Camelia, the Perl 6 bug

IRC log for #parrot, 2009-04-27

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:08 AndyA joined #parrot
00:11 eternaleye joined #parrot
00:13 Whiteknight joined #parrot
00:22 raiph joined #parrot
00:23 kid51 joined #parrot
00:24 Theory joined #parrot
00:50 bobke joined #parrot
00:50 bobke joined #parrot
01:23 samlh joined #parrot
01:23 eternaleye joined #parrot
01:36 bacek joined #parrot
01:40 dalek parrot: r38377 | jkeenan++ | trunk (2 files):
01:40 dalek parrot: Add tests for _set_intvalmaxmin() and _set_floatvalmaxmin().
01:40 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38377/
01:42 dukeleto joined #parrot
01:47 kid51 joined #parrot
02:17 raiph pmichaud: great to read npw/hackathon were outstanding. i just did a make realclean, git pull, configure, make, and got virtual memory exhausted as before. any ideas about what's up and/or how to debug? tia.
02:18 raiph exit
02:19 Infinoid building perl6.pbc does take a *lot* of memory
02:19 Infinoid about 2GB on linux/x86-64, judging from the amount of disk cache it invalidates
02:20 Theory joined #parrot
02:20 bacek_ joined #parrot
02:35 janus joined #parrot
02:40 amoc joined #parrot
03:33 eternaleye joined #parrot
04:04 eternaleye joined #parrot
04:25 mikehh joined #parrot
04:43 mikehh Smolder is still down - at least for me
04:46 petdance joined #parrot
04:49 masak joined #parrot
04:52 Santtu joined #parrot
04:57 mikehh codetest - t/codingstd/c_function_docs.t - #   Failed test 'src/dynext.c' in addition to the whole bunch of TODOs
04:59 mikehh and t/codingstd/copyright.t still fails on the 4 compilers/imcc files
05:24 flh joined #parrot
05:33 gaurav joined #parrot
06:03 uniejo joined #parrot
06:36 mikehh make -k fulltest at r38376, Kubuntu Jaunty Amd64 - testb, C, f, g, r, S, src_, run_, perl_tests PASS, codetest FAIL,
06:37 mikehh benchmark_, manifest_tests PASS, example_tests FAIL, distro_tests PASS
06:38 iblechbot joined #parrot
06:40 mikehh codetest as above
06:41 mikehh examples_tests - t/examples/pod.t - Failed tests:  4, 82-84, 86, 88, 90-97, 99, 101-102, 104, 106-107, 109-111, 113, 118-120, 122, 154
06:43 mikehh all but one in docs/book/ch09_pasm.pod, last in docs/book/ch03_pir.pod
09:14 cognominal joined #parrot
09:19 register joined #parrot
10:02 register joined #parrot
10:02 register twek!
10:07 zostay_ joined #parrot
10:16 bacek good evening
10:33 eternaleye joined #parrot
10:52 LylePerl How fast is parrot? I've come acorss the parrotbench utility, but are there any graphs and examples online?
10:55 gaurav LylePerl: There's some comparisons at http://shootout.alioth.debian.org/debian​/benchmark.php?test=all&lang=parrot
10:55 shorten gaurav's url is at http://xrl.us/beqim6
11:10 amoc joined #parrot
11:28 kid51 joined #parrot
11:46 LylePerl Am I right in thinking that you can convert PIR files to pasm?
11:46 LylePerl And pasm to pbc?
12:06 bacek LylePerl: you can convert PIR to PBC directly.
12:07 bacek ./parrot -o foo.pbc foo.pir
12:15 masak joined #parrot
12:38 jsut joined #parrot
12:40 LylePerl gaurav, bacek: thanks :)
12:40 LylePerl I wonder if I can convince those debian peeps to do those benchmarks with pbc...
12:47 DietCoke joined #parrot
12:53 Coke cotto: ping.
