aha allison++ is working on that in a branch return() will make some things much simpler cizra: since I actually don't use ruby, the most helpful thing you could do for me would be to complain about things not working as you'd like, provide tests, open tickets, whatever medium you prefer. Of course, patches are always nice too. *nod* Waah. I ought to be working on the work work. 0.upto(5) {|i| puts i} Syntax error at line 1, near "0.upto(5) " there's something weird about line endings required in places they shouldn't be. I need to track it down. also... % bacek has left bacek!~bacek@123-243-38-218.tpgi.com.au % iblechbot has joined #parrot r28194 | tene++ | trunk: : [cardinal] : * upto() builtin for Ints (cizra++) diff: http://www.parrotvm.org/svn/parrot/revision?rev=28194 0.upto(5) Syntax error at line 1, near "0.upto(5)\n" Hm. Try assigning to a variable and calling upto() on that. Oh, right, I know why that's happening. There's some issues in the grammar there that I haven't worked out yet, so I restricted it to only allow method calls on variables for now to work around it. % baest has joined #parrot Wow, I should have tested that. I forgot to switch decrement to increment. irb(main):001:0>i = 5 irb(main):001:0>i.upto(10) Lexical 'i' not found Also, that irb prompt has a space after it. Should have, that is Right now, the interactive interpreter makes a new lexical environment for each line it compiles. pmichaud has plans to fix that, but has been too busy on other things to do it yet. OK, so how do I interpret whole files? just use a filename as an argument OK parrot cardinal.pbc file.rb Syntax error at line 2, near "i.upto(5) " a = 1; a.upto(5) { |i| puts i; } works for me That ; shouldn't be necessary. I'm trying to see if I can get it to parse without that required terminator right the second one Muhaha, upto counds downwards Another easy thing would be adding builtins. What are those? Almost as easy would be reworking the class hierarchy like upto() Hmm.. Are you aware that Ruby treats 5 as a full-blown object of the Fixnum class? Also, who's pmichaud? He's the main developer of the grammar engine and much of the compiler tools. OK. Who else is involved in this Ruby compiler? (Carrot would have been a nicer name) kjs wrote the first draft of the grammar, but that's it. OK Uh. Sure seems complicated. What seems complicated? r28195 | tene++ | trunk: : [cardinal] : * upto() should increment, not decrement (tene--) the compiler code : * allow a statement to lack a trailing terminator at the end of a block diff: http://www.parrotvm.org/svn/parrot/revision?rev=28195 The grammar? It is. I've cleaned it up a bit, but ruby's grammar is complicated in general. alias carrot='parrot cardinal.pbc' why 'carrot'? It sounds fine and rhymes with parrot. 'kay Umm.. Do I need the SVN access yet? r28196 | fperrad++ | trunk: : [Lua] : - more robust diff: http://www.parrotvm.org/svn/parrot/revision?rev=28196 Ruby's initialize methods don't need that "self" in 08-class-t % cosimo has joined #parrot Huh. I wasn't sure whether they did or not. I was just copying from the examples I could find on my system. Is the return value of initialize used for anything ever? I don't think so. Okay, I'll fix that. Also, why is running make test so damn slow? first make it run, then make it right, then make it fast ;) Parsing. The grammar engine hasn't been optimized much at all yet. Well, 10 lines of Ruby code shouldn't take like 10 seconds to run. By the way, the 08-class-t runs without the self. r28197 | tene++ | trunk: : [cardinal] : * The return value of initialize() isn't used. diff: http://www.parrotvm.org/svn/parrot/revision?rev=28197 cizra++ cizra: Parrot is a bit strict about commit access to their svn repo. If you want in, send in a few patches, then you'll have to sign a licensing agreement. Gosh. I'd also be open to the idea of moving cardinal to a separate repository, if you have objections to that paperwork and plan to pay a lot of attention to working on cardinal. It has to do with their sponsorhip by the perl foundation and other legal issues. The license agreement basically says "You give the perl foundation the right to release all code you commit under an open source license, it's actually your code so you can give us copyright of it, you're not under any other weird legal agreements, etc." Time for me to sleep now. g'night. ciao % Zero_Yao has left Zero_Yao!~Zero_Yao@221.221.157.5 % jisom has left jisom!