Camelia, the Perl 6 bug

IRC log for #parrot, 2009-06-05

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:07 snarkyboojum joined #parrot
00:10 Austin_Hastings Okay, I'm reading this paper: http://www.soi.city.ac.uk/~​ross/papers/FingerTree.pdf on Finger Trees, and my head just exploded.
00:20 Austin_Hastings And FWIW, finger trees are not O(1). They are O(1) amortized for deque operations. (The perl list ops). For other operations, they have to be annotated (an O(n) operation) before becoming O(log n) objects.
00:21 Austin_Hastings Now I am a fraction more knowledgeable. Thanks, dan + darbelo.
00:24 bacek_ joined #parrot
00:29 Austin_Hastings rehai, bacek
00:30 bacek_ rehai Austin_Hastings
00:31 Austin_Hastings FYI: Okay, I'm reading this paper: http://www.soi.city.ac.uk/~​ross/papers/FingerTree.pdf on Finger Trees, and my head just exploded.
00:31 Austin_Hastings Now ur brane will hurts too
00:35 cotto It feels like forever since I've looked at an actual academic paper.
00:35 cotto I look forward to the head 'splodage.
00:35 Austin_Hastings Not long enough. I invested a lot of money in booze and drugs to get those brain cells killed off.
00:36 cotto I don't look at them often because I get lost really quickly, but that's beside the point.
00:49 cotto ping gandi.net
00:49 purl 10 packets transmitted, 10 received, 0% packet loss, time 9011ms, rtt min/avg/max/mdev = 115.286/115.300/115.319/0.012 ms
00:51 cotto and yet I don't get to connect
00:57 cotto Hmmm.  A lightning talk about pmc_pct and making Parrot more self-hosting might be interesting.
01:05 snarkyboojum joined #parrot
01:06 Austin_Hastings I just realized that class attributes are untyped.
01:06 Austin_Hastings How lame.
01:07 dalek parrot: r39398 | whiteknight++ | branches/io_rewiring/src/io/api.c:
01:07 dalek parrot: [io_rewiring] Convert most occurances of my lousy pseudoroles to using base_type checks with PCCINVOKE fallbacks.
01:07 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39398/
01:21 dalek TT #738 created by grantbow++: ICU download URL change in README
01:22 cotto trac does not win at url preservation
01:52 preflex joined #parrot
01:55 kid51 joined #parrot
02:41 janus joined #parrot
04:30 Austin_Hastings left #parrot
04:42 tetragon joined #parrot
05:08 snarkyboojum joined #parrot
05:25 Infinoid bacek_: eh, I think branch commits deserve full karma too
05:27 Infinoid I wouldn't mind using the #perl6 pugsbot format, except that some folks here do use the review links (and I want to encourage that)
05:40 cotto I use and like them
06:06 eternaleye joined #parrot
06:16 viklund joined #parrot
06:23 dalek parrot: r39399 | fperrad++ | trunk (2 files):
06:23 dalek parrot: [doc] update ICU download URL
06:23 dalek parrot: see TT #738
06:23 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39399/
06:25 dalek TT #738 closed by fperrad++: ICU download URL change in README
07:06 dalek parrot: r39400 | cotto++ | branches/pmc_pct/compilers/vtdumper/vtdumper.pir:
07:06 dalek parrot: [vtdumper] rename frozen vtable dump to vtable.frozen to avoid conflicts with the perl-generated vtable.dump
07:06 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39400/
07:11 barney joined #parrot
07:11 cotto All this self-hosting is going to make for some interesting makefile dependencies.
07:11 cotto Are frozen PMCs cross-platform?
07:13 cotto i.e. if I checked in a frozen past, would other platforms be able to use it?
07:14 * cotto goes off to investigate
07:16 * cotto hopes someone will pipe up before he has to dig into the freeze/thaw code too much
07:16 * cotto hears crickets
07:21 cotto Hmm.  I'm not confident that freeze/thaw will work.  It doesn't appear to be very cross-platform.
07:34 cotto can someone on non-x86 (x64 would be fine) checkout the pmc_pct branch, build it, compile vtable.tbl and post the md5sum of the resulting vtable.tbl?