12:53 rdice joined #parrot
12:56 whoppix joined #parrot
12:57 Coke msg Infinoid: no, docs.parrot.org is not updated for 1.1
12:57 purl Message for infinoid stored.
12:58 Coke msg cotto: I would be grateful for a patch for partcl to update its PMCs to deal with the lack of *_type_val macros.
12:58 purl Message for cotto stored.
12:58 Coke msg cotto (my naive attempt to do so still results in build warnings)
12:58 purl Message for cotto stored.
13:01 Coke trying to wget either release of parrot (1.0, 1.1) from the urls on the download page from OSU fails.
13:02 LylePerl I put through a request for them to benchmark PBC as well, finger crossed
13:03 jhorwitz joined #parrot
13:05 gryphon joined #parrot
13:05 Coke When folks hit the download links now, do they get a download? or an ftp folder with a single item?
13:07 Infinoid an ftp folder with a single item
13:09 Coke that might explain why that download never works for me.
13:09 Coke why is not to the /actual file/ ?
13:16 Infinoid I have no idea.
13:17 Infinoid A link to the actual file would make more sense to me, and so would using http (because we already know the client who clicked the link can speak that)
13:21 PacoLinux joined #parrot
13:23 Coke Infinoid: docs.parrot.org updated.
13:26 Infinoid Coke++
13:28 Infinoid regarding #584, I can update the url redirects to point to the file pretty easily.  I can even make it use http instead of ftp.  But release_manager_guide.pod should also be updated
13:29 Infinoid Do you want me to do this, or were you hoping for some discussion first?
13:37 Coke discussion.
13:37 purl discussion is lively and amusing and utterly stupid in here this morning! or on freenode #kwid
13:37 Infinoid Ok
13:40 Coke on docs.parrot.org, we have devel, latest, and stable links.
13:40 iblechbot joined #parrot
13:40 Coke I can see no reason not to make devel just point to latest.
13:41 Coke (right now, devel and latest independently point to 1.1.0, while stable is stuck at 1.0.0)
13:41 Coke I'd just just kill devel, but someone probably already linked to it.
13:42 Coke (done. devel -> latest -> 1.1.0; stable -> 1.0.0)
13:43 Coke (If we eventually use 'latest' to mean 'svn-latest', I can separate it back out again.)
13:45 Infinoid daily snapshots would be nice.  But if by "latest" you mean "release/current", I don't think they qualify for that particular link
13:46 AndyA joined #parrot
13:46 rg joined #parrot
13:47 rg oooh. op. shiny :)
13:48 Coke rg: ?
13:49 rg mode (+o rg) by slavorg
13:49 Coke slavorg: trust no one.
13:49 slavorg huh?
13:52 slavorg joined #parrot
13:55 Coke opbots, trust dukeleto
13:55 slavorg Ok
13:55 clunker3 But I do not trust you Coke
14:03 Coke clunker3: oh, it's ON.
14:42 flh joined #parrot
14:55 dukeleto Coke: danke
15:05 Coke you're human, right? =-)
15:06 dukeleto Coke: last time I checked
15:13 Infinoid That's what counts
15:36 bkuhn joined #parrot
15:43 Theory joined #parrot
15:44 particle don't bother checking again, your word is good enough
16:14 jhorwitz joined #parrot
16:18 mikehh Abyone any idea what has happened to smolder?
16:19 mikehh Any
16:52 uniejo joined #parrot
17:01 ruoso joined #parrot
17:09 barney joined #parrot
17:11 contingencyplan joined #parrot
17:23 dalek parrot: r38378 | NotFound++ | trunk/src/jit.c:
17:23 dalek parrot: [cage] fix c++ build, limit scope of some vars
17:23 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38378/
18:00 cotto Coke, will do
18:00 cotto messages erase
18:10 davidfetter joined #parrot
18:20 rdice Is there a known segfault from the current rakudo/parrot combo?
18:20 rdice Ever since I've been trying to built it this morning, it's been not happy.
