrot: Update PLATFORMS with netbsd platforms tested by heidnes++.
18:09 dalek parrot: There may be another update for netbsd5.0-ppc-gcc-4.1.3 soon, pending test results.
18:09 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38212/
18:10 rg ok, i can reproduce the Null PMC access in get_string.
18:11 bsdz joined #parrot
18:12 bsdz LylePerl: have you tried string replace. i.e. $P0 = new 'String'; $P0 = "...."; $P0.'replace'('foo','bar') ?
18:19 helozjisky joined #parrot
18:27 he Hmm... It looks like src/asmfun.s could / should be named src/asmfun.S, to more clearly indicate that the C preprocessor needs to deal with it...
18:28 Infinoid does that cause a build problem for you?
18:29 he Yep.  I'm having a little difficulty understanding exactly what is going on, though.
18:32 he Manually worked around it for now.
18:32 he (not a patch, ln src/asmfun.s src/asmfun.S, and did a "gmake -n src/asmfun.o" and replaced .s with .S and executed...)
18:33 Infinoid with that kind of change, I worry about causing unforeseen breakage on other platforms
18:34 he I thought .S was fairly well understood to be "pass through preprocessor", as opposed to .s?
18:34 he At least for gcc?
18:35 Infinoid yeah, it is.  I just don't understand this area of parrot at all
18:35 bsdz i think windows would ignore the case of the extension
18:35 he Running tests on NetBSD/macppc 5.0_RC2 now.
18:36 Infinoid I'm not sure windows uses it anyway.  so far ppc is the only place I've ever heard of src/asmfun.s being used
18:37 Infinoid x86 and x86-64 don't seem to have that file
18:37 he I beleive it's being copied there (by Configure.PL?)
18:37 bsdz i have done much asm recently but i i used to embed it directly in my c files in the past (on windows)
18:37 Infinoid yeah, looks like config/auto/jit.pm
18:38 he In the ppc case, it comes from src/jit/ppc/asm.s
18:38 Infinoid and I don't think any other arches use this feature.
18:39 he True, I think only ppc needs the C preprocessor.
18:39 he (that was part of one of my earlier patches)
18:41 Infinoid I think only ppc needs src/asmfun.s, too
18:42 Infinoid regardless of extension, I don't think any other arches copy or build that file.
18:42 Infinoid ok.  so without the preprocessor, how did your previous patches work?
18:42 he Oh, it does? Oh, yes, indeed.
18:43 he Hm, I don't know how to explain that right now.
18:44 Infinoid it would help me to know exactly where it breaks.  does it break on netbsd 5 but works on 3.1 and 4?
18:45 he yep, but now I'm building outside of pkgsrc, while my earlier tests were all with pkgsrc, which brings its own set of wrappers.
18:45 he I suspect that might have something to do with it.
18:46 he NetBSD/macppc 5.0_RC2 test results: All tests successful (3 subtests UNEXPECTEDLY SUCCEEDED), 17 tests and 576 subtests skipped.
18:47 dukeleto joined #parrot
18:52 Infinoid oh, hmm
18:55 he I'm re-trying the build with a patch to config/auto/jit.pm and Makefile.
18:55 Infinoid in pkgsrc, is it devel-parrot?  I don't seem to find it in http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/devel/
18:55 Infinoid uh, devel/parrot I mean
18:56 Infinoid ah, lang/parrot
18:57 he lang/parrot, and yes, I've not yet updated patches/patch-ab.
18:57 Infinoid I don't see anything in the pkgsrc patches which could affect asm preprocessing
18:58 Infinoid oh.  you're the maintainer?
18:58 he True.  However, there's "magic" happening in the pkgsrc infrastructure, e.g. PATH is toyed with, and wrappers are being used for the compiler.
18:58 he Infinoid: yep :)
18:58 he me?
18:58 purl you are saying that we are clearly cooler then Ruby, maybe even cooler then Scala, and quite possible cooler then Twitter itself!
