Camelia, the Perl 6 bug

IRC log for #parrot, 2010-06-01

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:38 snarkyboojum joined #parrot
00:38 dalek parrot: r47243 | darbelo++ | branches/gsoc_nfg (2 files):
00:38 dalek parrot: Let stringinfo know about STRING's new 'extra' pointer.
00:38 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47243/
00:41 kid51 joined #parrot
00:55 dalek parrot: r47244 | darbelo++ | branches/gsoc_nfg/src/ops/core_ops.c:
00:55 dalek parrot: Boostrap ops.
00:55 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47244/
01:01 abqar joined #parrot
01:28 Essobi joined #parrot
01:32 davidfetter joined #parrot
02:25 TiMBuS joined #parrot
02:26 snarkyboojum joined #parrot
02:32 kurahaupo joined #parrot
02:37 khairul joined #parrot
02:42 janus joined #parrot
02:47 cotto khairul, you ready?
02:48 khairul cotto: sure, see privmsg
03:17 JimmyZ joined #parrot
03:38 LoganLK joined #parrot
04:20 tcurtis joined #parrot
04:24 parthm joined #parrot
04:42 parthm left #parrot
05:02 plobsing joined #parrot
05:19 jsut_ joined #parrot
06:02 uniejo joined #parrot
06:25 dalek parrot: r47245 | NotFound++ | trunk/t/pmc/addrregistry.t:
06:25 dalek parrot: tests for AddrRegistry get_bool
06:25 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47245/
06:37 fperrad joined #parrot
06:41 fperrad_ joined #parrot
06:41 dalek parrot: r47246 | plobsing++ | branches/constant_unfolding (5 files):
06:41 dalek parrot: put common const variants of find_cclass and find_not_cclass back
06:41 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47246/
06:42 cotto Why doesn't --target=pbc spit out pbc?
06:42 cotto post works, but pbc just spits out ""pbc" => PMC 'Eval' { ... }"
06:53 aukjan joined #parrot
07:02 sorear --target=(anything at all that isn't pir post past parse evalpmc) means --target=evalpmc
07:02 cotto that's surprising
07:03 sorear the PCT railroad model of compilation makes pbc not easy to implement
07:03 sorear my "better idea" was shot down in #parrot (for several good reasons, iirc); I'm waiting for someone to come up with a replacement
07:04 cotto sounds like a tricky problem
07:06 plobsing the general case sure, but it should be fairly easy to change the default target-not-found behaviour to be "fail" in stead of "give me an eval"
07:07 moritz +1
07:07 purl 1
07:08 plobsing (note I am likely not an appropriate person to do this having little investment in, experience with, or interest in PCT's inner workings)
07:39 eternaleye joined #parrot
07:48 sorear as a cheap hack, implementing "just pbc" as an alternate final stage would not be too hard
07:48 sorear note that evalpmc = list of packfilesegment = basically just parsed pbc
07:49 sorear the main gotcha is wiring it up to the rest of pct in a way that doesn't make people sick
07:55 cotto It's not a big deal.  I was just looking to avoid a potentially unnecessary intermediate step in the build.
07:59 GeJ I'm experiencing failures in t/pmc/eval.t and t/pmc/threads.t
08:00 cotto plobsing, I'm surprised about the register allocator not being real.  I thought we at least had an algorithmically inefficient one.
08:00 GeJ Has this been reported by others or is this a my-box-only issue.
08:01 plobsing register allocator: int get_available_reg() { static regno = 0; return regno++; }
08:01 cotto ouch
08:01 plobsing pretty much
08:02 plobsing we did up until recently have a graph colouring register allocator. but it only ran on -O2 and caused extreme bugginess and occasional crashiness.
08:03 cotto so reg_alloc.c is basically useless?
08:03 nopaste "GeJ" at 192.168.1.3 pasted "Failures with latest trunk on FreeBSD 7/8" (47 lines) at http://nopaste.snit.ch/20795
08:04 cotto GeJ, looks fine here.  It appears that your special system is back in action.
08:04 plobsing eval has issues caused by packfile problems
08:05 cotto I am seeing the eval failure
08:05 GeJ Hah!
08:05 * GeJ feels less lonely.
08:07 GeJ I'll realclean for good measure and try again.
08:08 sorear cotto: the step is necessary in any case; all --target=pbc would save is makefile lines
08:08 cotto yes
08:08 cotto time for bed.
08:08 cotto night, humans
08:08 GeJ good night cotto.
08:09 GeJ clock?
08:09 plobsing cotto: looking at coverage for reg_alloc.c, much of it is to support dead simple register allocation
08:09 moritz is too-late-o'clock :-)
08:09 plobsing and the rest runs a single pass of optimization apparently
08:09 GeJ servus moritz.
08:09 moritz hi GeJ :-)
08:20 bacek aloha, humans
08:21 GeJ G'Day bacek
08:21 bacek G'Day GeJ, howizgoin'?
08:25 GeJ bacek: I've known better days. You?