18:22 Tene lemme try on my system.
18:23 nopaste "rdice" at 72.137.84.213 pasted "monad:/usr/local/src/rakudo# m" (7 lines) at http://nopaste.snit.ch/16414
18:25 rdice Oh, bloody hell.
18:25 purl Crikey!
18:25 rdice Now it works.
18:25 rdice Well, not that I should be unhappy about this I suppose.
18:26 pmichaud so, it's working, or not working?  ;-)
18:27 rblasch joined #parrot
18:28 NotFound pmichaud: yes ;)
18:28 rdice Linked: perl6
18:28 rdice monad:/usr/local/src/rakudo# ls -l ./perl6
18:28 rdice -rwxr-xr-x 1 root staff 6502848 2009-04-27 14:26 ./perl6
18:28 rdice monad:/usr/local/src/rakudo# ./perl6 -v
18:28 rdice This is Rakudo Perl 6, revision 38367 built on parrot 1.1.0-devel
18:28 rdice for x86_64-linux-gnu-thread-multi.
18:28 rdice Copyright 2006-2008, The Perl Foundation.
18:28 rdice monad:/usr/local/src/rakudo# date
18:28 rdice Mon Apr 27 14:30:08 EDT 2009
18:28 rdice Working.
18:28 purl i heard working was a good approximation.
18:29 pmichaud yay.
18:29 pmichaud Although we need to fix that -V option to be something more useful.  :-)
18:30 particle -v, yes indeed.
18:30 japhb joined #parrot
18:36 flh joined #parrot
18:43 jrockway_ joined #parrot
18:45 rdice Is Simon's work on sqlite for rakudo bundled in rakudo or parrot, or not bundled?
18:45 pmichaud parrot, I think
18:45 pmichaud in the examples/ subdir
18:46 pmichaud I think we should probably move that to a rakudo or perl6 repository
18:47 Infinoid Do we have a 6pan yet?
18:47 Tene We don't.
18:47 Tene There have been many suggestions, but nobody has done any work on any of them.
18:47 pmichaud officially, no.  But masak++ 's "proto" project is serving somewhat in that capacity at the moment.
18:48 Infinoid If tuits are what's needed, I might just throw something together
18:48 Tene Ah, haven't looked at that yet.
18:49 Coke cotto: danke.
18:49 Infinoid cpan6.org looks fairly motivated
18:49 Coke cotto: happy to give you commit privs to partcl.
18:50 jrockway joined #parrot
18:50 jhorwitz wow pbc_to_exe is a pig
18:50 * Coke remembers a long, hard to follow thread about the difference between cpan6 and 6pan.
18:50 pmichaud jhorwitz: yes, because we have to build the entire code in memory (because I/O is b0rken)
18:51 jhorwitz yeah cc eats up > 1GB of my 512 MB on slicehost.  :-P
18:51 jhorwitz fortunately i don't need the exe
18:52 pmichaud oh, it's the cc process that eats up all the space?  Not a whole I can do about that one.  :-|
18:52 pmichaud *whole lot
18:52 particle you know c. fix gcc!
18:53 * jhorwitz removes his glove and slaps particle
18:53 Infinoid it's just compiling an enormous data structure and a header stub to bootstrap it, right?
18:54 Infinoid fsvo "structure"
18:55 jhorwitz it would seem so.  a 14 MB source file.  :-P
18:55 cotto Coke, sure.  Do you know my google username?
18:56 Coke I fail to recall
18:56 cotto pm'd
18:59 Coke pmichaud: I wonder if changing PGE's core rules should require a deprecation notice.
19:00 dalek parrot: r38379 | pmichaud++ | trunk (2 files):
19:00 dalek parrot: [pge]:  underscore is considered a match for <alpha> (RT #65138).
19:00 dalek parrot: review: https://trac.parrot.org/parrot/changeset/38379/
19:00 Coke msg pmichaud I wonder if changing PGE's core rules should require a deprecation notice.
