Camelia, the Perl 6 bug

IRC log for #parrot, 2009-10-31

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:20 japhb cotto_work, well, I'm definitely getting some improvement, but hopefully more to come
00:20 Whiteknight joined #parrot
00:20 cotto_work I'm very happy to hear that.  Thanks!
00:22 darbelo japhb: btw, did the nqp-rx metadata work out or should I try again.
00:23 japhb darbelo, got stuck on 'make test' not working
00:23 darbelo Not a metadata thing problem, is it?
00:26 japhb darbelo, nope, I think the metadata is fine
00:26 darbelo Also, it passed for me just now.
00:27 darbelo I guess I'll look into the sudo stuff then.
00:27 japhb cotto_work, in store_stats, should it be incrementing $stats->{global_stats}{total_time} every time, even when $extra->{no_hits} is true?
00:27 japhb That feels like a bug
00:28 darbelo japhb: Could that cause >100% numbers on the output? 'cause I've seen those a few times.
00:28 cotto_work lemme check
00:29 japhb darbelo, It looks like the global time is being overcounted, so possibly
00:29 Eevee joined #parrot
00:29 dalek left #parrot
00:29 cotto_work That does sound like a bug.
00:29 dalek joined #parrot
00:31 kthakore joined #parrot
00:31 kthakore chromatic: morning
00:31 kthakore chromatic: or night :P
00:31 chromatic early evening
00:31 purl early evening is the best luxury money can't buy.
00:32 kthakore chromatic: close enough
00:32 kthakore chromatic: SDL::Video at 42% yay
00:32 chromatic I see that.  Very nice.
00:33 kthakore chromatic: I need updated PIR/PM tuts or what ever that is
00:33 kthakore chromatic: looking into SWIG to Parrot
00:35 kthakore hmm looks like you guys are busy
00:36 kthakore I am going to watch some mind rot (TV) and go to sleep
00:36 kthakore chromatic: gnight
00:42 Zak joined #parrot
00:43 cotto_work I'm out.  japhb, feel free to msg cotto with any questions.  I'll probably be on later tonight.
00:45 japhb cotto_work, OK
00:45 japhb I've got to go to daughter's soccer game anyway,
00:45 japhb so I'll BBL
00:47 patspam joined #parrot
00:55 darbelo left #parrot
00:55 diakopter left #parrot
01:08 kid51 joined #parrot
01:30 kiwichris joined #parrot
01:33 PacoLinux joined #parrot
02:00 PacoLinux joined #parrot
02:49 dalek nqp-rx: d35a707 | pmichaud++ | src/ (2 files):
02:49 dalek nqp-rx: The Regex and HLL libraries were inadvertently loading PGE.pbc (via PCT.pbc).
02:49 dalek nqp-rx: We don't need PGE, so don't load it.
02:49 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/d​35a707896ffdeff3822de70429fe76b14434e9a
02:59 dalek parrot: r42176 | jkeenan++ | trunk/t/examples/shootout.t:
02:59 dalek parrot: Don't need to maintain references to RTs that have been resolved.
02:59 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42176/
03:42 janus joined #parrot
03:44 * japhb is bak at da hak
03:48 pmichaud japhb: I'm writing up a proposal for the parrot-dev list about including nqp-rx in core parrot releases
03:49 japhb pmichaud, thank you
03:53 dalek nqp-rx: 2e786f4 | pmichaud++ | src/ (2 files):
03:53 dalek nqp-rx: Include a '&' on operator names by default.
03:53 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/2​e786f445e0a5fdcc2feffa9c9e7724543e25756
04:01 Coke so, reading jonathan's blog, is the nqprx rakudo on master, or in a branch?
04:03 pmichaud branch
04:03 pmichaud we want master to stay "working" for people who want to play with Perl 6
04:03 pmichaud (branch is named for "ng")
04:03 pmichaud s/for//
04:09 japhb cotto, first tuning commit coming your way
04:12 dalek parrot: r42177 | japhb++ | trunk/tools/dev/pprof2cg.pl:
04:12 dalek parrot: [tools] pprof2cg.pl: Stats bug fix + first tuning (42.5% less time on one benchmark)
04:12 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42177/
04:33 * Coke thinks that japhb is profiling the profiler.
04:33 pmichaud yay
04:35 japhb Coke: Well, I'm profiling the profile converter ....
04:35 japhb cotto said it was getting seriously in his way, so ...
04:38 pmichaud it is very slow, yes.
04:50 preflex joined #parrot
05:02 darbelo joined #parrot
05:18 japhb OK, second tuning pass done.