08:26 bacek GeJ, same shit :)
08:46 clinton joined #parrot
09:11 dalek parrot: r47247 | bacek++ | branches/gc_massacre (11 files):
09:11 dalek parrot: Merge branch 'master' into gc_massacre
09:11 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47247/
09:30 gaz joined #parrot
10:18 dalek parrot: r47248 | mikehh++ | branches/gc_massacre/MANIFEST:
10:18 dalek parrot: regen MANIFEST
10:18 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47248/
10:34 dalek parrot: r47249 | bacek++ | branches/gc_massacre_no_constant:
10:34 dalek parrot: Branch to get rif of 'constant' stuff
10:34 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47249/
10:34 dalek parrot: r47250 | mikehh++ | branches/gc_massacre/src/gc (5 files):
10:34 dalek parrot: add svn properties
10:34 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47250/
10:34 dalek parrot: r47251 | mikehh++ | branches/gc_massacre/src/gc/gc_tms.c:
10:34 dalek parrot: fix codetest failure - there should be at least one space between a C keyword and any subsequent open parenthesis
10:34 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47251/
10:40 snarkyboojum joined #parrot
10:49 gbacon joined #parrot
10:51 dalek parrot: r47252 | mikehh++ | branches/gc_massacre/src/gc/system.c:
10:51 dalek parrot: fix codetest failure - line length
10:51 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47252/
11:03 mikehh bacek: codetest - just ASSERT_ARGS, function docs and of course C++ comments left
11:03 bacek mikehh, ignore c++ comments. They are my own markers.
11:04 bacek But ASSERT_ARGS are worth to fix
11:04 mikehh bacek: yes I know - which is why I left them :-}
11:04 bacek :)
11:05 bacek mikehh, But if you can write docs based on wiki/code/whatevar it will be extremely useful
11:06 mikehh bacek: ok I'll have a look
11:06 dalek parrot: r47253 | mikehh++ | branches/gc_massacre/src/gc/pool_allocator.c:
11:06 dalek parrot: fix pod syntax - still missing documentation
11:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47253/
11:17 darbelo icu--
11:26 Infinoid karma icu
11:26 purl icu has karma of -10
12:03 mikehh bacek: I am getting a failure in make (both g++ and gcc) (make corevm seems ok) - runtime/parrot/library/PGE/Perl6Grammar.pbc
12:04 mikehh get_integer() not implemented in class 'CallContext'
12:05 bluescreen joined #parrot
12:16 snarkyboojum joined #parrot
12:29 dalek rakudo: 18d9960 | (Solomon Foster)++ | src/core/metaops.pm:
12:29 dalek rakudo: Add mixed hash / scalar hypers.
12:29 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/1​8d99603b2e546702ebe7ff6133b63d3f17e83f8
12:35 * mikehh gotta take one of my gransons to a doctors appointment - bbl
12:35 Coke bwahahaha. I have managed to volunter to release both parrot & rakudo in July.
12:36 JimmyZ joined #parrot
12:36 darbelo Double the blame!
12:43 ruoso joined #parrot
13:05 atrodo joined #parrot
13:11 davidfetter joined #parrot
13:14 plobsing joined #parrot
13:22 Coke_ joined #parrot
13:36 whiteknight joined #parrot
13:41 gbacon joined #parrot
13:41 whiteknight good morning, #parrot
13:41 darbelo o/
13:44 Coke_ rant: ops_massacre STILL not resolved.
13:44 Coke_ how did this get merged back?
13:44 * Coke opens a new ticket.
13:45 darbelo Coke: We don't have a Standard Waiting Time after asking on the list.
13:45 Coke how about "make corevm; make coretest" passes?
13:45 Coke that'd be a good start.
13:46 moritz how about "HLLs can compile on it"?
13:46 darbelo Do we advertise that corevm has extra magical properties that aren't covered by the larger 'test' targets?
13:47 Coke developers should know this; I'll mention it again in parrotsketch.
13:47 darbelo I know it does, and why it's there. I was here when it was added. People that came late to that party might not.
13:53 Coke clearly it's not getting run that often just now, or more complaints would be raised.
13:54 Coke I just updated my "build with testing" script to use corevm.
13:54 Coke (so I'll complain more frequently)
13:54 darbelo mikehh is probably our only reliable corevm tester
13:54 whiteknight what is the problem?
13:55 dalek TT #1669 created by coke++: t/pmc/eval.t  and t/pmc/threads.t fail during 'make coretest'
13:55 dalek TT #1669: http://trac.parrot.org/parrot/ticket/1669
13:55 darbelo Ops that are used in our corevm target got kicked out of the core.
13:56 darbelo Plus, dynops are not built as part of the core build.
13:56 darbelo Equals, coretest going kaboom.
13:56 Coke so the simple update to just .loadlib 'foo_ops' fails under coretest.
13:57 Coke the tests actually need to examined (do they really test things that require dynoplibs? if so, move them, if possible, rewrite them, if not, move them.)
13:57 darbelo I think I saw plobsing do that to a few tests already.
13:58 darbelo Let me check the irclogs...
14:00 darbelo Looks like r47167 is the one I'm thinking of.
14:01 JimmyZ_ joined #parrot
14:06 Coke yup. in future, we should probably have that be a condition of merging back a big branch. "make corevm && coretest && make && make test" (and perhaps fulltest.)
14:10 NotFound A big branch theory?
14:11 moritz and a few HLLs running, please
14:12 NotFound And harness running will not be so bad.
14:12 moritz make ueberfulltest # runs all these tests, and tries to compile lua, tcl and rakudo
14:14 bubaflub joined #parrot
14:14 Coke moritz: well, tcl's been broken for ages, so that doesn't help. =-)
14:14 darbelo It motivates people to fix it ;)
14:15 * Coke did not end up finding any time to hack on parrot/rakudo/partcl this long weekend.