19:00 purl Message for pmichaud stored.
19:00 pmichaud Coke: I'm not sure this counts as a deprecation.
19:00 pmichaud but I can see the argument.
19:04 HG` joined #parrot
19:08 particle1 joined #parrot
19:20 klapperl joined #parrot
19:32 Tene jonathan: ping
19:33 nopaste "tene" at 97.117.72.236 pasted "Weird array behavior..." (22 lines) at http://nopaste.snit.ch/16415
19:33 Tene pmichaud: any idea what's going on in that paste?
19:33 Tene I'm having a lot of trouble finding a way to work around that bug.
19:36 Tene pmichaud: it also shows up when I access any array element at all.  If I invoke @events[0] or [*-1], it just invokes the most-recently-pushed (or unshifted) item
19:36 acajou joined #parrot
19:37 acajou left #parrot
19:39 Theory joined #parrot
19:42 darbelo joined #parrot
19:44 pmichaud Tene: it looks to me as though it thinks all of the $_'s are the same.
19:44 Tene Yeah, looks like to me.
19:44 Tene If I push on something outside of the loop, that item works fine.
19:44 Tene Does this have to do with the auto-close semantics?
19:44 pmichaud no
19:45 pmichaud 1..10 returns an iterator
19:45 pmichaud and $_ is being set to the value of the iterator
19:45 pmichaud a new value isn't being created each time -- what is happening is that the .from value is incremented each time
19:45 pmichaud but it's the same .from PMC each time
19:45 nopaste "tene" at 97.117.72.236 pasted "another intersting variation of this bug..." (17 lines) at http://nopaste.snit.ch/16416
19:45 pmichaud so all of the $_'s are bound to the iterators .from
19:46 Tene I think this new example demonstrates something else is going on
19:47 pmichaud what do you get from output from that?
19:47 nopaste "tene" at 67.137.148.144 pasted "oops, here's the output" (10 lines) at http://nopaste.snit.ch/16417
19:48 pmichaud no, that still seems to fit with what I expect.
19:49 Tene howso?
19:49 pmichaud s/I expect/the way I described what is happening earlier/
19:50 pmichaud I'm not saying it's correct -- I'm just saying that what you're seeing is due to the way the range iterator works
19:50 pmichaud if you change it from 1..5 to 1,2,3,4,5 I suspect you'll see something different.
19:50 Tene look at the two pushes before the loop
19:50 Tene they also print '5' when invoked
19:51 Tene and it's the same when I change it to 1,2,3,4,5
19:51 pmichaud the two pushes before the loop don't surprise me.
19:51 pmichaud what are you expecting them to do?
19:52 pmichaud oh wait, I might've misread the output.
19:52 pmichaud just a sec
19:52 pmichaud yes, I did -- the 'foo' and 'bar' confused me.
19:52 pmichaud there shouldn't be any autoclose semantics taking place here.
19:52 pmichaud so I don't think that's the issue.
19:53 pmichaud I'm guessing that the closure isn't being cloned.
19:53 pmichaud in fact, if you change it to be
19:53 pmichaud my $x = { say $msg; }
19:53 pmichaud return $x;
19:53 pmichaud I bet it works.
19:54 Tene It does.
19:54 purl if you say so...
19:54 pmichaud so the problem is that .push( )  isn't cloning its values.
19:54 pmichaud I think this is a known bug.
19:55 pmichaud I'm not sure it should be doing an explicit clone -- perhaps it should just be assigning to the correct element.
19:55 pmichaud i.e., using the equivalent logic of infix:=
20:30 * Coke has an interesting cfmx coding problem. Sadly, he can't figure out how to do it.
20:45 * rg is pretty sure even cf is turing complete ;)
20:48 Coke I have code that looks like <transaction><try>stuff<catch>rollback;other stuff</catch></try></transaction> - I'm trying to make that look like <foo>stuff</foo>
20:48 cotto The best solution isn't always "write a better compiler for a better language".
