% Ademan has joined #parrot % iblechbot has joined #parrot % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru % Zaba has joined #parrot % Coke has left Coke!~coke@cpe-72-228-52-192.nycap.res.rr.com % Zaba_ has joined #parrot % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru % whisper has left whisper!~whisper@60.28.210.23 % mj41 has joined #parrot % kj has joined #parrot % Coke has joined #parrot % IllvilJa has joined #parrot % Zaba has joined #parrot % Zaba_ has left Zaba_!~zaba@ip102.148.adsl.wplus.ru % ruoso has joined #parrot % Zaba_ has joined #parrot % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru % AndyA has left AndyA!~andy@82.152.157.85 % AndyA has joined #parrot % wknight8111 has joined #parrot % muixirt has joined #parrot % wknight8111 has left wknight8111!~nobody@c-71-230-33-251.hsd1.pa.comcast.net % Zaba has joined #parrot % Zaba_ has left Zaba_!~zaba@ip102.148.adsl.wplus.ru % iblechbot has left iblechbot!~iblechbot@ppp-62-216-205-4.dynamic.mnet-online.de % skids has left skids!~bri@c-71-233-204-100.hsd1.ma.comcast.net % wknight8111 has joined #parrot bah, my isp is crap % kj has left kj!~IceChat7@193.1.100.110 % kj has joined #parrot hmm does Configure.pl generate makefiles? Yes. Then I assume I need to re-run it after added/removed/moved source files.. Usually. i guess usually is the key word hmm I wonder whether use v6; is necessary in all perl6 programs.. % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru % Zaba has joined #parrot % iblechbot has joined #parrot % Zaba_ has joined #parrot % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru % skids has joined #parrot Zaba_: (use p6) I believe the goal is "assume perl5 until you know otherwise", and 'use v6;' is the easiest way to enforce that. (however, rakudo doesn't support "perl5 mode" yet. % avar has left avar!avar@u.nix.is % tetragon has left tetragon!~seneca@216.126.67.44 Coke, I see. % Zaba_ is now known as Zaba % avar has joined #parrot % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru % gryphon has joined #parrot % Zaba has joined #parrot % tewk has left tewk!~tewk@ekstrom.org % tewk has joined #parrot % whisper has joined #parrot Is there a deinstall target in the Makefile ? nope. as "make install" implies, we're not entirely happy with the install process at the moment. * Zaba did not make install his svn checkout most of the developers run out of the build dir for now. ok, it seems i should remove it manually i am quite familiar to the make install process of FreeBSD ports also, the t/src/compiler.t fails in gmake test is there a list of what it installed into my box ? % paco has joined #parrot I believe MANIFEST.generated is as close as we come. (And that includes filenames from multiple platforms, so you'd see parrot and parrot.exe there.) ok, i got it, really a lot of stuff % rdice has joined #parrot jonathan: was your "PCT patch" email applied? hmm seems like the executable perl6 is broken, but not the .pbc.. with code like "sub test {}\ntest();\n" Coke: I think you mean PerlJam there (PCT patch). I'm reviewing/applying it this morning. is pbc_to_exe known to sometimes work wrong? or am I doing something wrong.. pbc_to_exe is still a little experimental, I think still, it may be worthwhile to submit a ticket to rakudobug@perl.org for this case % ccube has joined #parrot * pmichaud considers having PGE and PCT use the MD5.pbc library hurm. wrong jonathan. pmichaud: my concern is that it wasn't in RT. it's on my internal list -- it won't get warnocked. zaba: try remaking the perl6 executable first before reporting. (if you've made any local changes) Coke, actually the problem comes from parrot's internals and does not occur when using the perl6.pbc.. and no I haven't made local changes it could be a problem with the code that PCT is generating in this instance. I don't see in docs, examples, or exceptions.t how to return from exception handler to normal code (i.e. some label after pop_eh (or should you return to just before the pop_eh?)) ccube: just branch. There's no stack cleanup necesasry? after an exception is caught, it's already popped from the handler stack just a sec (nopasting example) The why do NONE of the tests in