14:15 GodFather joined #parrot
14:16 Coke hll's definitely have to be tested at some point at least soon after the merge - any failures should either be identified as reliance on a misfeature, or a regression test should be added. (which I think we're much better about these days)
14:17 Coke (note that I would never have guessed that the small change made that broke partcl would have broken partcl. =-)
14:18 NotFound Coke: I've recently added some tests for surprinsingly uncovered basic functionality.
14:19 NotFound I discovered it was uncovered by breaking things without make test detecting any problem X-)
14:21 moritz is there a list of previous parrot release managers somewhere?
14:22 darbelo The list of previous releases lists the managers.
14:22 darbelo docs/parrothist.pod
14:22 moritz thanks
14:27 dalek parrot: r47254 | coke++ | trunk/docs/parrothist.pod:
14:27 dalek parrot: note release status
14:27 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47254/
14:43 dalek parrot: r47255 | NotFound++ | trunk/t/pmc/lexpad.t:
14:43 dalek parrot: tests for LexPad elements and keyed access
14:43 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47255/
14:44 dalek TT #1670 created by doughera++: t/library/lwp.t fails after parallel build
14:44 dalek TT #1670: http://trac.parrot.org/parrot/ticket/1670
15:05 Andy joined #parrot
15:16 dalek parrot: r47256 | NotFound++ | trunk/t/pmc/resizablebooleanarray.t:
15:16 dalek parrot: test for clone empty RBA
15:16 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47256/
15:18 whiteknight NotFound++
15:19 darbelo NotFound++ # More tests now is less breakage later.
15:20 moritz any parrot foundation members around? gerd++ is looking for somebody to nominate him as a member... he's done two parrot releases for us so far (and lots of small patches)
15:20 moritz I'm not sure if I'm a member :-)
15:21 particle +1
15:21 purl 1
15:21 particle i consider gerd and moritz both members already
15:22 moritz because we have commit access?
15:23 clinton joined #parrot
15:23 particle because you're obviously both engaged in foundation-related activities, including ongoing code and other contributions
15:24 particle if you were a committer this time last year, you're already a member
15:24 moritz I was, yes
15:24 particle we need to make the membership roll public, even i haven't seen it yet :)
15:25 darbelo That would make me a non-member. (committership < 1 year)
15:26 Coke smash is handling that.
15:26 particle well, you've got my nomination
15:26 Coke (the membership roll)
15:26 darbelo Not that I know what menber vs non-member status means, really.
15:26 Coke member == you can vote on the board elections coming up soon.
15:26 particle darbelo: it means you get to vote for board members
15:26 Coke and any other foundation business, though I don't think there is anything else. =-)
15:27 NotFound Foundational members have the right to get the first beers at meetings.
15:27 particle you can nominate folks for membership
15:27 moritz darbelo: I'll just nominate you, since I'm writing a mail to smash just now anyway :-)
15:28 darbelo ZOMG! I get to participate in democracy!
15:31 whiteknight when do nominations start? is there a defined period?
15:32 moritz afaiu nominations can always be made
15:33 particle membership may be granted to individuals at any time
15:35 whiteknight I mean, when do nominations for the board start?
15:36 Coke was mentioned in a recent email or post, I thin.
15:36 particle ah, i suppose they have started already.
15:37 particle i'm sorry, we're still a little disorganized
15:37 Coke email was to parrot-members.
15:38 Coke (which prompted the drive to make sure all folks eligble to be members are, which smash is working on. If you're subscribed to parrot-members, you're a member. if you're not sure, ping smash.
15:38 Coke smash?
15:38 Coke purl, smash?
15:38 purl i haven't a clue, coke
15:38 bubaflub joined #parrot
15:39 whiteknight the emails that I have are that we are preparing for nominations, but first we have to make sure the membership roster is up-to-date
15:39 Coke smash is mestre.smash at gmail.com
15:39 Chandon joined #parrot
15:39 whiteknight I haven't seen information about when we can actually start making nominations, or what the method is for doing that
15:39 darbelo me?
15:39 darbelo purl: me?
15:39 purl bugger all, i dunno, darbelo
15:40 moritz darbelo is our Unicode overlord
15:40 darbelo Meh. Forgetful bot.
15:40 Coke ... what happened to purl?
15:40 moritz purl: darbelo?
15:40 purl no idea, moritz
15:40 Coke she's been lobotomized.
15:40 Coke coke?
15:40 whiteknight he got "fixed" with a large electromagnet
15:40 moritz purl: darbelo is our Unicode overlord
15:40 purl OK, moritz.
15:40 Coke ... wtf.
15:40 Coke purl, master?
15:40 purl no idea, coke
15:40 Coke purl, owner?
15:40 purl coke: wish i knew
15:40 Coke it's hachi, innit?
15:40 darbelo purl: purl ?
15:40 purl darbelo: i haven't a clue
15:40 whiteknight purl, whiteknight
15:40 purl whiteknight: i'm not following you...
15:40 whiteknight purl, whiteknight?
15:40 purl wish i knew, whiteknight
15:40 Coke msg hachi purl seems to have been wiped.
15:40 purl Message for hachi stored.
15:41 particle yah, hachi
15:41 particle purl, be mjd
15:41 purl particle: sorry...
15:41 darbelo Not that there was a lot of signal in the noise pit.