~jisom@c-98-212-164-35.hsd1.il.comcast.net % Zero_Yao has joined #parrot % ruoso has joined #parrot % Zero_Yao has left Zero_Yao!~Zero_Yao@123.112.114.207 Weird stuff. I replied by mail to my bug report AGES ago, but it hasn't shown up in the tracker. Does Parrot optimize away tail recursion? % ^conner has left ^conner!~dog@cpe-66-91-57-54.hawaii.res.rr.com I think it's trivial with cps What's cps? continuation passing style actually not 100% correct but sub calls are done through continuations, not stack so to do tail recursion optimatzion, you just have to re-use the current continuation OK Sounds like I'll learn a bunch. Is there any difference between using registers or local variables? (performance, I meant) Yay! I just wrote my very own Fibonacci numbers calculator, using only registers and register operations! \o/ re performance, dunno YaY, exam passed. 2/4 done. Mine are all done :-p * moritz envies cizra % ank has joined #parrot % cognominal has left cognominal!~cognomina@82.67.232.89 % nopaste has joined #parrot % iblechbot has left iblechbot!~iblechbot@ppp-62-216-200-102.dynamic.mnet-online.de % tetragon has joined #parrot % TonyC has joined #parrot % skv has left skv!~skv__@87.242.97.68 % AndyA has left AndyA!~andy@82.152.157.85 % tetragon has left tetragon!~seneca@216.126.67.44 % AndyA has joined #parrot % ank has left ank!~ank@ppp59-167-200-77.lns1.hba1.internode.on.net % bacek has joined #parrot is there any reason why i cant seem to pass a 'package' declared pmcarray to a function? ...package declared pmcarray? in PIR or ... ? i get "Null PMC access in find_method()" yes, pir mmd-- # For referencing NULL PMC in dispatch for dereferencing. I'm not sure what a 'package declared pmcarray' is. get_global $P12, "funstack" unless_null $P12, vivify_10 new $P12, "ResizablePMCArray" that okay, got that. Where is it being used? im just pushing a pmc onto it get_global $P14, "funstack" $P14."push"($P13) the first set of lines didn't actually create funstack as a ResizablePMCArray oh.. it did a lookup for funstack, and if no value was found it uses a (temporary) ResizablePMCArray in its place. im using the compiler toolkit here so you may need to set 'isdecl' somewhere =/ its set works when i change it to lexical can I see the code that is creating the past structure? A friend asked me: how could parrot be as fast as for instance ruby, if it needs to translate the ruby source into pir and then/or directly into parrot assembler and run that then? What's a good answer? ruby does that as well? ruby does translation also, just not into PIR no cizra: tell him that compile time normally doesn't really matter compared to run time Well, we were talking about Ruby-on-Parrot unless you had a 'ruby machine' heh at some point I hope/expect that we'll be translating PAST to bytecode and skipping the PIR intermediate step. I know a Haskell machine was build at some point. lisp machine as well cizra: and that once parrot does very good optimizations all high level languages will profit from it, so there's more incentive to do it how long till we have a parrot machine? :D TiMBuS: That was one-of-a-kind fun project. moritz: sounds fun oh, pmichaud % iblechbot has joined #parrot the code creating the stack is just: PAST::Var.new( :name('funstack'), :viviself('ResizablePMCArray'), :scope('package'), :isdecl(1), ), i bet im missing something arent i D: :isdecl(1) its there.. or is the comma a problem? :isdecl(1) isn't in what you just pasted... did it mispaste =/ What are "stack machines" and "register machines"? Are stack machines just with the stack like Forth, without any registers? Is a stack available in register machines? im no expert but i would think a stack machine would have registers, probably less registers though. the intel arch would be a stack machine, right? oh no, intel is definitely a register machine. really? % tetragon has joined #parrot http://en.wikipedia.org/wiki/Stack_machine # reasonable explanation in a stack machine, there's not really an 'add dest, src1, src2' instruction instead, it would be push s1 push s2 add like forth does and postscript and the result of the 'add' would be left on the top of the stack well there you go. OK So the main difference lies in the opcode arguments. in the way operations obtain their arguments and store results, yes. Thanks pmichaud, heres the code again, on one line so it isnt cut off: PAST::Var.new(:name('funstack'),:viviself('ResizablePMCArray'),:scope('package'),:isdecl(1)) see im not crazy its totally there the code for that should be get_global $P12, "funstack" unless_null vivify_10 $P12 = new "ResizablePMCArray" set_global "funstack", $P12 vivify_10: is it not generating the set_global? nope sounds like a bug then. :-| checking... oh, try :lvalue(1) that does it wait what. % Zaba has joined #parrot % Zaba_ has left Zaba_!