exceptions.t return to normal code flow? because we aren't testing that...? Hmm okay (yes, we probably should be) What's all this pushmark, popmark stuff? "pmichaud" at 76.183.97.54 pasted "example push_eh handling sequence" (9 lines) at http://nopaste.snit.ch/12745 I think pushmark/popmark may be deprecated they're from an older exception handling model The docs also referred to a "handled" opcode that doesn't exist correct, the exceptions implementation isn't complete yet What part's not functional yet? the parts that aren't implemented. :-) :-) sorry, I don't have a list handy. lol okay, thanks for the guidance exceptions in parrot? uh huh, my parrot svn build fails some tests By the way, the reason I'm fooling around with exceptions (besides the fun of it) is that not all PMCs honor the default set of methods (e.g. 'isa'). I was putting together a recursive container displayer and checking whether the value 'isa' Hash or Array in order to recursively iterate, but at least String PMCs don't do isa raising an exception Zaba: which failures? can you use http://nopaste.snit.ch/ to paste them? Infinoid, I've been talking about it yesterday, perl6 executable crashes with SIGABRT on program like "sub test {}\ntest();" % uniejo has left uniejo!~uniejo@langebro.adapt.dk Zaba: I remember that. does something in the testsuite fail because of that? Infinoid, I was talking about parrot test suite, and I don't think it's related so, what's failing in the parrot test suite? ccube; String implements isa. Can you provide a test case that shows your failure? "Coke" at 72.228.52.192 pasted "this prints 0" (8 lines) at http://nopaste.snit.ch/12747 Infinoid, lemme pastebin it Infinoid, http://rafb.net/p/FmgdvB89.html Coke, yes but how do I use nopaste, or would you prefer I mail it? nopaste? nopaste is at http://nopaste.snit.ch/ (ask TonyC for new channels) or http://rafb.net/paste or http://paste.husk.org/ or http://poundperl.pastebin.com/ or http://paste.scsys.co.uk/ or don't bother me while I'm eating (first one.) Zaba: what's the output of "prove -v t/compilers/tge/basic.t"? Infinoid, http://rafb.net/p/aMs9gQ16.html "ccube" at 74.70.96.161 pasted "Output: name: String Method 'i" (77 lines) at http://nopaste.snit.ch/12748 ccube: isa isn't a method should be $I1 = isa $P1, 'Hash' Zaba: yeah, that's pretty broken. what's your platform? (sorry if I asked this already) Infinoid, linux, amd64 GCC 4.1.2 zaba: what distro of linux? pmichaud, gentoo I might re-install my 64-bit os later today (I'm currently running 32-bit os to keep vmware happy.) vmware can work on 64bit systems, if you supply all needed 32bit libs ccube, there is no direction invocation of vtables, though they all (I think) have an opcode or PIR syntax you can use to invoke them. methods are a little higher level, but there isn't a guaranteed 1:1 correspondance between them. (Though, confusingly, it's possible to make that mapping in your own PMCs if you want.) I used to use vmware-server for a while oh, vmware "works" fine under 64-bit, except the clock is totally skewed and since I'm using vmware to run windows in order to run Captivate, clock skew is a big no-no it worked fine here I'm also running gentoo amd64, but I don't get that test failure sorry :) Infinoid, I didn't get it yesterday no clue what changed all tests pass for me with svn r27079 when captivate is running, windows puts itself into a "hires timer" mode, and vmware's interrupt system can't keep up with it (it's a long-standing and well known problem with vmware) * Coke stares at the "progress report" thread. oddly, vmware can handle it under a 32-bit os, just not a 64-bit one. * Zaba will make clean and rebuild any special Configure.pl arguments? (I've used the default config.) Infinoid, so did I ok, that works here, we can compare systems when I get back... dentist & % rafl has left rafl!~rafl@62.75.161.67 ccube: everything ok now? Playing around with new info - thanks for the help % rafl has joined #parrot hmm % whisper has left whisper!