05:18 japhb Time to go watch some TV for a while ...
05:20 patspam joined #parrot
05:21 dalek parrot: r42178 | japhb++ | trunk/tools/dev/pprof2cg.pl:
05:21 dalek parrot: [tools] pprof2cg.pl: Second tuning pass
05:21 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42178/
05:48 mikehh joined #parrot
05:53 davidfetter joined #parrot
05:54 TiMBuS joined #parrot
06:04 kiwichris joined #parrot
06:04 pmichaud message sent to parrot-dev, we'll see what sort of feedback it gets.
06:04 purl Message for sent stored.
06:34 darbelo left #parrot
06:46 japhb Replied with support for your proposal.
07:02 Zak joined #parrot
07:06 japhb Third tuning pass done.
07:06 dalek parrot: r42179 | japhb++ | trunk/tools/dev/pprof2cg.pl:
07:06 dalek parrot: [tools] pprof2cg.pl: Third tuning pass
07:06 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42179/
07:28 Tene japhb: you wouldn't need to write plumage in old NQP.  You could write it, compile it from NQP to PIR with nqp-rx, and just ship the pbc.
07:28 Tene japhb: But, yes, I mostly agree that it would be much better to have a version in Parrot.
07:29 japhb Tene, I would need the regex libraries, not just the compiled PIR from my code itself.
07:29 Tene Just don't use regexes! ;)
07:29 japhb Oh yeah, that would be *easy*.  ;-{
07:29 japhb er
07:29 japhb :-P
07:33 dalek parrot: r42180 | japhb++ | trunk/tools/dev/pprof2cg.pl:
07:33 dalek parrot: [tools] pprof2cg.pl: Fourth tuning: reorder if ladder by pass frequency
07:33 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42180/
07:39 dalek parrot: r42181 | japhb++ | trunk/tools/dev/pprof2cg.pl:
07:39 dalek parrot: [tools] pprof2cg.pl: Fifth tuning: undo singleton for alias trick; causes whitespace outdent
07:39 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42181/
07:41 desertm4x joined #parrot
07:44 fperrad joined #parrot
07:49 dalek parrot: r42182 | japhb++ | trunk/tools/dev/pprof2cg.pl:
07:49 dalek parrot: [tools] pprof2cg.pl: Sixth tuning: inline store_stats_stack()
07:49 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42182/
07:49 cotto japhb, thanks a bunch for that work.
07:50 japhb np
07:50 japhb by this point, you should be noticing the difference.  ;-)
07:50 cotto what kind of overall improvement are you seeing?
07:51 japhb Haven't done a full before and after yet, but I'd estimate it's a few times faster at this point.
07:51 japhb Care to give it a try on your box?
07:52 mokurai joined #parrot
07:52 cotto I'd love to
07:57 cotto not bad.  It looks like takes about 50% as long to run after your changes.
07:57 japhb Ah, then not as much of an improvement on your system as it is on mine.
07:58 japhb Too bad.  :-/
07:58 cotto how much on yours?
08:00 japhb Don't know exactly, but given that I took off 42.5% of the time in the first pass, then about another third in the second pass ... I was getting on to a total of 3-4x faster.
08:09 cotto I wonder how fast it'd be in nqp-rx
08:14 cotto but there's no point in throwing more time at an alternate implementation that *might* be faster when there are more reliable improvements to be gained
08:14 japhb nod
08:15 cotto as shiny as self-hosted tools are
08:15 * cotto resists the shininess
08:17 * japhb continues tuning
08:17 japhb :-)
08:17 cotto thanks
08:18 cotto night
08:19 japhb good night
08:19 chromatic joined #parrot
08:40 xenoterracide joined #parrot
08:41 dalek parrot: r42183 | japhb++ | trunk/tools/dev/pprof2cg.pl:
08:41 dalek parrot: [tools] pprof2cg.pl: Seventh tuning: CS path of process_line()
08:42 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42183/
08:46 xenoterracide joined #parrot
08:53 patspam joined #parrot
08:53 japhb All right, eight pass committed.
08:53 japhb er eighth
08:54 japhb And with that, I need to get some sleep.
08:54 japhb (Oh and yes, I definitely think there is more speed to be won.  I'm just getting too blurry-eyed to keep this up.)
08:55 dalek parrot: r42184 | japhb++ | trunk/tools/dev/pprof2cg.pl:
08:55 dalek parrot: [tools] pprof2cg.pl: Eighth tuning: Move input loop into process_input() [renamed from process_line()]
08:55 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42184/
08:56 Tene okay, I think I have a good plan for a select() PMC now.
