% IllvilJa has joined #parrot % ambs has joined #parrot % ambs has left ambs!~ambs@siglab.di.uminho.pt % mire has joined #parrot % Psyche^ has joined #parrot % Patterner has left Patterner!~Psyche@e177238008.adsl.alicedsl.de % Psyche^ is now known as Patterner % wknight8111 has joined #parrot % cosimo has joined #parrot % kid51 has joined #parrot % Ivatar has joined #parrot % wknight8111 has left wknight8111!~nobody@c-71-230-33-251.hsd1.pa.comcast.net % wknight8111 has joined #parrot % Coke has joined #parrot echo 'my Int $x = undef' | ../../parrot perl6.pbc Type check failed Jonathan: that should be allowed I think Jonathan: only low level types like int should disallow undef % iblechbot has left iblechbot!~iblechbot@ppp-62-216-197-248.dynamic.mnet-online.de moritz: Agree, please file an RT ticket if you have chance. Won't have chance to look at Rakudo stuff today, but will be spending all of tomorrow on it. Jonathan: ok with parrotbug? or is there a web interface? purl: rt? rt is probably just RT (http://bestpractical.com/rt) or (:rt3) or (: rt bugs) or Obra's trouble ticketing system or the first IBM RISC workstation (http://www.contrib.andrew.cmu.edu/~shadow/ibmrt.html) or the bombsquad or the Right Thing or very very capable and open-source or an application framework that bundles a ticketing system or obra's baby or SOOOO slow :-S or email mailto:perlbug-owner@perl.org for access purl: parrotbug? hmmm... parrotbug is mailto:parrotbug@parrotcode.org or http://svn.perl.org/parrot/trunk/docs/submissions.pod moritz: No, it's email...well, just send to perl6-compiler@perl.org if you like, I'll see it there and I'll remind me. *it'll remind me... Jonathan: ok, I'll do it Thanks. done % wknight8111 has left wknight8111!~nobody@c-71-230-33-251.hsd1.pa.comcast.net % kid51 has left kid51!~jkeen@pool-71-247-52-224.nycmny.east.verizon.net % iblechbot has joined #parrot % tetragon has left tetragon!~seneca@216.126.67.44 * Coke ponders "This week on parrot" again. % gryphon has joined #parrot % paco has joined #parrot % teknomunk has joined #parrot % mire has left mire!~Frodo@252-172-222-85.adsl.verat.net % teknomunk has left teknomunk!~teknomunk@r74-195-239-111.stl1cmta01.stwrok.ok.dh.suddenlink.net % Andy has joined #parrot % jhorwitz has joined #parrot 72+681+86 839 1/0 * Coke moves 53666 into the perl6 queue. % rdice has joined #parrot 71+681+86-23 815 # of tickets, if you subtract out the 23 remaining tcl tickets. Coke: What's the email address for the Perl 6 queue? Is it distinct from the Parrot one? perl 6 bugs? perl6 bugs? it has been said that perl6 bugs is Yeah, to go into RT. Aha, OK. perl 6 bugs is perl6 bugs yes, the queues are distinct. We need to get that, and probably some other content, onto rakudo.org I have no control over that, but yes. Yes you do. =-) Well, I'd rather delegate it. ;-) But if nobody will take it on, it needs doing so I'll step in and do it. % mj41 has joined #parrot Who has control over that? I'm sure there was plans to do this before... % mj41 has left mj41!chatzilla@pc-jurosz.ro.vutbr.cz % mj41 has joined #parrot Andy? rumour has it Andy is the man with the feathered broom or mailto:andy@petdance.com or AIM:petdance or OK cat-xeger is BACK or trying to get Sun Studio 12 installed so I can lint lint lint Registrant Name:Andy Lester whois coleda.com ? guess purl doesn't know whois. Andy: I was just commenting on how it would be good if rakudo.org had more than just a blog there. Jonathan: You need an account on rakudo? I have an account to blog there. Indeed. You have content to put out there? I'm just looking to have some static content there, like where to send bug reports, or to list what is done and the roadmap. static content++ # if maintained Not much, just some essential bits, plus instructions on how to build it and stuff. % mj41 has left mj41!chatzilla@pc-jurosz.ro.vutbr.cz note that parrotcode has the same issue: static good, maintaining hard. esp. a list of what is done would be useful for me as a spectator % mj41 has joined #parrot Whatever I can do, let me know. Juerd: Yes, I agree. Andy: What's the process? I'm willing to spend some time getting said content together. Andy: rgarding your question about the audience... I too am curious. It has to be someone who isn't subscribed to the mailing list, I think. though there is slightly more information not available to those just on list. Jonathan: Just write up some stuff and you can upload a page. At least I think you can. And if you can't, I'll make it so you can. how-to-build.html or whatever. OK, thanks. Andy: It's Movable Type right? If we get fancy we can move to a CMS like Drupal or something, but for now, uploading pages in MT will probably be enough it is, Juerd, yes. Andy: That has support for non-blog articles built in. Juerd: Yes, that's what I'm referring to. Will be flashing the URL around a bit at some conferences at the end of the month, so just want to have some clear building instructions and where to submit bugs and what you can expect to work etc on there for folks. Andy: Ah. I was thinking of the useful wysiwyg editor and such. Andy: As opposed to uploading Oh, I don't know of this editor you mean. Er, the thing a default MT installation gives you when you edit a blog entry or page... Or at least it does on mine :) % mj41 has left mj41!chatzilla@pc-jurosz.ro.vutbr.cz % mj41 has joined #parrot % rdice has left rdice!