~whisper@60.28.210.23 Infinoid, I'll try the test suite again.. Infinoid, oh there's also a warning I get in one of tests, http://rafb.net/p/4Q9LZ362.html oh, hey, someone installed Moose. =-) hmm? All tests successful, 25 tests and 576 subtests skipped. after rebuilding it works zaba; t/tools/smartlinks depends on Moose; I for one don't have it installed. Coke, I have Moose for one project that makes heavy use of CPAN r27080 | pmichaud++ | trunk: : [pge]: : * chromatic's use.perl.org journal post showed that PGE was generating : lots of runtime concat operations that could be performed at compile : time - this change removes those for a slight (1%) speedup. diff: http://www.parrotvm.org/svn/parrot/revision?rev=27080 % rotty` has joined #parrot what's pge? pge is, like, Parrot Grammar Engine, the regexp, er rule compiler or dealing with globs. or pacific gas & electric or (: pge park) or very smart in that it knows about bsr tailcalls or Portland General Electric one of core components of parrot, or..? yes * Zaba isn't really familiar with parrot's internals and architecture PGE is the regex engine for regular expression handling in Parrot % kj has left kj!~IceChat7@193.1.100.110 % DarkWolf84 has joined #parrot it rules cuz it parses rules. hehe * Zaba can't find explanation for all the compilers/ anywhere in docs/ I think each of the compilers/ has its own README some have, some don't Zaba, have you read http://www.parrotblog.org/ cognominal, no it gives you an idea of the whole chain of compilation start with that before diving in a specific subcompiler the problem with parrot and perl6 is that info is scattered in many sites that's the problem with *the internet* well, haskell is doing better than us they have a lot of thing under haskell.org haskell's at 6.8, we're not at 6.0 yet ;) so particle's answer is a cop out. I think haskell's a bit older, yes. haskell also has a smaller target community. 27065 broke my build crap. well, but they have even more compiler than we do cognominal: so, what are you suggesting as a fix? Regarding my proposal to change the ops syntax: would folks prefer :flag1,flag2 or :flag1 :flag2 ? the latter. or batter or letter or butter :flag1:flag2 looks better to me.. :flag1 :flag2 seems more consistent with other things we've been doing pmichaud: I expect that with with our parsing fu we will have some day a meta site that will convert all others in pod6 and back to whatever we want :) order out of chaos :) this is very perlish indeed, it is :-) zaba: do you mean :flag1:flag2 over :flag1 :flag2 or simply "more colons, less commas"? % seanstickle has joined #parrot I like :flag1:flag2 I prefer the spaces. makes it easier to spot the flags for people. fair enough. but I definitely prefer colons over commas Infinoid: what don't you like about the space? good morning. morning jackass if it'll parse correctly with spaces, I don't have any problem with it (I just somehow got under the impression that it wouldn't) I like :flag1 :flag2 and I don't think the whitespace should be required. but that's just me * Infinoid agrees with PerlJam I like hamburgers and I approve of this message. Infinoid: requires a small change to the RE that parses the op sig. pm: you're running for office?!? ;) * Infinoid is pro-hamburger * Infinoid is anti-badthings space++ pj: I'm aiming to be impeached. :-) not imbroiled? ;) heh I dont suppose anyone has written a book on Parrot assembly programming... ? anything but battered and fried? seanstickle: "Perl 6 and Parrot Essentials" seanstickle: there's a book, but it's out of date. seanstickle: it's not really about "assembly programming" though PerlJam: yeah see also http://svn.perl.org/parrot/trunk/docs/book/ Perfect! La perfection est atteinte non quand il ne reste rien à ajouter, mais quand il ne reste rien à enlever see also http://svn.perl.org/parrot/trunk/examples/tutorial/ seanstickle: Why do you want a book on asm programming? % Theory has joined #parrot PerlJam: because I like asm programming on various machines It's good fun! seanstickle: you're not interested in building a compiler or anything? parrot assembly is a ton of fun, yes. PerlJam: not as yet okay, just checking but we have free savings accounts! Although I have my eye on the APL compiler some people have even been known to write regular expression engines in Parrot assembly. :-) the apl of your eye? parrot asm is the perl of