08:56 Tene maybe maybe maybe have time to work on it tomorrow.  I hope.
08:56 Tene for now, goodnight all.
09:05 bacek joined #parrot
09:42 Psyche^ joined #parrot
09:45 TiMBuS joined #parrot
09:46 dalek tracwiki: v11 | fperrad++ | ListOfPlatforms
09:46 dalek tracwiki: https://trac.parrot.org/parrot/wiki/Lis​tOfPlatforms?version=11&action=diff
09:47 Psyche^_ joined #parrot
10:06 sjohnson joined #parrot
10:48 xenoterracide joined #parrot
10:52 baest joined #parrot
10:53 xenoterracide joined #parrot
10:57 xenoterracide joined #parrot
11:18 fperrad joined #parrot
11:40 mikehh trunk - pre/post-config, smoke (#29549) PASS, fulltest FAIL at r42184 - Ubuntu 9.10 amd64
11:40 mikehh t/benchmark/benchmarks.t - Failed tests:  18-20
11:40 mikehh the remainder of fulltest passes
11:53 Austin joined #parrot
11:54 jsut_ joined #parrot
11:55 Austin seen pmichaud
11:55 purl pmichaud was last seen on #parrot 5 hours, 50 minutes and 19 seconds ago, saying: message sent to parrot-dev, we'll see what sort of feedback it gets.
11:57 xenoterracide joined #parrot
12:00 xenoterracide joined #parrot
12:04 patspam joined #parrot
12:10 dalek TT #1166 created by mikehh++: benchmark_tests FAIL from r42170 - Failed tests:  18-20
12:11 darbelo joined #parrot
12:11 mikehh messages
12:37 fperrad joined #parrot
12:42 joeri joined #parrot
12:42 Whiteknight joined #parrot
12:44 Austin Good morning, Andrew.
12:46 bacek joined #parrot
12:47 plobsing joined #parrot
12:53 kid51 joined #parrot
12:55 Whiteknight Hello Austin
12:59 dalek parrot: r42185 | bacek++ | trunk (2 files):
12:59 dalek parrot: Rewrite examples/benchmarks/oo1.pasm into pir.
12:59 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42185/
13:01 darbelo Austin: ping
13:02 dalek parrot: r42186 | bacek++ | trunk (2 files):
13:02 dalek parrot: Rewrite examples/benchmarks/oo2.pasm into pir. Closes #1166
13:02 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42186/
13:02 xenoterracide joined #parrot
13:05 xenoterracide joined #parrot
13:05 dalek parrot: r42187 | bacek++ | trunk (4 files):
13:05 dalek parrot: Rewrite examples/benchmars/oo3.pasm into pir.
13:05 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42187/
13:08 Austin joined #parrot
13:08 darbelo Welcome back Austin.
13:08 Austin Thanks, I think.
13:09 darbelo Saw your mailing list message about MDD. I think I can explain why you see an extra PMC.
13:11 darbelo Most parrot arithmetic vtables ignore whatever 'dest' argument you pass them and unconditionally create a new destination.
13:14 darbelo The PMC originally in $P2 got discarded and the new PMC containing the result of the add operation got placed there.
13:18 Austin But after the op, the addr was the same, no?
13:19 bacek Austin, no
13:19 bacek at least it shouldn't
13:19 Austin You're right. $p2-before != $p2-during != $p2-after
13:19 Austin There's three P2 values, then.
13:19 Austin http://nopaste.snit.ch/18506
13:20 Austin before=...88, after=...12, and the set_native got made to ...32.
13:20 darbelo Austin: during?
13:21 Austin And now that I think about it, that's still a little weird.
13:21 Austin There are two extra PMCs.
13:22 Austin One is the replacement for <dest>. Add is doing "create a new object and bind" rather than "add and assign".
13:22 bacek Austin, before=..88 is Null PMC
13:23 darbelo One of them is part of the buxing/unboxing/type_conversion, it wouldn't be there for an all-Integer case.
13:23 Austin I think I can make an argument that that behavior is wrong. (Better still, I think I'll let the two matrix math guys in the channel make that argument for me. :)
13:23 bacek 134982632 == $P0 == self in VTABLE
13:24 Austin Bacek: You're right. I forgot to set P2 in this example. (My original case had $P2 = new integer)
13:24 bacek Austin, it will be ignored anyway.
13:24 darbelo Austin: The argument has alredy been made. The guy doing math with big decimal number lost. He nows wastes valuable memory on every arithmetic operation.
13:25 Austin Bacek: ..2632 is self in vtable, but it does not correspond to $P1, which is the object in this case. So what object is 2632?