20:50 particle1 coke: is there libxslt access?
20:52 Coke particle-: it's java under the hood, so there could be.
21:04 dolmen joined #parrot
21:18 NotFound Sometimes you can write a poor compiler for the language
21:30 cotto It does work for simple enough regexes.
21:33 Whiteknight joined #parrot
21:50 HG` joined #parrot
21:51 bacek good morning
21:51 pmichaud bacek: good morning
21:51 pmichaud does tomorrow look better than today?  ;-)
21:51 bacek pmichaud: definitely. Very warm Autumn :)
21:52 bacek pmichaud: I didn't quite understand about "self should be lexical"...
21:52 pmichaud bacek:  if you have
21:53 pmichaud if 1 { self.foo() }
21:53 pmichaud then the { self.foo() } part goes into a separate subroutine
21:53 pmichaud it's not a method, so there is no 'self' defined in Parrot
21:54 bacek I thinks that is marked ad "immediate" block. No?
21:55 bacek no...
21:55 jonathan immediate in PAST, not immediate at Parrot level.
21:55 jonathan self is stored in a lexical var, anyways.
21:55 bacek Ok, got the point.
21:55 bacek jonathan: not in NQP
21:55 jonathan So we can get at it.
21:55 pmichaud jonathan: right -- we're looking at NQP here.
21:56 jonathan NQP maybe not, no. But Rakudo, yes.
21:56 jonathan Ah, OK.
21:56 jonathan Adding self to NQP?
21:56 pmichaud bacek wrote a patch in his branch to add 'self' to NQP
21:56 jonathan Aha.
21:56 pmichaud but it does the naieve "map self to register self"
21:56 jonathan OK, I didn't have the context. :-)
21:56 pmichaud right.
21:56 bacek :)
21:56 jonathan Yeah, I did that once in the early days of Rakudo, and the ticket the next day quickly told me I was wrong. ;-)
21:56 pmichaud heh
21:56 pmichaud deja-vu, eh?
21:56 jonathan aye.
21:57 jonathan Well, it was the First Obvious Solution. ;-)
21:57 bacek But there is no objections about adding self to NQP?
21:57 moritz let me guess, masak wrote that ticket ;-)
21:57 jonathan moritz: I don't remember, but there's a high probability. :-)
21:57 jonathan bacek: No objections, if there's a use case...
21:57 bacek jonathan: PMC compiler in PCT :)
21:57 jonathan Though I'm also in favor of keeping NQP lightweight.
21:58 jonathan Hey, we've managed much of a Perl 6 compiler in PCT without self. ;-)
21:58 bacek jonathan: Me either. But underlying parrot provide more stuff that available from NQP
21:59 bacek jonathan: no, I don't want to write it in PIR. NQP is much better :)
21:59 jonathan Guess it's pmichaud's call really. But I see no reason why not if it's useful.
21:59 pmichaud I think I'd prefer to not create a lexical 'self' on every method when it's not needed.
22:00 * bacek wishes to have full feature optimizer for PIR...
22:00 bacek Even DCE will help...
22:01 pmichaud so I'm trying to see if I can think of a way to only add lexical 'self' when it's needed.
22:01 pmichaud unfortunately, that might involve the equivalent of get_outer_symbol in NQP... and I'm not sure I want to go there.
22:02 pmichaud bacek: anyway, the "correct" fix for now is that every method declaration should start with   .lex 'self', self
22:02 bacek pmichaud: simple flag "our $want_self := 0" on starting method. Set it in "noun".
22:03 pmichaud bacek: not _quite_ that simple.
22:03 bacek pmichaud: in Perl6 - yes. In NQP it should be enough.
22:03 bacek NQP is simple, isn't it? :)
22:03 pmichaud bacek: I'm not a big fan of flag variables.  Since I don't think NQP does any "our" flag variables as yet, I don't want this to introduce the concept.
22:04 jonathan Will be easier to do it neatly once we have context vars, perhaps.