15:43 whiteknight ...and nothing of value was lost
15:45 pmichaud did ops.num disappear?
15:45 darbelo Yes, it got folded into opsc.
15:45 darbelo Or something like it. cotto should know the details-
15:46 pmichaud Hmm.  I always found it to be handy to be able quickly find out what variants of an opcode were available.
15:47 darbelo That should still be available somewhere.
15:48 darbelo You could try src/ops/core_ops.c as a last resort.
15:48 pmichaud the keyword was "quickly"  :-)
15:49 pmichaud oh, that's not too bad
15:49 pmichaud I can probably use it.  Thanks!
15:49 pmichaud darbelo++
15:49 whiteknight karma darbelo
15:49 purl darbelo has karma of 789
15:49 darbelo It even has the numbers as comments!
15:49 whiteknight oh, nice. Karma numbers weren't wiped
15:50 darbelo Oh, the karma is still here?
15:50 whiteknight karma whiteknight
15:50 purl whiteknight has karma of 1997
15:50 darbelo karma c
15:50 purl c has karma of 8624
15:50 whiteknight yep, there's no way I have done ~2k things worthy of karma since Purl was wiped
15:50 darbelo karma g
15:50 purl g has karma of 1201
15:51 darbelo whiteknight: You are better than g++!
15:52 whiteknight g-- # balance
15:52 darbelo heh
15:53 darbelo Oh, crap. git-svn isn't caching my credentials.
15:54 moritz huh? it usually uses the svn cache
15:55 moritz so if you commit once with svn, it's surely in the cache
15:55 darbelo That explains it. The svn cache doesn't work for me :)
15:58 darbelo It's probably some sort of overly paranoid measure taken by whoever did the OpenBSD port.
16:02 whiteknight svn's cache is plain-text passwords only. So if your security profile prhibits that, it wont work
16:03 whiteknight SVN can be made to use the gnome keyring, but you need to compile it with support
16:03 hercynium joined #parrot
16:03 * darbelo has no gnomes on his computer.
16:04 whiteknight then I think you are out of luck
16:05 particle what's the other keyring svn cache supports?
16:05 darbelo It's still the same number of password-typings, but having it happen once on every commit is a bit different from typing it abunch of times on dcommit.
16:05 particle kwallet
16:06 theory joined #parrot
16:06 dalek parrot: r47257 | darbelo++ | branches/gsoc_nfg/src/string/grapheme.h:
16:06 dalek parrot: Move the grapheme table definition into grapeheme.h
16:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47257/
16:06 dalek parrot: r47258 | darbelo++ | branches/gsoc_nfg/src/string/grapheme.h:
16:06 dalek parrot: Add headerizer directives to grapheme.h
16:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47258/
16:06 dalek parrot: r47259 | darbelo++ | branches/gsoc_nfg/src/string/grapheme.h:
16:06 dalek parrot: Add a macro to help us determine if a given codepoint is a combinig character or not.
16:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47259/
16:06 dalek parrot: r47260 | darbelo++ | branches/gsoc_nfg/src/string/encoding/nfg.c:
16:06 dalek parrot: First stab at NFG encoding. Trancode-only for now.
16:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47260/
16:06 dalek parrot: r47261 | darbelo++ | branches/gsoc_nfg/src/string/grapheme.c:
16:06 dalek parrot: Add grapheme.c to the repo. Apparently I forgot to do this earlier.
16:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47261/
16:10 cotto pmichaud, there's also include/parrot/opsenum.h
16:14 pmichaud cotto++
16:14 pmichaud I'm more likely to remember src/ops/core_ops.c though, and it's shorter to type :-)
16:15 particle they're about equally easy to tab-complete
16:23 dalek parrot: r47262 | NotFound++ | trunk/src/pmc/fixedstringarray.pmc:
16:23 dalek parrot: use STRING_IS_NULL instead of checking NULL in FSA is_equal
16:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47262/
16:23 dalek parrot: r47263 | fperrad++ | trunk/config/gen/makefiles/root.in:
16:23 dalek parrot: dynpmcs need to be builded (not in core)
16:23 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47263/
16:23 Chandon What's the difference between set/get_global and set/get_hll_global?
16:23 pmichaud Chandon: set_global is in the current namespace
16:23 pmichaud set_hll_global is in the current hll root namespace
16:24 Chandon In a pir file where I typed ".sub blah" a couple times, is that the same thing?
16:25 pmichaud I don't understand.
16:26 Chandon In much of the sample code people like to type "get_hll_global" to get a reference to a sub, even in files where the word "namespace" doesn't appear at all. Is there some reason for that, or are people typing four extra characters (vs. get_global) out of habit?
16:28 whiteknight probably habit
16:29 Chandon Neat. Is there some reason why it's a good habit I should be copying?
16:29 darbelo Won't that change behaviour if the file get's loaded from different HLLs?
16:29 darbelo Or from two modules in the same HLL?
16:30 darbelo s/modules/namespaces/
16:30 darbelo In one case you do stuff to the current namespace, whatever that is. In the other you always do it to the HLL root namespace.
16:33 cotto_work joined #parrot
16:33 cotto_work aloha, humans
16:33 darbelo aloha, cotto_work
16:34 darbelo I like the new bot better than purl already ;)
16:35 patspam joined #parrot
16:43 dukeleto 'ello from OSBridge
16:44 Coke msg plobsing - ok. with your patch for partcl, using parrot 45624, partcl mostly works. I must have been one rev behind (that rev clearly states un-break the build)
16:44 purl Message for plobsing stored.