assembly languages ;) Because I needs me some APL And all the ones I've found are $$$ PerlJam: excellent quote -- I need to save that for a presentation somewhere seanstickle: the version of APL we have is incomplete, to warn you. Oh yeah, I know oddly, APL isn't listed at http://www.perlfoundation.org/parrot/index.cgi#parrot_targeted_compilers pmichaud's url is at http://xrl.us/bjnfu But once I can find the APL spec, I can start building it more The amount of documentation on APL on the Internet is pathetic I am trying to report a bug and there's a bug in the bug-reporting mechanism! I execute ./parrotbug as directed in docs/submissions.pod and it just freezes I tend to just send email to parrotbug@perl.org (because my local system isn't set up for sending email) Is Cardinal still being worked on? yes, cardinal is under development, but stalled in the past few months just waiting for round tuits well at least change submissions.pod so there is a working way for people to report bugs seanstickle: you could dive right in to almost any of the language implementations and I'm sure that you contributions would be welcomed :) Fair enough I just wanted to coordinate with the maintainer, if there is one :) seanstickle: I think all of the languages have a MAINTAINER file that tells you who to talk to check languages/LANGUAGES_STATUS.pod for APL, that's me and pmichaud PerlJam: stupid me, I just found that thanks ccube: you could post a bug report about parrotbug, even. :-) But perhaps nopaste a copy of the session where you're seeing the problem. will it get into RT if I just nopaste it? no, to get it into RT one sends to parrotbug@parrotcode.org.... HOWEVER (more) there are already some RT tickets on the subject http://rt.perl.org/rt3/Ticket/Display.html?id=41601 (open) http://rt.perl.org/rt3/Ticket/Display.html?id=52374 (says it's resolved) Okay, I am talking about the original bug, a segmentation fault stack overflow it has been said that stack overflow is before hachi's time purl, forget stack overflow Coke: I forgot stack overflow hachi machi Coke, I didn't pay attention at whitespace zaba: ok. bt in gdb give endless frames on key_string in src/key.c and Parrot_Key_get_string in src/pmc/key.pmc ie, they just alternate as far back as I'm willing to see in the backtrace OOC, does the error go away if you run parrot with -G ? no k. I will send the bug report to parrotbug@perl.org % seanstickle has left seanstickle!~seanstick@vpn6.customink.com danke * particle has seen that before * Zaba likes perl6 as a language afk, lunch Infinoid: which other regex parsers? (RT # 52570) ISTR there were a few. let me look I see other tags, just nothing that uses them, necessarily. tempting to just convert all the :flags into a hash of defined flags "Coke" at 72.228.52.192 pasted "defined :flags" (17 lines) at http://nopaste.snit.ch/12749 lib/Parrot/Op.pm 391: if ( $self->flags =~ /:pic/ also tempting to declare the supported types. there's one. (hash of defined flags)++ that's exactly what I meant by "first class citizen" :) it's the only one I see. Coke: If it causes breakage somewhere else, I'm sure someone will scream or fix it or something :) Mmm. I am even running 'make test' to verify I'm not breaking anything, still. =-) oh, of course. You might even be the someone to whom I was referring :) any idea what the :base_core flag is (or will be) used for? % Theory has left Theory!~Theory@67-42-221-224.ptld.qwest.net s/to/of/ (weird grammar) * particle suspects the docs for ops2c may have something to say about that hmm infinoid@chirp parrot-trunk % ack :base_core docs infinoid@chirp parrot-trunk % ack :base_core lib infinoid@chirp parrot-trunk % * Coke sees that one of the language is claiming to be :base_core dotnet *would*. it'd be nice to see the pmc and ops grammars get closer, and start sharing parser code meh * rotty` heads home Slacker! :-) % rotty` has left rotty`!~user@81.223.119.66 * Coke ponders creating YAB to hold an "ops syntax cleanup" party. % bphillips has left bphillips!~bphillips@nat-003.digitalriver.com question about rt.perl.org - does a bug report sent to parrotbug@perl.org have to be reviewed by somebody before it show up with a ticket # here ? I sent my bug report but I don't see a ticket # for it. parrotbug@parrotcode.org Oops good thing I asked % mj41 has left mj41!chatzilla@pc-jurosz.ro.vutbr.cz % AndyA has left AndyA!