07:35 cotto s/dump/frozen/
07:35 cotto Once it's built, you just need to run ./parrot compilers/vtdumper/vtdumper.pbc src/vtable.tbl
07:36 cotto and post the md5sum of the vtable.frozen, not vtable.tbl
07:37 cotto smolder?
07:37 purl smolder is probably http://sourceforge.net/projects/smolder or web-based smoke test aggregator used by developers and testers to upload (automated or manually) and view smoke/regression tests using the Test Anything Protocol (TAP). or http://smolder.plusthree.com/app​/public_projects/smoke_reports/8
07:46 masak joined #parrot
08:02 dalek parrot: r39401 | cotto++ | trunk/src/pmc_freeze.c:
08:02 dalek parrot: [freeze] minor freeze code readability improvement
08:02 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39401/
08:23 AndyA joined #parrot
08:23 donaldh joined #parrot
08:30 riffraff joined #parrot
08:42 bacek joined #parrot
08:44 krunen joined #parrot
08:47 snarkyboojum joined #parrot
09:04 cognominal joined #parrot
09:18 bacek joined #parrot
09:18 viklund_ joined #parrot
09:21 cotto bacek, do you know if a PMC's frozen representation is portable?
09:22 bacek cotto: it's not.
09:22 cotto ok.  in that case the vtable dump will probably have to use json
09:22 bacek we don't have anything "portable"... Even PBC isn't portable
09:23 bacek +1 for JSON
09:23 cotto Good.  I just finished writing a json dumper.
09:24 pjcj joined #parrot
09:24 cotto karma json
09:24 purl json has karma of 29
09:28 bacek btw, why do you need portable vtable dump?
09:28 cotto chicken and egg problem:
09:29 cotto a vtable dump is needed to build Parrot, but Parrot is needed to build the vtable dumper.
09:29 bacek but for parsing json dump you still need parrot
09:30 cotto Perhaps I should be sleeping instead of thinking through build dependencies.
09:30 bacek and there is no conceptual difference between parsing vtable.tbl and vtable.json
09:31 * cotto needs to think through what the eventual build process will look like
09:31 dalek parrot: r39402 | cotto++ | branches/pmc_pct/compilers/vtdumper/vtdumper.pir:
09:31 dalek parrot: [vtdumper] use JSON as the dump format since frozen PMCs aren't portable (yet?)
09:31 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39402/
09:31 bacek cotto: sleep on it :)
09:31 cotto yes
09:31 cotto good night
09:32 bacek good night
10:10 eiro joined #parrot
10:13 eiro joined #parrot
10:28 snarkyboojum joined #parrot
11:04 HG` joined #parrot
11:20 donaldh joined #parrot
11:45 burmas joined #parrot
11:50 snarkyboojum joined #parrot
11:53 nopaste "bacek" at 114.73.179.59 pasted "Victory! (Almost...) Emitting real PBC from within Parrot in NQP." (168 lines) at http://nopaste.snit.ch/16791
11:53 bacek Hooray!
11:53 bacek Infinoid: what about implementing PIR compiler in PCT? :)
12:11 rdice joined #parrot
12:27 Infinoid bacek: What about it? :)
12:28 bacek It's possible now :)
12:28 Infinoid awesome.  bacek++
12:28 bacek Check nopaste
12:28 bacek http://nopaste.snit.ch/16791
12:28 Infinoid I can check it tonight; have to leave soon
12:28 Infinoid This week has been crazy, sorry.  I will have time this weekend
12:29 bacek No worries.
12:30 bkuhn joined #parrot
12:30 bacek Infinoid: did you found new $dayjob?
12:52 skids joined #parrot
13:02 rdice joined #parrot
13:02 Whiteknight joined #parrot
13:06 dalek parrot: r39403 | bacek++ | branches/pmc_i_ops (4 files):
13:06 dalek parrot: Merge branch 'master' into pmc_i_ops
13:06 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39403/
13:06 dalek parrot: r39404 | bacek++ | branches/pmc_i_ops/src/pmc/bigint.pmc:
13:06 dalek parrot: [pmc] Reimplement most of BigInt math ops in terms of i_ops
13:06 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39404/
13:06 dalek parrot: r39405 | bacek++ | branches/pmc_i_ops/src/pmc/bigint.pmc:
13:06 dalek parrot: [pmc] BigInt bitwise ops use i_ops now.