~richard_d@CPE0014bfafbbd5-CM0011e6ecf48a.cpe.net.cable.rogers.com % mj41 has left mj41!chatzilla@pc-jurosz.ro.vutbr.cz % uniejo has left uniejo!~uniejo@langebro.adapt.dk % uniejo has joined #parrot anyone here interested in porting parrot to ibm's power architecture? i'm talking to someone at ibm who may be able to get us access to some hardware either running AIX or RHEL ppc64? yep dual-proc I'm interested, but doubt I have time to contribute much i have a rs/6k with aix .. Ooh, hardware... hardware is old fashioned :-) i knew the hw geeks would come crawling out of the closet... purl is just jealous Wait wait wait, would this wind up with me having to understand and hack on the JIT? bah, we don't support JIT on our supported platforms yet, you don't have to support their jit yet either. % uniejo has left uniejo!~uniejo@langebro.adapt.dk so true Don't support JIT? I agree it's probably far from complete and has issues, but there's something there for our supported platforms? x86 and ppc have working jit And we need it for? % uniejo has joined #parrot osx? Coke probably means "Mac OS X". % cotto_work has joined #parrot x64 Ah. coke: jit is for processors, not for oses particle: not exactly. it's for osx/processors. er, os+processor. the jit for linux/x86 doesnt' work on osx/x86, AIUI. that seems like a config thing. no, it's a code thing. well. config is code, so ok. =-) but just because it works one place doesn't mean it works on that processor out of the box everywhere, which is my point. =-) % uniejo has left uniejo!~uniejo@langebro.adapt.dk My take is that it's mostly a CPU thing, but it likely needs per-OS tweaks to make it work correctly. The majority of the work is CPU-specific rather than OS-specific. ...assuming its possible to set up the same environment for it to execute in, on all OSes (I'm not convinced this will be the case for linux vs windows on amd64, it sounds like they set up the CPU in very different modes) % rdice has joined #parrot anyone running opensolaris? * particle wonders what it would take to get parrot compiling on paco's rs/6000 did i say i miss working on aix? no? i don't. it reminds me too much of the days before i learned perl. particle: i have to install the CL compiler, but I have gcc working in the aix .. nice! a modern gcc? % rdice has left rdice!~richard_d@CPE0014bfafbbd5-CM0011e6ecf48a.cpe.net.cable.rogers.com yes I think so, but I don remember the version .. 4.x would be wonderful two months ago I reinstalled aix .. and I have the cl compiler waiting for install .. :) if someone wants I can give shell .. is a power3-II also I have an alpha machine with vms and some sgi's with irix. I can give shell in these machines too .. do you have my entire past history of machines available? sheesh. i think so :) .. i have some VAX too * particle misses his sgi onyx with realityengine^2 graphics 3b3? eh 3b2? no :( but have a fiend who onws one .. still, your computer museum sounds interesting :) yes, I have interesting machines .. % gryphon has left gryphon!~gryphon@dsl-209-221-185-54.zipcon.net % gryphon has joined #parrot 0.037*60 2.22 % Theory has joined #parrot no, perl6 bugs is okay, pmichaud. parrotbug? parrotbug is mailto:parrotbug@parrotcode.org or http://svn.perl.org/parrot/trunk/docs/submissions.pod parrotbug is also see also "rakudobug" okay, pmichaud. parrotbug? parrotbug is probably mailto:parrotbug@parrotcode.org or http://svn.perl.org/parrot/trunk/docs/submissions.pod or see also "rakudobug" rakudobug? rakudobug is mailto:rakudobug@perl.org (rakudo.org content) -- there's also the possibility of putting static content for rakudo onto the perl6 wiki with links from rakudo.org Possible, though having some content on rakudo.org itself would make it feel a little more "unified". well, I'm a little bugged at the moment that we have content in so manyplaces there's dev.perl.org/perl6, the perl 6 wiki, rakudo.org, etc. I don't mind having it in so manyplaces as long as it's consistent and people can find it I don't propose to have anything non-specific to Rakudo on rakudo.org itself. Just what it is, how to build it, where to send bug reports, what's done, roadmap. well, if it's easy to keep information up-to-date on rakudo.org, then it'd probably be okay with me I'm just concerned