13:26 Austin That's the second extra PMC. It seems to be a temporary Foo object.
13:27 bacek Austin, ah. It's different registers. In different Context.
13:27 bacek hang on. nope.
13:28 darbelo Austin: .sub 'set_num' :vtable('set_number_native')
13:28 darbelo .param pmc arg1
13:28 darbelo The extra PMC is arg1
13:29 Austin darbelo: I don't print the address of arg1. I print the address of self.
13:29 * darbelo had a parsing FAIL
13:31 bacek Austin, Integer.add(DEFAULT) falling to get_number_naative. Line 354 in integer.pmc.
13:31 bacek It creates new PMC of lhs type
13:31 bacek And than call set_number_native(self.value + other.get_number)
13:35 kid51 Can someone confirm for me the following error:
13:35 kid51 line 409 docs/book/pir/ch04_variables.pod
13:35 kid51 $S2 = << "End_Token"
13:35 kid51 should be:
13:35 kid51 $S2 = <<"End_Token"
13:35 kid51 i.e., no wordspace after <<
13:36 Austin I'm thinking that is a "bug" - in the sense of not DTRT.
13:36 dalek TT #1166 closed by bacek++: benchmark_tests FAIL from r42170 - Failed tests:  18-20
13:36 kid51 I'm working my way through the printed PIR book and am spotting errors.
13:37 bacek Austin, why it's not "DTRT"?
13:37 Austin Because it doesn't know what kind of object it has. (it's the default case)
13:37 kid51 If I run a program with code as typed, I get:
13:37 kid51 error:imcc:syntax error, unexpected SHIFT_LEFT ('<<')
13:37 kid51 in file 'strings.pir' line 9
13:39 bacek Austin, and? You can always specify your own MMD function for "add".
13:39 Austin kid51: I get correct behavior with no space, and a parse error with the space. It's a bug.
13:39 bacek .sub 'add' :multi(_, 'Foo')
13:40 nopaste "Austin" at 98.235.55.43 pasted "Confirm kid51's book bug" (6 lines) at http://nopaste.snit.ch/18518
13:44 Austin Bacek: Yes, you can override MMD. But you shouldn't have to. If I have a RPA and I try to compute 7  $need = 15 + $rpa , is it going to create a new array and resize it for me?
13:45 Austin Will it do the same thing for temporary files if I try to compute a file offset?
13:45 dalek parrot: r42188 | jkeenan++ | trunk/docs/book/pir/ch04_variables.pod:
13:45 dalek parrot: Correct documentation bug:  wordspace after '<<' impeded proper heredoc parsing.
13:45 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42188/
13:46 bacek Austin, what is your expected behaviour? Throw an exception? Die horribly?
13:46 jonathan I'd hope it might think about telling you there's no applicable candidate. ;-)
13:46 Austin Sorry, 7 $rpa  above should be '@rpa'
13:46 jonathan otoh the default probably numifies the two...
13:46 Austin Bacek: ask me for a number. Do the math. Return the result.
13:47 bacek "Result" is?
13:47 jonathan So you'd geet $need = 15 + get_integer($rpa) or so
13:48 nopaste "Austin" at 98.235.55.43 pasted "Integer.pmc default add" (7 lines) at http://nopaste.snit.ch/18519
13:49 Austin jonathan: We're discussing the fact that it fails to do that. The code in question is in integer.pmc@354:  create a new other-type, and set-number-native the sum
13:49 Austin bacek: The result for an 'add' opcode? I'm hoping it's the sum of two numerical values.
13:50 bacek and result of "15 + @rpa" is?
13:51 jonathan bacek: I very much hope *not* an RPA.
13:51 allison joined #parrot
13:52 jonathan Maybe I'm just seeing this through Perl 6 goggles, but it looks to me like 15 + @rpa and @rap + 15 in Parrot would perhaps do something different?
13:52 jonathan That feels very wrong to me.
13:53 jonathan also...why is dest being passed in here, but whatever is passed being ignored?
13:53 Austin jonathan: I'm being picky about the specification because it has an impact on operator MMD at the PIR level. I don't know what @rpa+15 would do, because we haven't chased that down.
13:54 Austin That is one of life's little mysteries.
13:54 Austin Everyone asks the same question, but I don't remember seeing a good answer.
13:56 * jonathan tries to find said spec
13:56 pmichaud good morning, #parrot
13:57 jonathan Austin: Is it PDD 27 that you're looking at?
13:58 Austin bacek: Since rpa's numify inherits from FPA, both the num and int versions return #elts. So 15 + @rpa is 15 more than the number of elements currently in @rpa.