16:45 Coke msg plobsing so I'll probably branch partcl at the version I have, apply your patch, then start moving parrot forward and applying patches to the branch, then eventually merging it back to master. Danke.
16:45 purl Message for plobsing stored.
16:47 kurahaupo1 joined #parrot
16:48 bline joined #parrot
16:50 Coke going to yapc::na 2010?
16:50 Coke I hate you, purl.
16:50 purl Coke: huh?
16:51 Coke going to yapc::na 2010 is qw(coke packy pmichaud)
16:52 cotto_work going to yapc::na 2010?
16:53 Coke I hate you, purl.
16:53 purl Coke: excuse me?
16:54 * kurahaupo1 wonders what Purl ever did to Coke
16:54 particle coke hates everything.
16:55 Coke kurahaupo: purl seems to be suffering from amensia.
16:55 Coke particle: only things that suck.
16:56 Chandon purl: bacon is delicious
16:56 purl OK, Chandon.
16:56 kurahaupo1 BTW "hi everyone". I've been off doing $DAYJOB for a couple of months.
16:56 kurahaupo1 Bacon?
16:56 Chandon purl: bacon?
16:56 purl i haven't a clue, chandon
16:56 Chandon Good times. Not only is purl broken, it lies about it.
16:57 cotto_work purl, status
16:57 purl Since Sun May 30 08:59:35 2010, there have been 477 modifications and 84 questions.  I have been awake for 2 days, 7 hours, 57 minutes, 29 seconds this session, and currently reference 844608 factoids. Addressing is in optional mode.
16:57 cotto_work purl, asldfjkasdlfjsklf is laksdjfiwewlksldj
16:57 purl OK, cotto_work.
16:57 cotto_work purl, status
16:57 purl Since Sun May 30 08:59:35 2010, there have been 478 modifications and 84 questions.  I have been awake for 2 days, 7 hours, 57 minutes, 58 seconds this session, and currently reference 844609 factoids. Addressing is in optional mode.
16:57 particle hee
16:57 kurahaupo1 asldfjkasdlfjsklf?
16:58 kurahaupo1 purl: asldfjkasdlfjsklf
16:58 purl kurahaupo1: excuse me?
16:59 cotto_work at least the factoids are being stored
16:59 mikehh purl has definately lost track
16:59 purl mikehh: excuse me?
17:02 kurahaupo1 purl: kurahaupo is in UTC+12
17:02 purl OK, kurahaupo1.
17:02 lucian joined #parrot
17:02 kurahaupo1 purl: I am an insomniac
17:02 purl OK, kurahaupo1.
17:02 kurahaupo1 purl: status
17:02 purl Since Sun May 30 08:59:35 2010, there have been 482 modifications and 84 questions.  I have been awake for 2 days, 8 hours, 3 minutes, 3 seconds this session, and currently reference 844612 factoids. Addressing is in optional mode.
17:02 kurahaupo1 purl: who am i
17:02 purl i don't know, kurahaupo1
17:03 kurahaupo Purl: who am I
17:03 purl no idea, kurahaupo
17:03 theory joined #parrot
17:45 dalek parrot: r47264 | darbelo++ | branches/gsoc_nfg/config/gen/makefiles/root.in:
17:45 dalek parrot: Add the grapheme files to the build.
17:45 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47264/
17:48 bkuhn joined #parrot
17:49 snarkyboojum joined #parrot
17:50 tcurtis joined #parrot
17:51 PerlJam joined #parrot
18:09 patspam joined #parrot
18:14 ruoso joined #parrot
18:18 bluescreen joined #parrot
18:19 darbelo clock?
18:19 darbelo Dammit, forgot that the bot lobotomy.
18:20 whiteknight I accidentally the whole bot
18:23 hercynium joined #parrot
18:26 Coke I can't believe I bot the whole thing.
18:28 darbelo For the TZ-impaired. How long until #ps ?
18:32 PerlJam if it's at 20:30 utc, then ~2 hours
18:32 ash_ joined #parrot
18:40 aukjan joined #parrot
18:41 joeri joined #parrot
18:41 Coke that's about right.
18:42 estrabd joined #parrot
18:45 whiteknight purl msg allison Do we make any guarantees about atomicity of various operations? Specifically I'm worried about a thread switch in the middle of a VTABLE override leaving a potentially-shared PMC in an inconsistent internal state. With PMC._sync gone and STM gone, do we have any mechanism to prevent those kinds of problems, or does it need to be written from the ground-up?
18:45 purl Message for allison stored.
18:45 * darbelo guesses ground up.
18:46 whiteknight me too.
18:46 whiteknight I can think of a few ways around the problem but none of them are very good
18:47 whiteknight if we set a "is busy" flag on the Object when we enter a vtable override along with the thread handle, if we try to access a vtable on that Object from a different thread while the flag is set we force a thread context switch back to the owner thread
18:48 darbelo Threading is never simple.
18:48 whiteknight And we would have to perform that check from PIR, since doing it from C would create a huge problem with longjmp and other garbage
18:49 whiteknight so we would need to do compile-time code modification on VTABLEs to add a "do not enter here from multiple threads" lock implicitly to every vtable override
18:49 whiteknight Once we have Lorito, much of this goes away, but it's still a little ugly
18:50 darbelo And here I was thinking "Dammit our strings suck". Should I feel comforted that our threads suck worse?