13:06 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39405/
13:22 whoppix joined #parrot
13:31 AndyA joined #parrot
13:35 gryphon joined #parrot
13:39 dalek rakudo: 056847f | pmichaud++ | src/classes/Grammar.pir:
13:39 dalek rakudo: Add .parse to base Perl6::Grammar and other non-Rakudo grammars. Fixes #66270.
13:40 dalek rakudo: Eventually PGE will provide a .parse method by default, at which point
13:40 dalek rakudo: we can inherit that one and omit this one.
13:40 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/0​56847f74c18e342b2defe6c303aa4ff67db036d
13:46 dalek parrot: r39406 | bacek++ | branches/pmc_i_ops/src/pmc/scalar.pmc:
13:46 dalek parrot: [pmc] Scalar ops reimplemented in terms of i_ops.
13:46 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39406/
13:57 gaz joined #parrot
13:58 ruoso joined #parrot
14:09 snarkyboojum joined #parrot
14:14 davidfetter joined #parrot
14:15 Andy joined #parrot
14:27 dalek TT #739 created by gerd++: path of installed config.fpmc file on a 64bit system
14:31 dalek rakudo: 636a2b6 | jnthn++ | src/parser/actions.pm:
14:31 dalek rakudo: Fix one half of what is reported in RT#65128 - our T1 T2 sub f() { ... } now gives a compile time error like STD.
14:31 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/6​36a2b6f7cb4d0b5c0d4944482918137a0713615
14:31 dalek rakudo: 7a40779 | jnthn++ | src/setting/ (2 files):
14:31 dalek rakudo: Fix up various return type constraints in teh built-ins.
14:31 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/7​a40779c46512e8aee9721993e98bfb46c34eef0
14:31 dalek rakudo: 8c09b53 | jnthn++ | src/parser/actions.pm:
14:31 dalek rakudo: Refactor setting of return types to eliminate duplicate code, and implement checking of implicitly returned values for the my T sub foo() { ... } and sub foo(--> T) { ... } cases.
14:31 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/8​c09b534e9e5e7c85a07cbf2f19229850ae97150
14:31 dalek rakudo: a89161f | jnthn++ | :
14:31 dalek rakudo: Merge branch 'master' of git@github.com:rakudo/rakudo
14:31 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/a​89161fd83a47163c51b97ed46ea30e63ed98f8e
15:07 iblechbot joined #parrot
15:20 donaldh joined #parrot
15:24 dalek rakudo: fb2fd43 | jnthn++ | src/parser/actions.pm:
15:24 dalek rakudo: Make sure we type check implicit return values for of and returns also. Completely resolves RT#65128.
15:24 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/f​b2fd43d4096fe46d5a988d0700a1dccc3b39900
15:57 Theory joined #parrot
16:00 Austin_Hastings joined #parrot
16:01 Austin_Hastings w00t! Not every namespace contains code.
16:19 darbelo joined #parrot
16:25 gryphon_ joined #parrot
16:30 darbelo Is there a way to make tokens case-insensitive in PCT grammars?
16:32 Whiteknight I don't know if PCT supports it, but normal P6 grammars should have the :i flag
16:32 Whiteknight what kinds of case-insensitivity do you need, just the first letter?
16:33 Psyche^ joined #parrot
16:33 particle :i is supported by pge
16:33 darbelo I have "token context_field { [ 'maxexponent' | 'minexponent' ] }"
16:34 darbelo In want 'minExponent' to match too.
16:34 particle token foo { :i 'm' [ 'in' | 'ax' ] 'exponent' } # just for fun :)
16:36 Austin_Hastings token m{<[Mm]>} token e{<[Ee]>} token foo { <.m> in <.e> xponent }
16:36 darbelo He, I trimmed the example a bit, ":i" is what I wanted.
16:37 Austin_Hastings :)
16:37 Austin_Hastings Does anyone know to what extend the wrong reporting of pir line numbers is a "known bug" versus something else?