22:04 pmichaud if we already had a bunch, I might agree.  But there are times when I'm not a big fan of expedient fixes, because then everyone starts to use them.
22:04 pmichaud and then they get out of control and it's hard to get rid of them.
22:04 dduncan joined #parrot
22:05 iblechbot joined #parrot
22:06 bacek ok, I'll benchmark ".lex 'self', self". If it is not very slow I'll implement it.
22:06 bacek gotta go to $work...
22:07 bacek see you soon.
22:07 pmichaud it won't be slow.
22:07 pmichaud it's okay to implement it.
22:14 cotto If anyone's interested in darbelo's GCoC project, our discussion takes place on http://groups.google.com/group/decnum-dynpmcs .
22:16 davidfetter joined #parrot
22:37 kid51 joined #parrot
22:40 cotto I wonder why bkuhn hangs out here.
22:41 Infinoid We're awesome?
22:43 cotto purl really should have a response to that question.
22:43 purl cotto: i'm not following you...
22:46 kid51 Is there anybody available right now who builds Parrot on Win32?
22:47 davidfetter HAI
22:47 davidfetter does rakudo build against any released parrot atm?
22:47 Infinoid I build parrot on win32 when people report bugs on mingw
22:47 Infinoid but I'm leaving momentarily
22:48 kid51 Infinoid:  Can you tell me whether or not on Win32 'make' creates a file:  ./installable_pirc.exe ?
22:49 Infinoid I'll give it a try (booting pc)
22:49 dduncan question, is there still plan to release Parrot on CPAN like v1.0.0 and earlier, or is it only being released on parrot.org now?
22:50 Infinoid I don't think so, dduncan.  Since parrot no longer bundles perl6, we're less likely to get away with that.
22:50 Infinoid (but I'm not authoritative)
22:51 dduncan I thought the rules for CPAN were loose and that projects loosely connected with Perl or from its community could also be released there
22:52 Infinoid I don't see anything about CPAN in the release manager guide
22:54 Infinoid hmm
22:54 Infinoid [release] A few cleanups in release instructions/tools, preparing for
22:54 Infinoid the next release. Remove CPAN from the release process (it's another
22:54 Infinoid package distribution, like Debian/Cygwin, etc).
22:54 Infinoid -- svn log, r36851
22:54 dduncan okay
22:56 Infinoid cotto: http://www.ebb.org/projects/parrotemacs/wiki
22:58 Theory joined #parrot
22:59 cotto Infinoid, that 'splains it.
23:04 kid51 dduncan:  rules for CPAN are loose and none of them prevent us from placing Parrot there.
23:05 kid51 But right now our focus is on demonstrating that Parrot is good tool for *all* dynamic languages -- not just Perl 6.
23:05 dduncan I understand, this is just about how the Parrot team decided to do releaes
23:05 kid51 ... even if Rakudo is best use of Parrot ATM
23:05 kid51 Yup, it's Parrot policy, not CPAN policy
23:08 kid51 Infinoid:  https://trac.parrot.org/parrot/ticket/586 is the reason for my Win32 question.
23:08 Infinoid kid51: No installable_pirc.exe built by "make", no installable_pirc.exe make rule to build it directly
23:08 Infinoid (it just finished building)
23:09 Infinoid are all those things renamed by make install?
23:09 kid51 Are there *other* installable_*.exe files created by 'make'?
23:10 kid51 ... in top-level dir
23:10 Infinoid no.
23:10 kid51 Hrrm, that's not the result I was hoping for.
23:10 Infinoid I have the standard fare of binary tools, none of them have the installable_ prefix
23:11 kid51 Thanks for looking at this.
23:11 Infinoid I have ... miniparrot.exe, parrot.exe, parrot_config.exe, parrot_debugger.exe, pbc_disassemble.exe, pbc_dump.exe, pbc_info.exe, pbc_merge.exe, pbc_to_exe.exe
23:11 Infinoid np.  that's on mingw/strawberry/winxp
23:11 Infinoid back later &
23:14 edd joined #parrot
23:14 edd hai
23:14 edd anyone awake?