18:51 ash_ are there any internal string parsers? I am asking for my gsoc project, I can parse the string myself, but if there are facilities to do it already, i wouldn't mind taking advantage of those
18:51 particle if you do, you're a sick puppy.
18:52 NotFound darbelo: Our strings suck a lot less recently.
18:52 darbelo NotFound: I know, but there's still a lot of work there.
18:53 NotFound darbelo: yeah, but at least we know what length and bytelength are.
18:55 NotFound BTW, someone has started to draft a ByteBuffer?
18:55 darbelo And I'm very grateful for that.
18:59 darbelo Although character length gets kind of nebulous in some places...
18:59 ash_ NotFound: isn't there a GSoC for a byte buffer?
19:00 NotFound darbelo: tell me where, and I'll take a look.
19:00 darbelo I mean as a concept.
19:00 khairul joined #parrot
19:00 darbelo $S0 = unicode:"a\u0308n\u0308"
19:00 darbelo How long is that string?
19:01 NotFound darbelo: that's not a problem with length, is a problem with pir escape rules for string literals.
19:01 darbelo That too.
19:03 darbelo My point is that one string can have more than one length, depending on how 'composed' it is.
19:03 NotFound darbelo: that's not correct. The encoding is part of the string.
19:05 NotFound And if you talk about grapheme length in normalized form, that is an upper level of abstraction, not a problem with the basic levels.
19:19 whiteknight if the string isn't normalized it could have any number of lengths
19:21 NotFound whiteknight: if you change the string content.
19:22 whiteknight change it when, after the last normalization?
19:22 purl whiteknight: that doesn't look right
19:24 NotFound whiteknight: Normalization of what? Of the string buffer content?
19:25 whiteknight yes
19:25 NotFound If you change the string buffer content, is a different string.
19:26 NotFound So it can have a different length.
19:26 darbelo I was expressing myself badly on this. NotFound is right here, the suck comes when different strings compare equal.
19:27 NotFound If you check grapheme equality, you can't count on equal byte or codepoint length.
19:30 NotFound Same as you can't count en equal bytelength when comparing strings with different encodings.
19:38 tcurtis joined #parrot
19:48 pjcj joined #parrot
19:50 cotto_work Coke, ping
19:51 khairul cotto_work: i found out the ans to my qn about events. eventhandlers compare task subtypes with their own type , so no 'allevents' (IIUC). subclassing it and overriding the 'can_handle' method works as an alternative.
19:51 cotto_work ok
19:52 cotto_work I hope you don't find too many bugs to impede your progress.
19:52 Chandon khairul: Do you have an example of setting an event handler and getting it to trigger?
19:52 khairul see t/pmc/scheduler.t
19:55 whiteknight NotFound: Maybe it's time stdhandle wasn't experimental anymore?
19:59 NotFound whiteknight: allison don't like it.
20:01 whiteknight NotFound: then we need to figure out what she would prefer. Because we need some kind of standard, built-in way to get a hold of the STD* filehandles
20:01 NotFound whiteknight: fully agree.
20:01 Coke cotto_work: poing
20:02 Coke whiteknight: getstdout?
20:02 Coke (or was that moved too?)
20:02 NotFound There is another posibiliy, adding some methods to FileHandle.
20:02 cotto_work Coke, what would it take to set up a test trac site on parrot.org for testing the git plugin?
20:02 Psyche^ joined #parrot
20:02 Coke the biggest problem here is that the original request to move stuff was pretty vague. I got the impression it needed review before someone jumped in. (which is why I never finished working on that ticket.)
20:03 cotto_work (and letting everyone see how it works (or doesn't)
20:03 Coke cotto_work: didn't we have a guy that said he'd do that already?
20:03 Coke I'd just reply to that ticket where he said he'd do it and ask if he's made any progress yet.
20:03 NotFound $P0 = new 'FileHandle'  | $P0.'openstdin'() # for example
20:03 cotto_work wfm
20:04 bacek aloha, humans
20:04 NotFound Coke: all get/set std in/out/err has moved.
20:05 Coke is src/ops/core_ops_cg.c dead?
20:05 bacek aloha, going to yapc::na 2010 is qw(coke packy pmichaud)
20:05 aloha bacek: Okay.
20:05 bacek aloha, going to yapc::na 2010?
20:05 aloha bacek: going to yapc::na 2010 is qw(coke packy pmichaud)
20:05 Coke bacek: why have 2 bots?
20:06 Coke (and why duplicate that in channel?)
20:06 bacek cluster!
20:06 cotto_work aloha is for tab completion
20:06 bacek aloha, aloha?
20:06 aloha bacek: I'm little purl's sister which can memoize things
20:06 cotto_work purl is broken
20:06 Coke and now one of the bots will almost always be wrong. =-)
20:06 dmalcolm joined #parrot
20:06 Coke cotto_work: aloha was here before that.
20:06 tcurtis purl, going to yapc::na 2010?
20:06 purl i haven't a clue, tcurtis
20:07 bacek Coke, I added Infobot module to it after purl has amnesia.
20:08 bacek Initially it was pure tab-completion bot :)
20:09 aukjan1 joined #parrot
20:09 Coke I ask about src/ops/core_ops_cg.c because it contains a reference to getstdout and isn't showing up in svn st.