18:58 Infinoid ok, you obviously know a lot more than I do :)
18:58 he hmm...
18:59 Infinoid it would be nice if parrot worked when built standalone, too.  if you can make a patch that works on all platforms, I'll happily check it in
18:59 Infinoid We can harass kid51 to test on darwin/ppc, btw.
19:00 he oh, cool.
19:00 he he is Havard Eidnes <he@NetBSD.org>
19:00 Infinoid he?
19:00 purl he is saying that we are clearly cooler then Ruby, maybe even cooler then Scala, and quite possible cooler then Twitter itself!
19:00 he me?
19:00 purl i think he is saying that we are clearly cooler then Ruby, maybe even cooler then Scala, and quite possible cooler then Twitter itself!
19:00 Infinoid no, he is Havard Eidnes <he@NetBSD.org>
19:00 purl okay, Infinoid.
19:00 he Heh, thanks.
19:02 parrot-poke joined #parrot
19:05 LylePerl joined #parrot
19:06 GeJ Good morning everyone
19:09 LylePerl bsdz: thanks I'll try that... actually is there an evalbot here?
19:09 Theory joined #parrot
19:10 LylePerl parrot: $P0 = new 'String'; $P0 = "this;that;other"; $P0.'replace'(';',':')
19:10 bsdz LylePerl: not really. we nopaste stuff occasionally...
19:11 nopaste "bsdz" at 78.105.78.157 pasted "string replace" (8 lines) at http://nopaste.snit.ch/16320
19:19 cotto perl6: say "hi"
19:20 cotto rakudo: say "hi"
19:20 moritz ENOPOLYGLOTBOT
19:20 cotto Hmm.  We used to have one.
19:20 Infinoid ENOFEATHER3
19:34 Whiteknight joined #parrot
19:39 tetragon_ joined #parrot
19:39 rdice joined #parrot
19:42 tetragon__ joined #parrot
19:42 HG` joined #parrot
19:42 rg whois p6eval
19:42 purl p6eval is probably misc/irclog/ in the pugs repo
19:47 rg actually it's evalbot in the repo, but close enough :P
19:50 nopaste "he" at 158.38.152.119 pasted "Change to copy to asmfunc.S and to handle it, only used on ppc" (40 lines) at http://nopaste.snit.ch/16321
19:52 Infinoid he++
19:54 Infinoid msg kid51 Does http://nopaste.snit.ch/16321 build for you correctly on darwin/ppc?  It's needed for building on netbsd/ppc.
19:54 purl Message for kid51 stored.
19:55 dukeleto joined #parrot
20:05 Tene moritz: the problem is that the languages have migrated out of the parrot repo, and nobody ever updated polyglotbot to build those other languages too
20:06 rg tene: is there some repository of the polyglotbot code so one could submit patches?
20:07 Tene rg: I can give you an account on feather3
20:07 Infinoid Tene: The problem as I see it is simpler: feather3 hasn't responded to pings or ssh in a month or more.
20:07 rg isn't feather3 being down part of the problem?
20:07 uniejo joined #parrot
20:08 Tene Infinoid: polyglotbot was broken for quite a while before feather3 died
20:08 Infinoid I noticed lolcode wasn't working, but rakudo still seemed to work
20:10 Infinoid if/when feather3 comes back up, I can work on external language repositories
20:11 Tene feather is still up...
20:11 Tene but I guess juerd has been out of contact
20:12 Infinoid isn't feather3 a vm?  I think it just needs to be started.
20:14 moritz yes
20:14 moritz write a mail to Juerd and ask him
20:15 moritz I don't think anybody else has access to the controller
20:15 moritz (but I don't know for sure)
20:32 donaldh joined #parrot
20:35 donaldh left #parrot
20:50 dalek rakudo: 0d55159 | pmichaud++ | t/spectest.data:
20:50 dalek rakudo: Small regression on autothread.t.