it'll never get updated, so I want it to be on a place with easy updates (for non-committers and non-developers) I think that it's better for people who just want to play with Rakudo to have all the essentials in one place. agreed I'm thinking more end users than developers here. I'm not sure rakudo.org is that place I'm thinking the perl6 wiki might be easier to keep up-to-date i.e., I think that "all of the essentials" for rakudo users goes beyond just specifics about rakudo itself (maybe) Oh, sure, there will be info on Perl 6 language stuff, and I don't think we should have that on rakudo.org. It's just that if you google rakudo now, you end up at rakudo.org, plus it's the URL that I give out to people at conferences and so forth. that's a reasonable point. but rakudo.org can have links that point to the perl6 wiki, yes? and we can keep the content there. We can do that. I think we can, anyway. given the choice of "content on rakudo.org but difficult to update" and "easy to update content but not on rakudo.org", I vote for the latter. If we can have easy-to-update content on rakudo.org, then that's even better. =-) I think it's easy to update in the sense of not hard to actually do the changes, it's just that you need an account on rakudo.org to do that. well, whoever does the work gets to choose the forum, at this point. :-) if we want this to be focused towards users rather than developers, I'd like to be able to encourage users to be able to create the content so that developers don't have to do it (as much) True. pmichaud++ # turning my mail into ticket yes, rt#53748. easy to forward. :-) is that the undef assignment one? yes. Ah, I was planning to fix that tomorrow anywya. *anyway what's the fix? the fix is the financial information exchange protocol or at http://www.fixprotocol.org/ or in or Just One Fix or dngor has the start of a POE filter for it (ask him about it) * Jonathan doesn't like the overhead of doing RT too % ambs has joined #parrot I haven't decided yet. I have an answer (make the type checker happy assigning undef to anything), but before I go down that path I need to go read up on the whole, undef is no more kinda thing. 53750 seems to be a perl6 bug, not a parrot bug. I know there were changes. I'm wondering if it should instead create an undef'd Int a-la a protoobject Jonathan: but you can't assign undef to native types moritz: Sure, though those aren't implemented yet. moritz: we're not worrying about native types yet. (encourage users to create content) then just create pages on the wiki and link to them from rakudo.org pmichaud: If you write my Int $x; then $x is the Int proto-object today. I *think* that's the Right Thing. right, so I'm thinking that my Int $x = undef; should do a similar thing we may need a p6l ruling on this one. Yeah, that would make sense. I'm not sure if the spec says something equivalent. that sounds like a confusion of container and content I don't know that it makes sense for $x to become a Failure object. IMHO but I'm not very familiar with implementing type systems ;) moritz: We don't actually have two separate PMCs, one for container and one for value, for scalars. Thus discussions can get fun on such things. :-) for my Int $x; does $x bind directly to the Int proto-object or is a clone made? * moritz has to run, sorry I *think* directly at the moment. hmmm, that seems like it could be problematic Yeah. my Int $x; $x = 5; # changes the protoobject? * Jonathan tests it possibly we want to adjust assignment so that assigning to a protoobject de-protos it my Int $x = 5; say $x; my Int $x; say $x; 5 Int you used $x for both? Hmm, the compiler should maybe moan about that? coke, for #53750, parrot or perl6, why is the motivation for not listing the whole name? my Int $x = 5; say $x; my Int $y; say $y; 5 Int try my Int $x; $x = 5; my Int $y; say $x; say $y; my Int $x; $x = 5; say $x; my Int $y; say $y; 5 Int looks okay to me then :-) Yeah. I do remember doing *something* to make us not damage the proto object. okay, great. One unanswered question I have though. role Foo { ... }; my Foo $x; What is $x? i've stored the hottest stuff in $xxx that's being discussed on p6l at the moment Same for subset EvenInt of Int where { $_ % 2 == 0 }; but I remember chromatic saying that we don't instantiate roles my EvenInt $x; # what is $x now too nope, can't instantiate roles. they're class-construction-time only for EvenInt, I'd guess that $x is an EvenInt OK, but when you are writing that declaration you are saying "we can assign something to $x that does this role" no So I guess you just get an Undef with the correct type entry in %!properties er, there is no Undef we're saying "automagically create a class with the same name as this role, and instantiate an object of that type" s/type/class/ as not to mix syntax So how to we say "I want a variable that can hold any class that does this role"? s/to/do/ Other than going to write a subset type for it... i don't know any other way than subset my $x does FooRole ?? pmichaud: Yes, but what can I do with $x? If I print it, is it a failure? if it's not instantiated, it's a Failure (iiuc) you can assign to it, of course. yes. failure is still somewhat undefined in the spec OK, so we need *something* that we can have $x be. why not an undefined Int with an EvenInt role constraint? i was wondering if failure should keep some sort of state cognominal: if you can show me a case that doesn't invoke ".WHAT()", I might be able to answer that from a parrot standpoint. (but Iunno why perl6 does what it does. =-) * pmichaud looks at rt#53750 pmichaud: Not sure what you mean - I don't see EvenInt as a role in any way... Jonathan: okay, I should be quiet now then. I haven't looked at roles in great detail in the spec yet. Jonathan: take your best guess and ask questions on p6c or p6l or #perl6 or wherever gets you an answer :-) cognominal: PGE::Match reports 'Match' because that's the way .WHAT is defined in S12 pmichaud: My feeling is that a lot of this is kinda semi-spec'd. :-) so I should call .name instead? from S12: WHAT the prototype object of the type, stringifies to short name Jonathan: I agree. In which case I'm not the one who resolves the semi-ness -- that takes place on p6l. and there's been a discussion on classes and roles there recently, so we might just want to see how that resolves s/want/wait/ pmichaud: Sure, but I was interested to hear your thoughts and understanding on it too. got it :-) so in "my Int $x = undef;", $x contains a Failure object, which contains (or is) an exception. after that statement, it's an exception of type 'undef' (or whatever). postfix:<++> catches that exception, and sets $x to the default initial value of type Int, which is 0... then increments $x to 1. that's my understanding, anyway. particle: is that recent? my understanding? yes :) pmichaud, so according to S12, I should use WHO to get to the long name? not that it is implemented. cognominal: iiuc, the long name comes from .WHO i may be wrong, but that's the way i picture it. correct, it's not implemented % barney has joined #parrot particle: I was wanting to ask p6c/p6l if ++ still worked on undef objects cognominal: Implement it, or file a ticket if you want it in Rakudo, I may be able to do it. ok cognominal: Looks like FPW accepted three of my talks. pmichaud: my expectation for Int is that ++ should Just Work and i think the way i described it, it will. with an Int containing a Failure of type NaN, ++ won't work Jonathan, indeed ...because that type of Failure isn't handled by ++ particle: I don't disagree at all with your interpretation; I just know that failure handling of this sort is still a bit nebulous according to the spec yes, indeed it is we can certainly go ahead and implement it this way for now, until the spec is made clearer but we should also post a message to p6c or p6l asking "what's the official answer?" re: 53750 -- it's not specific to rakudo, so it belongs in the parrot queue nopast nopaste pmichaud: how is "WHAT" not specific to perl6? because it's a part of PGE and PCT in this case, it's PGE and PCT that are implementing .WHAT But WHAT is specifically a perl6 thing, neh? That's now how we'd normally expect to interrogate core parrot, is it? s/now/not/ since PGE is very perl 6ish at its core, it tends to follow the Perl 6 specs k. same for pct Also, I think you answered his question and can close the ticket. =-) well, perhaps the ticket needs to morph into "provide a .WHO method" "cognominal" at 82.67.232.89 pasted "missing WHO in Protoobject" (17 lines) at http://nopaste.snit.ch/12890 cognominal: what does .name() return if the original class is ['PGE';'Match'] instead of 'PGE::Match' ? the later is going away well, WHO and WHAT should not return strings but objects that stringifies to string. But thas a start. *latter Is PGE using "::" to indicate parrot's ;-separator? (should it not then return.... there you go.) cognominal: you're correct -- I haven't had an opportunity to get Protoobject.pir to catch up to doing true protoobjects and I'm wondering if I should just unify Protoobject.pir with rakudo somehow (so that one implementation handles both) I update the ticket to mention what you have said if I unserstand well, rakudo has his own protoobjects? s/his/its/ hmm. sending Parrot-related email to @cpan.org addresses bounces; they think "Configure.pl" is a spam URL. at the moment rakudo has its own protoobjects, yes but I'm thinking that we can unify the two % purl has left purl!purl@sentient.life the non-rakudo protoobject was basically to get things working to the point that I could implement the rakudo one :-) * Jonathan goes for dinner Doing my Rakudo day tomorrow. I should be around all day tomorrow OK, cool. might even get up a bit earlier to more closely match Jonathan-time I can get up a bit later to match yours too... ;-) enjoy dinner. :-) Thanks :-) this month's release is on the 20th, yes? yes dump run, followed by lunch & so there's a little time to do some major pge refactor/breakage between then good. % AndyA has left AndyA!