20:11 plobsing_work joined #parrot
20:11 GeJ Good morning everyone.
20:11 Util Good morning, GeJ
20:13 cotto_work coke, that file looks like one that simply didn't get deleted when we nuked the other runcores.
20:14 cotto_work It's a generated file that's no longer generated.
20:15 Coke it's explicitly ignored atm.
20:15 Coke svn:ignore should probably be updated, then.
20:15 cotto_work probably a leftover from when it was generated
20:15 cotto_work Mmm.  instareply
20:16 cotto_work +1
20:16 purl 1
20:19 dalek plparrot: eb8c67f | dukeleto++ | plparrot (2 files):
20:19 dalek plparrot: Call .loadlib from the correct place
20:19 dalek plparrot: review: http://github.com/leto/plparrot/commit/e​b8c67f943a1f7bcdcda2fe3304f5076c404f2cc
20:23 davidfetter joined #parrot
20:27 cotto_work #ps in 3
20:28 Coke darbelo: ping
20:29 Coke is there anything to be saved from simon's strings branch?
20:29 Coke seen chromatic?
20:29 purl chromatic was last seen on #parrot 2 days, 22 hours, 16 minutes and 49 seconds ago, saying: That seems workable.  [May 29 22:12:45 2010]
20:29 Coke purl: help
20:29 purl Coke: sorry...
20:30 darbelo Coke: Not at the code level.
20:30 Coke As I recall, it was perl6-like psuedocode only in that branch.
20:30 darbelo Coke: It's all pseudo code. Yes.
20:31 darbelo I looked it over when drafting my proposal, but I'm defining the C datastructures now, perlish pseudocode can't help me here...
20:32 Coke anyone awake and willing to run #parrotsketch?
20:33 * cotto_work would if he didn't have a meeting in 27 minutes
20:33 * Coke has to leave in 27m, so we'll be done by then. =-)
20:34 snarkyboojum joined #parrot
20:39 Andy joined #parrot
20:42 dalek plparrot: 1b192c8 | dukeleto++ |  (2 files):
20:42 dalek plparrot: Cleanup and move bombing test to the end of the suite
20:42 dalek plparrot: review: http://github.com/leto/plparrot/commit/1​b192c89d88857931fa186126661ff4466138bce
20:46 dalek website: Chandon++ | Threads are continuations.
20:46 dalek website: http://www.parrot.org/conten​t/threads-are-continuations.
20:51 cotto_work moritz++ for adding that documentation
20:51 moritz I figured it was less work than opening a ticket for it :-)
20:56 tcurtis If I "$P0 = 0" a RPA, will it be emptied?
20:58 bacek tcurtis, yes
20:59 cotto_work tcurtis: it does so lazily by setting the size to 0
21:00 tcurtis Thanks.
21:03 dalek parrot: r47265 | moritz++ | trunk/docs/running.pod:
21:03 dalek parrot: [docs] --hash-seed option
21:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47265/
21:03 dalek parrot: r47266 | bacek++ | branches/gc_massacre_no_constant/src/hll.c:
21:03 dalek parrot: Get rid of pmc_new_constants usage in HLL related stuff. We mark it Parrot_gc_trace_roots anyway.
21:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47266/
21:03 dalek parrot: r47267 | bacek++ | branches/gc_massacre_no_constant/t/pmc/eval.t:
21:03 dalek parrot: Fix test.
21:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47267/
21:03 dalek parrot: r47268 | bacek++ | branches/gc_massacre_no_con​stant/src/multidispatch.c:
21:03 dalek parrot: Don't use "constant" in multidispatch.
21:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47268/
21:03 dalek parrot: r47269 | bacek++ | branches/gc_massacre_no_constant/src/dynext.c:
21:03 dalek parrot: Move dynext from "constant" usage.
21:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47269/
21:03 dalek parrot: r47270 | bacek++ | branches/gc_massacre_no_constant/l​ib/Parrot/Pmc2c/UtilFunctions.pm:
21:03 dalek parrot: Don't use constants in generated dynpmcs.
21:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47270/
21:03 dalek parrot: r47271 | bacek++ | branches/gc_massacre_no_constant​/src/pmc/fixedintegerarray.pmc:
21:03 dalek parrot: Get rid of constant cloning in FIA
21:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/47271/
21:06 plobsing_work left #parrot
21:09 theory joined #parrot
21:14 tcurtis joined #parrot
21:23 whiteknight joined #parrot
21:33 tcurtis joined #parrot
21:34 kurahaupo joined #parrot
21:38 cotto_work http://docs.google.com/viewer?url=http%3A%2F%​2Fwww.azulsystems.com%2Fevents%2Fjavaone_2009​%2Fsession%2F2009_J1_HardwareCrashCourse.pdf
21:44 theory joined #parrot
21:46 kurahaupo_mobi joined #parrot
21:48 * dukeleto is learning about JIT with LLVM at OSBridge
21:50 tcurtis Chandon: don't we already have cooperative green threads in the form of Coroutine?
21:51 GeJ Anyone else witnessing failures on t/pmc/eval.t ?
21:52 cotto_work GeJ: yes
21:54 GeJ cotto_work: Oh, now I remember you said something about it last night. What's your platform?