16:38 darbelo Now you've told them, they know the bug is there ;)
16:39 Austin_Hastings That's what I'm afraid of.
16:39 darbelo I would search for a TT.
16:39 Whiteknight it's so hard to keep up with the list of supported features because development happens so quickly
16:41 darbelo Whiteknight: There's a simple solution to that. But i doubt you'd like it.
16:42 Whiteknight and what's that?
16:42 purl that is probably about the level that you were asking for
16:42 Austin_Hastings Why not? It's working for me.
16:42 darbelo Whiteknight: develop slower :)
16:43 Austin_Hastings Oh.
16:43 Austin_Hastings My approach was just to ignore what else is going on.
16:43 Austin_Hastings They're doing it wrong, anyways.
16:44 Whiteknight darbelo: Oh, I thought you were going to say that I needed to pay more freaking attention
16:45 darbelo "Paying attention" is not on the 'easy' category of my book :)
16:46 NotFound I think there is a RT, not a TT
16:46 NotFound The pir line numbers thing
16:47 Austin_Hastings NotFound: Really?
16:47 Austin_Hastings How do I search rt's?
16:47 NotFound Not sure, let me check...
16:50 clinton joined #parrot
16:50 Austin_Hastings Found this RT page. I'll search here. http://rt.perl.org/rt3/Public/Search/Simple.h​tml?Query=Queue%20=%20%27parrot%27%20AND%20(S​tatus%20=%20%27open%27%20OR%20Status%20=%20%2​7new%27%20OR%20Status%20=%20%27stalled%27)
16:51 NotFound http://rt.perl.org/rt3/Publi​c/Bug/Display.html?id=38594
16:51 NotFound Uh, no, this is solved
16:51 NotFound This one: http://rt.perl.org/rt3/Publi​c/Bug/Display.html?id=40204
16:52 clinton left #parrot
16:54 Austin_Hastings Not quite the same problem I'm having, I think. But it's good to know that life sucks for other people, too.
16:59 NotFound Austin_Hastings: feel free to add a comment to that  ticket, or to create one in TT
17:01 Austin_Hastings Yeah, I'm a little leery. There's always the chance that the error was "somewhat right", since the bogus number in this case points to a die inside an exception handler. Maybe the line numbers are right and the exception handling stack is wrong. :( I've snapshotted the code, so I think I can post an overbloated test case if need be.
17:07 Austin_Hastings And what does "positional inside named args" mean?
17:27 dalek TT #740 created by doughera++: [PATCH] PLATFORMS update for 1.3 release for Solaris 8/SPARC
17:55 PacoLinux joined #parrot
18:04 dalek TT #741 created by Austin_Hastings++: Line numbers in backtraces are substantially wrong.
18:07 PacoLinux joined #parrot
18:10 cotto Austin_Hastings++ for filing that
18:12 sekimura joined #parrot
18:25 Austin_Hastings left #parrot
18:37 Theory joined #parrot
18:39 NotFound Someone here reads german Linux Magazine ?
18:53 ruoso joined #parrot
18:54 rdice joined #parrot
18:57 masak joined #parrot
18:57 burmas joined #parrot
19:06 burmas joined #parrot
19:21 PacoLinux joined #parrot
19:24 particle joined #parrot
19:32 cotto L1?
19:32 purl L1 is a hypothetical language that would be used to implement PMCs and PIR-visible ops so that they could all be easily jitted. or http://irclog.perlgeek.de/p​arrot/2009-04-21#i_1083550 or http://rt.perl.org/rt3/Ticket/D​isplay.html?id=39313#txn-471982
20:37 nopaste "tene" at 97.117.57.131 pasted "example of importing in PIR for japhb++" (8 lines) at http://nopaste.snit.ch/16793
20:45 dalek parrot: r39407 | tene++ | trunk/runtime/parrot/languages/parrot/parrot.pir:
20:45 dalek parrot: Add a utility method to the parrot compiler to load libraries into the local namespace
20:45 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39407/
20:50 particle joined #parrot
20:58 Austin_Hastings joined #parrot
21:29 burmas joined #parrot
21:32 eternaleye joined #parrot
21:33 Whiteknight joined #parrot
21:45 mikehh joined #parrot
21:46 * Coke fails to review
21:46 particle left #parrot
21:46 burmas joined #parrot
21:51 bacek joined #parrot
22:02 cotto bacek, ping
22:16 snarkyboojum joined #parrot
22:24 bacek cotto: pong
22:24 cotto I'd appreciate your comments on my post to parrot-dev about the build process
22:25 bacek good morning
22:25 purl Here I am, brain the size of a planet, and all they say is 'Good Morning'
22:25 bacek Just read it. Thinking in progress...