23:14 purl nobody here but us chickens!
23:15 edd ok, so i found your web-page just 5 minutes ago
23:15 edd i have just finished writing a JIT compiler using LLVM for my dissertation
23:15 edd i am looking to comare llvm to parrot
23:15 edd compare*
23:16 edd (your current release link on your web page is broken)
23:16 edd making it hard to get running to try it out :p
23:18 kid51 edd:  (This doesn't directly address your report, but is relevant):  http://use.perl.org/~Whiteknight/journal/38827
23:18 kid51 Which web page did you go to?
23:19 edd http://www.parrot.org/
23:19 purl http://www.parrot.org/ is the new homepage and feather aka http://www.parrotvm.org/
23:20 kid51 Another relevant link:  http://socghop.appspot.com/student_project/​show/google/gsoc2009/dukeleto/t124022225777
23:20 shorten kid51's url is at http://xrl.us/beqmt4
23:22 kid51 edd:  Was your problem with the link marked "Current developer release"?
23:23 edd both
23:23 edd Not found - http://www.parrot.org/download
23:23 edd The page you were opening doesn't exist.
23:23 edd Try to load the page again, or move on to another page.
23:23 edd (says midori)
23:23 kid51 Well, when I clicked on Current supported release, an ftp connection was opened and I was able to download parrot-1.0.0.tar.gz
23:24 edd ok so its midori being an ass
23:24 edd i have firefox too, lemme see
23:24 kid51 Well, I was *definitely* able to get to:  http://www.parrot.org/download
23:24 edd (only switched yesterday to midori, promising, but lacking in some places)
23:24 kid51 ... but i did have a problem with "Current developer release"
23:25 edd yeh works in ff3
23:25 edd my bad
23:25 edd kid51: the links you sent me are not selling your project to me!
23:25 edd tell me something good about parrot
23:25 kid51 "The Finder cannot complete the operation because some data in 'ftp://ftp.parrot.org/releases/devel/1.1.0' could not be read or written. (Error code -36).
23:26 edd the asm looks much simpler
23:26 kid51 How about:  svn co https://svn.parrot.org/parrot/trunk
23:26 edd i could
23:26 edd :p
23:26 kid51 I personally have never downloaded from the web page (until 5 min ago).  So I'm unfamiliar with problems people might encounter there.
23:27 edd releases are better milestones for documentational purposes, dont you think :p
23:27 kid51 I agree.  But we're an all-volunteer project and cannot guarantee that every aspect of our web site will always be up-to-date.
23:28 edd sure no problems
23:28 edd now sell the project to me :p
23:28 kid51 But I just discarded the 1.0.0 tarball and was able to download the 1.1.0 tarball.
23:28 kid51 So it may be more of a problem with our ftp service than with Parrot.
23:29 rg there has already been some discussion about offering http links rather than ftp. i think wheels are in motion.
23:30 edd so, im looking to compare llvm to parrot, whats the key differences
23:30 kid51 edd:  I'm not knowledgeable about the sort of issues you're likely to be interested in/have expertise in.  That's why I suggested those links.
23:30 edd i see you have a parse tree format
23:31 edd does this imply parrot has a parser built in?
23:31 edd and tokeniser perhaps?
23:31 kid51 Once you've read them, I'd suggest posting to our mailing list, so that people like whiteknight, allison and tewk -- who are expert -- can respond fully.
23:31 edd i looked at your links, but they seem to put the vm down a lot
23:31 kid51 Do you have link to your dissertation?