21:55 cotto_work at work it's ubuntu 10.04 x64
21:55 cotto_work at home it's 10.04 x86
22:02 Chandon tcurtis: I can't immediately think of a program you could write with my blog post that you can't write with Coroutine, but the mental models are a bit different.
22:11 tcurtis Chandon: I see that you are correct. I misunderstood what coroutines are.
22:12 theory joined #parrot
22:33 ash_ dukeleto: have you looked into the llvm much yet?
22:46 bubaflub joined #parrot
22:51 dalek winxed: r486 | julian.notfound++ | trunk/winxedst1.winxed:
22:51 dalek winxed: refactor common parts of simple argument lists
22:51 dalek winxed: review: http://code.google.com/p/w​inxed/source/detail?r=486
22:55 jsut joined #parrot
22:59 whiteknight darbelo++ #blog post, work
23:00 whiteknight Chandon++ # blog post, working code snippet
23:00 * whiteknight lives to see all the gsoc students here, working
23:00 darbelo whiteknight: thanks, it's still rough around the edges, but I'm confident I cn make it work on schedule.
23:02 cotto_work khairul posted an update at http://parrot.mangkok.com/.  He seems to have forgotten to mention it in #ps.
23:02 ash_ I am working on a GSoC week 1 update, i haven't gotten as much done as i had hoped, but i should do better this week (and in the weeks to come)
23:04 darbelo On that note... I'm off to run errands.
23:04 darbelo See y'all later.
23:04 cotto_work That's what you think.
23:05 whiteknight ...and I just got vomited on
23:05 ash_ my sister has a new baby (2 months old) so far i have dodged that bullet
23:10 whiteknight khairul is having a segfault? I haven't tracked one of those down in a while
23:10 cotto_work apparently not one bad enough to prompt him to ask for help
23:10 whiteknight ash_: yeah, I get shot about twice a week
23:11 ash_ i got a segfault from using nci to call a function with the wrong signature, :P
23:11 ash_ but that one makes sense
23:11 ash_ (imo)
23:12 whiteknight purl msg khairul: The Opcode and OpLib PMCs have problems with dynops? That sounds like a bug, they should be able to see all ops. Can you file a ticket?
23:12 purl Message for khairul stored.
23:13 cotto_work http://code.google.com/events/io/2010/sess​ions/jit-compiler-androids-dalvik-vm.html (no video yet, hopefully someone will notice)
23:14 whiteknight purl msg khairul .loadlib is a compiler directive that runs at compile time. See compilers/imcc/imcc.y:do_loadlib (lines 1055-1064). "loadlib" is an opcode that runs at runtime.
23:14 purl Message for khairul stored.
23:15 * whiteknight is reading his blog post, answering questions as he gets to them
23:16 whiteknight khairul++ # blog post
23:17 cotto_work It's amusing to see how different browsers render "n̈"
23:19 whiteknight tracer.pir doesn't appear to be present in khairul's branch
23:19 cotto_work Is this Spın̈al Tap?
23:19 whiteknight ah, it's tracer.nqp now. Needs to update manifest
23:21 cotto_work easy fix
23:22 whiteknight true
23:23 darbelo cotto_work: close. It's a dotless i.
23:23 darbelo http://www.imdb.com/title/tt0088258/
23:23 cotto_work That was a copy/paste from the blog post.
23:23 cotto_work To me it looks like a dotless i.
23:24 darbelo Then it must be something in my irssi. 'cause I see a dot.
23:25 whiteknight LOWERCASE I MINUS COMBINING DOT
23:25 whiteknight SMALL I BUT THE DOT DONE GOT STOLED
23:25 whiteknight EMPTY SPACE WITH COMBINING LOWER PART OF THE I
23:26 whiteknight LOWERCASE I BUT THE DOT GOT USED IN DARBELOS BLOG POST INSTEAD
23:27 whiteknight my unicode-foo is strong
23:27 darbelo I wonder if there's a unicode 'CLOSED FIST WITH UPTURNED MIDDLE FINGER'
23:27 darbelo or something like it.
23:27 whiteknight I'm sure I saw something like that in wingdings
23:28 whiteknight it's better than CLOSED FIST IN THE SHAPE OF JODIE FOSTERS KNUCKLE
23:29 whiteknight family guy references are rarely appropriate, but always hilarious
23:29 * darbelo wonders how is he supposed fo fire unicode u+2022 'BULLET's if there no unicode 'GUN'
23:30 darbelo But, really. Errands now. For real.
23:30 cotto_work Can't you compose a gun out of something?
23:30 whiteknight CAPITAL GUN WITH COMPOSING BULLETS
23:33 * whiteknight just added a macro "exit" to GDB. Expects his usage of cursewords to decrease dramatically
23:41 whiteknight hmmm... the backtrace in khairul's segfault looks weird.
23:46 cotto_work what are you running
23:46 cotto_work ?
23:47 whiteknight examples/pir/io.pir, from his blogpost
23:48 whiteknight The program uses the fast_core, descends into the Instrument core, and then seems to be using the "victim" interpreter instead of the "supervisor" one for calls after that
23:48 whiteknight shouldn't it be using he instrument core for everything?
23:48 cotto_work I need to look at that when I can think about it clearly.
23:49 whiteknight ok
23:49 whiteknight I'll look at it until I get bored
23:50 theory joined #parrot
23:50 whiteknight ...which will happen in 5...4...3....2...1...
23:50 cotto_work or puked on again

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

Parrot | source cross referenced