13:59 jonathan Right, that's kinda what I'd expect. But looking at this, I fear that @rpa + 15 would do something different.
14:00 Austin Except that it is currently implemented as "Create an RPA, and resize it to be 15 elements larger than this other array, then return that."
14:01 jonathan It'd kinda expect that we'd have something like: (Integer, Integer) and (Any, Integer) and (Integer, Any) all create an Integer (or whatever the current HLL's hll_map says Integer is mapped to) holding the result.
14:01 jonathan And we'd have similar for (Float, Float), (Float, Any) and (Any, Float)
14:01 jonathan But doing the same for Float instead.
14:02 jonathan Or perhaps consider taking the type of the thing *known* to be of this PMC's type as the thing to create.
14:02 jonathan But certainly not just go and create something that isn't numeric in any way at all as the result to an MMD.
14:02 jonathan erm, the result of an "add" I meant.
14:03 pmichaud parrots opcodes and vtable ops are by no means commutative
14:03 bacek jonathan, who win in (Integer,Any) vs (Any,Float)?
14:03 Austin jonathan: I'm not looking at any pdd. I started this trying to understand how to call an objects .get_integer vtable. Everything snowballed from there.
14:03 pmichaud Austin: are you wanting to call it from PIR?  It's  $I0 = ...
14:04 jonathan bacek: With what arguments?
14:04 bacek Int + Float
14:05 pmichaud $P0 = box 1;  $P1 = box 2.3;   $P2 = add $P0, $P1   #   Integer.add
14:05 pmichaud $P2 = add $P1, $P0  # Float.add
14:05 jonathan Right, Parrot curently biases leftwards
14:06 Austin pmichaud: No, I was trying to figure how to cause it to be called from NQP.
14:06 pmichaud more precisely, Parrot treats    add $P2, $P0, $P1   as being     $P2 := $P0.'add'($P1)
14:06 pmichaud Austin: there's not presently a way to do it from NQP.  In the new version of NQP, it will be
14:06 pmichaud my int $x := $y;
14:07 xenoterracide joined #parrot
14:07 pmichaud one can also do    pir::set__IP($y)
14:08 jonathan bacek: Those two would need some kind of tie-breaker, or the conflict would cause a fallback to some default.
14:09 jonathan bacek: Or just a left-bias.
14:09 allison left #parrot
14:11 xenoterracide joined #parrot
14:17 bacek jonathan, left-bias doesn't work... 1 + 1.2 will result to 2.
14:17 jonathan bacek: nod
14:18 bacek Easiest way is to add explicit Integer.add(Float) etc for core pmcs.
14:18 kid51 Here's another documentation problem:
14:18 bacek And Integer.add(Any) producing integer
14:19 kid51 line 831 of docs/book/pir/ch04_variables.pod:    $P0 = new "Array"
14:20 kid51 I get:  Array index out of bounds!
14:20 kid51 current instr.: 'main' pc 91 (strings.pir:46)
14:20 kid51 The example only works if I start with:      $P1 = new 'ResizablePMCArray'
14:21 kid51 $P0 = new "Array"    appears to be insufficient
14:22 bacek kid51, it should be RPA (or RSA)
14:23 kid51 RPA being ... ResizablePMCArray ?
14:23 bacek yes
14:23 bacek S for String
14:23 kid51 thanks
14:27 dukeleto 'ello
14:27 dalek parrot: r42189 | jkeenan++ | trunk/docs/book/pir/ch04_variables.pod:
14:27 dalek parrot: 'Array' must be either 'ResizablePMCArray' or 'ResizableString Array' for example of 'join' to work.  bacek++.
14:27 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42189/
14:27 kid51 dukeleto:  You were at pdx hackathon, correct?
14:29 dukeleto kid51: Yes.
14:30 kid51 I'm hoping that some one who participated and do a post about it on the pmgroups mailing list.
14:31 kid51 That list currently has a thread about what local .pm groups can do ...
14:31 kid51 ... and such a hackathon would be a good example.
14:32 kid51 pm_groups@pm.org
14:36 bacek Can someone with more native English than my review r42190, please?
14:37 dalek parrot: r42190 | bacek++ | trunk/src/pmc/hash.pmc:
14:37 dalek parrot: Add little bit of documentation about using different key and values types in
14:37 dalek parrot: Hash.
14:37 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42190/
14:39 * bacek fall asleep
14:43 kid51 bacek:  Done
14:43 kid51 afk
14:48 dalek parrot: r42191 | jkeenan++ | trunk/src/pmc/hash.pmc:
14:48 dalek parrot: Minor documentation touchups.