22:32 rg1 joined #parrot
22:34 bacek cotto: I think second approach is more sane
22:34 bacek and simpler
22:35 cotto Can you think of anything that needs to be addressed or fleshed out better?
22:36 bacek not now. I didn't have my coffee yet...
22:37 cotto no rush.  it's worth taking time to figure it out well
22:37 Austin_Hastings Umm, how about "why?"
22:37 bacek Austin_Hastings: "why" what?
22:37 purl "why" is yet unanswered.
22:37 Austin_Hastings Why does the pmc compiler need to be parrot hosted?
22:37 cotto Austin_Hastings, a couple reasons:
22:38 cotto First, we're trying to get rid of perl5 as a build dependency.
22:38 cotto Second, having a legitimate compiler will make it much easier to switch to L1
22:38 Austin_Hastings What's L1?
22:38 purl L1 is probably a hypothetical language that would be used to implement PMCs and PIR-visible ops so that they could all be easily jitted. or http://irclog.perlgeek.de/p​arrot/2009-04-21#i_1083550 or http://rt.perl.org/rt3/Ticket/D​isplay.html?id=39313#txn-471982
22:39 Austin_Hastings Ok.
22:39 Austin_Hastings So again, why? There's already a working programming language at the bottom of the build stack: C
22:40 cotto By using L1, we can generate either C, LLVM bytecode or something else.
22:40 bacek Austin_Hastings: think about JIT
22:40 Tene Writing PMCs in terms of Parrot ops allows for JIT.
22:42 Austin_Hastings That's not my question. My question is, why write another PMC compiler; (you say: remove perl5 from build stack, okay) and given that, why write it in something other than C?
22:43 cotto We've got some nice compiler tools.  Why not use them?
22:43 Austin_Hastings Because your target environment doesn't have them, but you know it does have a C compiler?
22:45 snarkyboojum joined #parrot
22:46 cotto If you want to write one that's fine, but it's *much* easier to use Parrot
22:46 cotto 's tools, since we know we'll have them.
22:46 cotto s/you/someone/
22:47 Austin_Hastings Speaking as someone who's trying to implement a language using parrot's tools now, I'm not sure I entirely agree with that statement. :-X
22:48 cotto There's a learning curve, but it's easier. ;)
22:48 kid51 joined #parrot
22:48 bacek cotto: btw, after bootstrapping we can replace IMCC with PCT based PIR compiler :)
22:48 cotto and involves fewer segfaults
22:49 cotto bacek, that'd be a pretty cool
22:49 cotto although kj would be disappointed
22:49 bacek why?
22:49 eternaleye joined #parrot
22:50 bacek We will use his pir compiler as base
22:50 cotto he's been working on pirc, which is (was?) meant to replace imcc as a from-scratch less crufty alternative
22:50 bacek He also implemented PCT base PIR parser :)
22:50 bacek s/base/based/
22:52 bacek So, we can merge two approaches - reimplement kj' pirc in NQP using kj's PIR grammar
22:52 bacek ...
22:52 bacek 3. Profit!
22:54 cotto so we'd have two pir compilers?
22:54 cotto or we'd require a working parrot to bootstrap?
22:56 tetragon joined #parrot
22:57 Tene He hasn't been working on pirc in quite a while.
22:58 bacek imcc will be used to bootstrap only.
22:59 bacek PCT based PIR compiler for everything else
23:00 Tene I started trying to work on pirc, but ran into some issues and los tmomentum.
23:01 cotto pirc does have the disadvantage of a very low bus number.
23:02 dalek TT #733 closed by jkeenan++: Branch cleanup:  pdd30install_stage4
23:02 Tene argh... I hate composing mails to parrot-dev
23:02 cotto reply oddness?