~andy@82.152.157.85 % AndyA has joined #parrot cognominal: the WHO patch is fine; but I think I may try to do my Protoobject refactor tonight if not I'll go ahead and apply the WHO patch (unless you need it fairly soon) % contingencyplan has left contingencyplan!~contingen@cpe-76-186-27-146.tx.res.rr.com pmichaud : the refactor is the right thing to do Jus tsent mail off to someone wanting to be a cage cleaner. I am interested in helping with the parrot project and thought that starting with cage cleaning woulf be a good place. I have a few patches based on warnings during compilation already. Are you still the one to contact about this? Is there something that you know specifically needs help? Whooo! nice \o/ rurban@cpan.org | Parrot Development on Windows: link: http://www.perlfoundation.org/parrot/index.cgi?parrot_development_on_windows dalek's url is at http://xrl.us/bjcdh rurban@cpan.org | Parrot Development on Windows: link: http://www.perlfoundation.org/parrot/index.cgi?parrot_development_on_windows dalek's url is at http://xrl.us/bjcdh % grim_fandango has joined #parrot pmichaud: I am not sure about the meaning of long name in the definition of WHAT in S012, this may be related to versionning of modules? I really don't see the interest of WHAT being the last part of a name with '::' % contingencyplan has joined #parrot % Andy has left Andy!~AndyL@host3130.follett.com cognominal: I'm simply going by what Pugs returns in that case for example, say (1==1).WHAT; returns "True" and not "Bool::True" similarly, that's what several of the test files seem to indicate ok % Andy has joined #parrot pmichaud: Can I specify a namespace when calling a sub in PAST ? :pirop("call"), :name("my_sub") barney: :namespace, and give it an array Or maybe a string will do. er, that's for defining a sub Oh? for calling a sub I don't know if :namespace works Ah. :-) but one can always do a PAST::Var node that does the lookup :-) Oh yes, it's postcircumfix stuff in Rakudo... * Jonathan shouldn't try and do #parrot while working on Other Stuff. barney: in other words, for the PAST::Op node with :pasttype('call'), have the first child be a PAST::Var node that looks up the sub in another namespace r27326 | bernhard++ | trunk: : [Plumhead] : decode Base64 encoded string, before using them in setting up PAST diff: http://www.parrotvm.org/svn/parrot/revision?rev=27326 pmichaud: I see Andy could headerize make a pretty signature in comments for functions? All the visual cruft headerize adds really hurts skimability r27327 | infinoid++ | trunk: I'm not sure what you're sking. : [pmc_freeze] r24867 removed a potential bug, but didn't remove the : XXX comment about it. Remove that too. diff: http://www.parrotvm.org/svn/parrot/revision?rev=27327 give me a specific. a specific file/funciton src/packfile.c:71 k, hold on. still doing an update on this machine, which I haven't for eons. No rush just a suggestion. (could also look on http://svn.perl.org/parrot/trunk/src/packfile.c :-) OK, so you're talking about the function declaration on statics? because the function declaration at the body is much less cluttery. yeah there's not much that can be stripped from there. not much = none because the compiler/splint needs to see all those modifiers. tewk: did you want it stripped or just formatted differently? What do you see as a "signature" on it? Its not that big a deal, but a list of statics can be seen as a table of contents for the file, I was thinking of a comment section /* static PackFile_Segment * byte_code_new(interp, PackFile *pf, const char *name, int add); ... */ Sorry I just have to complain every once in a while:) I also think code lines need to be 200+ characters before they should be line wrapped. Just ignore me. On a lighter note, I'm investigating using gcc-dehydra from mozilla for parsing c header files to make NCI signatures. % davidfetter has left davidfetter!~davidfett@start.fetter.org particle, pmichaud et al: imo, thinking of Int (the protoob) as a generic integer, Int++ should still be Int; and NaN should do the same; and C and C (a rational NaN) should be (close to?) interchangeable, at least arithmetically. Replicating gcc parsing in c99 could turn into a rabbit hole. Once I'd recursed through 3 or 4 different rabbit holes I started looking for alternatives. dehydra is pretty simple/cool spinclad: what about the general case of: my $x; $x++; tewk: oh, nice... too bad its gcc-specific