~zaba@ip102.148.adsl.wplus.ru % Whiteknight has joined #parrot % Whiteknight has left Whiteknight!~nobody@c-71-230-33-251.hsd1.pa.comcast.net summon pmichaud % Whiteknight has joined #parrot yapc:eu is looking less and less likely. pong pmichaud, #55506. I'm very tired of NULL pmc access in type()... patching Parrot to ignore NULL probably isn't the answer, though. pmichaud, ok. I'll try to find more info what happens... the problem here is that the assignment is returning a NULL value. wait, that's not quite it. $a=' % gryphon has joined #parrot $a='foo'; generates an exception which try {...} catches ...so then what's the return value of C? Non initialised $P $P60 = 'foo'() pop_eh: ... 'is'($P60,...) what should the value of try { ... } be if an exception occurs? * bacek reading spec % acmoore has joined #parrot % tetragon has left tetragon!~seneca@69-196-141-26.dsl.teksavvy.com pmichaud, probably 'undef' I can't find it in spec. But in S02-builtin_data_types/type.t: is((try{$foo = 'xyz'}), undef, 'Int restricts to integers'); Is has $.a, $.b even valid? I guess syntax error... Anyway, returning NULL is bad idea for me. BTW, is I have sub foo(_) and sub foo(_,_) I should able to call it with NULL or two NULLs... Getting type of argument not always required for MMD As I recall, there was at one point a dispatch problem with NULL and _ s/is/if % paco has joined #parrot % Andy has joined #parrot I'm guessing that has $.a, $.b should generate something like "$.b not found in scope" % jhorwitz has joined #parrot % cognominal has joined #parrot % cognominal has left cognominal!~cognomina@82.67.232.89 % cognominal has joined #parrot Does C89 allow bitfields? I mean, I assume it does, bitfields were even present in K&R I dont think I've seen any bitfields in use so far in Parrot pmichaud: ran into a repeated zero-length match last night with: rule stmts { [ + ]+ }; token term { ';' | \n | > } % Limbic_Region has joined #parrot pmichaud: I found a workaround, but you asked me to let you know whenever I see it. % rdice has joined #parrot Tene: got it, thanks! % Zero_Yao has joined #parrot my $b = try { die 'xyz'; }; say $b; wrong chan, sorry hi, guys, I want to set break point at static funtion such as 'compile_to_bytecode' in IMCC, but I got: ''compile_to_bytecode has no debugger info'. What is wrong? I have '-g' option to CC. does gdb normally allow breakpoints to static functions? pmichaud, I use dbx on solaris, and I can set break point at static funciton before. i don't remember ever having a problem with that with gdb then I don't know. :-) Zero_Yao: what options did you use for Configure.pl ? DietCoke, I tried both 'perl Configure.pl' and 'perl Configure.pl --inline' and I just write a simple test function, and dbx can stop at static function correctly. % uniejo has left uniejo!~uniejo@langebro.adapt.dk % TiMBuS has left TiMBuS!~Hurf@123-243-167-27.static.tpgi.com.au % davidfetter has joined #parrot % masak has left masak!~user@130.238.45.242 % Zero_Yao has left Zero_Yao!~Zero_Yao@123.112.114.207 % jjore is now known as zz_jjore % zz_jjore is now known as jjore odd I'd throw it at the list and see if anyone can help. * davidfetter missed the context % sjansen has joined #parrot ... check the logs, man. =-) logs? parrot logs? irc log? purl? parrot logs are at http://irclog.perlgeek.de/parrot/today * moritz acts as dipspurl * DietCoke yawns * davidfetter peeks at /topic and blushes % Zaba_ has joined #parrot % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru % obra is now known as SmokyTheBear % SmokyTheBear is now known as obra Jeff Horwitz | mod_parrot: link: http://www.perlfoundation.org/parrot/index.cgi?mod_parrot % Limbic_Region has left Limbic_Region!www-data@feather.perl6.nl % Zaba_ has left Zaba_!~zaba@ip102.148.adsl.wplus.ru % Zaba has joined #parrot % pdcawley has joined #parrot % cosimo has left cosimo!~cosimo@pat-tdc.opera.com % cosimo has joined #parrot % clunker3 has joined #parrot % jjore is now known as zz_jjore % zz_jjore is now known as jjore r28198 | Whiteknight++ | gsoc_pdd09: : [gsoc_pdd09] adding some details about flags, both internal and external diff: http://www.parrotvm.org/svn/parrot/revision?rev=28198 % bacek has left bacek!