23:02 Tene anybody want to do it for me?
23:03 cotto "Tene was going to say something."
23:03 Tene about segfaults!  make sure to mention segfaults!
23:03 Tene And GC!
23:03 cotto The segfaults are causing GC.
23:03 cotto How's that?
23:05 Theory joined #parrot
23:05 kid51 Good evening.
23:05 kid51 If I could get some feedback on https://trac.parrot.org/pa​rrot/ticket/266#comment:9, I could (re-)close that ticket.
23:06 Tene GC is causing segfaults.
23:06 Austin_Hastings Tene, aside from parrot-dev, who cares about this?
23:06 Tene kid51: I can confirm.
23:06 Austin_Hastings Pretend you're leaving them a message on here.
23:07 dalek parrot: r39408 | chromatic++ | trunk/src/embed.c:
23:08 dalek parrot: [src] Tidied code; no functional changes.
23:08 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39408/
23:08 dalek parrot: r39409 | chromatic++ | trunk/src/gc/alloc_register.c:
23:08 dalek parrot: [GC] Added code to null out array elements when free()ing contexts from the
23:08 dalek parrot: sized context array.  If our memory management were perfect, we wouldn't ever
23:08 dalek parrot: wonder if there's a potential double free problem here.  As it is, paranoia
23:08 dalek parrot: rules.
23:08 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39409/
23:08 dalek parrot: r39410 | chromatic++ | trunk/src/packfile.c:
23:08 dalek parrot: [src] Freed PackFile members that might get overlooked.
23:08 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39410/
23:08 dalek parrot: r39411 | chromatic++ | trunk/src/pmc/eventhandler.pmc:
23:08 dalek parrot: [PMC] Hoisted a SUPER() call in EventHandler's mark entry into an if block
23:08 dalek parrot: where it only makes sense to call it.  Avoiding needless work is good.
23:08 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39411/
23:08 Tene Mostly, I'm hoping chromatic will swoop in from the skies and fix it.
23:08 Tene ... that's a lot of GC-related commits from chromatic.
23:08 Tene o.O
23:08 * Tene updates parrot
23:08 Austin_Hastings Perhaps he has...
23:11 dalek parrot: r39412 | chromatic++ | trunk/src (3 files):
23:11 dalek parrot: [src] Tidied code; no functional changes.
23:11 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39412/
23:13 Tene Nope, it still breaks.
23:14 cotto bacek, is the idea about using imcc only to bootstrap and using the PCT-based PIR compiler for everything else yours?
23:15 bacek cotto: ahm. May be mine. But it is pretty obvious idea
23:16 chromatic joined #parrot
23:16 darbelo bacek: You mean maintaning to separate compilers or turn IMCC into a 'Not Qite PIR" of sorts?
23:17 bacek darbelo: actually turn pirc to nqp. imcc is awful...
23:17 skids joined #parrot
23:18 darbelo s/IMCC/"The one written in c"/
23:19 cotto karma imcc
23:19 purl imcc has karma of -1
23:19 darbelo karma pirc
23:19 purl pirc has neutral karma
23:20 darbelo That's all the evidence I need of pirc's superiority. QED
23:21 dalek parrot: r39413 | chromatic++ | trunk (4 files):
23:21 dalek parrot: [misc] Fixed references to runcore command-line option (TT #639, Klaus).
23:21 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39413/
23:21 bacek darbelo: just check the code :)
23:21 chromatic I wouldn't use IMCC or a C-based compiler to bootstrap.
23:22 chromatic We could distribute generated files, as noted, or even cross-platform PBCs if necessary.
23:22 dalek TT #639 closed by chromatic++: Usage information for command line options
23:23 bacek chromatic: our pbc isn't very stable. But I prefer to have single PCT based compiler too
23:24 chromatic When Allison and I last discussed this, she seemed comfortable checking in generated files as a fallback.
23:25 bacek "generated pbc" and "generated C" are slightly different
23:25 bacek But... We can generate C from PIR :)
23:25 Tene chromatic: I'm getting a segfault somewhere in the GC right now... I've been having trouble tracking it down.  Think you'll have a chance to look at it if I pos tto parrot-dev?