20:50 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/0​d5515989e68702a8b71bf084dd36b97475e481d
20:51 shorten dalek's url is at http://xrl.us/bepjxx
20:52 szabgab joined #parrot
20:56 moritz is there another hackathon tomorrow?
20:57 wayland76 joined #parrot
21:01 moritz oops, wrong channel
21:20 szabgab I was trying to build pynie now with parrot r38208 and I got:
21:20 szabgab File "./setup.py", line 28, in <module>  PARROT        = "%s/parrot%s" % (pc['bin_dir'], pc['exe']) KeyError: 'bin_dir'
21:20 szabgab can anyone help?
21:25 szabgab pipp also blew up
21:35 DietCoke not me, anyway. I'd probably go to the respective sites for those languages and file bugs.
21:45 dalek parrot: r38213 | allison++ | trunk/ports/cpan (2 files):
21:45 dalek parrot: [cage] Some clarifications on the CPAN process.
21:45 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38213/
21:54 jan joined #parrot
21:55 bsdz szabgab: i wrote setup.py. it tries to find parrot's bindir by running parrot_config --dump. perhaps that has been deprecated in recent parrot builds.
21:56 bsdz szabgab: oh hold on. it looks like it's bindir without an underscore. there used to both bin_dir and bindir.
22:04 Coke there was a ticket to get rid of one of those, yes.
22:05 dalek parrot: r38214 | Infinoid++ | trunk/config (2 files):
22:05 dalek parrot: [doc] Apply patches from he++ in TT #555, to allow Configure.pl to detect perldoc when run as root.
22:05 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38214/
22:07 Coke Infinoid: what was the point of the change in docs.in ?
22:08 Coke ah, it's the same change in both.
22:09 * Coke reads the ticket and is satisfied.
22:09 Coke running 'make fulltest'.
22:10 Coke does the benchmarks test take ETOOLONG for anyone else?
22:11 Tene Ah, looks like planet.parrotcode.org is subscribed to my full feed, not my parrot-specific feed.
22:12 Tene that's why my Perl 6 Web-stuff post showed up there
22:14 Infinoid pmichaud: Did you take ownership of TT #291, or was it just created that way?  I think the reporter needs to be pinged (we've recently fixed some stuff for {net,open}bsd), but I was wondering if you knew more about the issue
22:25 szabgab bsdz, oh good,  I hope someone fix that soon so I can try to go on
22:26 bsdz szabgab: about to commit a patch. one mo ...
22:27 bsdz szabgab:  should be there on your next svn update.
22:28 szabgab now line 30
22:28 baest_ joined #parrot
22:28 szabgab KeyError: 'lib_dir'
22:28 bsdz szabgab:  not another one!
22:29 bsdz szabgab: perhaps i should ditch all the underscores :)
22:31 bsdz szabgab: are you running it using parrot trunk?
22:31 szabgab parrot r38208
22:31 szabgab the one that rakudo pulls for me
22:32 dalek parrot: r38215 | Infinoid++ | trunk/config/gen/platform/openbsd/math.c:
22:32 dalek parrot: [config] Apply patch from darbelo++ in TT #499, to fix a math.h-related build failure on openbsd 4.5.
22:32 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38215/
22:32 bsdz szabgab: i've commited another one.
22:33 szabgab thanks, another issues
22:33 bsdz szabgab: out of interest, are you only looking at pynie out of curiosity?
22:34 szabgab as I am not using an installed parrot it freaks out
22:34 szabgab it is looking for parrot here: /usr/local/bin/parrot
22:34 bacek joined #parrot
22:35 bsdz szabgab: yeah tell me about it. the keys in parrot_config don't dynamically depend on whether parrot is installed or still sitting in the compilation directory.