14:48 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42191/
15:01 dalek parrot: r42192 | mikehh++ | trunk/examples/benchmarks/oo3.pir:
15:01 dalek parrot: set svn properties
15:01 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42192/
15:09 Psyche^ joined #parrot
15:11 payload joined #parrot
15:12 mikehh All tests PASS (pre/post-config, smoke (#29552), fulltest) at r42192 - Ubuntu 9.10 amd64
15:19 desertm4x_ joined #parrot
15:23 tetragon joined #parrot
15:33 Austin_away joined #parrot
15:36 dukeleto kid51: yes, I could write something up.
15:53 kid51 dukeleto:  Great!
15:55 theory joined #parrot
15:58 mikehh NotFound: ping
15:59 NotFound mikehh: pong
16:00 mikehh NotFound: did you see my last comment on TT #1110
16:00 mikehh NotFound: I was just wondering if you had any advice on the matter
16:01 NotFound mikehh: I was not paying much attention to that ticket, expecting to update my ubuntu.
16:02 NotFound But I think that the root of the problem is not the compiler version: is that we are using as non const pointer a const char * pointer.
16:02 mikehh NotFound: It is just that the definition of strchr has changed re g++ (not gcc) in Ubuntu 9.10 in string.h
16:03 mikehh NotFound: I put the differences in the last comment in TT #1110
16:03 NotFound mikehh: strchr is a mess, the best we can do is not use it.
16:04 NotFound The definition looks correct for correct code: if it takes a const, it return a const
16:05 mikehh NotFound: I agree - it is referenced in  12 files in parrot - ack strchr
16:05 NotFound And with all logic, the pointed by a const char* can't be modified
16:06 mikehh with g++ now strchr has a const char * and a char * definition (9.10)
16:07 NotFound mikehh: and the previous definition was a shit for bacwards compatibility with wrong code.
16:12 NotFound I'm going to upgrade my laptop, will take a look after.
16:12 mikehh NotFound: yeah - but it doesn't help with compatability across different versions of the definition :-}
16:13 mikehh anyway - got to go into town - bbl
16:13 NotFound mikehh: and that is the reason why I say that the best solution is to drop strchr
16:19 particle joined #parrot
16:36 dukeleto particle: mornin'
16:55 iblechbot joined #parrot
17:27 particle1 joined #parrot
17:34 Coke seggggggfaulty
17:45 darbelo Good release name for 1.8
17:45 darbelo ;)
17:45 jonathan Dares you!
17:46 * darbelo doesn't trust himself to do a release.
17:46 darbelo And nether should you ;)
17:47 jsut joined #parrot
18:08 Austin_away joined #parrot
18:15 dukeleto Parrot 1.8.0 "Segfaulty" released! does have a nice ring to it
18:16 darbelo Of course, we'd have to make damm sure you *can't* make Segfaulty segfault before releasing it.
18:18 darbelo It stops being fun when it becomes an accurate description :)
18:27 Zak joined #parrot
18:27 jan joined #parrot
18:42 zostay joined #parrot
19:02 chromatic joined #parrot
19:07 dukeleto darbelo: touche!
19:08 mokurai joined #parrot
19:17 dalek parrot-plumage: 3594c61 | japhb++ | :
19:17 dalek parrot-plumage: [META] More TODOs from CPAN Meta Spec Proposals process
19:17 dalek parrot-plumage: review: http://gitorious.org/parrot-plumage/parrot-plumag​e/commit/3594c612afd35511e680f841c8abd04b2ba40368
19:29 mokurai1 joined #parrot
19:32 plobsing hi #parrot
19:57 darbelo msg dukeleto http://www.boingboing.net/2009/​10/26/another-hardcore-mom.html
19:57 purl Message for dukeleto stored.
19:57 darbelo msg dukeleto Parrot 2.0.0 "Killer Kea" released! Run for your lives!
19:57 purl Message for dukeleto stored.
19:57 darbelo
19:58 darbelo_ joined #parrot
19:58 darbelo_ msg dukeleto Parrot 2.0.0 "Killer Kea" released! Run for your lives!
19:58 purl Message for dukeleto stored.
20:12 jonathan Heh, they trimmed the story title nicely in that URL. :-)
20:22 Coke happy halloween.
20:26 dalek TT #1167 created by fperrad++: some new types not handled by gen::opengl step
20:33 Austin_away joined #parrot
20:57 chromatic Hm, the hanoi.pir benchmark seems to have increased in speed dramatically in the past year.
20:58 japhb chromatic, when you say "dramatically", what kind of improvement are you talking about?
20:59 chromatic Three times faster in trunk than at 0.9.0, per dukeleto's benchmark script.