23:26 chromatic We have to generate C for now.
23:26 chromatic Tene, I'm working on TT #730 right now, but I can look at yours next.
23:29 Tene chromatic: you want me to send to parrot-dev, or where?
23:30 cotto So if we use a PCT-based pir compiler without a C-based compiler too bootstrap, would the parrot binary be just a pbc interpreter?
23:30 cotto s/too/to/
23:30 chromatic If you have a test case, nopaste it and if I can't get to it soon, parrot-dev is fine.
23:30 chromatic cotto, it'll also need all of the backing libraries in src/
23:31 cotto right, but nothing about compiling will be built-in
23:32 cotto that raises the importance of portable pbc
23:32 chromatic I don't follow.
23:33 chromatic Oh.  I see what you're saying.
23:33 chromatic I don't think we can get rid of the C code to implement ops and PMCs until we have working L1.
23:33 chromatic Until then the Parrot binary won't be just a PBC interpreter.
23:33 bacek we can't checkin generated PBCs into svn because thay are not very portable
23:34 chromatic Right.
23:34 bacek chromatic: it's not about ops, it's about PIR...
23:35 bacek implementation of ops and pmcs in terms of L1 is different task
23:35 Limbic_Region joined #parrot
23:35 nopaste "tene" at 97.117.57.131 pasted "GC segfault with steme and rakudo for chromatic++" (101 lines) at http://nopaste.snit.ch/16796
23:35 Tene chromatic: is that enough of a test case for you?
23:36 chromatic One sec.
23:36 chromatic Does anyone find this code sane?
23:36 chromatic $P1.unicode:"foo\x{b1}"()
23:37 Tene Is that "Do you know if this works with the current imcc?" or "Does this sound sane conceptually?" ?
23:38 chromatic The latter, because I can change the answer to the former depending on answers to the latter.
23:38 Infinoid chromatic: How would you feel about reducing the INITIAL_BUCKETS #define from 16 to 4?  It causes a significant improvement in perl6 -e 'my $i = 0; while ( $i < 1000000) { $i++ }' and seems to have either positive or no effect on every other benchmark I can find.
23:38 Tene It looks fine to me.
23:38 chromatic Infinoid, no argument here.
23:41 cotto +1 to the unicode method call as long as we don't end up with a lot of code that looks like that
23:41 chromatic Okay, that's three votes for.  I'll make it work.
23:44 dalek parrot: r39414 | Infinoid++ | trunk/src/hash.c:
23:44 dalek parrot: [core] Tweak initial hash buckets (from 16 to 4).
23:44 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39414/
23:44 dalek parrot: r39415 | Infinoid++ | trunk/PLATFORMS:
23:44 dalek parrot: [PLATFORMS] Apply patch from doughera++ in TT #740, to remove sol8-sparc-gcc from PLATFORMS.
23:44 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39415/
23:44 dalek parrot: r39416 | jkeenan++ | trunk/t/compilers/imcc/syn/file.t:
23:44 dalek parrot: Delete DIR => cwd() key-value pair from one instance of create_tempfile(), in hope that this clears up a testing problem on Cygwin.
23:44 dalek parrot: review: https://trac.parrot.org/parrot/changeset/39416/
23:45 dalek TT #740 closed by Infinoid++: [PATCH] PLATFORMS update for 1.3 release for Solaris 8/SPARC
23:46 Infinoid flight &
23:49 dalek TT #742 created by jkeenan++: Parrot::Test::Util::create_tempfile() cannot handle DIR option on Cygwin
23:51 japhb I cannot manage to subscribe myself to parrot-dev ... any mail admins around?
23:53 kid51 japhb:  I am not a mail admin ... but what is the specific problem?
23:55 japhb I never get the subscribe confirmation email ... even though I'm using the same address that I subscribe to all the perl6-* lists, so it's not like it's failing across the board.
23:55 particle1 see you again soon, infinoid....
23:55 particle finally, no more of that - suffix.
23:58 kid51 japhb:  Hmm, yeah, maybe there's a mail problem.  FWIW, if you prefer a news interface, you can subscribe to gmane.comp.compilers.parrot.devel at news.gmane.org.

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

Parrot | source cross referenced