~bacek@123-243-38-218.tpgi.com.au % Theory has joined #parrot % IllvilJa has joined #parrot * davidfetter wonders whether oracle's using parrot somewhere IllvilJa, any hints? Not AFAIK, but that does not necessary mean "no" :-). heh * davidfetter hoping to make parrot part of postgres at some point well, get pg to use parrot in some intelligent way % kj has joined #parrot r28199 | kjs++ | trunk: : [tutorial] add solution to exercise in ep 5 diff: http://www.parrotvm.org/svn/parrot/revision?rev=28199 % purl has joined #parrot I've worked with oracle-the-company a few times, and I'm very confident saying "no chance at all". I know of one place that ever used parrot in production, and we pretty much broke it with backwards incompatible changes (not unsurprising given our version #s) r28200 | kjs++ | trunk: : [tutorial] add solution to ex in ep 5 diff: http://www.parrotvm.org/svn/parrot/revision?rev=28200 r28201 | kjs++ | trunk: : [tutorial] add solution to ex in ep 7. (and whoops: previous commit should have said 'ep 6' , not 'ep 5') diff: http://www.parrotvm.org/svn/parrot/revision?rev=28201 r28202 | Whiteknight++ | gsoc_pdd09: : [gsoc_pdd09] changes to structure diff: http://www.parrotvm.org/svn/parrot/revision?rev=28202 r28203 | kjs++ | trunk: : [tutorial] add solution to ex in ep 8 diff: http://www.parrotvm.org/svn/parrot/revision?rev=28203 % kj has left kj!~IceChat7@193.1.100.109 (was dan's workplace, btw.) % particle has joined #parrot % barney has joined #parrot r28204 | chromatic++ | trunk: : [PMC] Changed isa_str member of the vtable to a hash. This allows more reuse : of constant strings within the system and simplifies isa checks to a hash : lookup, not repeated string manipulation. MRO setup has changed slightly too, : and there's an opportunity for refactoring here by calling a new function to : initialize MRO and the isa hash from the PMC class initializer. : Note that you need to reconfigure Parrot and probably make realclean after : applying this patch. If you see segfaults in dynpmcs (especially in : languages/), you didn't make realclean. diff: http://www.parrotvm.org/svn/parrot/revision?rev=28204 % rdice has left rdice!~richard_d@CPE0014bfafbbd5-CM0011e6ecf48a.cpe.net.cable.rogers.com r28205 | chromatic++ | trunk: : [src] Copied some logic from string_equal into STRING_compare; the latter can : make a few assumptions that the former cannot, and this plus the isa_hash : commit speed up the Rakudo-building benchmark by 4.89%. diff: http://www.parrotvm.org/svn/parrot/revision?rev=28205 the latest build for perl6 explodes, even with make realclean looks like a fakecutable issue barney, ping hi cotto do you know of a good example of a language spec that I could use as a basis for the PHP spec I'm writing? Not really. Perl 6 ? heh perl6 is a whole different class of language there's a spec for lolcode... is that the same class as php? ;) I can haz languidge spek? In spirit, HQ9+ might be closer. are you thinking of a syntax spec, or full language including semantics? for syntax, i'd use ebnf a full spec, although there are a lot of places where "do the obvious thing" would be sufficient You could ask Whiteknight whether he can recomment a Wikibook as a template vim + POD seems to be working I meant as a template for content Funny. In PHP $ref1 = &$var; seems to be the same as $ref1 =& $var; The more I read the code, the more I find wacky stuff like that. I'm not quite desensitized yet, though. Do you mean that this might not have been a sound a design decision? =:) Funny. In Perl $ref1 =- $var; seems to be the same as $ref1 = -$var; what about wikibooks? wouldn' it wouldn't hurt I've only got about 7KB of documentation written, so reformatting it wouldn't be too painful. % grim_fandango has joined #parrot jonathan: I just thought of a potential lazy-list blocker -- it will be difficult to overload set_pmc_keyed from ResizablePMCArray. Our alternative would be that List has an RPA attribute instead of inheriting from RPA. and we'd have to get the various RPAs that occur to magically morph themselves into Lists at every opportunity. I guess that's not too bad, all things considered. % rdice has joined #parrot % sjansen has left sjansen!~sjansen@hq-nat2.gurulabs.com pmichaud: that's what HLL mapping is for, neh? well, to avoid RPAs from occuring the first place. does HLL mapping also map :slurpy's ? ISTR it used to, at least. if it doesn't, it should. at any rate, we still end up with RPA's coming from PGE and/or PCT. any place where parrot is creating a PMC for you, you can expect it to use the right mapping. ;(if it doesn't it's a bug) PCT/PGE, perhaps you can have them do a lookup on the HLL mapping before instantiating. that sounds a bit tricker. We have to know who the hll caller actually is. *trickier anyway, while I think we can get most RPAs to HLL map to List, I'm not convinced we can reliably get all of them. and if we want to call Parrot libraries, we probably need to be able to handle RPAs anyway. (parrot libraries) that is a whole missing parrot of parrot HLL interop, I think. "missing part" based on what was done with P6object I don't think getting RPA to work will be a bit issue -- it's just something we have to be aware of. s/bit/big/ # can't type anymore :-( pmichaud: I was pondering that lists would separately maintain an evaluated an unevaluated portion, as separate arrays. So it's cheaper if you have an iterator of some kind in there. Otherwise you have to splice all of the time. Which is probably costly. I didn't think we needed to separately maintain it, though unless you're thinking that push would be a lot faster than splice I think so; isn't a splice a case of, copy the iterator and all after it along some elements? I was thinking that it can just be a list of mixed evaluated and unevaluated portions, where we keep track of the initial evaluated portion but having it as separate lists makes sense, I guess We can do. But splicing just feels a tad expensive. Unless we work out a way to make splicing cheat under the hood and be fast. I think I'll go with the two list approach first one nice thing about the single list approach is that adding/removing generators is still a simple pop/push and I think that we build lists far more often than we index them Hmm, true. Well, maybe. :-) also, we get the right semantics from :slurpy for "free" and I'm pretty sure we'll have a lot of :slurpy % Ivatar has joined #parrot True. Hmmm. Will think on it some. :-) well, once we have the basics in place we can try different models and see what works. Wouldn't surprise me if we ultimately end up with PMCs for a lot of these :-) I think having an implementation that *works* is more a priority than one that's fast. exactly. And I guess benchmarks are the way to work out what is fast. can always throw someone at the conversion of PIR to PMC later. There's a few too many factors, to be able to just guess what will be fastest. :-) OK, going to grab some food. r28206 | fperrad++ | trunk: : [configure] : fix on Windows : patch of kid51++ diff: http://www.parrotvm.org/svn/parrot/revision?rev=28206 r28207 | Whiteknight++ | gsoc_pdd09: : [gsoc_pdd09] moving stuff around, fixing my big mess. diff: http://www.parrotvm.org/svn/parrot/revision?rev=28207 % jan has left jan!~chatzilla@89-253-66-101.customers.ownit.se % jisom has joined #parrot % jan has joined #parrot % grim_fandango has left grim_fandango!~matt@bas2-kingston08-1167934364.dsl.bell.ca % NotFound has joined #parrot Hello. r28208 | bernhard++ | trunk: : [Plumhead] : Mention --run-nqp in the synopsis. diff: http://www.parrotvm.org/svn/parrot/revision?rev=28208 Jeff Horwitz | Parrot: link: http://www.perlfoundation.org/parrot/index.cgi?parrot Jeff Horwitz | Articles of Incorporation: link: http://www.perlfoundation.org/parrot/index.cgi?articles_of_incorporation dalek's url is at http://xrl.us/bkyfj will@coleda.com | Articles of Incorporation: link: http://www.perlfoundation.org/parrot/index.cgi?articles_of_incorporation dalek's url is at http://xrl.us/bkyfj r28209 | bernhard++ | trunk: : [Plumhead PHC] : Try to add variable name binding, but not succeeding. diff: http://www.parrotvm.org/svn/parrot/revision?rev=28209 % purl has left purl!purl@sentient.life % purl has joined #parrot hi purl! % jisom has left jisom!~jisom@c-98-212-164-35.hsd1.il.comcast.net % Zaba has left Zaba!~zaba@ip102.148.adsl.wplus.ru % Zaba has joined #parrot % AndyA has left AndyA!~andy@82.152.157.85 % AndyA has joined #parrot % davidfetter has left davidfetter!~chatzilla@start.fetter.org % barney has left barney!~bernhard@dslb-084-058-167-186.pools.arcor-ip.net pmichaud: inside a pge closure {{ }} is there any way to print the current match object for debug reasons? _dumper(match) (assuming dumper is loaded.) so I need to load dumper inside the closuer {{ }} oh, it can be loaded outside of the closure, too.