22:35 szabgab today pmichaud applied some changes so I could integrated with Padre and start to provide syntax highlighting to perl6 based on the parse tree I get back from rakudo
22:35 szabgab I am trying to do the same with pynie and pipp
22:37 szabgab it should work with any language using the Parrot compiler toolkit
22:37 bsdz szabgab: can you do a "make install" for your parrot but that won't necessarily mean you won;t need to start pushing pbc files to get things to work.
22:37 szabgab I can create some symlinks from /usr/local/...
22:37 szabgab I'd rather not sudo make install
22:38 bsdz szabgab: it would be really cool to have a visual editor/debugger that could even peek at the pir code. i was flirting with the idea of building a debugger interface into MS Visual Studio but you need a special (expensive) license from MS to that :(
22:41 bsdz szabgab: if you wanted to avoid those simlinks you could write a simple perl parrot_debugger script that spits out the needed keys.
22:41 bsdz s/parrot_debugger/parrot_config/
22:43 szabgab omg another one
22:43 szabgab so I created two symlinks
22:43 szabgab one to parrot
22:44 szabgab and one to /home/gabor/work/rakudo/parrot/runtime/parrot/ /usr/local/lib/parrot/1.0.0-devel
22:44 bsdz yeah but you also need all those pbc files too. it will get untidy :/
22:44 szabgab but the 1.0.0-devel did not contain a languages directory
22:45 szabgab where it was looking for the nqp.pbc
22:45 bsdz i was more thinking about other pbcs from the runtime library.
22:46 szabgab maybe others are missing too but this was the first one it failed on
22:46 NotFound joined #parrot
22:47 NotFound hi
22:47 bsdz evening NotFound
22:47 NotFound Night here
22:47 bsdz yes nearly MidNight here :)
22:48 NotFound I win, nearly 1 am ;)
22:48 wayland76 8:45am tomorrow here, and I've been up since midnight :)
22:48 bsdz ha ha you must be on the continent. France?
22:49 wayland76 Do I win the mink-lined whatsits?
22:49 bsdz oh hold or South Aftrica?
22:49 wayland76 Nah, Australia
22:49 NotFound Spain
22:49 bsdz nice and warm down there
22:50 bsdz those Ozzies always seem to start so early ;-)
22:51 wayland76 Well, I'm on a weird sleeping schedule.  And I've got 2 heaters on because it's cold :)
22:52 wayland76 ...and because my computer crashes if it gets cold (true -- it's weird)
22:52 bsdz of course, must be approaching winter down there. old pcs are a good source of heat in winter
22:54 nopaste "bsdz" at 78.105.78.157 pasted "szabgab: an example parrot_config that might help" (9 lines) at http://nopaste.snit.ch/16322
22:55 wayland76 Well, it's not winter like they get some places.  My Google Gadget says that today is going up to 20 from a low of 7, and it's currently 13
22:56 Infinoid odd, I thought that said "uncle Gadget" for a second
22:56 bsdz we expect those types of temperatures in our summer here..
22:58 wayland76 Google is more like Big Brother than uncle :)
22:58 wayland76 bsdz: But you probably have buildings designed for it :)
22:58 NotFound I bought a 42" TV, now I don't care the weather ;)
23:00 szabgab bsdz, thanks but it still cannot find nqp.pbc
23:01 szabgab it is in ...rakudo/parrot/compilers/nqp/nqp.pbc
23:02 bsdz szabgab: hmm. you might need to adjust the script. it's expecting to find nqp.pbc at $libdir/$versiondir/languages/nqp/nqp.pbc, or perhaps that's were you create a simlink.
23:03 bsdz szabgab:  don't forget your $libdir has been set to /home/gabor/work/rakudo/parrot/runtime/parrot/ in that script i nopasted.
23:03 szabgab yeah, I'll play with it a bit tomorrow
23:03 bsdz szabgab: okay good luck :)
23:03 szabgab I'll need that :-)
23:28 wayland joined #parrot

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

Parrot | source cross referenced