~andy@82.152.157.85 and it will be visible without human review % AndyA has joined #parrot % tlb has joined #parrot It's there now : #53156: [BUG] Segmentation violation in parrot +1 1 % Theory has joined #parrot % wolverian has left wolverian!wolverian@feather.perl6.nl I believe that get_class should not throw an exception with a non-existant classname. It should just return the null PMC and let the calling code deal with it. % Zaba_ has joined #parrot In general parrot shouldn't be throwing internal exceptions that can be detected by code. You make it more difficult to write flexible code. % wolverian has joined #parrot % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru Let the HLL writers decide when they want to thrown an exception ccube: I agree with you. ccube: we've been slowly moving to that model in Parrot. that said, trapping exceptions in Parrot isn't that difficult Hmm, except exceptions look like they are a moving target to code to if you're just talking about trapping get_class, it's not a big issue if you're talking about exceptions in general, yes, there's still a bit of a moving target about it. But that's true for any development effort of this sort. It is certainly easier to test for a null PMC with defined than to push_eh, pop_eh and have a handler oh, I think it's about equal. Just a sec "pmichaud" at 76.183.97.54 pasted "get_class, exception versus null" (14 lines) at http://nopaste.snit.ch/12751 they seem to be equally flexible. The only difference is that the firsst one requires a few extra lines of code. when you're concerned about a tight loop, wouldn't a defined be cheaper than handling an exception? depends on the loop if I can push the exception handler onto the loop once (before the loop) and then pop it off when all of the items have completed, then I save the defined checks inside the loop i.e., the loop would be tighter with the exception model than it would be by checking each result for definedness % Zaba_ is now known as Zaba % mj41 has joined #parrot Okay, so that means you're NOT in favor of removing the thrown exception from get_class eventually? no, see my first comment 17:57 ccube: I agree with you. I'm just pointing out that it's not always as black-and-white as it might seem :-) ok as I said, we're slowly moving to that model in Parrot, where opcodes return null as opposed to throwing exceptions. But some changes come slowly, it seems. oh wow, I just found out why one of my messages from last year got warnocked. It never got sent. :-( that would make it difficult to reply to. "pmichaud" at 76.183.97.54 pasted "for ccube -- how I think get_class should work (June 2007)" (54 lines) at http://nopaste.snit.ch/12752 in fact, we have: http://www.nntp.perl.org/group/perl.perl6.internals/2007/10/msg40440.html pmichaud's url is at http://xrl.us/bjnke and: http://www.nntp.perl.org/group/perl.perl6.internals/2007/10/msg40477.html pmichaud's url is at http://xrl.us/bjnkg which seems to indicate that get_class should be returning null and not throwing an exception. ...and that's what Parrot currently does. "pmichaud" at 76.183.97.54 pasted "get_class returns null" (11 lines) at http://nopaste.snit.ch/12753 ccube: so, is this a documentation bug? don't know, let me check documentation bug only in the sense that it doesn't mention throwing the exception (at least in src/ops/object.ops) ... it's not throwing an exception. get_class doesn't throw an exception. at least not in my or pmichaud testing just now. =-) do you have a sample of code where it is? hold on let me create a shorter example k ccube: OOC, what are you working on with parrot ATM? Just trying to teach myself parrot very good. =-) % mj41 has left mj41!