21:03 cotto It's nice that something's getting faster.
21:04 cotto japhb, ping
21:04 dukeleto yep
21:04 japhb cotto, pong
21:04 japhb sup?
21:04 purl Yo dawg
21:04 dukeleto hanoi is a good benchmark
21:05 chromatic Are you passing any arguments to it?
21:06 cotto japhb, need to check something quickly
21:06 japhb cotto, shoot
21:06 * cotto shoots first and asks questions later
21:07 cotto I see that the profile for oofib is goofy, but I also see that it's goofy with the original output from pprof2cg.
21:08 cotto I'd originally thought that your optimizations were to blame.  Clearly not.
21:09 japhb "Supposedly" the only functional change I made was to correct the handling of cumulative time as per our discussion.
21:09 japhb It's possible that as it got really late here I made a mistake.
21:13 cotto japhb, you're golden.  The new and old versions are broken in the same way.
21:13 japhb wheee
21:13 japhb I had started taking md5sums of the output to make sure it was the same run to run, but I only thought of that halfway through.
21:14 bacek joined #parrot
21:14 * japhb afk again, ping me if you need me cotto
21:23 eternaleye joined #parrot
21:32 dukeleto the RTEMS devs have compiled Parrot and have it working on RTEMS!!!
21:32 cotto rtems?
21:32 purl rtems is the Real-Time Operating System for Multiprocessor Systems, see http://www.rtems.com/ for more.
21:36 NotFound Parrots in Space!
21:45 bacek seen chromatic
21:45 purl chromatic was last seen on #parrot 39 minutes and 20 seconds ago, saying: Are you passing any arguments to it?
21:45 bacek Good morning, btw :)
21:51 bacek msg mikehh oo1pir and oo2.pir probably require svn properties too...
21:51 purl Message for mikehh stored.
21:58 mikehh bacek: they didn't fail codetest, but I'll check
22:00 bacek mikehh, strange... I definitely didn't set svn properties for this files...
22:04 dukeleto cotto: i met the RTEMS core devs at the google mentor summit and had many cross-pollinations of good ideas
22:04 dukeleto they have given me a makefile that makes parrot able to cross-compile to RTEMS
22:05 dukeleto we now have to make our configure system learn how to spit that out
22:08 mikehh bacek: both oo1.pir and oo2.pir in examples/benchmarks have the properties set but also have csv2svn:cvs-rev set (1.9 and 1.5) respectively
22:09 bacek mikehh, interesting. Looks like those files were migrated from cvs.
22:09 mikehh joined #parrot
22:12 mikehh yeah - when did we last use cvs? - long before my time
22:14 bacek mikehh, few years ago.
22:14 dukeleto back in the olden days, we had to carry our branches up hill, by hand, both ways!
22:14 dukeleto in the snow
22:23 mikehh tags/REL_0_0_5 was 8 years ago
22:25 dukeleto interesting times on parrot-dev!
22:25 kid51 joined #parrot
22:27 mikehh joined #parrot
22:28 mikehh bah - my internet connection is dropping in and out
22:29 mikehh anyway I think I am going to install the i386 version of Ubuntu 9.10 - bbl
22:40 mikehh joined #parrot
22:42 * mikehh changed my mind - want to backup some stuff first - probably install i386 tomorrow
23:22 dalek parrot: r42193 | bacek++ | branches/context_auto_attrs:
23:22 dalek parrot: Branch for converting Context to use auto_attr
23:22 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42193/
23:24 bacek mikehh, around?
23:25 dalek parrot: r42194 | bacek++ | branches/context_auto_attrs/src/pmc/context.pmc:
23:25 dalek parrot: Add ATTRibutes to Context.
23:25 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42194/
23:25 dalek parrot: r42195 | bacek++ | branches/context_auto_attrs (8 files):
23:25 dalek parrot: Get rid of Parrot_Context structure.
23:25 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42195/
23:29 dalek parrot: r42196 | bacek++ | branches/context_auto_attrs (5 files):
23:29 dalek parrot: Split allocating Context from registers
23:29 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42196/
23:29 dalek nqp-rx: 1e17347 | pmichaud++ | src/HLL/ (2 files):
23:29 dalek nqp-rx: Rename <escape> subrule to <quote_escape>.
23:29 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/1​e17347f617d117fb643f349b91efe4c5dd0de81
23:29 dalek nqp-rx: dcee671 | pmichaud++ |  (3 files):
23:29 dalek nqp-rx: [nqp]:  Add interpolation of scalars into double-quoted strings.
23:29 dalek nqp-rx: review: http://github.com/perl6/nqp-rx/commit/d​cee6715974147f12467a37293678bd156634492
23:29 Topic for #parrotis now Parrot 1.7.0 "African Grey" is out! | Fix issues caused by the pcc_reapply merge | find out what's up with the slice opcode | Please test context_auto_attrs branch.
23:30 NotFound ETOOMUCHTITLE ;)
23:30 bacek msg mikehh Can you test context_auto_attrs branch, please? On something strange like Ubuntu/amd64? :)
23:30 purl Message for mikehh stored.
23:31 bacek NotFound :)
23:32 * NotFound poders to add '| please test winxed'
23:32 bacek "| replace NQP with nqp-rx" :)
23:33 bacek NotFound, are you on amd64?
23:33 NotFound bacek: yes, debian
23:33 bacek (or something less common than Linux/i386)
23:33 bacek NotFound, good-good. Can you test context_auto_attrs branch?
23:34 * bacek don't expect any problems anyway.
23:35 NotFound bacek: https://svn.parrot.org/parrot​/branches/context_auto_attrs ?
23:35 bacek NotFound, something like this :)
23:35 NotFound svn co'ing...
23:36 NotFound Slow, I'm distro upgrading ubuntu on the laptop
23:37 bacek to proper debian?
23:37 NotFound To 9.10
23:37 NotFound Buliding
23:37 joeri joined #parrot
23:39 NotFound Testing
23:39 * bacek found how to merges branches in git-svn with preserving history. Yay!
23:41 NotFound Winxed tests pass
23:41 NotFound parrot tests pass
23:41 NotFound Looks good :)
23:41 bacek Hooray!
23:41 * bacek merging it back to trunk.
23:42 NotFound Doing fulltest
23:44 * bacek typed in "git svn dcommit" and waiting to hit enter.
23:45 NotFound Looks a little faster than trunk
23:45 dukeleto 00:24:30 <@pmichaud> (rakudo would like nqp-rx in parrot core asap)
23:45 dukeleto taken straight from the mouth of the river
23:46 dukeleto that is from #perl6, btw
23:46 bacek NotFound, it should be slower...
23:46 chromatic I can benchmark as soon as it lands.
23:47 dukeleto so the new version of nqp-rx installs itself as "nqp" instead of "parrot_nqp", so all we need to do to use NQP is switch to that binary
23:47 dukeleto except when we use the .pbc version of nqp ...
23:47 bacek chromatic, I do more allocations than before. But this branch is required to merge Context and CallSig
23:47 NotFound bacek: I was just timing the wixed tests while runing parrot make test, not a very good benchmarking frame
23:48 chromatic Performance depends on *which* allocations.
23:48 bacek chromatic, indeed. One fixed size for attributes, one variable size for registers.
23:48 NotFound Test Summary Report
23:48 NotFound -------------------
23:48 NotFound t/pmc/eval.t                       (Wstat: 0 Tests: 17 Failed: 0)
23:48 NotFound TODO passed:   12
23:48 NotFound t/pmc/threads.t                    (Wstat: 256 Tests: 14 Failed: 1)
23:48 NotFound Failed test:  8
23:48 NotFound Non-zero exit status: 1
23:49 NotFound This is from make fulltest
23:49 bacek sigh...
23:49 NotFound The thread failure, I'm also having it intermitently on trunk
23:50 NotFound Always in amd64
23:50 bacek Ah! Hooray :)
23:52 bacek landed
23:52 dalek parrot: r42197 | bacek++ | trunk (12 files):
23:52 dalek parrot: Merge branch context_auto_attrs back to trunk.
23:52 dalek parrot: review: https://trac.parrot.org/parrot/changeset/42197/
23:52 bacek NotFound, can you remove old branch from svn?
23:52 dukeleto bacek++
23:53 NotFound bacek: better wait to some test of the merge
23:54 dukeleto bacek: your famous now http://twitter.com/parrotvm/status/5325786286
23:55 bacek dukeleto, :) It was very small branch
23:56 dukeleto bacek: praise in public, punish in private ;)
23:56 dukeleto bacek: i want the @parrotvm to announce all branch merges. they are important and it is a good way for people using parrot to know when they should check if their stuff still works
23:57 NotFound trunk looks good
23:58 ttbot Parrot trunk/ r42197 MSWin32-x86-multi-thread make error http://tt.ro.vutbr.cz/file/cmdout/128918.txt ( http://tt.ro.vutbr.cz//buil​dstatus/pr-Parrot/rp-trunk/ )
23:59 bacek Bah! Poor win32...

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

Parrot | source cross referenced