23:32 edd http://students.dec.bmth.ac.uk/e​barrett/wiki/doku.php?id=edd:3c
23:34 edd i have not made source available yet
23:35 kid51 http://planet.parrotcode.org/ is also available
23:35 Whiteknight edd: LLVM is a very low-level virtual machine, it doesn't support high-level languages like Parrot does
23:37 edd there is a gcc frontend, which seems to be doing well, but other than that yes... correct
23:37 edd however it jits on many platforms
23:38 edd hi Whiteknight, just read your blog link kid51 sent
23:39 Theory joined #parrot
23:40 edd sorry to come on in here banging on about llvm by the way. I am just looking for a neutral comparison
23:41 Whiteknight edd: Let me put it to you this way: Parrot will probably use LLVM as a jit backend
23:41 Whiteknight LLVM is lower-level, Parrot is higher-level
23:41 kid51 edd:  I think we are interested in such comparisons and in learning more about LLVM -- particularly if it helps resolve our JIT problems.
23:41 edd source -> vm -> vm -> JIT/codegen
23:41 edd interesting
23:42 Whiteknight yes, comparisons and talking are very good things
23:42 edd well what i can tell you from initial looks
23:42 edd is that your vm aims to cover many languages as frontends
23:42 Whiteknight the term "Virtual Machine" is really a very broad term that covers lots of completely different things
23:42 edd llvm only really provides an API for other devs to use
23:43 edd llvm is very strongly typed compared to parrot from what i seee
23:43 edd llvm has no strings
23:43 Whiteknight edd: LLVM was originally designed to be a replacement C compiler to GCC, as a more modern thing
23:43 edd only char arrays
23:43 edd which have differing type if the array is longer
23:43 Whiteknight right, it's just a compiler for C and machine-independent assembly
23:43 edd ie the type of i32 x 4 != the type of i32 x 5
23:44 edd gotcha
23:45 edd is there an official reference to parrot changing its jit to llvm
23:45 dduncan about: kid51: "The Finder cannot complete the operation because some data in 'ftp://ftp.parrot.org/releases/devel/1.1.0' could not be read or written. (Error code -36).
23:45 dduncan I got that problem too
23:45 dduncan so I used Fetch instead to go to the ftp server and that worked
23:45 Whiteknight edd: No official reference, it is a GSOC project this summer though
23:45 edd or is Whiteknight's blog authorotive in any way? Whiteknight, are you a core developer?
23:46 Whiteknight my blog is certainly not "authoritative", but I am a committer, yes
23:46 edd ok
23:48 edd Whiteknight: mind if i ask your real name
23:48 edd (for reference)
23:48 Whiteknight Andrew
23:50 edd second name, only if you dont mind
23:50 edd i have to bibtex that article
23:50 Whiteknight bibtex? I've never been in a citation before!
23:50 kid51 edd:  Try the IRC /w command
23:50 tetragon joined #parrot
23:51 Whiteknight Andrew Whitworth
23:51 kid51 tetragon ping
23:51 tetragon pong
23:51 kid51 Still in TO?  I haven't been there in > 1 year now.  Am experiencing withdrawal symptoms.
23:51 bacek_ joined #parrot
23:51 tetragon I'm still there
23:52 Limbic_Region joined #parrot
23:52 kid51 Still diving into the innards of Darwin and Macs?
23:53 tetragon I still deal with them and get annoyed with assumptions that all the world's GNU
23:53 cotto Whiteknight, how exCITING
23:53 kid51 Do you attend BSDCAN?
23:54 tetragon No
23:54 tetragon I may go to PGCon, though
23:54 kid51 I don't, but people in my NYC BSD user group will.
23:54 kid51 What/where is PGCon?
23:55 tetragon It's a PostgreSQL conference that's the week after BSDCAN at the University of Ottawa
23:55 kid51 I liked Ottawa when we had YAPC::CA at Carlton U there 6 years ago next month.
23:57 edd tetragon: know the felling re: GNU as an OpenBSD user
23:57 edd Whiteknight: thanks
23:57 Whiteknight no problem.
23:58 edd Whiteknight: sounds like an english name?
23:59 Whiteknight Welsh, I'm told

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

Parrot | source cross referenced