~chatzilla@pc-jurosz.ro.vutbr.cz pmichaud, Coke - I can't reproduce it now - looks like I was mistaken % mj41 has joined #parrot any chance you were accidentally using 'getclass'? (i.e., w/o underscore) getclass (no underscore) is the old opcode that we deprecated... but it might still be in Parrot somewhere possibly - I haven't been using any source code archiving while I'm noodling around it does appear that Parrot still has the getclass opcode Yup. imcc uses it, so it's not a simple rip out. say, what do you guys do to step through Parrot code. It looks like pdb is not currently up to snuff real time? nothing atm. ./parrot -t 1 foo.pir might give you something useful. (warning: copious output if you're using a library) also the 'trace' opcode trace 1 to enable tracing trace 0 to turn it off In a pinch I have been using gdb with a breakpoint on DO_OPS in the slow runcore % ruoso has left ruoso!~ruoso@195.23.92.2 % Zaba_ has joined #parrot % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru seen ambs? ambs was last seen on #parrot 1 day and 2 hours ago, saying: in the *correct way [Apr 20 10:17:18 2008] % jalbo has joined #parrot Hello. hi jalbo anyone want me to avoid working on RT#52838 and RT#52842 (user stack ops, saveall/restoreall) ? Based on chromatic's message I think we can get a big performance win if we do those and refactor bsr/ret handling. pmichaud: I'll take it. I'm sure that'd be a huge boost to tcl, too. you'll take me working on it or you want to do it? ;-) in that "I'm all for it", not "I will implement it". ^_^ okay I'm not sure that's something I'd do compentently atm. I'll grab the tickets and start on it. I think I can do it relatively quickly -- i.e., within a day or so what happened to working on the test harness? it's parrot bug whack-a-mole time again I got the impression that others were proceeding with it, so I decided to stand aside and let them go (e.g., gsoc project) if that's not the case, then I can also look at test harness there's no gsoc project (even proposed) for test harness Did you mean the failures in make test due to wrong Test::Harness version? jalbo: no oh, maybe it was a tpf grant. anyway, I had the impression that others are looking at it. maybe coke knows about a tpf grant submission i don't I think it's a retarget of an existing grant having to do with whatever cool html thingy coke mentioned a couple of weeks ago (which I've forgotten the name of by now) there's something with smolder and parrot that's for smokes right, that's it. yes, it's a TPF grant. % Ivatar has joined #parrot ok. i'm not sure how it fits in. my point is that I wanted a public display of rakudo test progresses er, progress % ambs has joined #parrot I could try doing that with TAP::HTML::Matrix, but that seems to be going by the wayside in favor of smolder to me, that's +1 if someone else is already working on smolder+parrot integration, then I'm willing to let them go a bit farther before worrying about it for rakudo Test::TAP::HTMLMatrix-- smolder++ ok, now i get it. thanks so, what happened to the roadmap? prior to learning about smolder, I had the impression that we were going in circles with respect to html-posting-of-test-results and so I was trying to get us off the bubble (roadmap) -- it's pretty well in my head, just needs some typing work I ended up having to take care of something else late last week me too, vacation :) I could get a lot more done on Parrot if I didn't have relatives. :-) well... you do live in texas. there are many options. % rhr has joined #parrot still on vacation, particle, or back already? seen rdice? rdice was last seen on #perl 1 hour and 2 minutes ago, saying: waltman: I prefer C-x M-c M-butterfly r27081 | chromatic++ | trunk: : [src] Added reference counting to stack chunks managed through : stack_push/stack_pop. This allows an optimization of recycling stack chunks : manually rather than waiting for the GC to collect them, producing an 11.73% : improvement on the Rakudo building benchmark. : Anything that uses bsr/ret and direct continuation invocation will also see : improvements. diff: http://www.parrotvm.org/svn/parrot/revision?rev=27081 pmichaud: i'm back now oh, looks like chromatic's working on it as well purl, seen merlyn merlyn was last seen on #moose 31 days and 22 hours ago, saying: ... http://methodsandmessages.vox.com/library/post/the-year-of-smalltalk.html [Mar 20 15:09:20 2008] % mj41 has left mj41!chatzilla@pc-jurosz.ro.vutbr.cz % rblackwe_ has joined #parrot r27082 | pmichaud++ | trunk: : [core]: