r26765 | chromatic++ | type_ids: : [PMC] Really remove type_keyed_str vtable entry, with which we have always been : at war. We love Oceana, for now. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26765 r26766 | chromatic++ | trunk: : [docs] Fixed a POD formatting error in PDD template. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26766 r26767 | chromatic++ | trunk: : [Lua] Added copyright and SVN id to Parse::Yapp template, so that it appears in : the generated (and checked-in) file. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26767 % chromatic has left chromatic!~chromatic@sub17-30.member.dsl-only.net % silug has joined #parrot % barney has joined #parrot r26768 | bernhard++ | trunk: : [codingstd] : languages/lua/Lua/parser.pm is gone, so there is no need to excempt it. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26768 % paco has left paco!~chatzilla@139.Red-80-36-122.staticIP.rima-tde.net % iblechbot has joined #parrot % jan has joined #parrot % Psyche^ has joined #parrot % Patterner has left Patterner!~Psyche@f054000150.adsl.alicedsl.de % Psyche^ is now known as Patterner % Senaka has joined #parrot hi is dod.h THE INCLUDE for gc stuff? or are there other interface files that I should be looking at? Hmm. You might want to look at docs/pdds/pdd09_gc.pod % kj has joined #parrot Tene: yes I did look. Thanks. But, the requirement is to find the methods (signatures) etc. % barney has left barney!~bernhard@dslb-084-058-141-170.pools.arcor-ip.net % barney has joined #parrot ping anyone knows? * purl *knows* Senaka in that oh so special way :) Senaka: i believe dod.h is the include for gc. spinclad: thanks % Senaka has left #parrot % barney has left barney!~bernhard@dslb-084-058-141-170.pools.arcor-ip.net % barney has joined #parrot % barney has left barney!~bernhard@dslb-084-058-141-170.pools.arcor-ip.net % Senaka has joined #parrot % Senaka has left #parrot % Senaka has joined #parrot % Senaka has left #parrot % Senaka has joined #parrot Hi hey, Senaka. hi blocking or parrot == locking ? blocking on parrot == locking ? I'm talking interms of the dod.h What I want to know is whether calling Parrot_block_xx() is like acquiring a lock? Senaka: I don't know much about the GC, but I think when you're blocking it, the GC will stop but I'm not entirely sure what you're working on, and what's "xx" xx means any word(s) just a generic way to represent a set of methods ooh ok :) in that case I won't be of much help ok maybe check out an implementaton of block_xx? and see what it's doing hmm.... like yes I see what I want to clarify is whether it is essentially the same as telling the GC.... I point to the object don't sweep it? I can't say. but it doesn't sound strange looks like the 'xx_block_level's act as locks on the whole arena, blocking GC or DOD not on single objects ok... yep that's what I meant I think it has to do with the term "infant mortality", but that's only what I read ages ago like telling the whole arena is locked so I 'might be horribly wrong hmm If the block level is 0, a DOD run will take place. Otherwise, it's blocked from doing so. The reason it's a level is in case nested things increment and decrement it. Or rather, nested things block and unblock it. ok that makes perfect sense :-) It's because sometimes a DOD run might kill stuff you haven't anchored anywhere yet. So occasionally you don't want to block it. ... and that is what is meant by "infant mortality", isnt' it? so is this a stop-the-world scenario that perhaps the incremental GC would not have? erm occasionally you want to block it kj: yes, same issue. hmm seems Senaka: Well, it is an issue for any GC scheme I guess. If some code has unanchored stuff around, then you want the GC to not run for the short bit of time while you anchor them. yes I'm trying to match the interfaces between Harmony and Parrot GC so wanted to be sure that I got things right perhaps what might be different for incremental GC is the separate phases of DOD and GC hmm... not exactly sure about it DOD and the sweeping is separate today in Parrot. yep... so even the locking is separate? yes I guess but I think the blocking should always happen, no matter what GC algorithm; the initialization of objects should always be done as an atomic operation, as far as the GC is concerned yes my issue is to identify how you do things and find the corresponding match in Harmony the issue is that you have no single way of implementing it So wanted to clarify rather so, i don't know much about GCs, but it might be worth the trouble to list all API functions of one implementation, describe what each does and then do the same for another implementation we have a single primary implementation we use yeah, but I meant the harmony one which is not used by parrot, right? Yes, once this is done. It should make someone's life pretty easy in understanding Parrot's GC kj: no kj: I'm supposed to get it working yes, that's what I thought but it might be hard to map harmony functions to parrot's gc functions because their implementation (of the gc's) might be very different kj: :( yep... I'm having a time so that the APIs are used differently not very but hard to find patterns these tasks are typically not completed in short time unfortunately, but it takes time for brains to find patterns. In my experience anyway :0 :) one more question, one more question, is there a way to represent a '/' in a url when sending it to a cgi prog? eg http://server/CGI/add?/index.html like say can you run this in parrallel concurrent threads? no? well, maybe. purl, forget one more question spinclad, I didn't have anything matching one more question purl, forget one more question, spinclad: I forgot one more question, % AndyA has joined #parrot % tewk has joined #parrot we intend GC to be pluggable, but you may well uncover hidden assumptions in how we've done it to date. you may have to understand both harmony and our current GC in some detail to reconcile them yes read the code and ask here and on the list. when someone who knows the gc system is around, they can answer... ok you do have another channel right? no, this is best for parrot questions ok parrotsketch? i heard parrotsketch was 18:30 UTC, Tuesdays or a weekly status meeting for parrot design team mailing list is ... parrot-porters@perl.org ? yep I'm on it and what purl said will ask on list IRC is kinda dependent on, who's around. :) I think I'm a bit early today I only know the GC roughly...I know what it does, but at a code level I'm less familiar. ok * spinclad -> bed r26769 | jonathan++ | trunk: : [rakudo] The first little bits of captures. Parse the \(...) syntax and create a new capture based upon that. Uses the Capture PMC, though it needs some fixes. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26769 % rdice has joined #parrot % kid51 has joined #parrot % cosimo has joined #parrot r26770 | jonathan++ | trunk: : [rakudo] Implement my Type $x parsing, and assignment of the prototype object. That means we can do my Foo $x; $x.new(), or $x ~~ Foo == true, and such stuff. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26770 % askie has left askie!~askie@81.171.100.208 jonathan: so cooool! % askie has joined #parrot r26771 | kjs++ | trunk: : [pirc] update grammar spec for pasm: also allow .lex directives in PASM (but this grammar is not used, it's meant to be a "clean reference"; maybe later it can be used.) diff: http://www.parrotvm.org/svn/parrot/revision?rev=26771 % kj has left #parrot got on my macbook svn: Failed to add directory 'languages/hq9plus': object of the same name already exists seems a case insensitivity problem indeed cognominal: if you're not working in that area, remove it and re-up. cognominal: We saw that problem last week and anyone who updates *across* that set of revisions will always see it. :| Yes, Coke's advice is what we had to do; it will clear up. that's waht I did thx we know who did it if you wish to throw rotten tomatoes. =-) * Coke yawns. % cosimo has left cosimo!~cosimo@87.238.46.253 * kid51 passes Coke a mug of French roast coffee cognominal proposes an irish coffee irish coffee at 7am... wuf. seems to me that Parrot_(un?)block_GC_.+ has semantics similar to an rwlock % tetragon has joined #parrot or other PV semaphores but not quite. these functions won't block if a dod/gc run is already running, they'll just prevent a new one from spawning that sounds racey r26772 | jkeenan++ | ports: : Add test to guarantee that auto::ports is skipped on non-Darwin OSes. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26772 r26773 | jkeenan++ | ports: : Create test of case where Macports directories cannot be located. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26773 % barney has joined #parrot % Senaka has left #parrot r26774 | jkeenan++ | ports: : Create test of case where directory structure needed for Macports is located. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26774 r26775 | jonathan++ | trunk: : [rakudo] Implement .= operator (call method and assign result). diff: http://www.parrotvm.org/svn/parrot/revision?rev=26775 % Senaka has joined #parrot t/doc/pod.t warns about docs/pdds/pdd_template.pod, and says I should use podchecker to check the failure. but podchecker doesn't seem to have any problem with it. r26776 | jkeenan++ | ports: : Add files to test auto::ports in cases where verbose output has been : requested. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26776 r26777 | bernhard++ | trunk: : Remove trailing whitespace. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26777 Infinoid: I had seen something like that too, but don't remember the reason Might be a config picked up by pod.t Infinoid: I'll take a look at that. I was making files conform to t/codingstd/pdd_format.t last night and may have broken something. Is there anything to paste? r26778 | jonathan++ | trunk: : [core] Fix capture PMC enough to allow assignment of captures in Rakudo. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26778 let me reconfigure and find out kid51: I have a crude patch to drop the arch flags from the Perl 5 config data on Darwin that shouldn't affect people explicitly passing them into Configure r26779 | bernhard++ | trunk: : [Pheme] : Regenerate Makefile when template has changed. : Remove $(TOOL_DIR) as it had misleading name. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26779 Does this fit into the subject matter of one of the RTs I or you opened since to.pm? If so, then you can respond to that item on list and include patch as email attachment (recommend: file name ending in .txt). It's the lipo ticket you opened kid51: that test failure disappeared after I did a realclean/configure. false alarm? Infinoid: I'll check in a minute. kid51: I was having difficulties earlier with emailed responses to a ticket not getting through Let's diagnose. % Senaka has left Senaka!~senaka@124.43.203.54 tetragon: I see that you made an RT entry on http://rt.perl.org/rt3/Ticket/Display.html?id=52214 this morning. How did you effect that? The account I created after my email failure tetragon: What type of account? Bitcard? it has been said that Bitcard is our new auth system. Bitcard Okay, so you were working from the RT's GUI interface, correct? Yes % Senaka has joined #parrot Ok, then do another post there, remembering (a) to click on the 'Click to toggle Cc of perl6-internals' link; and (b) click on the Browse button to locate the patch you wish to post. Hrm... I didn't get an email for my reply this morning r26780 | bernhard++ | trunk: : Coding std fixes. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26780 You don't get an email confirmation for every post -- just in response to the email that *creates* a new RT I also know that my current patch is probably not in the form that would be used Infinoid: After make realclean and svn up, t/doc/pod.t passes cleanly for me. yeah, me too. tetragon: Understood. These issues peer into the murky depths of Darwin, which we don't understand nearly as well as we need to. In particular, as you have documented, there are weird differences between ppc and intel OTOH and between 10.3/10.4/10.5 OTOH that we don't understand. So don't worry about producing a finished product. Quite a few people will have to weigh in before something like this is applied. The patch is rather simple % Senaka has left Senaka!~senaka@124.43.203.54 I've sent it up % Senaka has joined #parrot But getting it to work on the various Darwins may not be simple. It took > 2 months for us to understand why a patch Allison, the Project Architect wanted, wasn't working for me -- and that was a very short patch as well. * kid51 would really welcome having Parrot persuade Apple to have 1 developer put 10 hours/month into porting to darwin. % Limbic_Region has joined #parrot tetragon: I see the patch. And since you hit the "Cc" toggle, a copy will go to list. The patch looks well constructed, but we'll have to consider 2 questions: 1. Is ignoring the '-arch' flags the correct thing to do in all circumstances? 2. Should this code be in config/init/hints/darwin.pm -- a Darwin-specific module -- rather than in config/init/defaults. It only ignores the '-arch' flags obtained via Perl 5's configure, it doesn't deal with those from the command line (if my understanding of Configure.pl is correct) 2. Should this code be in config/init/hints/darwin.pm -- a Darwin-specific module -- rather than in config/init/defaults.pm -- which is a trans-OS module? And as for config/init/hints/darwin.pm, that's the thing I was wondering about I don't want people who want to pass in -arch flags to lose that option % malte has joined #parrot Okay. As I said, there are a number of people who I expect will have thoughts on this. Thanks for investigating the issue. % kid51 is now known as kid51_afk * barney trying to rip PAST-pm from pheme * barney only two failures left % davidfetter has joined #parrot % grim_fandango has left grim_fandango!~matt@bas2-kingston08-1167933634.dsl.bell.ca r26781 | jkeenan++ | ports: : Configuration element is now known as ports_root rather than macports_root. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26781 Hrm... t/stm/runtime.t test 4 has stopped failing consistently Erm, crashing It didn't crash a couple of the times when I ran it Configure.pl --cc=g++ is not successful "Verifying that the compiler supports function pointer casts...Although it is not required by the ANSI C standard, Parrot requires the ability to cast from void pointers to function pointers for its JIT support. Your compiler does not appear to support this behavior with the flags you have specified. You must adjust your settings in order when I try it, I get a bunch of "Linker failed" messages. maybe ld needs -lstdc++ on the command line yes they are among the other messages Use of uninitialized value in numeric eq (==) at config/auto/pack.pm line xx are some others % Andy has joined #parrot Infinoid:yes undefined reference to `__gxx_personality_v0' is the error r26782 | fperrad++ | trunk: : [install] : - dotnet builds again diff: http://www.parrotvm.org/svn/parrot/revision?rev=26782 ok done. Configure.pl --cc=g++ --link=g++ r26783 | fperrad++ | trunk: : [Lua] : - distro diff: http://www.parrotvm.org/svn/parrot/revision?rev=26783 use of C++ keywords as formal parameter names breaks build on C++ using the '_' (underscore) convention in patch is it ok? r26784 | jonathan++ | trunk: : [rakudo] Initial implementation of runtime type-checking of assignments. Missing a trivial compile-time optimization and certainly doesn't work for PMC types and constraints, but it works OK for checking what you assign is compatible with the named class for class types or, if it's a role name specified as the type, that what is being assigned does that role. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26784 Senaka: that sounds fine. jonathan: cool, proceeding with patch nice :-) * jonathan wonders just how wrong the patch he just checked in is. jonathan: patch sent cool btw it is not @ all a good habit to use keywords in any place other than where it is supposed to. Especially in headers. This is why we find OS headers/source awfully complicated, since they use _ all over :) I think t/codingstd/cppcomments.t could use some cage cleaner attention it says it's looking for cuddled elses (lines 40 and 43) static op_lib_t core_op_lib redefined in core_ops.c lines 22 and 29933 any reasons? seen Infinoid Infinoid was last seen on #parrot 37 minutes and 16 seconds ago, saying: maybe ld needs -lstdc++ on the command line * Infinoid didn't do it! infinoid: hi privet, Senaka. hey Configure.pl --cc=g++ --link=g++ does it then came several issues :) during make I sent a patch for the keyword use scenario #52500 I'm guessing compiling with g++ is needed for your GC work? no I think We'll be connecting an adapter probably I dunno how this could be addressed... depends on what it would be after interfaces are aligned but normally say someone wants a C++ parrot... in that case you need to get these fixed * davidfetter contemplates Cb parrots C# also, some people only have C++ I think. (most of the compilers listed in README_win32.pod are C++) oh! davidfetter: *seizure* C#?? i think C# is the bastard child of Perl and Java according to Simon heh davidfetter: I already embedded Parrot in C#. Well, kinda. do tell :) purl: wow Senaka: what? I did enough to instantiate an interpreter, and then destroy it. ;-) "we had to destroy the interpreter in order to save it." C# how did you get it running? Use of C#'s somewhat lesser used extren keyword, the unsafe keyword and the [DllImport(...)] attribute. :-) AFAIK borland C++ and MS VC++ has C variants ain't it? MS VC++ will certainly compile C. compile as C code I mean not as C++ code I thought the .c extension triggers this? nope jonathan: yes it does http://msdn2.microsoft.com/en-us/library/032xwy55(VS.80).aspx But, when using VS 2005 I've noticed that there is a default compile as C++ code option that's what made me wonder * davidfetter wonders whether Senaka ever worked @ afilias direct invocation of cl.exe wouldn't cause this anyhow davidfetter:nope :) davidfetter: Senaka or Seneca? Because I did work at Afilias ah i see :) :) * davidfetter wonders whether he can get tetragon interested in pl/parrot Seneca Cunningham ? yep yep that's not me :) ah, sorry That's me roight :) * tetragon laughs at the confused identities :D so about this pl/parrot thing... I'm still at the point of parrot segfaulting on me great! * davidfetter more into team-gathering at this stage btw, there is yet an un-answered question static op_lib_t core_op_lib redefined in core_ops.c ?? g++ is not willing to go any further :) it has to be a difference between gcc and g++ looks like its declared once, used in the interim, and then declared with an initializer later on I guess gcc allows redeclaration of static variables. it makes sense, because you can't use "extern" for those. yes that's ANSI C I guess C++ 1998 wont isn't it? srry C99 is what I meant the extern/static conflict is still there in C99, as far as I know hmm... IIRC gcc is C99 isn't it? => gcc still has that prob gcc isn't very strict ok we can turn on warnings for some C99 things, but there are some things we can't force its also possible that this behavior is a "GNU extension" anyway, for g++ to work, the core_ops.c generation would need to be rearranged a bit, I guess wonder whether someone tried reading warnings when building using VS VS gives some good warning especially with the /W3 and /W4 options we've had a few recent bug reports regarding warnings for MSVC I don't remember this being one of them. but I don't have that platform to test with hmm.... I do but not @ the moment... perhaps on Monday I've ordered a book on GC, so hopefully I will become less clueless in the future :) I wish that I'm not going to get bombarded with warnings as VS has a good reputation for flooding terminals book on GC ?? may I know the name? http://www.amazon.com/Garbage-Collection-Algorithms-Automatic-Management/dp/0471941484/ref=si3_rdr_bb_product Infinoid's url is at http://xrl.us/bix8k if you do get bombarded with warnings, please nopaste them or post them to a ticket :) sure there are already a couple of MSVC warning tickets open but more info is always better % contingencyplan has left contingencyplan!~contingen@cpe-76-186-27-146.tx.res.rr.com % contingencyplan has joined #parrot regarding the redefinition of the static variables you are free to redefine it inside the method block isn't it? wouldn't that help? perhaps two static vars? if this doesn't help no, the definition (with initializer) needs to be moved up how are we to accept that all the source files result in a in-order execution? Infinoid: yes and that's an ops2c hack % AndyA has left AndyA!~andy@87.238.46.251 Infinoid: puzzled core_ops.c is generated from core/ops/*.ops by ops2c.pl (and lib/Parrot/Ops2c/*) oh i see I can look at that later today Infinoid: thanks % Limbic_Region has left Limbic_Region!~Limbic_Re@c-68-49-236-220.hsd1.md.comcast.net % malte has left malte!~chatzilla@c182135.adsl.hansenet.de is DOD related stuff deprecated? r26785 | jkeenan++ | ports: : Corrections in _handle_darwin_for_macports() and related changes in tests. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26785 * Infinoid tries to build on armv5b r26786 | bernhard++ | trunk: : #43753: [TODO] $language should be the name of the test Module : Remove idle comment. diff: http://www.parrotvm.org/svn/parrot/revision?rev=26786 * davidfetter ph34rz teh d4l3kz % barney has left barney!~bernhard@dslb-084-058-141-170.pools.arcor-ip.net % kid51_afk has left kid51_afk!~jkeen@pool-71-247-59-51.nycmny.east.verizon.net % rblackwe_ has left rblackwe_!rblackwe@where.is.allmydata.net % rblackwe has joined #parrot "EX-TER-MI-NATE!" * tetragon finds a convenient set of stairs % rdice has left rdice!~richarddi@CPE001217e365c7-CM00159a01d44c.cpe.net.cable.rogers.com % guru has joined #parrot % guru has left guru!~guru@67-208-9-120.ip.tor.radiant.net % guru has joined #parrot % Ademan has joined #parrot % rdice has joined #parrot % Psyche^ has joined #parrot % Patterner has left Patterner!~Psyche@d122167.adsl.hansenet.de % Psyche^ is now known as Patterner % purl has left purl!purl@sentient.life % purl has joined #parrot hi do you trace the PMC object roots on the Stack within the GC? or in other words is there any Stack enumeration? % camgirl29 has joined #parrot % Dave has left Dave!~dave@pool-141-153-244-48.mad.east.verizon.net % guru has left #parrot % Davemon has joined #parrot when is 1.0 of Parrot due? tomorrow o/~ the sun will come out.. tomorrow.. o/~ or the National Day of Slayer, and the National Emo Kid Beatdown day, by http://www.nationaldayofslayer.org/ and http://community.livejournal.com/wtf_inc/2805832.html or tomorrow and tomorrow and tomorrow creeps in this petty pace or mañana or free for all or another day wolwerian: are you serious? wolverian: you sound like a volunteer :) But tomorrow /never/ comes? seems every day, I ask myself "Self, is it tomorrow yet?" and the answer is no but there may be some exploitable race conditions at midnight